{
  "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": 12.935441478144865,
      "learning_rate": 5.3475935828877005e-08,
      "loss": 1.4455,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 14.7622354944256,
      "learning_rate": 1.0695187165775401e-07,
      "loss": 1.4577,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.745943568788293,
      "learning_rate": 1.6042780748663104e-07,
      "loss": 1.3129,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 13.9302655173877,
      "learning_rate": 2.1390374331550802e-07,
      "loss": 1.473,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.694705289156342,
      "learning_rate": 2.6737967914438503e-07,
      "loss": 1.4385,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.720888398162117,
      "learning_rate": 3.208556149732621e-07,
      "loss": 1.4653,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.812943011118795,
      "learning_rate": 3.7433155080213904e-07,
      "loss": 1.3959,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.821988569940946,
      "learning_rate": 4.2780748663101604e-07,
      "loss": 0.7875,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.182259847820657,
      "learning_rate": 4.812834224598931e-07,
      "loss": 1.5014,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.026573876602576,
      "learning_rate": 5.347593582887701e-07,
      "loss": 1.4184,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 11.672855803818452,
      "learning_rate": 5.882352941176471e-07,
      "loss": 1.5295,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.5219344415093214,
      "learning_rate": 6.417112299465242e-07,
      "loss": 0.7443,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.822765399632402,
      "learning_rate": 6.951871657754011e-07,
      "loss": 1.4695,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.572331406231754,
      "learning_rate": 7.486631016042781e-07,
      "loss": 1.3112,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.50054132155673,
      "learning_rate": 8.021390374331551e-07,
      "loss": 1.3738,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.490496835595879,
      "learning_rate": 8.556149732620321e-07,
      "loss": 1.362,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.8960541730994915,
      "learning_rate": 9.090909090909091e-07,
      "loss": 1.1374,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.612965836506855,
      "learning_rate": 9.625668449197862e-07,
      "loss": 1.1762,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.233528018489697,
      "learning_rate": 1.0160427807486633e-06,
      "loss": 1.1174,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.816254368834543,
      "learning_rate": 1.0695187165775401e-06,
      "loss": 1.1701,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.997004404002938,
      "learning_rate": 1.1229946524064172e-06,
      "loss": 1.105,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.25413563828559,
      "learning_rate": 1.1764705882352942e-06,
      "loss": 1.1117,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.254001012967801,
      "learning_rate": 1.2299465240641713e-06,
      "loss": 0.6391,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.8588334655369434,
      "learning_rate": 1.2834224598930483e-06,
      "loss": 1.1637,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.2963170527733574,
      "learning_rate": 1.3368983957219254e-06,
      "loss": 1.042,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.282541430114072,
      "learning_rate": 1.3903743315508022e-06,
      "loss": 0.6116,
      "step": 26
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.7293564908374353,
      "learning_rate": 1.4438502673796793e-06,
      "loss": 0.9941,
      "step": 27
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.824192721701949,
      "learning_rate": 1.4973262032085562e-06,
      "loss": 1.0038,
      "step": 28
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.8125014928177339,
      "learning_rate": 1.5508021390374334e-06,
      "loss": 1.0333,
      "step": 29
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.7184486102384333,
      "learning_rate": 1.6042780748663103e-06,
      "loss": 0.952,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.8407635577341634,
      "learning_rate": 1.6577540106951873e-06,
      "loss": 0.9006,
      "step": 31
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.703457385982276,
      "learning_rate": 1.7112299465240642e-06,
      "loss": 0.8891,
      "step": 32
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6214136372997996,
      "learning_rate": 1.7647058823529414e-06,
      "loss": 0.9676,
      "step": 33
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5284451424310843,
      "learning_rate": 1.8181818181818183e-06,
      "loss": 0.9191,
      "step": 34
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.433591334843516,
      "learning_rate": 1.8716577540106954e-06,
      "loss": 0.8436,
      "step": 35
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4811324547013833,
      "learning_rate": 1.9251336898395724e-06,
      "loss": 0.8871,
      "step": 36
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6693464083370484,
      "learning_rate": 1.9786096256684497e-06,
      "loss": 0.9509,
      "step": 37
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5227273449265926,
      "learning_rate": 2.0320855614973265e-06,
      "loss": 0.9546,
      "step": 38
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4756603584692545,
      "learning_rate": 2.0855614973262034e-06,
      "loss": 0.9534,
      "step": 39
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4729605341959016,
      "learning_rate": 2.1390374331550802e-06,
      "loss": 0.8574,
      "step": 40
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4308867149250293,
      "learning_rate": 2.1925133689839575e-06,
      "loss": 0.8769,
      "step": 41
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2843916639976503,
      "learning_rate": 2.2459893048128343e-06,
      "loss": 0.7959,
      "step": 42
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3303200637988284,
      "learning_rate": 2.2994652406417116e-06,
      "loss": 0.7755,
      "step": 43
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3981022102143286,
      "learning_rate": 2.3529411764705885e-06,
      "loss": 0.8467,
      "step": 44
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4070369115048074,
      "learning_rate": 2.4064171122994653e-06,
      "loss": 0.8869,
      "step": 45
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3826852752536778,
      "learning_rate": 2.4598930481283426e-06,
      "loss": 0.8539,
      "step": 46
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2994263971268318,
      "learning_rate": 2.5133689839572194e-06,
      "loss": 0.8226,
      "step": 47
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4029755203913823,
      "learning_rate": 2.5668449197860967e-06,
      "loss": 0.8063,
      "step": 48
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4640857513245897,
      "learning_rate": 2.6203208556149735e-06,
      "loss": 0.8219,
      "step": 49
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3400580068888166,
      "learning_rate": 2.673796791443851e-06,
      "loss": 0.7591,
      "step": 50
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4271256402645964,
      "learning_rate": 2.7272727272727272e-06,
      "loss": 0.8773,
      "step": 51
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2593254014363953,
      "learning_rate": 2.7807486631016045e-06,
      "loss": 0.7871,
      "step": 52
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3018443477097152,
      "learning_rate": 2.8342245989304818e-06,
      "loss": 0.8449,
      "step": 53
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3182813983078345,
      "learning_rate": 2.8877005347593586e-06,
      "loss": 0.7531,
      "step": 54
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.315026501483241,
      "learning_rate": 2.9411764705882355e-06,
      "loss": 0.731,
      "step": 55
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2645000378266102,
      "learning_rate": 2.9946524064171123e-06,
      "loss": 0.8118,
      "step": 56
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3939481835820742,
      "learning_rate": 3.0481283422459896e-06,
      "loss": 0.8107,
      "step": 57
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3722036495691252,
      "learning_rate": 3.101604278074867e-06,
      "loss": 0.7851,
      "step": 58
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4013348097015585,
      "learning_rate": 3.1550802139037433e-06,
      "loss": 0.8707,
      "step": 59
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3945354126187168,
      "learning_rate": 3.2085561497326205e-06,
      "loss": 0.7455,
      "step": 60
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3615958450470937,
      "learning_rate": 3.262032085561498e-06,
      "loss": 0.8429,
      "step": 61
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3363841680546134,
      "learning_rate": 3.3155080213903747e-06,
      "loss": 0.8495,
      "step": 62
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3891357284948158,
      "learning_rate": 3.368983957219252e-06,
      "loss": 0.9195,
      "step": 63
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.271946386715015,
      "learning_rate": 3.4224598930481284e-06,
      "loss": 0.749,
      "step": 64
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.219667288144616,
      "learning_rate": 3.4759358288770056e-06,
      "loss": 0.6372,
      "step": 65
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2601525896049124,
      "learning_rate": 3.529411764705883e-06,
      "loss": 0.7609,
      "step": 66
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4258547561284391,
      "learning_rate": 3.5828877005347597e-06,
      "loss": 0.8498,
      "step": 67
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3384436860521522,
      "learning_rate": 3.6363636363636366e-06,
      "loss": 0.7567,
      "step": 68
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.237169349467392,
      "learning_rate": 3.6898395721925134e-06,
      "loss": 0.7743,
      "step": 69
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2628384486054505,
      "learning_rate": 3.7433155080213907e-06,
      "loss": 0.7615,
      "step": 70
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3039911911016937,
      "learning_rate": 3.796791443850268e-06,
      "loss": 0.7763,
      "step": 71
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4536181475187755,
      "learning_rate": 3.850267379679145e-06,
      "loss": 0.8278,
      "step": 72
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3389444939627722,
      "learning_rate": 3.903743315508022e-06,
      "loss": 0.7661,
      "step": 73
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2578830820664857,
      "learning_rate": 3.957219251336899e-06,
      "loss": 0.7756,
      "step": 74
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2616404615407286,
      "learning_rate": 4.010695187165775e-06,
      "loss": 0.725,
      "step": 75
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2731557167858414,
      "learning_rate": 4.064171122994653e-06,
      "loss": 0.6998,
      "step": 76
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3484428672133917,
      "learning_rate": 4.11764705882353e-06,
      "loss": 0.7932,
      "step": 77
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2909478782594066,
      "learning_rate": 4.171122994652407e-06,
      "loss": 0.7236,
      "step": 78
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7994965665573495,
      "learning_rate": 4.224598930481284e-06,
      "loss": 0.4944,
      "step": 79
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5323738271913132,
      "learning_rate": 4.2780748663101604e-06,
      "loss": 0.837,
      "step": 80
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2559640178659703,
      "learning_rate": 4.331550802139038e-06,
      "loss": 0.7898,
      "step": 81
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2156880135548962,
      "learning_rate": 4.385026737967915e-06,
      "loss": 0.7338,
      "step": 82
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2886913112014688,
      "learning_rate": 4.438502673796792e-06,
      "loss": 0.7038,
      "step": 83
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.273326354954758,
      "learning_rate": 4.491978609625669e-06,
      "loss": 0.7561,
      "step": 84
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2916258313295492,
      "learning_rate": 4.5454545454545455e-06,
      "loss": 0.7931,
      "step": 85
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2597583771598537,
      "learning_rate": 4.598930481283423e-06,
      "loss": 0.6588,
      "step": 86
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2920208680762029,
      "learning_rate": 4.6524064171123e-06,
      "loss": 0.7691,
      "step": 87
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2720819534402383,
      "learning_rate": 4.705882352941177e-06,
      "loss": 0.8344,
      "step": 88
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2826249153403808,
      "learning_rate": 4.759358288770054e-06,
      "loss": 0.6944,
      "step": 89
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1577666803150262,
      "learning_rate": 4.812834224598931e-06,
      "loss": 0.728,
      "step": 90
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2905971202346285,
      "learning_rate": 4.866310160427808e-06,
      "loss": 0.6942,
      "step": 91
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3635752155344887,
      "learning_rate": 4.919786096256685e-06,
      "loss": 0.7546,
      "step": 92
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3662328636967387,
      "learning_rate": 4.973262032085562e-06,
      "loss": 0.7836,
      "step": 93
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4707915804076896,
      "learning_rate": 5.026737967914439e-06,
      "loss": 0.7372,
      "step": 94
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2849509484569301,
      "learning_rate": 5.0802139037433165e-06,
      "loss": 0.7776,
      "step": 95
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.428058767538607,
      "learning_rate": 5.133689839572193e-06,
      "loss": 0.8066,
      "step": 96
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.615922362802789,
      "learning_rate": 5.187165775401069e-06,
      "loss": 0.8649,
      "step": 97
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.319482758837517,
      "learning_rate": 5.240641711229947e-06,
      "loss": 0.8037,
      "step": 98
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6801244635264085,
      "learning_rate": 5.294117647058824e-06,
      "loss": 0.481,
      "step": 99
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.408729969060582,
      "learning_rate": 5.347593582887702e-06,
      "loss": 0.7608,
      "step": 100
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.517835000841765,
      "learning_rate": 5.4010695187165785e-06,
      "loss": 0.8009,
      "step": 101
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3021444696294928,
      "learning_rate": 5.4545454545454545e-06,
      "loss": 0.7761,
      "step": 102
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3926050404746577,
      "learning_rate": 5.508021390374332e-06,
      "loss": 0.7868,
      "step": 103
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6832263550818355,
      "learning_rate": 5.561497326203209e-06,
      "loss": 0.4697,
      "step": 104
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3721074757572662,
      "learning_rate": 5.614973262032086e-06,
      "loss": 0.7854,
      "step": 105
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3098264456149247,
      "learning_rate": 5.6684491978609635e-06,
      "loss": 0.6518,
      "step": 106
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3750556327247605,
      "learning_rate": 5.7219251336898395e-06,
      "loss": 0.7788,
      "step": 107
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.368039595492315,
      "learning_rate": 5.775401069518717e-06,
      "loss": 0.7205,
      "step": 108
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3117106157324974,
      "learning_rate": 5.828877005347594e-06,
      "loss": 0.7234,
      "step": 109
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4473178635658313,
      "learning_rate": 5.882352941176471e-06,
      "loss": 0.7516,
      "step": 110
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.396316738422691,
      "learning_rate": 5.935828877005349e-06,
      "loss": 0.8079,
      "step": 111
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1715901374545499,
      "learning_rate": 5.989304812834225e-06,
      "loss": 0.5798,
      "step": 112
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.388524502842756,
      "learning_rate": 6.0427807486631015e-06,
      "loss": 0.7906,
      "step": 113
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2982019855974833,
      "learning_rate": 6.096256684491979e-06,
      "loss": 0.7176,
      "step": 114
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.338159896784674,
      "learning_rate": 6.149732620320856e-06,
      "loss": 0.7134,
      "step": 115
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2613799535421117,
      "learning_rate": 6.203208556149734e-06,
      "loss": 0.7265,
      "step": 116
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.472526413330938,
      "learning_rate": 6.25668449197861e-06,
      "loss": 0.8516,
      "step": 117
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3388925592171055,
      "learning_rate": 6.3101604278074865e-06,
      "loss": 0.7541,
      "step": 118
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3693232759630902,
      "learning_rate": 6.363636363636364e-06,
      "loss": 0.7319,
      "step": 119
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.217116268775871,
      "learning_rate": 6.417112299465241e-06,
      "loss": 0.6994,
      "step": 120
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.383252591520476,
      "learning_rate": 6.470588235294119e-06,
      "loss": 0.7199,
      "step": 121
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2247793941666851,
      "learning_rate": 6.524064171122996e-06,
      "loss": 0.7144,
      "step": 122
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2817725807213516,
      "learning_rate": 6.577540106951872e-06,
      "loss": 0.7443,
      "step": 123
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2541881712292917,
      "learning_rate": 6.631016042780749e-06,
      "loss": 0.7827,
      "step": 124
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4904141052411564,
      "learning_rate": 6.684491978609626e-06,
      "loss": 0.785,
      "step": 125
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.455037889026677,
      "learning_rate": 6.737967914438504e-06,
      "loss": 0.7385,
      "step": 126
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3587712236420801,
      "learning_rate": 6.791443850267381e-06,
      "loss": 0.7268,
      "step": 127
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3989330276607206,
      "learning_rate": 6.844919786096257e-06,
      "loss": 0.7274,
      "step": 128
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7253566527353816,
      "learning_rate": 6.898395721925134e-06,
      "loss": 0.7787,
      "step": 129
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.296949252584265,
      "learning_rate": 6.951871657754011e-06,
      "loss": 0.6882,
      "step": 130
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2826085284747382,
      "learning_rate": 7.005347593582889e-06,
      "loss": 0.7839,
      "step": 131
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8080921523697212,
      "learning_rate": 7.058823529411766e-06,
      "loss": 0.4443,
      "step": 132
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3800525896378077,
      "learning_rate": 7.112299465240642e-06,
      "loss": 0.7355,
      "step": 133
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.363840912423529,
      "learning_rate": 7.1657754010695195e-06,
      "loss": 0.6729,
      "step": 134
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2649157806773565,
      "learning_rate": 7.219251336898396e-06,
      "loss": 0.7261,
      "step": 135
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3214263402465884,
      "learning_rate": 7.272727272727273e-06,
      "loss": 0.7623,
      "step": 136
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.288570252898323,
      "learning_rate": 7.326203208556151e-06,
      "loss": 0.7265,
      "step": 137
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2363944963995177,
      "learning_rate": 7.379679144385027e-06,
      "loss": 0.655,
      "step": 138
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4156804170487287,
      "learning_rate": 7.433155080213904e-06,
      "loss": 0.7961,
      "step": 139
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2384359160180567,
      "learning_rate": 7.486631016042781e-06,
      "loss": 0.7878,
      "step": 140
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3774744960531309,
      "learning_rate": 7.540106951871658e-06,
      "loss": 0.8762,
      "step": 141
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2386418951519893,
      "learning_rate": 7.593582887700536e-06,
      "loss": 0.7585,
      "step": 142
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3251484755237808,
      "learning_rate": 7.647058823529411e-06,
      "loss": 0.7012,
      "step": 143
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3869648836737731,
      "learning_rate": 7.70053475935829e-06,
      "loss": 0.7384,
      "step": 144
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2839819282631588,
      "learning_rate": 7.754010695187166e-06,
      "loss": 0.6685,
      "step": 145
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.380634963113264,
      "learning_rate": 7.807486631016043e-06,
      "loss": 0.7416,
      "step": 146
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3516287254672021,
      "learning_rate": 7.86096256684492e-06,
      "loss": 0.6661,
      "step": 147
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.476190394846508,
      "learning_rate": 7.914438502673799e-06,
      "loss": 0.8217,
      "step": 148
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3277349424446394,
      "learning_rate": 7.967914438502674e-06,
      "loss": 0.7373,
      "step": 149
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3610384964315203,
      "learning_rate": 8.02139037433155e-06,
      "loss": 0.7429,
      "step": 150
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1415142233931699,
      "learning_rate": 8.07486631016043e-06,
      "loss": 0.7019,
      "step": 151
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3150801069322091,
      "learning_rate": 8.128342245989306e-06,
      "loss": 0.6895,
      "step": 152
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0193872623524336,
      "learning_rate": 8.181818181818183e-06,
      "loss": 0.4849,
      "step": 153
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4244765728424607,
      "learning_rate": 8.23529411764706e-06,
      "loss": 0.7747,
      "step": 154
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3007303504370338,
      "learning_rate": 8.288770053475937e-06,
      "loss": 0.763,
      "step": 155
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2697650681744665,
      "learning_rate": 8.342245989304813e-06,
      "loss": 0.7737,
      "step": 156
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3963893836361554,
      "learning_rate": 8.39572192513369e-06,
      "loss": 0.7515,
      "step": 157
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3363467946483825,
      "learning_rate": 8.449197860962567e-06,
      "loss": 0.7604,
      "step": 158
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3274245686332713,
      "learning_rate": 8.502673796791444e-06,
      "loss": 0.8,
      "step": 159
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7717008133929879,
      "learning_rate": 8.556149732620321e-06,
      "loss": 0.5089,
      "step": 160
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.291861767521327,
      "learning_rate": 8.609625668449198e-06,
      "loss": 0.7105,
      "step": 161
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2952891407984455,
      "learning_rate": 8.663101604278076e-06,
      "loss": 0.6938,
      "step": 162
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3635723537878723,
      "learning_rate": 8.716577540106953e-06,
      "loss": 0.7749,
      "step": 163
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4160132017337088,
      "learning_rate": 8.77005347593583e-06,
      "loss": 0.768,
      "step": 164
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3509135795582823,
      "learning_rate": 8.823529411764707e-06,
      "loss": 0.7451,
      "step": 165
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.351803385991898,
      "learning_rate": 8.877005347593584e-06,
      "loss": 0.7246,
      "step": 166
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.218080048051328,
      "learning_rate": 8.93048128342246e-06,
      "loss": 0.7194,
      "step": 167
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2640725233899797,
      "learning_rate": 8.983957219251337e-06,
      "loss": 0.6985,
      "step": 168
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2762845164377183,
      "learning_rate": 9.037433155080214e-06,
      "loss": 0.6798,
      "step": 169
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2810621410476162,
      "learning_rate": 9.090909090909091e-06,
      "loss": 0.6819,
      "step": 170
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3779930981716855,
      "learning_rate": 9.144385026737968e-06,
      "loss": 0.7396,
      "step": 171
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3778386495198012,
      "learning_rate": 9.197860962566846e-06,
      "loss": 0.7001,
      "step": 172
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4089047112988118,
      "learning_rate": 9.251336898395723e-06,
      "loss": 0.7553,
      "step": 173
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2853824849437605,
      "learning_rate": 9.3048128342246e-06,
      "loss": 0.7321,
      "step": 174
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3138606515454636,
      "learning_rate": 9.358288770053477e-06,
      "loss": 0.6608,
      "step": 175
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3648407254095352,
      "learning_rate": 9.411764705882354e-06,
      "loss": 0.7672,
      "step": 176
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3653948567640994,
      "learning_rate": 9.46524064171123e-06,
      "loss": 0.7931,
      "step": 177
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3093774998215235,
      "learning_rate": 9.518716577540108e-06,
      "loss": 0.7334,
      "step": 178
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.273571084537043,
      "learning_rate": 9.572192513368986e-06,
      "loss": 0.656,
      "step": 179
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5505377475596862,
      "learning_rate": 9.625668449197861e-06,
      "loss": 0.8257,
      "step": 180
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5729070011195827,
      "learning_rate": 9.679144385026738e-06,
      "loss": 0.8037,
      "step": 181
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4547553138374314,
      "learning_rate": 9.732620320855617e-06,
      "loss": 0.478,
      "step": 182
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2513348841143141,
      "learning_rate": 9.786096256684493e-06,
      "loss": 0.6765,
      "step": 183
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3627686835443136,
      "learning_rate": 9.83957219251337e-06,
      "loss": 0.7556,
      "step": 184
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.774634246259936,
      "learning_rate": 9.893048128342247e-06,
      "loss": 0.4749,
      "step": 185
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.397399587792225,
      "learning_rate": 9.946524064171124e-06,
      "loss": 0.7883,
      "step": 186
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.367808901959227,
      "learning_rate": 1e-05,
      "loss": 0.7913,
      "step": 187
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.25082090581326,
      "learning_rate": 9.999999318931088e-06,
      "loss": 0.6764,
      "step": 188
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.331214831242888,
      "learning_rate": 9.999997275724535e-06,
      "loss": 0.7612,
      "step": 189
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4249493714320949,
      "learning_rate": 9.999993870380897e-06,
      "loss": 0.6491,
      "step": 190
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1990182316559028,
      "learning_rate": 9.999989102901105e-06,
      "loss": 0.7281,
      "step": 191
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3438234474302149,
      "learning_rate": 9.999982973286455e-06,
      "loss": 0.7064,
      "step": 192
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2564493552885914,
      "learning_rate": 9.999975481538618e-06,
      "loss": 0.6813,
      "step": 193
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.272763795139657,
      "learning_rate": 9.999966627659635e-06,
      "loss": 0.6266,
      "step": 194
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2134451906590265,
      "learning_rate": 9.999956411651916e-06,
      "loss": 0.6801,
      "step": 195
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3448442838532517,
      "learning_rate": 9.999944833518248e-06,
      "loss": 0.6677,
      "step": 196
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3835673263159791,
      "learning_rate": 9.999931893261783e-06,
      "loss": 0.8111,
      "step": 197
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4606603551203912,
      "learning_rate": 9.999917590886046e-06,
      "loss": 0.7676,
      "step": 198
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.374405927822037,
      "learning_rate": 9.999901926394932e-06,
      "loss": 0.7414,
      "step": 199
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3730441365852943,
      "learning_rate": 9.99988489979271e-06,
      "loss": 0.5905,
      "step": 200
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3257743411629859,
      "learning_rate": 9.999866511084021e-06,
      "loss": 0.7766,
      "step": 201
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1854186441967431,
      "learning_rate": 9.999846760273873e-06,
      "loss": 0.7184,
      "step": 202
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3439684911943945,
      "learning_rate": 9.999825647367643e-06,
      "loss": 0.7614,
      "step": 203
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.335934880678688,
      "learning_rate": 9.999803172371088e-06,
      "loss": 0.786,
      "step": 204
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4174458318046552,
      "learning_rate": 9.999779335290328e-06,
      "loss": 0.7289,
      "step": 205
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8889855242307784,
      "learning_rate": 9.999754136131855e-06,
      "loss": 0.491,
      "step": 206
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2834334506221448,
      "learning_rate": 9.999727574902538e-06,
      "loss": 0.6817,
      "step": 207
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3013454465054761,
      "learning_rate": 9.999699651609611e-06,
      "loss": 0.7477,
      "step": 208
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2342997822910966,
      "learning_rate": 9.999670366260682e-06,
      "loss": 0.7318,
      "step": 209
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4344589829959105,
      "learning_rate": 9.999639718863728e-06,
      "loss": 0.743,
      "step": 210
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1760729067482503,
      "learning_rate": 9.9996077094271e-06,
      "loss": 0.7114,
      "step": 211
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2317896448799757,
      "learning_rate": 9.999574337959514e-06,
      "loss": 0.7212,
      "step": 212
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.387643872824135,
      "learning_rate": 9.999539604470068e-06,
      "loss": 0.7768,
      "step": 213
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5577030539795724,
      "learning_rate": 9.999503508968218e-06,
      "loss": 0.7212,
      "step": 214
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2210243091792707,
      "learning_rate": 9.9994660514638e-06,
      "loss": 0.6915,
      "step": 215
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.275254204000143,
      "learning_rate": 9.99942723196702e-06,
      "loss": 0.7326,
      "step": 216
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2571978728202617,
      "learning_rate": 9.999387050488451e-06,
      "loss": 0.5764,
      "step": 217
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1723890682837121,
      "learning_rate": 9.99934550703904e-06,
      "loss": 0.6662,
      "step": 218
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.225134245438625,
      "learning_rate": 9.999302601630106e-06,
      "loss": 0.7359,
      "step": 219
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2198619424501254,
      "learning_rate": 9.999258334273338e-06,
      "loss": 0.6699,
      "step": 220
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8867795004181716,
      "learning_rate": 9.999212704980792e-06,
      "loss": 0.4979,
      "step": 221
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.398722762579555,
      "learning_rate": 9.999165713764902e-06,
      "loss": 0.7379,
      "step": 222
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5025694669167649,
      "learning_rate": 9.999117360638469e-06,
      "loss": 0.7499,
      "step": 223
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.497817607359326,
      "learning_rate": 9.999067645614666e-06,
      "loss": 0.8253,
      "step": 224
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1717524039805163,
      "learning_rate": 9.999016568707036e-06,
      "loss": 0.7384,
      "step": 225
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6730693707993181,
      "learning_rate": 9.998964129929493e-06,
      "loss": 0.4905,
      "step": 226
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2642786064229417,
      "learning_rate": 9.998910329296322e-06,
      "loss": 0.789,
      "step": 227
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2876519614802493,
      "learning_rate": 9.998855166822186e-06,
      "loss": 0.7156,
      "step": 228
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3026579801479032,
      "learning_rate": 9.998798642522105e-06,
      "loss": 0.7563,
      "step": 229
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5017436593551474,
      "learning_rate": 9.998740756411483e-06,
      "loss": 0.7207,
      "step": 230
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.266400053072259,
      "learning_rate": 9.998681508506087e-06,
      "loss": 0.648,
      "step": 231
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1864505088783366,
      "learning_rate": 9.998620898822059e-06,
      "loss": 0.6951,
      "step": 232
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3855125754652082,
      "learning_rate": 9.998558927375909e-06,
      "loss": 0.6922,
      "step": 233
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2175747229109015,
      "learning_rate": 9.998495594184523e-06,
      "loss": 0.6695,
      "step": 234
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.276390781773472,
      "learning_rate": 9.998430899265152e-06,
      "loss": 0.6748,
      "step": 235
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2407184959699031,
      "learning_rate": 9.998364842635422e-06,
      "loss": 0.6653,
      "step": 236
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2556164713929225,
      "learning_rate": 9.998297424313327e-06,
      "loss": 0.6495,
      "step": 237
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4368173970617737,
      "learning_rate": 9.998228644317235e-06,
      "loss": 0.77,
      "step": 238
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3309990949955843,
      "learning_rate": 9.998158502665884e-06,
      "loss": 0.7261,
      "step": 239
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2526749234192738,
      "learning_rate": 9.99808699937838e-06,
      "loss": 0.701,
      "step": 240
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3372315522059617,
      "learning_rate": 9.998014134474207e-06,
      "loss": 0.6976,
      "step": 241
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3471462521866626,
      "learning_rate": 9.997939907973212e-06,
      "loss": 0.7213,
      "step": 242
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2306654806999384,
      "learning_rate": 9.997864319895616e-06,
      "loss": 0.7087,
      "step": 243
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.858364792336332,
      "learning_rate": 9.997787370262012e-06,
      "loss": 0.4779,
      "step": 244
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.76013363356788,
      "learning_rate": 9.997709059093364e-06,
      "loss": 0.4854,
      "step": 245
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2988792592854377,
      "learning_rate": 9.997629386411006e-06,
      "loss": 0.7282,
      "step": 246
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4542539405542785,
      "learning_rate": 9.997548352236644e-06,
      "loss": 0.7274,
      "step": 247
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.320286546988568,
      "learning_rate": 9.99746595659235e-06,
      "loss": 0.7365,
      "step": 248
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.309978405328875,
      "learning_rate": 9.997382199500577e-06,
      "loss": 0.7659,
      "step": 249
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1707539596309837,
      "learning_rate": 9.997297080984136e-06,
      "loss": 0.71,
      "step": 250
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3798793183963456,
      "learning_rate": 9.997210601066218e-06,
      "loss": 0.6374,
      "step": 251
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4811551224463737,
      "learning_rate": 9.997122759770386e-06,
      "loss": 0.7478,
      "step": 252
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3058697719571093,
      "learning_rate": 9.997033557120567e-06,
      "loss": 0.7421,
      "step": 253
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.244869644724416,
      "learning_rate": 9.996942993141063e-06,
      "loss": 0.6853,
      "step": 254
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.450253976156751,
      "learning_rate": 9.996851067856546e-06,
      "loss": 0.519,
      "step": 255
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1749160681368627,
      "learning_rate": 9.996757781292058e-06,
      "loss": 0.6864,
      "step": 256
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5588788302091203,
      "learning_rate": 9.996663133473017e-06,
      "loss": 0.6967,
      "step": 257
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3148715392780708,
      "learning_rate": 9.996567124425201e-06,
      "loss": 0.6698,
      "step": 258
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3213244835498843,
      "learning_rate": 9.996469754174772e-06,
      "loss": 0.7199,
      "step": 259
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2947213163042959,
      "learning_rate": 9.996371022748251e-06,
      "loss": 0.7128,
      "step": 260
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3156315734077115,
      "learning_rate": 9.99627093017254e-06,
      "loss": 0.6567,
      "step": 261
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3371295027852197,
      "learning_rate": 9.996169476474902e-06,
      "loss": 0.75,
      "step": 262
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2616705578376781,
      "learning_rate": 9.996066661682981e-06,
      "loss": 0.6927,
      "step": 263
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3571985209697057,
      "learning_rate": 9.995962485824783e-06,
      "loss": 0.7451,
      "step": 264
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.230141529118125,
      "learning_rate": 9.995856948928688e-06,
      "loss": 0.7215,
      "step": 265
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2267964009778596,
      "learning_rate": 9.99575005102345e-06,
      "loss": 0.651,
      "step": 266
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2887761207584736,
      "learning_rate": 9.99564179213819e-06,
      "loss": 0.7504,
      "step": 267
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2962040637810222,
      "learning_rate": 9.995532172302399e-06,
      "loss": 0.6948,
      "step": 268
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2769936535782778,
      "learning_rate": 9.995421191545942e-06,
      "loss": 0.7075,
      "step": 269
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4286561027893085,
      "learning_rate": 9.995308849899052e-06,
      "loss": 0.8251,
      "step": 270
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3072936066939465,
      "learning_rate": 9.995195147392335e-06,
      "loss": 0.7475,
      "step": 271
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2369123118560394,
      "learning_rate": 9.995080084056767e-06,
      "loss": 0.6701,
      "step": 272
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2085389329504574,
      "learning_rate": 9.994963659923695e-06,
      "loss": 0.6256,
      "step": 273
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1643315502173501,
      "learning_rate": 9.994845875024834e-06,
      "loss": 0.6593,
      "step": 274
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.286554009304387,
      "learning_rate": 9.994726729392272e-06,
      "loss": 0.7334,
      "step": 275
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8743332726561144,
      "learning_rate": 9.99460622305847e-06,
      "loss": 0.5115,
      "step": 276
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2680546273089717,
      "learning_rate": 9.994484356056256e-06,
      "loss": 0.7543,
      "step": 277
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7943082300961309,
      "learning_rate": 9.994361128418828e-06,
      "loss": 0.505,
      "step": 278
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2025177145253436,
      "learning_rate": 9.99423654017976e-06,
      "loss": 0.6858,
      "step": 279
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3822213681884803,
      "learning_rate": 9.99411059137299e-06,
      "loss": 0.7706,
      "step": 280
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2644026740402285,
      "learning_rate": 9.993983282032831e-06,
      "loss": 0.6906,
      "step": 281
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3744922785419906,
      "learning_rate": 9.993854612193967e-06,
      "loss": 0.7457,
      "step": 282
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2012060722321076,
      "learning_rate": 9.993724581891451e-06,
      "loss": 0.6845,
      "step": 283
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2314007641045293,
      "learning_rate": 9.993593191160704e-06,
      "loss": 0.682,
      "step": 284
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1716279672060668,
      "learning_rate": 9.993460440037525e-06,
      "loss": 0.7052,
      "step": 285
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2840643927613349,
      "learning_rate": 9.993326328558076e-06,
      "loss": 0.6667,
      "step": 286
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2553897101143476,
      "learning_rate": 9.993190856758892e-06,
      "loss": 0.6474,
      "step": 287
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2342484627628807,
      "learning_rate": 9.99305402467688e-06,
      "loss": 0.7231,
      "step": 288
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2506166392693183,
      "learning_rate": 9.99291583234932e-06,
      "loss": 0.7158,
      "step": 289
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2799363415300542,
      "learning_rate": 9.992776279813854e-06,
      "loss": 0.7628,
      "step": 290
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.319625011570069,
      "learning_rate": 9.992635367108505e-06,
      "loss": 0.7417,
      "step": 291
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0637254628784878,
      "learning_rate": 9.992493094271657e-06,
      "loss": 0.6807,
      "step": 292
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2890203754510048,
      "learning_rate": 9.992349461342073e-06,
      "loss": 0.7077,
      "step": 293
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2287410655645907,
      "learning_rate": 9.992204468358879e-06,
      "loss": 0.7278,
      "step": 294
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2638386137239284,
      "learning_rate": 9.992058115361578e-06,
      "loss": 0.7108,
      "step": 295
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2729022475181637,
      "learning_rate": 9.991910402390041e-06,
      "loss": 0.695,
      "step": 296
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2496925396617156,
      "learning_rate": 9.991761329484505e-06,
      "loss": 0.657,
      "step": 297
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2540515841751168,
      "learning_rate": 9.991610896685587e-06,
      "loss": 0.7011,
      "step": 298
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2247738160047426,
      "learning_rate": 9.991459104034262e-06,
      "loss": 0.5431,
      "step": 299
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.324998754958549,
      "learning_rate": 9.991305951571891e-06,
      "loss": 0.6445,
      "step": 300
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3194273049410405,
      "learning_rate": 9.99115143934019e-06,
      "loss": 0.6805,
      "step": 301
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1463422929971323,
      "learning_rate": 9.990995567381255e-06,
      "loss": 0.6931,
      "step": 302
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.186861454033246,
      "learning_rate": 9.990838335737551e-06,
      "loss": 0.692,
      "step": 303
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.188753937804532,
      "learning_rate": 9.990679744451909e-06,
      "loss": 0.6522,
      "step": 304
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3695028124417525,
      "learning_rate": 9.990519793567539e-06,
      "loss": 0.696,
      "step": 305
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.212641102232149,
      "learning_rate": 9.990358483128012e-06,
      "loss": 0.7645,
      "step": 306
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3418494191959847,
      "learning_rate": 9.990195813177272e-06,
      "loss": 0.6759,
      "step": 307
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1394146441246351,
      "learning_rate": 9.99003178375964e-06,
      "loss": 0.6743,
      "step": 308
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2084250380038932,
      "learning_rate": 9.989866394919795e-06,
      "loss": 0.649,
      "step": 309
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.26929032529933,
      "learning_rate": 9.9896996467028e-06,
      "loss": 0.6499,
      "step": 310
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1390710108289281,
      "learning_rate": 9.98953153915408e-06,
      "loss": 0.6479,
      "step": 311
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1445823212644295,
      "learning_rate": 9.989362072319431e-06,
      "loss": 0.6189,
      "step": 312
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2464565876993572,
      "learning_rate": 9.98919124624502e-06,
      "loss": 0.769,
      "step": 313
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2782427600503015,
      "learning_rate": 9.989019060977388e-06,
      "loss": 0.7443,
      "step": 314
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2435394254511745,
      "learning_rate": 9.988845516563437e-06,
      "loss": 0.7197,
      "step": 315
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.200401468957928,
      "learning_rate": 9.988670613050452e-06,
      "loss": 0.74,
      "step": 316
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1479829406518383,
      "learning_rate": 9.988494350486077e-06,
      "loss": 0.7472,
      "step": 317
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2188447733257157,
      "learning_rate": 9.988316728918332e-06,
      "loss": 0.6729,
      "step": 318
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2380019122646702,
      "learning_rate": 9.988137748395608e-06,
      "loss": 0.647,
      "step": 319
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2113814365732802,
      "learning_rate": 9.987957408966662e-06,
      "loss": 0.6883,
      "step": 320
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1836754608407634,
      "learning_rate": 9.987775710680622e-06,
      "loss": 0.6861,
      "step": 321
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2513729213480607,
      "learning_rate": 9.987592653586992e-06,
      "loss": 0.7145,
      "step": 322
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.148049920875932,
      "learning_rate": 9.987408237735638e-06,
      "loss": 0.7364,
      "step": 323
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1887268144403802,
      "learning_rate": 9.987222463176803e-06,
      "loss": 0.6854,
      "step": 324
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.110222455442517,
      "learning_rate": 9.987035329961092e-06,
      "loss": 0.7447,
      "step": 325
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7164103752519383,
      "learning_rate": 9.986846838139492e-06,
      "loss": 0.5111,
      "step": 326
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2586794855633905,
      "learning_rate": 9.986656987763348e-06,
      "loss": 0.7445,
      "step": 327
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2050419814495252,
      "learning_rate": 9.986465778884384e-06,
      "loss": 0.7948,
      "step": 328
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2529472693071761,
      "learning_rate": 9.986273211554689e-06,
      "loss": 0.6765,
      "step": 329
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.254394734298952,
      "learning_rate": 9.986079285826721e-06,
      "loss": 0.7061,
      "step": 330
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2710035041481733,
      "learning_rate": 9.985884001753317e-06,
      "loss": 0.6215,
      "step": 331
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2344446175458699,
      "learning_rate": 9.985687359387673e-06,
      "loss": 0.6688,
      "step": 332
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1742468375895199,
      "learning_rate": 9.98548935878336e-06,
      "loss": 0.7463,
      "step": 333
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1548595916777202,
      "learning_rate": 9.985289999994322e-06,
      "loss": 0.6193,
      "step": 334
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.319122302898996,
      "learning_rate": 9.985089283074867e-06,
      "loss": 0.724,
      "step": 335
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1855380992583031,
      "learning_rate": 9.984887208079675e-06,
      "loss": 0.6728,
      "step": 336
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2140268826406544,
      "learning_rate": 9.9846837750638e-06,
      "loss": 0.712,
      "step": 337
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1946992342967895,
      "learning_rate": 9.98447898408266e-06,
      "loss": 0.6431,
      "step": 338
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0848472049848554,
      "learning_rate": 9.984272835192047e-06,
      "loss": 0.6256,
      "step": 339
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.264192562721289,
      "learning_rate": 9.984065328448122e-06,
      "loss": 0.7262,
      "step": 340
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2867666404364067,
      "learning_rate": 9.983856463907415e-06,
      "loss": 0.6994,
      "step": 341
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2417529865863632,
      "learning_rate": 9.983646241626825e-06,
      "loss": 0.6457,
      "step": 342
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1971415238568486,
      "learning_rate": 9.983434661663625e-06,
      "loss": 0.6605,
      "step": 343
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2041269107631913,
      "learning_rate": 9.983221724075453e-06,
      "loss": 0.7163,
      "step": 344
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1926985043339906,
      "learning_rate": 9.983007428920322e-06,
      "loss": 0.7127,
      "step": 345
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2582674549204695,
      "learning_rate": 9.982791776256608e-06,
      "loss": 0.6459,
      "step": 346
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1460708432876627,
      "learning_rate": 9.982574766143063e-06,
      "loss": 0.757,
      "step": 347
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.123781321365631,
      "learning_rate": 9.982356398638807e-06,
      "loss": 0.7232,
      "step": 348
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2774526181564465,
      "learning_rate": 9.982136673803328e-06,
      "loss": 0.6972,
      "step": 349
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2228251539912751,
      "learning_rate": 9.981915591696484e-06,
      "loss": 0.7134,
      "step": 350
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.191353005764966,
      "learning_rate": 9.981693152378509e-06,
      "loss": 0.6868,
      "step": 351
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3557416164999287,
      "learning_rate": 9.981469355909996e-06,
      "loss": 0.7573,
      "step": 352
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2469418817573794,
      "learning_rate": 9.981244202351916e-06,
      "loss": 0.7498,
      "step": 353
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1984908296886687,
      "learning_rate": 9.981017691765606e-06,
      "loss": 0.7511,
      "step": 354
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2004419272130282,
      "learning_rate": 9.980789824212776e-06,
      "loss": 0.6611,
      "step": 355
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3811575729777732,
      "learning_rate": 9.980560599755498e-06,
      "loss": 0.6956,
      "step": 356
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1814425693792978,
      "learning_rate": 9.980330018456227e-06,
      "loss": 0.7598,
      "step": 357
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.198696177946858,
      "learning_rate": 9.980098080377771e-06,
      "loss": 0.653,
      "step": 358
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1270488798787484,
      "learning_rate": 9.979864785583325e-06,
      "loss": 0.611,
      "step": 359
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1764837161047377,
      "learning_rate": 9.979630134136438e-06,
      "loss": 0.663,
      "step": 360
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.242448318393586,
      "learning_rate": 9.979394126101039e-06,
      "loss": 0.7504,
      "step": 361
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1962281862047308,
      "learning_rate": 9.979156761541421e-06,
      "loss": 0.7183,
      "step": 362
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.146547701924311,
      "learning_rate": 9.978918040522249e-06,
      "loss": 0.6336,
      "step": 363
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1278651084233722,
      "learning_rate": 9.97867796310856e-06,
      "loss": 0.6716,
      "step": 364
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2563055824238085,
      "learning_rate": 9.978436529365757e-06,
      "loss": 0.7644,
      "step": 365
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.171095445121014,
      "learning_rate": 9.978193739359611e-06,
      "loss": 0.6609,
      "step": 366
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2087903631779031,
      "learning_rate": 9.977949593156264e-06,
      "loss": 0.6213,
      "step": 367
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2787873813634485,
      "learning_rate": 9.977704090822232e-06,
      "loss": 0.7989,
      "step": 368
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.21544841841805,
      "learning_rate": 9.977457232424394e-06,
      "loss": 0.6722,
      "step": 369
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2607813552972889,
      "learning_rate": 9.97720901803e-06,
      "loss": 0.6314,
      "step": 370
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.28256845878936,
      "learning_rate": 9.976959447706673e-06,
      "loss": 0.6381,
      "step": 371
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1297128466140272,
      "learning_rate": 9.976708521522403e-06,
      "loss": 0.6831,
      "step": 372
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1535140281730443,
      "learning_rate": 9.976456239545547e-06,
      "loss": 0.6002,
      "step": 373
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2673336966957607,
      "learning_rate": 9.976202601844834e-06,
      "loss": 0.6539,
      "step": 374
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3158001121530785,
      "learning_rate": 9.975947608489363e-06,
      "loss": 0.6368,
      "step": 375
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1545330178386846,
      "learning_rate": 9.975691259548598e-06,
      "loss": 0.7716,
      "step": 376
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3194081951710745,
      "learning_rate": 9.975433555092383e-06,
      "loss": 0.7104,
      "step": 377
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1234745072307746,
      "learning_rate": 9.975174495190915e-06,
      "loss": 0.6063,
      "step": 378
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2047813612524296,
      "learning_rate": 9.974914079914775e-06,
      "loss": 0.6314,
      "step": 379
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.138228514572019,
      "learning_rate": 9.974652309334904e-06,
      "loss": 0.7392,
      "step": 380
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.07569534927839,
      "learning_rate": 9.974389183522618e-06,
      "loss": 0.6504,
      "step": 381
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2343228426442545,
      "learning_rate": 9.9741247025496e-06,
      "loss": 0.6937,
      "step": 382
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1867179341443375,
      "learning_rate": 9.973858866487898e-06,
      "loss": 0.6007,
      "step": 383
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2396305234459037,
      "learning_rate": 9.973591675409934e-06,
      "loss": 0.6993,
      "step": 384
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2694710080055773,
      "learning_rate": 9.973323129388504e-06,
      "loss": 0.7064,
      "step": 385
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2080871696969717,
      "learning_rate": 9.97305322849676e-06,
      "loss": 0.6582,
      "step": 386
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1073413750342909,
      "learning_rate": 9.972781972808234e-06,
      "loss": 0.6735,
      "step": 387
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2541194215675302,
      "learning_rate": 9.972509362396825e-06,
      "loss": 0.6176,
      "step": 388
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.047829915296732,
      "learning_rate": 9.972235397336796e-06,
      "loss": 0.6028,
      "step": 389
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1984323437974636,
      "learning_rate": 9.971960077702785e-06,
      "loss": 0.745,
      "step": 390
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3560946586338647,
      "learning_rate": 9.971683403569795e-06,
      "loss": 0.7396,
      "step": 391
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1503300529728597,
      "learning_rate": 9.971405375013201e-06,
      "loss": 0.7073,
      "step": 392
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.31364021338241,
      "learning_rate": 9.971125992108746e-06,
      "loss": 0.639,
      "step": 393
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2196416543938773,
      "learning_rate": 9.97084525493254e-06,
      "loss": 0.6013,
      "step": 394
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.141087670297583,
      "learning_rate": 9.970563163561064e-06,
      "loss": 0.742,
      "step": 395
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2652218610208563,
      "learning_rate": 9.970279718071169e-06,
      "loss": 0.7436,
      "step": 396
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.141684970935003,
      "learning_rate": 9.969994918540071e-06,
      "loss": 0.6565,
      "step": 397
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3109552880370567,
      "learning_rate": 9.969708765045361e-06,
      "loss": 0.6552,
      "step": 398
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2450928216019495,
      "learning_rate": 9.96942125766499e-06,
      "loss": 0.6907,
      "step": 399
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1927129228959974,
      "learning_rate": 9.969132396477286e-06,
      "loss": 0.7036,
      "step": 400
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1982424181724343,
      "learning_rate": 9.968842181560943e-06,
      "loss": 0.677,
      "step": 401
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2059255747498785,
      "learning_rate": 9.968550612995023e-06,
      "loss": 0.6207,
      "step": 402
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2126575816916907,
      "learning_rate": 9.968257690858955e-06,
      "loss": 0.6298,
      "step": 403
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2782251670578761,
      "learning_rate": 9.967963415232544e-06,
      "loss": 0.7582,
      "step": 404
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1486482386443593,
      "learning_rate": 9.967667786195955e-06,
      "loss": 0.5538,
      "step": 405
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2451201136556358,
      "learning_rate": 9.967370803829725e-06,
      "loss": 0.6924,
      "step": 406
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.157220433884935,
      "learning_rate": 9.967072468214763e-06,
      "loss": 0.6856,
      "step": 407
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6575462354452141,
      "learning_rate": 9.966772779432343e-06,
      "loss": 0.5274,
      "step": 408
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2437295089429923,
      "learning_rate": 9.966471737564107e-06,
      "loss": 0.611,
      "step": 409
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2018431787770065,
      "learning_rate": 9.96616934269207e-06,
      "loss": 0.6813,
      "step": 410
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1984920313194274,
      "learning_rate": 9.965865594898608e-06,
      "loss": 0.6984,
      "step": 411
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.243301829001436,
      "learning_rate": 9.965560494266475e-06,
      "loss": 0.757,
      "step": 412
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.209554261657944,
      "learning_rate": 9.965254040878786e-06,
      "loss": 0.6858,
      "step": 413
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2837342854838043,
      "learning_rate": 9.964946234819028e-06,
      "loss": 0.7561,
      "step": 414
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.309896257034652,
      "learning_rate": 9.964637076171056e-06,
      "loss": 0.7213,
      "step": 415
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2783130591264011,
      "learning_rate": 9.964326565019094e-06,
      "loss": 0.7492,
      "step": 416
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1383239570314165,
      "learning_rate": 9.964014701447733e-06,
      "loss": 0.6721,
      "step": 417
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6854907832660053,
      "learning_rate": 9.963701485541935e-06,
      "loss": 0.5149,
      "step": 418
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.310379703447458,
      "learning_rate": 9.963386917387025e-06,
      "loss": 0.7655,
      "step": 419
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2287297265272992,
      "learning_rate": 9.963070997068702e-06,
      "loss": 0.7296,
      "step": 420
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0863892673254076,
      "learning_rate": 9.962753724673033e-06,
      "loss": 0.6211,
      "step": 421
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3027314630879676,
      "learning_rate": 9.96243510028645e-06,
      "loss": 0.7283,
      "step": 422
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2465231453810837,
      "learning_rate": 9.962115123995754e-06,
      "loss": 0.6911,
      "step": 423
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2753288352990937,
      "learning_rate": 9.961793795888118e-06,
      "loss": 0.6842,
      "step": 424
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2244577773415144,
      "learning_rate": 9.96147111605108e-06,
      "loss": 0.703,
      "step": 425
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0853815444012955,
      "learning_rate": 9.961147084572544e-06,
      "loss": 0.6685,
      "step": 426
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0980962163485015,
      "learning_rate": 9.96082170154079e-06,
      "loss": 0.7364,
      "step": 427
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1064609370198906,
      "learning_rate": 9.960494967044457e-06,
      "loss": 0.6961,
      "step": 428
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0993942272329889,
      "learning_rate": 9.960166881172558e-06,
      "loss": 0.6033,
      "step": 429
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1658126062639649,
      "learning_rate": 9.959837444014473e-06,
      "loss": 0.6522,
      "step": 430
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2436564638694596,
      "learning_rate": 9.95950665565995e-06,
      "loss": 0.6781,
      "step": 431
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.267446097500326,
      "learning_rate": 9.959174516199105e-06,
      "loss": 0.6473,
      "step": 432
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3963474814024208,
      "learning_rate": 9.95884102572242e-06,
      "loss": 0.7078,
      "step": 433
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1411297131165479,
      "learning_rate": 9.958506184320749e-06,
      "loss": 0.6852,
      "step": 434
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1953110134876261,
      "learning_rate": 9.95816999208531e-06,
      "loss": 0.6917,
      "step": 435
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6740677493231655,
      "learning_rate": 9.957832449107694e-06,
      "loss": 0.5036,
      "step": 436
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0962127480634218,
      "learning_rate": 9.957493555479856e-06,
      "loss": 0.6529,
      "step": 437
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2142417208917269,
      "learning_rate": 9.957153311294119e-06,
      "loss": 0.7142,
      "step": 438
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1695545476592806,
      "learning_rate": 9.956811716643173e-06,
      "loss": 0.6728,
      "step": 439
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6503216423736332,
      "learning_rate": 9.956468771620082e-06,
      "loss": 0.4809,
      "step": 440
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.364580037944365,
      "learning_rate": 9.956124476318271e-06,
      "loss": 0.6878,
      "step": 441
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1137815627733518,
      "learning_rate": 9.955778830831537e-06,
      "loss": 0.6228,
      "step": 442
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1886410527781188,
      "learning_rate": 9.955431835254044e-06,
      "loss": 0.6655,
      "step": 443
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1781882385729894,
      "learning_rate": 9.95508348968032e-06,
      "loss": 0.7544,
      "step": 444
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2434011269259,
      "learning_rate": 9.954733794205264e-06,
      "loss": 0.6882,
      "step": 445
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0866981248051866,
      "learning_rate": 9.954382748924148e-06,
      "loss": 0.617,
      "step": 446
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.288228778203744,
      "learning_rate": 9.9540303539326e-06,
      "loss": 0.7296,
      "step": 447
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2637737687522859,
      "learning_rate": 9.953676609326627e-06,
      "loss": 0.6472,
      "step": 448
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1264429975974244,
      "learning_rate": 9.953321515202597e-06,
      "loss": 0.6694,
      "step": 449
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2260658329930987,
      "learning_rate": 9.952965071657244e-06,
      "loss": 0.7322,
      "step": 450
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1341017298554932,
      "learning_rate": 9.952607278787679e-06,
      "loss": 0.6341,
      "step": 451
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1631348298519713,
      "learning_rate": 9.95224813669137e-06,
      "loss": 0.6119,
      "step": 452
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2072017277737226,
      "learning_rate": 9.95188764546616e-06,
      "loss": 0.6689,
      "step": 453
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3220061396349503,
      "learning_rate": 9.951525805210256e-06,
      "loss": 0.7461,
      "step": 454
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2753798221016264,
      "learning_rate": 9.951162616022234e-06,
      "loss": 0.7558,
      "step": 455
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.142408973872296,
      "learning_rate": 9.950798078001034e-06,
      "loss": 0.7406,
      "step": 456
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.291617994771683,
      "learning_rate": 9.950432191245968e-06,
      "loss": 0.7248,
      "step": 457
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0961719187655317,
      "learning_rate": 9.950064955856716e-06,
      "loss": 0.691,
      "step": 458
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1987380921034372,
      "learning_rate": 9.949696371933319e-06,
      "loss": 0.6827,
      "step": 459
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1134318087147919,
      "learning_rate": 9.94932643957619e-06,
      "loss": 0.6073,
      "step": 460
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1559799439068397,
      "learning_rate": 9.948955158886113e-06,
      "loss": 0.6522,
      "step": 461
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1394586289731738,
      "learning_rate": 9.94858252996423e-06,
      "loss": 0.6269,
      "step": 462
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2002732674749657,
      "learning_rate": 9.948208552912057e-06,
      "loss": 0.5459,
      "step": 463
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1300957542651615,
      "learning_rate": 9.947833227831477e-06,
      "loss": 0.6821,
      "step": 464
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1983979266121416,
      "learning_rate": 9.947456554824736e-06,
      "loss": 0.6699,
      "step": 465
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.100936896472829,
      "learning_rate": 9.947078533994454e-06,
      "loss": 0.6156,
      "step": 466
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2878974988182825,
      "learning_rate": 9.94669916544361e-06,
      "loss": 0.7358,
      "step": 467
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1367661059609175,
      "learning_rate": 9.94631844927556e-06,
      "loss": 0.6686,
      "step": 468
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1383793253055459,
      "learning_rate": 9.945936385594017e-06,
      "loss": 0.6506,
      "step": 469
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1854037914673214,
      "learning_rate": 9.945552974503065e-06,
      "loss": 0.7272,
      "step": 470
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2735080801907124,
      "learning_rate": 9.94516821610716e-06,
      "loss": 0.7674,
      "step": 471
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1734637162830088,
      "learning_rate": 9.944782110511119e-06,
      "loss": 0.6907,
      "step": 472
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3135960191278515,
      "learning_rate": 9.944394657820127e-06,
      "loss": 0.6557,
      "step": 473
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1441923990139509,
      "learning_rate": 9.944005858139737e-06,
      "loss": 0.6253,
      "step": 474
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1493882934214332,
      "learning_rate": 9.943615711575867e-06,
      "loss": 0.7193,
      "step": 475
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2282679978937754,
      "learning_rate": 9.943224218234809e-06,
      "loss": 0.6816,
      "step": 476
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7409637388717382,
      "learning_rate": 9.94283137822321e-06,
      "loss": 0.4966,
      "step": 477
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2195467870431518,
      "learning_rate": 9.942437191648095e-06,
      "loss": 0.7199,
      "step": 478
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1923691452498928,
      "learning_rate": 9.94204165861685e-06,
      "loss": 0.6553,
      "step": 479
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1411051632101619,
      "learning_rate": 9.94164477923723e-06,
      "loss": 0.6784,
      "step": 480
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1287377055137306,
      "learning_rate": 9.941246553617353e-06,
      "loss": 0.6421,
      "step": 481
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1636418813669103,
      "learning_rate": 9.940846981865709e-06,
      "loss": 0.6576,
      "step": 482
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1541497177807003,
      "learning_rate": 9.940446064091151e-06,
      "loss": 0.7463,
      "step": 483
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2150527017028636,
      "learning_rate": 9.940043800402903e-06,
      "loss": 0.7286,
      "step": 484
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1972714684104622,
      "learning_rate": 9.939640190910552e-06,
      "loss": 0.7284,
      "step": 485
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2537624272529804,
      "learning_rate": 9.939235235724047e-06,
      "loss": 0.7836,
      "step": 486
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1995277117548084,
      "learning_rate": 9.938828934953716e-06,
      "loss": 0.6392,
      "step": 487
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2421249029696795,
      "learning_rate": 9.938421288710242e-06,
      "loss": 0.7847,
      "step": 488
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1778886576856036,
      "learning_rate": 9.93801229710468e-06,
      "loss": 0.683,
      "step": 489
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6557273888295984,
      "learning_rate": 9.937601960248452e-06,
      "loss": 0.4832,
      "step": 490
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.298117776637159,
      "learning_rate": 9.937190278253345e-06,
      "loss": 0.6869,
      "step": 491
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1295745225258655,
      "learning_rate": 9.93677725123151e-06,
      "loss": 0.6866,
      "step": 492
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0762405120703018,
      "learning_rate": 9.936362879295471e-06,
      "loss": 0.6419,
      "step": 493
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1662411308757306,
      "learning_rate": 9.93594716255811e-06,
      "loss": 0.7446,
      "step": 494
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2022500818632824,
      "learning_rate": 9.935530101132683e-06,
      "loss": 0.7342,
      "step": 495
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1660846634723887,
      "learning_rate": 9.935111695132807e-06,
      "loss": 0.6947,
      "step": 496
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1146215157244876,
      "learning_rate": 9.934691944672468e-06,
      "loss": 0.5636,
      "step": 497
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.149770816378899,
      "learning_rate": 9.934270849866019e-06,
      "loss": 0.6526,
      "step": 498
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1777414548334122,
      "learning_rate": 9.933848410828175e-06,
      "loss": 0.6979,
      "step": 499
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2165433436804116,
      "learning_rate": 9.933424627674022e-06,
      "loss": 0.6992,
      "step": 500
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1072268965384175,
      "learning_rate": 9.932999500519011e-06,
      "loss": 0.6281,
      "step": 501
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1334663035461165,
      "learning_rate": 9.932573029478956e-06,
      "loss": 0.7025,
      "step": 502
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0774366164520455,
      "learning_rate": 9.932145214670042e-06,
      "loss": 0.6295,
      "step": 503
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.110742520894135,
      "learning_rate": 9.931716056208814e-06,
      "loss": 0.7078,
      "step": 504
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2138506637232074,
      "learning_rate": 9.931285554212189e-06,
      "loss": 0.621,
      "step": 505
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1057974112865008,
      "learning_rate": 9.930853708797448e-06,
      "loss": 0.727,
      "step": 506
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2747805720596324,
      "learning_rate": 9.930420520082238e-06,
      "loss": 0.6075,
      "step": 507
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.197063673279613,
      "learning_rate": 9.929985988184568e-06,
      "loss": 0.686,
      "step": 508
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3100121171197556,
      "learning_rate": 9.929550113222822e-06,
      "loss": 0.7487,
      "step": 509
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.179954704029963,
      "learning_rate": 9.929112895315739e-06,
      "loss": 0.6439,
      "step": 510
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2251553338649839,
      "learning_rate": 9.928674334582431e-06,
      "loss": 0.7287,
      "step": 511
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.232762139542434,
      "learning_rate": 9.928234431142376e-06,
      "loss": 0.6233,
      "step": 512
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2487549666358873,
      "learning_rate": 9.927793185115416e-06,
      "loss": 0.6658,
      "step": 513
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6951278285076156,
      "learning_rate": 9.927350596621753e-06,
      "loss": 0.532,
      "step": 514
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.241359735962445,
      "learning_rate": 9.926906665781967e-06,
      "loss": 0.7226,
      "step": 515
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.184216563449497,
      "learning_rate": 9.926461392716993e-06,
      "loss": 0.6841,
      "step": 516
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4427849618152664,
      "learning_rate": 9.926014777548139e-06,
      "loss": 0.7274,
      "step": 517
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2011617465505329,
      "learning_rate": 9.92556682039707e-06,
      "loss": 0.6985,
      "step": 518
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0721598703006259,
      "learning_rate": 9.925117521385827e-06,
      "loss": 0.6968,
      "step": 519
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1682133059523674,
      "learning_rate": 9.92466688063681e-06,
      "loss": 0.6654,
      "step": 520
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7037391719999192,
      "learning_rate": 9.924214898272783e-06,
      "loss": 0.4915,
      "step": 521
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2786868587542795,
      "learning_rate": 9.923761574416884e-06,
      "loss": 0.7331,
      "step": 522
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1865216965166312,
      "learning_rate": 9.923306909192608e-06,
      "loss": 0.6434,
      "step": 523
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6472122413131611,
      "learning_rate": 9.922850902723815e-06,
      "loss": 0.4969,
      "step": 524
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1423892338495318,
      "learning_rate": 9.922393555134739e-06,
      "loss": 0.6739,
      "step": 525
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1281429287655047,
      "learning_rate": 9.92193486654997e-06,
      "loss": 0.6577,
      "step": 526
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.130769436545356,
      "learning_rate": 9.921474837094473e-06,
      "loss": 0.7031,
      "step": 527
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1053623018996277,
      "learning_rate": 9.921013466893567e-06,
      "loss": 0.6365,
      "step": 528
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.121818177065228,
      "learning_rate": 9.920550756072945e-06,
      "loss": 0.6616,
      "step": 529
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1450067956094951,
      "learning_rate": 9.92008670475866e-06,
      "loss": 0.6778,
      "step": 530
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1033360143415212,
      "learning_rate": 9.919621313077135e-06,
      "loss": 0.701,
      "step": 531
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.058289274485389,
      "learning_rate": 9.919154581155152e-06,
      "loss": 0.641,
      "step": 532
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1889313501506664,
      "learning_rate": 9.918686509119867e-06,
      "loss": 0.7124,
      "step": 533
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2630487075319436,
      "learning_rate": 9.91821709709879e-06,
      "loss": 0.7433,
      "step": 534
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.219072434854954,
      "learning_rate": 9.917746345219805e-06,
      "loss": 0.7342,
      "step": 535
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1040858565196574,
      "learning_rate": 9.917274253611156e-06,
      "loss": 0.6516,
      "step": 536
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2105967303379057,
      "learning_rate": 9.916800822401457e-06,
      "loss": 0.6421,
      "step": 537
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1694288730726754,
      "learning_rate": 9.91632605171968e-06,
      "loss": 0.7354,
      "step": 538
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1333568245587693,
      "learning_rate": 9.915849941695167e-06,
      "loss": 0.6313,
      "step": 539
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.184053852232108,
      "learning_rate": 9.915372492457624e-06,
      "loss": 0.7035,
      "step": 540
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.084857145915072,
      "learning_rate": 9.914893704137123e-06,
      "loss": 0.565,
      "step": 541
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2339695853118111,
      "learning_rate": 9.914413576864094e-06,
      "loss": 0.6579,
      "step": 542
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1632436425485746,
      "learning_rate": 9.913932110769342e-06,
      "loss": 0.6434,
      "step": 543
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1630065352792258,
      "learning_rate": 9.91344930598403e-06,
      "loss": 0.6181,
      "step": 544
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3410053202359082,
      "learning_rate": 9.912965162639686e-06,
      "loss": 0.7359,
      "step": 545
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1338222943771186,
      "learning_rate": 9.912479680868205e-06,
      "loss": 0.7335,
      "step": 546
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1873556496658888,
      "learning_rate": 9.911992860801846e-06,
      "loss": 0.6602,
      "step": 547
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.228740007792514,
      "learning_rate": 9.911504702573232e-06,
      "loss": 0.7021,
      "step": 548
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.094285422410936,
      "learning_rate": 9.91101520631535e-06,
      "loss": 0.5907,
      "step": 549
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1994402125495436,
      "learning_rate": 9.910524372161554e-06,
      "loss": 0.6775,
      "step": 550
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1610362258753477,
      "learning_rate": 9.910032200245559e-06,
      "loss": 0.6217,
      "step": 551
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2273931029446528,
      "learning_rate": 9.909538690701448e-06,
      "loss": 0.7312,
      "step": 552
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2844182559917314,
      "learning_rate": 9.909043843663664e-06,
      "loss": 0.6205,
      "step": 553
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4034698382779167,
      "learning_rate": 9.908547659267019e-06,
      "loss": 0.7232,
      "step": 554
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1709127331685032,
      "learning_rate": 9.908050137646685e-06,
      "loss": 0.6718,
      "step": 555
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.063061417451581,
      "learning_rate": 9.907551278938203e-06,
      "loss": 0.6176,
      "step": 556
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1747705335839875,
      "learning_rate": 9.907051083277477e-06,
      "loss": 0.7033,
      "step": 557
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.175444179989701,
      "learning_rate": 9.90654955080077e-06,
      "loss": 0.6516,
      "step": 558
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2316346472906745,
      "learning_rate": 9.906046681644717e-06,
      "loss": 0.7115,
      "step": 559
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1633707899949712,
      "learning_rate": 9.905542475946312e-06,
      "loss": 0.6469,
      "step": 560
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.146568495624561,
      "learning_rate": 9.905036933842914e-06,
      "loss": 0.592,
      "step": 561
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3077577238127382,
      "learning_rate": 9.904530055472246e-06,
      "loss": 0.687,
      "step": 562
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2358972052507384,
      "learning_rate": 9.904021840972397e-06,
      "loss": 0.7368,
      "step": 563
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3106980093867937,
      "learning_rate": 9.90351229048182e-06,
      "loss": 0.7422,
      "step": 564
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1566044692963313,
      "learning_rate": 9.903001404139329e-06,
      "loss": 0.6838,
      "step": 565
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1644943085229642,
      "learning_rate": 9.902489182084101e-06,
      "loss": 0.7138,
      "step": 566
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1741870624884594,
      "learning_rate": 9.901975624455685e-06,
      "loss": 0.7217,
      "step": 567
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0519172418917817,
      "learning_rate": 9.901460731393984e-06,
      "loss": 0.5827,
      "step": 568
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1218434577276994,
      "learning_rate": 9.90094450303927e-06,
      "loss": 0.6013,
      "step": 569
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.177373214804677,
      "learning_rate": 9.900426939532178e-06,
      "loss": 0.6276,
      "step": 570
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2286618811684675,
      "learning_rate": 9.899908041013706e-06,
      "loss": 0.7373,
      "step": 571
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2366287760338548,
      "learning_rate": 9.899387807625217e-06,
      "loss": 0.6714,
      "step": 572
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3026809975438984,
      "learning_rate": 9.898866239508437e-06,
      "loss": 0.7736,
      "step": 573
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1520816039763897,
      "learning_rate": 9.898343336805456e-06,
      "loss": 0.716,
      "step": 574
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0900733072829238,
      "learning_rate": 9.897819099658726e-06,
      "loss": 0.6415,
      "step": 575
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.218270924805952,
      "learning_rate": 9.897293528211062e-06,
      "loss": 0.6468,
      "step": 576
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0200023789201833,
      "learning_rate": 9.896766622605649e-06,
      "loss": 0.6872,
      "step": 577
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1136163288424725,
      "learning_rate": 9.896238382986028e-06,
      "loss": 0.7393,
      "step": 578
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.236151668794328,
      "learning_rate": 9.895708809496106e-06,
      "loss": 0.6547,
      "step": 579
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8549241049806982,
      "learning_rate": 9.895177902280151e-06,
      "loss": 0.5048,
      "step": 580
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1784929771722537,
      "learning_rate": 9.8946456614828e-06,
      "loss": 0.691,
      "step": 581
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1972402589076063,
      "learning_rate": 9.894112087249049e-06,
      "loss": 0.7096,
      "step": 582
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.233755968771153,
      "learning_rate": 9.893577179724259e-06,
      "loss": 0.6047,
      "step": 583
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2058099789911745,
      "learning_rate": 9.893040939054152e-06,
      "loss": 0.6348,
      "step": 584
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1894890387326191,
      "learning_rate": 9.892503365384818e-06,
      "loss": 0.6789,
      "step": 585
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.273559098118028,
      "learning_rate": 9.891964458862702e-06,
      "loss": 0.7233,
      "step": 586
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1425957486655585,
      "learning_rate": 9.891424219634622e-06,
      "loss": 0.6412,
      "step": 587
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0745400895394557,
      "learning_rate": 9.89088264784775e-06,
      "loss": 0.5305,
      "step": 588
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8742768135738936,
      "learning_rate": 9.890339743649628e-06,
      "loss": 0.5204,
      "step": 589
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1540515846346784,
      "learning_rate": 9.889795507188156e-06,
      "loss": 0.6956,
      "step": 590
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0939616122277867,
      "learning_rate": 9.889249938611599e-06,
      "loss": 0.688,
      "step": 591
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6874176432520733,
      "learning_rate": 9.888703038068587e-06,
      "loss": 0.4784,
      "step": 592
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0524315957917807,
      "learning_rate": 9.88815480570811e-06,
      "loss": 0.6232,
      "step": 593
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.164761362071935,
      "learning_rate": 9.887605241679518e-06,
      "loss": 0.7593,
      "step": 594
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2079202270070422,
      "learning_rate": 9.887054346132534e-06,
      "loss": 0.7912,
      "step": 595
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0698576196714158,
      "learning_rate": 9.886502119217232e-06,
      "loss": 0.6369,
      "step": 596
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2028278733223143,
      "learning_rate": 9.885948561084056e-06,
      "loss": 0.7513,
      "step": 597
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1577396220768945,
      "learning_rate": 9.885393671883811e-06,
      "loss": 0.5627,
      "step": 598
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0432835497165727,
      "learning_rate": 9.884837451767662e-06,
      "loss": 0.5861,
      "step": 599
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1810660416020078,
      "learning_rate": 9.88427990088714e-06,
      "loss": 0.7421,
      "step": 600
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9088559897317271,
      "learning_rate": 9.883721019394136e-06,
      "loss": 0.5154,
      "step": 601
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.051312694435416,
      "learning_rate": 9.883160807440907e-06,
      "loss": 0.6475,
      "step": 602
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.134419693715016,
      "learning_rate": 9.88259926518007e-06,
      "loss": 0.5974,
      "step": 603
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1596943851834276,
      "learning_rate": 9.882036392764602e-06,
      "loss": 0.627,
      "step": 604
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1108763034258309,
      "learning_rate": 9.881472190347846e-06,
      "loss": 0.6636,
      "step": 605
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1798135078930199,
      "learning_rate": 9.880906658083507e-06,
      "loss": 0.6754,
      "step": 606
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1627964836285178,
      "learning_rate": 9.880339796125653e-06,
      "loss": 0.6587,
      "step": 607
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2117043357056776,
      "learning_rate": 9.87977160462871e-06,
      "loss": 0.5128,
      "step": 608
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1341039479983468,
      "learning_rate": 9.87920208374747e-06,
      "loss": 0.6298,
      "step": 609
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1498543020546936,
      "learning_rate": 9.878631233637086e-06,
      "loss": 0.6556,
      "step": 610
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.04641102779228,
      "learning_rate": 9.878059054453076e-06,
      "loss": 0.6047,
      "step": 611
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.071126347340601,
      "learning_rate": 9.877485546351312e-06,
      "loss": 0.6383,
      "step": 612
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1753241808683828,
      "learning_rate": 9.876910709488038e-06,
      "loss": 0.6768,
      "step": 613
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2292926282467858,
      "learning_rate": 9.876334544019855e-06,
      "loss": 0.6672,
      "step": 614
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2153076807372112,
      "learning_rate": 9.875757050103722e-06,
      "loss": 0.6448,
      "step": 615
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.067205415477389,
      "learning_rate": 9.87517822789697e-06,
      "loss": 0.5893,
      "step": 616
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0994707634423433,
      "learning_rate": 9.874598077557284e-06,
      "loss": 0.5902,
      "step": 617
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1401011224368014,
      "learning_rate": 9.874016599242711e-06,
      "loss": 0.603,
      "step": 618
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1003835302482134,
      "learning_rate": 9.873433793111665e-06,
      "loss": 0.5596,
      "step": 619
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2091003500915674,
      "learning_rate": 9.872849659322916e-06,
      "loss": 0.6302,
      "step": 620
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.194189383129039,
      "learning_rate": 9.872264198035599e-06,
      "loss": 0.7359,
      "step": 621
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1882669257254042,
      "learning_rate": 9.871677409409209e-06,
      "loss": 0.7466,
      "step": 622
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.048545848172781,
      "learning_rate": 9.871089293603604e-06,
      "loss": 0.675,
      "step": 623
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1772444243003835,
      "learning_rate": 9.870499850779003e-06,
      "loss": 0.6316,
      "step": 624
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0917820312479634,
      "learning_rate": 9.869909081095988e-06,
      "loss": 0.6267,
      "step": 625
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1056905288428078,
      "learning_rate": 9.8693169847155e-06,
      "loss": 0.6103,
      "step": 626
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1239072812755515,
      "learning_rate": 9.86872356179884e-06,
      "loss": 0.6771,
      "step": 627
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1603227876273414,
      "learning_rate": 9.868128812507675e-06,
      "loss": 0.5894,
      "step": 628
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9866016136427981,
      "learning_rate": 9.86753273700403e-06,
      "loss": 0.5816,
      "step": 629
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1409097107488921,
      "learning_rate": 9.866935335450293e-06,
      "loss": 0.7351,
      "step": 630
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0877914302948914,
      "learning_rate": 9.866336608009215e-06,
      "loss": 0.6099,
      "step": 631
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1477016855108522,
      "learning_rate": 9.865736554843904e-06,
      "loss": 0.7153,
      "step": 632
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2249509435579633,
      "learning_rate": 9.865135176117829e-06,
      "loss": 0.7152,
      "step": 633
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0584944801375353,
      "learning_rate": 9.864532471994823e-06,
      "loss": 0.5937,
      "step": 634
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0579573080148321,
      "learning_rate": 9.86392844263908e-06,
      "loss": 0.6522,
      "step": 635
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.130868722794895,
      "learning_rate": 9.863323088215155e-06,
      "loss": 0.673,
      "step": 636
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1358829015765082,
      "learning_rate": 9.862716408887962e-06,
      "loss": 0.6714,
      "step": 637
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.019101686143704,
      "learning_rate": 9.862108404822779e-06,
      "loss": 0.6388,
      "step": 638
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.128230021030712,
      "learning_rate": 9.861499076185241e-06,
      "loss": 0.6507,
      "step": 639
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0917670954400573,
      "learning_rate": 9.860888423141346e-06,
      "loss": 0.6943,
      "step": 640
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1808613522959637,
      "learning_rate": 9.860276445857456e-06,
      "loss": 0.7114,
      "step": 641
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.194345844631833,
      "learning_rate": 9.859663144500284e-06,
      "loss": 0.7589,
      "step": 642
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0992402143283686,
      "learning_rate": 9.859048519236915e-06,
      "loss": 0.5756,
      "step": 643
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.295352848148064,
      "learning_rate": 9.85843257023479e-06,
      "loss": 0.6805,
      "step": 644
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1402459000316831,
      "learning_rate": 9.85781529766171e-06,
      "loss": 0.6345,
      "step": 645
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1519704840342817,
      "learning_rate": 9.857196701685836e-06,
      "loss": 0.7206,
      "step": 646
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0864437251552281,
      "learning_rate": 9.85657678247569e-06,
      "loss": 0.6296,
      "step": 647
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2173719600753716,
      "learning_rate": 9.855955540200157e-06,
      "loss": 0.7159,
      "step": 648
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1503016699662922,
      "learning_rate": 9.85533297502848e-06,
      "loss": 0.7001,
      "step": 649
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.182056435634815,
      "learning_rate": 9.854709087130261e-06,
      "loss": 0.7288,
      "step": 650
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0603320690552864,
      "learning_rate": 9.854083876675466e-06,
      "loss": 0.6133,
      "step": 651
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2333423736364608,
      "learning_rate": 9.853457343834421e-06,
      "loss": 0.759,
      "step": 652
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7852678318821367,
      "learning_rate": 9.852829488777808e-06,
      "loss": 0.4909,
      "step": 653
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1790140422269264,
      "learning_rate": 9.852200311676675e-06,
      "loss": 0.6759,
      "step": 654
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2372068215621552,
      "learning_rate": 9.851569812702423e-06,
      "loss": 0.684,
      "step": 655
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0780591363053666,
      "learning_rate": 9.850937992026821e-06,
      "loss": 0.6337,
      "step": 656
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.24874562719759,
      "learning_rate": 9.850304849821992e-06,
      "loss": 0.6613,
      "step": 657
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.239132838668827,
      "learning_rate": 9.849670386260424e-06,
      "loss": 0.7061,
      "step": 658
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1465329065335992,
      "learning_rate": 9.84903460151496e-06,
      "loss": 0.6595,
      "step": 659
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1502214269235445,
      "learning_rate": 9.848397495758806e-06,
      "loss": 0.6677,
      "step": 660
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0893541842015229,
      "learning_rate": 9.847759069165528e-06,
      "loss": 0.6288,
      "step": 661
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0703048100044776,
      "learning_rate": 9.847119321909049e-06,
      "loss": 0.6834,
      "step": 662
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0830875806924167,
      "learning_rate": 9.846478254163657e-06,
      "loss": 0.6919,
      "step": 663
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1460320709487686,
      "learning_rate": 9.845835866103993e-06,
      "loss": 0.7148,
      "step": 664
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1177630643332157,
      "learning_rate": 9.845192157905063e-06,
      "loss": 0.6703,
      "step": 665
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1247522391186011,
      "learning_rate": 9.844547129742231e-06,
      "loss": 0.7139,
      "step": 666
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1841373416483318,
      "learning_rate": 9.84390078179122e-06,
      "loss": 0.6941,
      "step": 667
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1704535720211449,
      "learning_rate": 9.843253114228113e-06,
      "loss": 0.6625,
      "step": 668
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1612216617566975,
      "learning_rate": 9.842604127229353e-06,
      "loss": 0.6624,
      "step": 669
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1067298535057377,
      "learning_rate": 9.841953820971742e-06,
      "loss": 0.6945,
      "step": 670
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.09345165677508,
      "learning_rate": 9.841302195632441e-06,
      "loss": 0.6591,
      "step": 671
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1259759464358314,
      "learning_rate": 9.84064925138897e-06,
      "loss": 0.6408,
      "step": 672
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0662586892915495,
      "learning_rate": 9.839994988419208e-06,
      "loss": 0.6321,
      "step": 673
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.182905221340787,
      "learning_rate": 9.839339406901398e-06,
      "loss": 0.7055,
      "step": 674
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2768753281187153,
      "learning_rate": 9.838682507014138e-06,
      "loss": 0.7878,
      "step": 675
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7870188064085515,
      "learning_rate": 9.838024288936381e-06,
      "loss": 0.4757,
      "step": 676
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0550074565577388,
      "learning_rate": 9.837364752847447e-06,
      "loss": 0.6249,
      "step": 677
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.15199744650406,
      "learning_rate": 9.836703898927012e-06,
      "loss": 0.7149,
      "step": 678
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2233988556140745,
      "learning_rate": 9.83604172735511e-06,
      "loss": 0.6194,
      "step": 679
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0621113126186494,
      "learning_rate": 9.835378238312136e-06,
      "loss": 0.6313,
      "step": 680
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.091762318648744,
      "learning_rate": 9.834713431978842e-06,
      "loss": 0.5985,
      "step": 681
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7012729213887197,
      "learning_rate": 9.834047308536342e-06,
      "loss": 0.4795,
      "step": 682
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1695086984838274,
      "learning_rate": 9.8333798681661e-06,
      "loss": 0.666,
      "step": 683
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1379929483367186,
      "learning_rate": 9.83271111104995e-06,
      "loss": 0.6903,
      "step": 684
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1673677106963345,
      "learning_rate": 9.83204103737008e-06,
      "loss": 0.6772,
      "step": 685
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0782050856966119,
      "learning_rate": 9.831369647309033e-06,
      "loss": 0.7151,
      "step": 686
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0722181439698553,
      "learning_rate": 9.830696941049718e-06,
      "loss": 0.6869,
      "step": 687
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0771743670512253,
      "learning_rate": 9.830022918775399e-06,
      "loss": 0.6889,
      "step": 688
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.104686317721014,
      "learning_rate": 9.829347580669694e-06,
      "loss": 0.6168,
      "step": 689
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0959546815793324,
      "learning_rate": 9.828670926916588e-06,
      "loss": 0.6831,
      "step": 690
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1304581160333063,
      "learning_rate": 9.827992957700417e-06,
      "loss": 0.6638,
      "step": 691
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.163236390540185,
      "learning_rate": 9.82731367320588e-06,
      "loss": 0.6033,
      "step": 692
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1756678543760257,
      "learning_rate": 9.826633073618034e-06,
      "loss": 0.6585,
      "step": 693
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.096836231345986,
      "learning_rate": 9.825951159122291e-06,
      "loss": 0.6613,
      "step": 694
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1354930581695504,
      "learning_rate": 9.825267929904423e-06,
      "loss": 0.6716,
      "step": 695
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.092673470636174,
      "learning_rate": 9.824583386150563e-06,
      "loss": 0.6532,
      "step": 696
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1552783404769593,
      "learning_rate": 9.8238975280472e-06,
      "loss": 0.6029,
      "step": 697
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.018465462176151,
      "learning_rate": 9.823210355781177e-06,
      "loss": 0.6109,
      "step": 698
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.196860147792639,
      "learning_rate": 9.822521869539699e-06,
      "loss": 0.7328,
      "step": 699
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1127392945890429,
      "learning_rate": 9.821832069510333e-06,
      "loss": 0.7219,
      "step": 700
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1318025689002094,
      "learning_rate": 9.821140955880995e-06,
      "loss": 0.6095,
      "step": 701
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0724054236572238,
      "learning_rate": 9.820448528839965e-06,
      "loss": 0.6659,
      "step": 702
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1351209035612464,
      "learning_rate": 9.81975478857588e-06,
      "loss": 0.6478,
      "step": 703
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0787567327934542,
      "learning_rate": 9.819059735277731e-06,
      "loss": 0.6246,
      "step": 704
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0605840424997237,
      "learning_rate": 9.818363369134876e-06,
      "loss": 0.5663,
      "step": 705
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1255932353729787,
      "learning_rate": 9.817665690337017e-06,
      "loss": 0.6862,
      "step": 706
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1249921647757453,
      "learning_rate": 9.816966699074226e-06,
      "loss": 0.7629,
      "step": 707
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1550934119151437,
      "learning_rate": 9.816266395536924e-06,
      "loss": 0.6357,
      "step": 708
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1906576137877185,
      "learning_rate": 9.815564779915896e-06,
      "loss": 0.6904,
      "step": 709
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1579017686030542,
      "learning_rate": 9.81486185240228e-06,
      "loss": 0.6312,
      "step": 710
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1575139068449294,
      "learning_rate": 9.814157613187573e-06,
      "loss": 0.6666,
      "step": 711
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1139146494658285,
      "learning_rate": 9.81345206246363e-06,
      "loss": 0.7292,
      "step": 712
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2033822814357062,
      "learning_rate": 9.812745200422659e-06,
      "loss": 0.6855,
      "step": 713
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2527931288867737,
      "learning_rate": 9.812037027257234e-06,
      "loss": 0.65,
      "step": 714
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1196317449848967,
      "learning_rate": 9.811327543160276e-06,
      "loss": 0.6849,
      "step": 715
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0904044480536927,
      "learning_rate": 9.810616748325072e-06,
      "loss": 0.6313,
      "step": 716
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0372026342087108,
      "learning_rate": 9.809904642945261e-06,
      "loss": 0.537,
      "step": 717
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1888525468182856,
      "learning_rate": 9.809191227214838e-06,
      "loss": 0.6864,
      "step": 718
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0619869673768934,
      "learning_rate": 9.808476501328158e-06,
      "loss": 0.7105,
      "step": 719
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7241118451270379,
      "learning_rate": 9.807760465479934e-06,
      "loss": 0.4456,
      "step": 720
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2627579704941168,
      "learning_rate": 9.807043119865232e-06,
      "loss": 0.6565,
      "step": 721
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1371006135755681,
      "learning_rate": 9.806324464679477e-06,
      "loss": 0.665,
      "step": 722
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2864860733532468,
      "learning_rate": 9.80560450011845e-06,
      "loss": 0.7814,
      "step": 723
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0446563209593065,
      "learning_rate": 9.80488322637829e-06,
      "loss": 0.5876,
      "step": 724
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1552104668247112,
      "learning_rate": 9.804160643655494e-06,
      "loss": 0.6429,
      "step": 725
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2119777153951785,
      "learning_rate": 9.803436752146909e-06,
      "loss": 0.6365,
      "step": 726
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1257518650891913,
      "learning_rate": 9.802711552049746e-06,
      "loss": 0.6703,
      "step": 727
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.101282230316314,
      "learning_rate": 9.801985043561569e-06,
      "loss": 0.6417,
      "step": 728
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0110835759810834,
      "learning_rate": 9.801257226880297e-06,
      "loss": 0.6924,
      "step": 729
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.124466843713137,
      "learning_rate": 9.80052810220421e-06,
      "loss": 0.6736,
      "step": 730
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.101597531285294,
      "learning_rate": 9.79979766973194e-06,
      "loss": 0.701,
      "step": 731
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.077469693166165,
      "learning_rate": 9.799065929662478e-06,
      "loss": 0.6527,
      "step": 732
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1581554184086444,
      "learning_rate": 9.798332882195172e-06,
      "loss": 0.646,
      "step": 733
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1227707961734683,
      "learning_rate": 9.79759852752972e-06,
      "loss": 0.6983,
      "step": 734
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.164062495658345,
      "learning_rate": 9.796862865866183e-06,
      "loss": 0.7205,
      "step": 735
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2178982921311714,
      "learning_rate": 9.796125897404973e-06,
      "loss": 0.589,
      "step": 736
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0160301244125352,
      "learning_rate": 9.795387622346866e-06,
      "loss": 0.6328,
      "step": 737
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0672582895160998,
      "learning_rate": 9.794648040892983e-06,
      "loss": 0.668,
      "step": 738
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2228926893775223,
      "learning_rate": 9.793907153244808e-06,
      "loss": 0.7207,
      "step": 739
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0820212768172814,
      "learning_rate": 9.79316495960418e-06,
      "loss": 0.5672,
      "step": 740
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1554934015585892,
      "learning_rate": 9.792421460173294e-06,
      "loss": 0.7083,
      "step": 741
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0952142194969918,
      "learning_rate": 9.791676655154696e-06,
      "loss": 0.6947,
      "step": 742
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7018228248119613,
      "learning_rate": 9.790930544751297e-06,
      "loss": 0.487,
      "step": 743
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0923356573550314,
      "learning_rate": 9.790183129166351e-06,
      "loss": 0.6142,
      "step": 744
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1200044821085462,
      "learning_rate": 9.78943440860348e-06,
      "loss": 0.702,
      "step": 745
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1654238353641524,
      "learning_rate": 9.788684383266655e-06,
      "loss": 0.6778,
      "step": 746
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1419804082456895,
      "learning_rate": 9.787933053360203e-06,
      "loss": 0.6974,
      "step": 747
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1063097025805682,
      "learning_rate": 9.78718041908881e-06,
      "loss": 0.624,
      "step": 748
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.049643560220273,
      "learning_rate": 9.786426480657507e-06,
      "loss": 0.6779,
      "step": 749
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1054719518643539,
      "learning_rate": 9.785671238271694e-06,
      "loss": 0.5698,
      "step": 750
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1107476670906633,
      "learning_rate": 9.784914692137118e-06,
      "loss": 0.6779,
      "step": 751
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0825626936632151,
      "learning_rate": 9.784156842459882e-06,
      "loss": 0.6742,
      "step": 752
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1876552992394858,
      "learning_rate": 9.783397689446447e-06,
      "loss": 0.6925,
      "step": 753
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0935688099305556,
      "learning_rate": 9.782637233303624e-06,
      "loss": 0.6637,
      "step": 754
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1056973865756239,
      "learning_rate": 9.781875474238587e-06,
      "loss": 0.6508,
      "step": 755
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0251144668629986,
      "learning_rate": 9.781112412458858e-06,
      "loss": 0.681,
      "step": 756
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.118401847631918,
      "learning_rate": 9.780348048172315e-06,
      "loss": 0.6291,
      "step": 757
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0856948322979647,
      "learning_rate": 9.779582381587192e-06,
      "loss": 0.6764,
      "step": 758
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1167775815106917,
      "learning_rate": 9.778815412912078e-06,
      "loss": 0.69,
      "step": 759
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1156245136316403,
      "learning_rate": 9.778047142355917e-06,
      "loss": 0.6555,
      "step": 760
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.117224523287798,
      "learning_rate": 9.777277570128008e-06,
      "loss": 0.6188,
      "step": 761
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1400145150224619,
      "learning_rate": 9.776506696438002e-06,
      "loss": 0.6966,
      "step": 762
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1098635952580846,
      "learning_rate": 9.775734521495905e-06,
      "loss": 0.7047,
      "step": 763
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.176647613060308,
      "learning_rate": 9.774961045512082e-06,
      "loss": 0.7205,
      "step": 764
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.060312563904902,
      "learning_rate": 9.774186268697247e-06,
      "loss": 0.6392,
      "step": 765
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1332929406673735,
      "learning_rate": 9.773410191262471e-06,
      "loss": 0.704,
      "step": 766
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0878436519922638,
      "learning_rate": 9.772632813419181e-06,
      "loss": 0.6296,
      "step": 767
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1094079349670647,
      "learning_rate": 9.771854135379153e-06,
      "loss": 0.7008,
      "step": 768
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0867050871032815,
      "learning_rate": 9.771074157354521e-06,
      "loss": 0.6736,
      "step": 769
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.108129505299658,
      "learning_rate": 9.770292879557774e-06,
      "loss": 0.6275,
      "step": 770
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1733838763765414,
      "learning_rate": 9.769510302201751e-06,
      "loss": 0.6377,
      "step": 771
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2078112756030042,
      "learning_rate": 9.76872642549965e-06,
      "loss": 0.7529,
      "step": 772
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.188180707633336,
      "learning_rate": 9.767941249665022e-06,
      "loss": 0.7171,
      "step": 773
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.015452664559184,
      "learning_rate": 9.767154774911767e-06,
      "loss": 0.5639,
      "step": 774
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9867899983879593,
      "learning_rate": 9.766367001454144e-06,
      "loss": 0.5951,
      "step": 775
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1090135612693237,
      "learning_rate": 9.765577929506764e-06,
      "loss": 0.7219,
      "step": 776
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0934918207561355,
      "learning_rate": 9.764787559284592e-06,
      "loss": 0.6733,
      "step": 777
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0662334974604029,
      "learning_rate": 9.763995891002946e-06,
      "loss": 0.6458,
      "step": 778
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0481399885582665,
      "learning_rate": 9.763202924877502e-06,
      "loss": 0.6507,
      "step": 779
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1235560146026446,
      "learning_rate": 9.762408661124279e-06,
      "loss": 0.6719,
      "step": 780
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1621910675287144,
      "learning_rate": 9.761613099959663e-06,
      "loss": 0.7111,
      "step": 781
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0343618062802882,
      "learning_rate": 9.760816241600383e-06,
      "loss": 0.57,
      "step": 782
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1602436667974612,
      "learning_rate": 9.760018086263525e-06,
      "loss": 0.6901,
      "step": 783
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0690032226897987,
      "learning_rate": 9.759218634166531e-06,
      "loss": 0.6748,
      "step": 784
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.109745664602789,
      "learning_rate": 9.75841788552719e-06,
      "loss": 0.6241,
      "step": 785
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0897497460054035,
      "learning_rate": 9.757615840563654e-06,
      "loss": 0.682,
      "step": 786
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.035094430965384,
      "learning_rate": 9.756812499494417e-06,
      "loss": 0.6437,
      "step": 787
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.046623122788427,
      "learning_rate": 9.756007862538333e-06,
      "loss": 0.6591,
      "step": 788
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1518106250189126,
      "learning_rate": 9.755201929914607e-06,
      "loss": 0.7309,
      "step": 789
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1135561339059143,
      "learning_rate": 9.754394701842797e-06,
      "loss": 0.7424,
      "step": 790
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.990885023960312,
      "learning_rate": 9.753586178542815e-06,
      "loss": 0.617,
      "step": 791
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2891519186594527,
      "learning_rate": 9.752776360234925e-06,
      "loss": 0.793,
      "step": 792
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0976370897847914,
      "learning_rate": 9.751965247139743e-06,
      "loss": 0.6584,
      "step": 793
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0980152434001413,
      "learning_rate": 9.751152839478238e-06,
      "loss": 0.6525,
      "step": 794
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1686544447687142,
      "learning_rate": 9.750339137471733e-06,
      "loss": 0.6586,
      "step": 795
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0804798707012149,
      "learning_rate": 9.749524141341905e-06,
      "loss": 0.673,
      "step": 796
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2306899490406018,
      "learning_rate": 9.74870785131078e-06,
      "loss": 0.68,
      "step": 797
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0870759004278514,
      "learning_rate": 9.747890267600735e-06,
      "loss": 0.7081,
      "step": 798
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.027898436759447,
      "learning_rate": 9.747071390434507e-06,
      "loss": 0.6773,
      "step": 799
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1362753191845703,
      "learning_rate": 9.746251220035176e-06,
      "loss": 0.7214,
      "step": 800
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.142417329753885,
      "learning_rate": 9.745429756626184e-06,
      "loss": 0.6716,
      "step": 801
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0901056393581727,
      "learning_rate": 9.744607000431317e-06,
      "loss": 0.6324,
      "step": 802
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1392228290670723,
      "learning_rate": 9.743782951674715e-06,
      "loss": 0.6286,
      "step": 803
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.055072373051012,
      "learning_rate": 9.742957610580876e-06,
      "loss": 0.6666,
      "step": 804
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.047155710841041,
      "learning_rate": 9.742130977374643e-06,
      "loss": 0.6228,
      "step": 805
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.049592201301659,
      "learning_rate": 9.741303052281214e-06,
      "loss": 0.5766,
      "step": 806
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1677753500888257,
      "learning_rate": 9.740473835526139e-06,
      "loss": 0.7503,
      "step": 807
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.055867029592767,
      "learning_rate": 9.73964332733532e-06,
      "loss": 0.6741,
      "step": 808
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1829879144892457,
      "learning_rate": 9.738811527935008e-06,
      "loss": 0.593,
      "step": 809
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1435573110805033,
      "learning_rate": 9.737978437551812e-06,
      "loss": 0.6499,
      "step": 810
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0980371455634672,
      "learning_rate": 9.737144056412685e-06,
      "loss": 0.5825,
      "step": 811
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.060457506292837,
      "learning_rate": 9.736308384744935e-06,
      "loss": 0.6587,
      "step": 812
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0180825102620248,
      "learning_rate": 9.735471422776225e-06,
      "loss": 0.5964,
      "step": 813
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1984903285617077,
      "learning_rate": 9.734633170734568e-06,
      "loss": 0.7007,
      "step": 814
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0941518362757818,
      "learning_rate": 9.733793628848323e-06,
      "loss": 0.643,
      "step": 815
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1305095837908723,
      "learning_rate": 9.732952797346205e-06,
      "loss": 0.5793,
      "step": 816
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1076645657388142,
      "learning_rate": 9.73211067645728e-06,
      "loss": 0.6271,
      "step": 817
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.226772057164279,
      "learning_rate": 9.731267266410967e-06,
      "loss": 0.6866,
      "step": 818
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.034266121827543,
      "learning_rate": 9.73042256743703e-06,
      "loss": 0.6298,
      "step": 819
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0318928534724523,
      "learning_rate": 9.729576579765594e-06,
      "loss": 0.6544,
      "step": 820
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.074490253360686,
      "learning_rate": 9.728729303627124e-06,
      "loss": 0.6491,
      "step": 821
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0867944566176206,
      "learning_rate": 9.727880739252444e-06,
      "loss": 0.6981,
      "step": 822
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.109875300934573,
      "learning_rate": 9.727030886872724e-06,
      "loss": 0.6538,
      "step": 823
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1758615686974467,
      "learning_rate": 9.726179746719492e-06,
      "loss": 0.6619,
      "step": 824
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.145086451533505,
      "learning_rate": 9.725327319024618e-06,
      "loss": 0.6703,
      "step": 825
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1070722263596582,
      "learning_rate": 9.724473604020327e-06,
      "loss": 0.6395,
      "step": 826
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1352621847649027,
      "learning_rate": 9.723618601939197e-06,
      "loss": 0.611,
      "step": 827
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0469384773305612,
      "learning_rate": 9.722762313014152e-06,
      "loss": 0.6155,
      "step": 828
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1717381655097072,
      "learning_rate": 9.721904737478468e-06,
      "loss": 0.6281,
      "step": 829
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1094618445886826,
      "learning_rate": 9.721045875565774e-06,
      "loss": 0.744,
      "step": 830
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0072172503069594,
      "learning_rate": 9.720185727510047e-06,
      "loss": 0.6605,
      "step": 831
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1824748151331537,
      "learning_rate": 9.719324293545615e-06,
      "loss": 0.6994,
      "step": 832
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0453950970818084,
      "learning_rate": 9.718461573907158e-06,
      "loss": 0.6481,
      "step": 833
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6403298070341439,
      "learning_rate": 9.717597568829702e-06,
      "loss": 0.6364,
      "step": 834
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1493262552461538,
      "learning_rate": 9.716732278548628e-06,
      "loss": 0.6513,
      "step": 835
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1579344302878785,
      "learning_rate": 9.715865703299663e-06,
      "loss": 0.6794,
      "step": 836
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2249903414880976,
      "learning_rate": 9.714997843318887e-06,
      "loss": 0.7409,
      "step": 837
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0881109905088606,
      "learning_rate": 9.714128698842728e-06,
      "loss": 0.5861,
      "step": 838
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2070736888264695,
      "learning_rate": 9.713258270107966e-06,
      "loss": 0.7161,
      "step": 839
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0847757512329015,
      "learning_rate": 9.71238655735173e-06,
      "loss": 0.5815,
      "step": 840
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1386957887593023,
      "learning_rate": 9.711513560811498e-06,
      "loss": 0.7045,
      "step": 841
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1059070091102732,
      "learning_rate": 9.710639280725097e-06,
      "loss": 0.6515,
      "step": 842
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9714775195343888,
      "learning_rate": 9.709763717330709e-06,
      "loss": 0.63,
      "step": 843
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.113181883490431,
      "learning_rate": 9.708886870866856e-06,
      "loss": 0.6224,
      "step": 844
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.035446031914055,
      "learning_rate": 9.708008741572422e-06,
      "loss": 0.6571,
      "step": 845
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1563565652996073,
      "learning_rate": 9.707129329686626e-06,
      "loss": 0.6679,
      "step": 846
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0976526167786826,
      "learning_rate": 9.706248635449048e-06,
      "loss": 0.6834,
      "step": 847
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1267220131330384,
      "learning_rate": 9.705366659099617e-06,
      "loss": 0.623,
      "step": 848
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.171574879508651,
      "learning_rate": 9.704483400878602e-06,
      "loss": 0.6779,
      "step": 849
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0831137161215711,
      "learning_rate": 9.703598861026627e-06,
      "loss": 0.716,
      "step": 850
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1103037656427084,
      "learning_rate": 9.702713039784668e-06,
      "loss": 0.7062,
      "step": 851
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1627111593577728,
      "learning_rate": 9.701825937394045e-06,
      "loss": 0.7174,
      "step": 852
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.827784484577726,
      "learning_rate": 9.700937554096432e-06,
      "loss": 0.4967,
      "step": 853
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7681586313814103,
      "learning_rate": 9.700047890133845e-06,
      "loss": 0.4805,
      "step": 854
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1642378919689045,
      "learning_rate": 9.699156945748657e-06,
      "loss": 0.7068,
      "step": 855
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0724223629447225,
      "learning_rate": 9.698264721183584e-06,
      "loss": 0.6721,
      "step": 856
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6884331745793695,
      "learning_rate": 9.69737121668169e-06,
      "loss": 0.4977,
      "step": 857
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.056495437876798,
      "learning_rate": 9.696476432486395e-06,
      "loss": 0.644,
      "step": 858
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0891122881543713,
      "learning_rate": 9.695580368841462e-06,
      "loss": 0.6839,
      "step": 859
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0386168600187955,
      "learning_rate": 9.694683025991e-06,
      "loss": 0.5733,
      "step": 860
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1844881432130085,
      "learning_rate": 9.693784404179472e-06,
      "loss": 0.7426,
      "step": 861
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0289717955913023,
      "learning_rate": 9.692884503651687e-06,
      "loss": 0.6666,
      "step": 862
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0737368679428585,
      "learning_rate": 9.691983324652804e-06,
      "loss": 0.4978,
      "step": 863
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.028322598339251,
      "learning_rate": 9.691080867428328e-06,
      "loss": 0.6368,
      "step": 864
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.171478679657464,
      "learning_rate": 9.690177132224113e-06,
      "loss": 0.784,
      "step": 865
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1377306989337588,
      "learning_rate": 9.689272119286361e-06,
      "loss": 0.6456,
      "step": 866
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1727858762219754,
      "learning_rate": 9.688365828861625e-06,
      "loss": 0.7021,
      "step": 867
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1349317783807622,
      "learning_rate": 9.6874582611968e-06,
      "loss": 0.6557,
      "step": 868
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1713535969491309,
      "learning_rate": 9.686549416539135e-06,
      "loss": 0.7064,
      "step": 869
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8532548834613224,
      "learning_rate": 9.685639295136224e-06,
      "loss": 0.5009,
      "step": 870
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0712037023872192,
      "learning_rate": 9.684727897236008e-06,
      "loss": 0.6097,
      "step": 871
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1624935024369119,
      "learning_rate": 9.683815223086777e-06,
      "loss": 0.7184,
      "step": 872
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7720612323267094,
      "learning_rate": 9.68290127293717e-06,
      "loss": 0.5101,
      "step": 873
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1434917699888576,
      "learning_rate": 9.68198604703617e-06,
      "loss": 0.6994,
      "step": 874
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6899637140459434,
      "learning_rate": 9.681069545633113e-06,
      "loss": 0.4665,
      "step": 875
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1056463048579763,
      "learning_rate": 9.680151768977676e-06,
      "loss": 0.6286,
      "step": 876
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1516937702379184,
      "learning_rate": 9.679232717319887e-06,
      "loss": 0.6927,
      "step": 877
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0994712415279446,
      "learning_rate": 9.678312390910123e-06,
      "loss": 0.6683,
      "step": 878
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1830052145770937,
      "learning_rate": 9.677390789999106e-06,
      "loss": 0.7788,
      "step": 879
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.057782602742653,
      "learning_rate": 9.676467914837904e-06,
      "loss": 0.5236,
      "step": 880
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5359709510592465,
      "learning_rate": 9.675543765677935e-06,
      "loss": 0.5224,
      "step": 881
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.100154684112589,
      "learning_rate": 9.674618342770962e-06,
      "loss": 0.6958,
      "step": 882
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.136639220551466,
      "learning_rate": 9.673691646369094e-06,
      "loss": 0.6564,
      "step": 883
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2037814544313608,
      "learning_rate": 9.672763676724792e-06,
      "loss": 0.6566,
      "step": 884
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1032279079872478,
      "learning_rate": 9.671834434090861e-06,
      "loss": 0.6862,
      "step": 885
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.036831533197075,
      "learning_rate": 9.670903918720446e-06,
      "loss": 0.6109,
      "step": 886
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2209863702651758,
      "learning_rate": 9.669972130867053e-06,
      "loss": 0.675,
      "step": 887
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0165531027139059,
      "learning_rate": 9.66903907078452e-06,
      "loss": 0.5698,
      "step": 888
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.123683699176069,
      "learning_rate": 9.668104738727045e-06,
      "loss": 0.7357,
      "step": 889
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0449187369928445,
      "learning_rate": 9.667169134949158e-06,
      "loss": 0.6679,
      "step": 890
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.117650197986259,
      "learning_rate": 9.666232259705751e-06,
      "loss": 0.603,
      "step": 891
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0962511217947746,
      "learning_rate": 9.66529411325205e-06,
      "loss": 0.7291,
      "step": 892
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0411376733134625,
      "learning_rate": 9.664354695843632e-06,
      "loss": 0.6411,
      "step": 893
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1233724061884105,
      "learning_rate": 9.66341400773642e-06,
      "loss": 0.5903,
      "step": 894
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1429336410299331,
      "learning_rate": 9.662472049186688e-06,
      "loss": 0.7038,
      "step": 895
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1443299551680814,
      "learning_rate": 9.661528820451045e-06,
      "loss": 0.7106,
      "step": 896
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1385015234568867,
      "learning_rate": 9.660584321786456e-06,
      "loss": 0.5751,
      "step": 897
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0473321858348568,
      "learning_rate": 9.65963855345023e-06,
      "loss": 0.6439,
      "step": 898
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0886459659523429,
      "learning_rate": 9.658691515700016e-06,
      "loss": 0.693,
      "step": 899
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0793842832486205,
      "learning_rate": 9.657743208793818e-06,
      "loss": 0.584,
      "step": 900
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1213955090964036,
      "learning_rate": 9.656793632989976e-06,
      "loss": 0.7693,
      "step": 901
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0862173410923455,
      "learning_rate": 9.655842788547184e-06,
      "loss": 0.5919,
      "step": 902
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.112356874383914,
      "learning_rate": 9.654890675724478e-06,
      "loss": 0.6794,
      "step": 903
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2235856467663266,
      "learning_rate": 9.653937294781237e-06,
      "loss": 0.6724,
      "step": 904
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1662553926184664,
      "learning_rate": 9.652982645977193e-06,
      "loss": 0.6753,
      "step": 905
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1928430117441486,
      "learning_rate": 9.652026729572415e-06,
      "loss": 0.6826,
      "step": 906
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1172491333052608,
      "learning_rate": 9.651069545827321e-06,
      "loss": 0.6576,
      "step": 907
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1087359463284592,
      "learning_rate": 9.650111095002676e-06,
      "loss": 0.6077,
      "step": 908
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0960101508285884,
      "learning_rate": 9.649151377359588e-06,
      "loss": 0.7588,
      "step": 909
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.113640118578029,
      "learning_rate": 9.64819039315951e-06,
      "loss": 0.6631,
      "step": 910
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0757990384428555,
      "learning_rate": 9.647228142664241e-06,
      "loss": 0.6957,
      "step": 911
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0964473543020707,
      "learning_rate": 9.646264626135924e-06,
      "loss": 0.6412,
      "step": 912
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1738124788974702,
      "learning_rate": 9.645299843837047e-06,
      "loss": 0.723,
      "step": 913
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1149588254920535,
      "learning_rate": 9.644333796030444e-06,
      "loss": 0.6204,
      "step": 914
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0234351273863371,
      "learning_rate": 9.643366482979296e-06,
      "loss": 0.7208,
      "step": 915
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0846232740957282,
      "learning_rate": 9.64239790494712e-06,
      "loss": 0.5845,
      "step": 916
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1518984132232157,
      "learning_rate": 9.641428062197789e-06,
      "loss": 0.6871,
      "step": 917
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0898179301392574,
      "learning_rate": 9.640456954995509e-06,
      "loss": 0.6522,
      "step": 918
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1011862063784692,
      "learning_rate": 9.639484583604841e-06,
      "loss": 0.6956,
      "step": 919
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2751146477467272,
      "learning_rate": 9.638510948290684e-06,
      "loss": 0.6727,
      "step": 920
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0840926302595486,
      "learning_rate": 9.637536049318284e-06,
      "loss": 0.7329,
      "step": 921
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1685891887418123,
      "learning_rate": 9.63655988695323e-06,
      "loss": 0.7877,
      "step": 922
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.117801561495144,
      "learning_rate": 9.635582461461455e-06,
      "loss": 0.6997,
      "step": 923
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.130812199330516,
      "learning_rate": 9.634603773109235e-06,
      "loss": 0.6491,
      "step": 924
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0439990237422345,
      "learning_rate": 9.633623822163196e-06,
      "loss": 0.693,
      "step": 925
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0273413604250146,
      "learning_rate": 9.6326426088903e-06,
      "loss": 0.6889,
      "step": 926
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0719983580740586,
      "learning_rate": 9.631660133557858e-06,
      "loss": 0.6696,
      "step": 927
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1047675843873939,
      "learning_rate": 9.630676396433524e-06,
      "loss": 0.6232,
      "step": 928
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0785846392506815,
      "learning_rate": 9.629691397785294e-06,
      "loss": 0.6707,
      "step": 929
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0958993582790775,
      "learning_rate": 9.628705137881509e-06,
      "loss": 0.6236,
      "step": 930
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.152198655506274,
      "learning_rate": 9.627717616990853e-06,
      "loss": 0.6595,
      "step": 931
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0585660818456581,
      "learning_rate": 9.626728835382353e-06,
      "loss": 0.7122,
      "step": 932
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.070465798404608,
      "learning_rate": 9.625738793325384e-06,
      "loss": 0.6245,
      "step": 933
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.153897335597271,
      "learning_rate": 9.624747491089657e-06,
      "loss": 0.7077,
      "step": 934
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0864882772702205,
      "learning_rate": 9.623754928945233e-06,
      "loss": 0.6052,
      "step": 935
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.994726691565809,
      "learning_rate": 9.622761107162511e-06,
      "loss": 0.5298,
      "step": 936
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0469794163856567,
      "learning_rate": 9.621766026012236e-06,
      "loss": 0.5545,
      "step": 937
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.169758419362588,
      "learning_rate": 9.620769685765497e-06,
      "loss": 0.7218,
      "step": 938
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1602537149422971,
      "learning_rate": 9.619772086693721e-06,
      "loss": 0.6838,
      "step": 939
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.070276426000004,
      "learning_rate": 9.618773229068685e-06,
      "loss": 0.6879,
      "step": 940
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.025536597725996,
      "learning_rate": 9.617773113162505e-06,
      "loss": 0.6068,
      "step": 941
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7313660770269282,
      "learning_rate": 9.616771739247639e-06,
      "loss": 0.5093,
      "step": 942
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0953577298773127,
      "learning_rate": 9.615769107596888e-06,
      "loss": 0.6864,
      "step": 943
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0747299133367834,
      "learning_rate": 9.614765218483398e-06,
      "loss": 0.6503,
      "step": 944
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.066941078305158,
      "learning_rate": 9.613760072180656e-06,
      "loss": 0.6523,
      "step": 945
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.157118947739973,
      "learning_rate": 9.612753668962492e-06,
      "loss": 0.6954,
      "step": 946
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0060634229123426,
      "learning_rate": 9.611746009103077e-06,
      "loss": 0.6212,
      "step": 947
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1238563226085063,
      "learning_rate": 9.610737092876924e-06,
      "loss": 0.7266,
      "step": 948
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0458060523573245,
      "learning_rate": 9.609726920558893e-06,
      "loss": 0.639,
      "step": 949
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9720594327926683,
      "learning_rate": 9.60871549242418e-06,
      "loss": 0.6299,
      "step": 950
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0991738860570166,
      "learning_rate": 9.607702808748329e-06,
      "loss": 0.6112,
      "step": 951
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0095207334456053,
      "learning_rate": 9.60668886980722e-06,
      "loss": 0.5569,
      "step": 952
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1126848004583536,
      "learning_rate": 9.605673675877077e-06,
      "loss": 0.6149,
      "step": 953
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1911304343291462,
      "learning_rate": 9.604657227234468e-06,
      "loss": 0.7099,
      "step": 954
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0534403542883153,
      "learning_rate": 9.603639524156306e-06,
      "loss": 0.6823,
      "step": 955
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0516101331808927,
      "learning_rate": 9.602620566919834e-06,
      "loss": 0.6399,
      "step": 956
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0499763084025036,
      "learning_rate": 9.601600355802649e-06,
      "loss": 0.6496,
      "step": 957
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1293754980044328,
      "learning_rate": 9.600578891082683e-06,
      "loss": 0.6631,
      "step": 958
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0502904566139173,
      "learning_rate": 9.599556173038211e-06,
      "loss": 0.5549,
      "step": 959
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0759478002381275,
      "learning_rate": 9.59853220194785e-06,
      "loss": 0.5845,
      "step": 960
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.308552938369847,
      "learning_rate": 9.597506978090558e-06,
      "loss": 0.5742,
      "step": 961
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1206630000574285,
      "learning_rate": 9.596480501745635e-06,
      "loss": 0.701,
      "step": 962
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0318104664560217,
      "learning_rate": 9.59545277319272e-06,
      "loss": 0.5293,
      "step": 963
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1529080040459074,
      "learning_rate": 9.594423792711796e-06,
      "loss": 0.7055,
      "step": 964
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1809894168146542,
      "learning_rate": 9.593393560583184e-06,
      "loss": 0.7125,
      "step": 965
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.180935189197949,
      "learning_rate": 9.592362077087548e-06,
      "loss": 0.7648,
      "step": 966
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9612061868710139,
      "learning_rate": 9.591329342505894e-06,
      "loss": 0.6317,
      "step": 967
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0727807330395043,
      "learning_rate": 9.590295357119565e-06,
      "loss": 0.6076,
      "step": 968
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0055986147244516,
      "learning_rate": 9.589260121210249e-06,
      "loss": 0.7117,
      "step": 969
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0033832134441991,
      "learning_rate": 9.588223635059973e-06,
      "loss": 0.5656,
      "step": 970
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0814408716538169,
      "learning_rate": 9.587185898951104e-06,
      "loss": 0.6167,
      "step": 971
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.074292401768141,
      "learning_rate": 9.58614691316635e-06,
      "loss": 0.6202,
      "step": 972
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9129956297990014,
      "learning_rate": 9.585106677988756e-06,
      "loss": 0.5198,
      "step": 973
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0685695120202705,
      "learning_rate": 9.584065193701717e-06,
      "loss": 0.5342,
      "step": 974
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.053562906867665,
      "learning_rate": 9.583022460588956e-06,
      "loss": 0.6798,
      "step": 975
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9782225534476784,
      "learning_rate": 9.581978478934547e-06,
      "loss": 0.5582,
      "step": 976
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1962585343056356,
      "learning_rate": 9.580933249022896e-06,
      "loss": 0.6083,
      "step": 977
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.05228409684923,
      "learning_rate": 9.579886771138754e-06,
      "loss": 0.6743,
      "step": 978
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1170726552830834,
      "learning_rate": 9.578839045567212e-06,
      "loss": 0.7001,
      "step": 979
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9112665215625259,
      "learning_rate": 9.577790072593696e-06,
      "loss": 0.5801,
      "step": 980
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0661693287778133,
      "learning_rate": 9.576739852503976e-06,
      "loss": 0.6034,
      "step": 981
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3235536582507976,
      "learning_rate": 9.575688385584162e-06,
      "loss": 0.6702,
      "step": 982
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0138726072680782,
      "learning_rate": 9.574635672120705e-06,
      "loss": 0.6531,
      "step": 983
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9893404239322373,
      "learning_rate": 9.573581712400386e-06,
      "loss": 0.5796,
      "step": 984
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1221063417929105,
      "learning_rate": 9.57252650671034e-06,
      "loss": 0.6461,
      "step": 985
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1184053021868707,
      "learning_rate": 9.57147005533803e-06,
      "loss": 0.6353,
      "step": 986
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1394894590771216,
      "learning_rate": 9.570412358571264e-06,
      "loss": 0.6956,
      "step": 987
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.196597060480913,
      "learning_rate": 9.569353416698187e-06,
      "loss": 0.7122,
      "step": 988
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0919614188821092,
      "learning_rate": 9.568293230007284e-06,
      "loss": 0.6799,
      "step": 989
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0789266655103467,
      "learning_rate": 9.56723179878738e-06,
      "loss": 0.6733,
      "step": 990
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.054446385079856,
      "learning_rate": 9.566169123327638e-06,
      "loss": 0.6982,
      "step": 991
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0575308473552738,
      "learning_rate": 9.565105203917559e-06,
      "loss": 0.6354,
      "step": 992
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0279614479300914,
      "learning_rate": 9.564040040846984e-06,
      "loss": 0.6113,
      "step": 993
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0083167847563335,
      "learning_rate": 9.562973634406096e-06,
      "loss": 0.5922,
      "step": 994
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1168000958897386,
      "learning_rate": 9.561905984885407e-06,
      "loss": 0.6101,
      "step": 995
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0410854175900213,
      "learning_rate": 9.560837092575781e-06,
      "loss": 0.5763,
      "step": 996
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.704859720733422,
      "learning_rate": 9.55976695776841e-06,
      "loss": 0.4741,
      "step": 997
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1653599862752817,
      "learning_rate": 9.558695580754828e-06,
      "loss": 0.7356,
      "step": 998
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1021709436890195,
      "learning_rate": 9.55762296182691e-06,
      "loss": 0.6582,
      "step": 999
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0842460978639399,
      "learning_rate": 9.556549101276865e-06,
      "loss": 0.6645,
      "step": 1000
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0987064323932816,
      "learning_rate": 9.555473999397242e-06,
      "loss": 0.6218,
      "step": 1001
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.139750477899937,
      "learning_rate": 9.55439765648093e-06,
      "loss": 0.7272,
      "step": 1002
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0699884904351982,
      "learning_rate": 9.553320072821154e-06,
      "loss": 0.7348,
      "step": 1003
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9952557290081818,
      "learning_rate": 9.552241248711478e-06,
      "loss": 0.5712,
      "step": 1004
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0647379138297899,
      "learning_rate": 9.551161184445801e-06,
      "loss": 0.6133,
      "step": 1005
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.017581941066395,
      "learning_rate": 9.550079880318363e-06,
      "loss": 0.6248,
      "step": 1006
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.728936553347363,
      "learning_rate": 9.548997336623744e-06,
      "loss": 0.4895,
      "step": 1007
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9948388468302773,
      "learning_rate": 9.547913553656856e-06,
      "loss": 0.6214,
      "step": 1008
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1437496215668703,
      "learning_rate": 9.54682853171295e-06,
      "loss": 0.6694,
      "step": 1009
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1952011541009888,
      "learning_rate": 9.545742271087622e-06,
      "loss": 0.728,
      "step": 1010
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.081529900865028,
      "learning_rate": 9.54465477207679e-06,
      "loss": 0.6773,
      "step": 1011
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.050731103556642,
      "learning_rate": 9.543566034976728e-06,
      "loss": 0.5817,
      "step": 1012
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0321591924352482,
      "learning_rate": 9.542476060084035e-06,
      "loss": 0.6412,
      "step": 1013
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0273157733773564,
      "learning_rate": 9.541384847695645e-06,
      "loss": 0.5903,
      "step": 1014
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0304904001691007,
      "learning_rate": 9.540292398108842e-06,
      "loss": 0.6062,
      "step": 1015
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.081837612606692,
      "learning_rate": 9.539198711621234e-06,
      "loss": 0.6252,
      "step": 1016
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1295474336579325,
      "learning_rate": 9.538103788530773e-06,
      "loss": 0.6425,
      "step": 1017
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0194068114574404,
      "learning_rate": 9.537007629135745e-06,
      "loss": 0.5517,
      "step": 1018
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0325828575722746,
      "learning_rate": 9.53591023373478e-06,
      "loss": 0.5903,
      "step": 1019
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0260765734962127,
      "learning_rate": 9.53481160262683e-06,
      "loss": 0.5698,
      "step": 1020
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0951006092215407,
      "learning_rate": 9.5337117361112e-06,
      "loss": 0.6442,
      "step": 1021
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1621859396641459,
      "learning_rate": 9.532610634487519e-06,
      "loss": 0.6824,
      "step": 1022
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2389190815244484,
      "learning_rate": 9.531508298055758e-06,
      "loss": 0.793,
      "step": 1023
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.087497083191494,
      "learning_rate": 9.530404727116225e-06,
      "loss": 0.6303,
      "step": 1024
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6777898595721662,
      "learning_rate": 9.529299921969563e-06,
      "loss": 0.5097,
      "step": 1025
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0915995902746736,
      "learning_rate": 9.528193882916753e-06,
      "loss": 0.6837,
      "step": 1026
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1085655973232171,
      "learning_rate": 9.52708661025911e-06,
      "loss": 0.6219,
      "step": 1027
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0452463858932728,
      "learning_rate": 9.525978104298282e-06,
      "loss": 0.6222,
      "step": 1028
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.052396718240974,
      "learning_rate": 9.524868365336259e-06,
      "loss": 0.7177,
      "step": 1029
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0489006976855053,
      "learning_rate": 9.523757393675365e-06,
      "loss": 0.5991,
      "step": 1030
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.994170797456951,
      "learning_rate": 9.522645189618258e-06,
      "loss": 0.5957,
      "step": 1031
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0648631326669196,
      "learning_rate": 9.521531753467935e-06,
      "loss": 0.5667,
      "step": 1032
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.024249069294412,
      "learning_rate": 9.520417085527725e-06,
      "loss": 0.6865,
      "step": 1033
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0323248495750048,
      "learning_rate": 9.519301186101297e-06,
      "loss": 0.5894,
      "step": 1034
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1872508562299484,
      "learning_rate": 9.518184055492648e-06,
      "loss": 0.7089,
      "step": 1035
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0752819051439941,
      "learning_rate": 9.517065694006118e-06,
      "loss": 0.666,
      "step": 1036
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1105561466253857,
      "learning_rate": 9.51594610194638e-06,
      "loss": 0.6737,
      "step": 1037
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8154925649063165,
      "learning_rate": 9.514825279618443e-06,
      "loss": 0.5242,
      "step": 1038
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1074637562604694,
      "learning_rate": 9.513703227327646e-06,
      "loss": 0.6536,
      "step": 1039
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0737546944299396,
      "learning_rate": 9.512579945379668e-06,
      "loss": 0.6429,
      "step": 1040
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1286206779146506,
      "learning_rate": 9.511455434080525e-06,
      "loss": 0.5987,
      "step": 1041
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1319256860563078,
      "learning_rate": 9.51032969373656e-06,
      "loss": 0.6941,
      "step": 1042
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9797067007811695,
      "learning_rate": 9.50920272465446e-06,
      "loss": 0.644,
      "step": 1043
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1730949121799528,
      "learning_rate": 9.508074527141243e-06,
      "loss": 0.6962,
      "step": 1044
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0604995218014863,
      "learning_rate": 9.506945101504256e-06,
      "loss": 0.6512,
      "step": 1045
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1151795995324953,
      "learning_rate": 9.50581444805119e-06,
      "loss": 0.7242,
      "step": 1046
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7311448785681954,
      "learning_rate": 9.504682567090064e-06,
      "loss": 0.5115,
      "step": 1047
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0365243238319133,
      "learning_rate": 9.503549458929236e-06,
      "loss": 0.6794,
      "step": 1048
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0044721735542828,
      "learning_rate": 9.502415123877393e-06,
      "loss": 0.5497,
      "step": 1049
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1242288466236354,
      "learning_rate": 9.501279562243561e-06,
      "loss": 0.6752,
      "step": 1050
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1331670185085372,
      "learning_rate": 9.500142774337097e-06,
      "loss": 0.7197,
      "step": 1051
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0103550192543755,
      "learning_rate": 9.499004760467694e-06,
      "loss": 0.603,
      "step": 1052
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0533246463330384,
      "learning_rate": 9.49786552094538e-06,
      "loss": 0.5605,
      "step": 1053
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0451492548509471,
      "learning_rate": 9.496725056080512e-06,
      "loss": 0.7136,
      "step": 1054
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9921862442775142,
      "learning_rate": 9.495583366183788e-06,
      "loss": 0.6157,
      "step": 1055
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1750595671905533,
      "learning_rate": 9.494440451566232e-06,
      "loss": 0.6459,
      "step": 1056
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9992534350081556,
      "learning_rate": 9.493296312539206e-06,
      "loss": 0.6489,
      "step": 1057
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0321907251505391,
      "learning_rate": 9.492150949414408e-06,
      "loss": 0.6748,
      "step": 1058
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.045928752077126,
      "learning_rate": 9.491004362503862e-06,
      "loss": 0.6338,
      "step": 1059
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1068064325353986,
      "learning_rate": 9.489856552119934e-06,
      "loss": 0.6598,
      "step": 1060
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0620415062035822,
      "learning_rate": 9.488707518575319e-06,
      "loss": 0.5837,
      "step": 1061
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0663443799558874,
      "learning_rate": 9.487557262183043e-06,
      "loss": 0.6156,
      "step": 1062
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0861266500955866,
      "learning_rate": 9.486405783256467e-06,
      "loss": 0.6903,
      "step": 1063
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0823357865648187,
      "learning_rate": 9.485253082109289e-06,
      "loss": 0.6214,
      "step": 1064
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0753546883658682,
      "learning_rate": 9.484099159055534e-06,
      "loss": 0.6628,
      "step": 1065
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0724229101600617,
      "learning_rate": 9.482944014409563e-06,
      "loss": 0.6598,
      "step": 1066
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0135985503263316,
      "learning_rate": 9.481787648486069e-06,
      "loss": 0.6221,
      "step": 1067
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.086275601865181,
      "learning_rate": 9.480630061600079e-06,
      "loss": 0.7074,
      "step": 1068
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0964530637608785,
      "learning_rate": 9.47947125406695e-06,
      "loss": 0.6518,
      "step": 1069
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0723795638785913,
      "learning_rate": 9.478311226202375e-06,
      "loss": 0.6972,
      "step": 1070
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0992302190855479,
      "learning_rate": 9.477149978322378e-06,
      "loss": 0.6634,
      "step": 1071
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1613175510587674,
      "learning_rate": 9.475987510743311e-06,
      "loss": 0.6912,
      "step": 1072
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.027713959737752,
      "learning_rate": 9.474823823781866e-06,
      "loss": 0.6543,
      "step": 1073
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.128922780945922,
      "learning_rate": 9.473658917755063e-06,
      "loss": 0.6255,
      "step": 1074
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0984362213611925,
      "learning_rate": 9.472492792980252e-06,
      "loss": 0.6751,
      "step": 1075
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0502364975895737,
      "learning_rate": 9.471325449775119e-06,
      "loss": 0.6352,
      "step": 1076
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0327652001917453,
      "learning_rate": 9.470156888457682e-06,
      "loss": 0.5715,
      "step": 1077
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0220506068336401,
      "learning_rate": 9.468987109346288e-06,
      "loss": 0.636,
      "step": 1078
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1005100810369575,
      "learning_rate": 9.467816112759616e-06,
      "loss": 0.6752,
      "step": 1079
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0560512286705024,
      "learning_rate": 9.46664389901668e-06,
      "loss": 0.565,
      "step": 1080
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1426094207887838,
      "learning_rate": 9.465470468436822e-06,
      "loss": 0.649,
      "step": 1081
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.025384478031677,
      "learning_rate": 9.464295821339715e-06,
      "loss": 0.6952,
      "step": 1082
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9452834167643365,
      "learning_rate": 9.46311995804537e-06,
      "loss": 0.5813,
      "step": 1083
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0134587988017134,
      "learning_rate": 9.46194287887412e-06,
      "loss": 0.6274,
      "step": 1084
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0463946367818822,
      "learning_rate": 9.460764584146635e-06,
      "loss": 0.644,
      "step": 1085
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0740947551913984,
      "learning_rate": 9.459585074183919e-06,
      "loss": 0.6753,
      "step": 1086
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1403711895037398,
      "learning_rate": 9.458404349307295e-06,
      "loss": 0.6609,
      "step": 1087
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7380717786674862,
      "learning_rate": 9.457222409838433e-06,
      "loss": 0.4881,
      "step": 1088
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1102071189666116,
      "learning_rate": 9.456039256099321e-06,
      "loss": 0.5692,
      "step": 1089
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.139399781337805,
      "learning_rate": 9.454854888412285e-06,
      "loss": 0.7134,
      "step": 1090
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0099436525935124,
      "learning_rate": 9.453669307099978e-06,
      "loss": 0.6956,
      "step": 1091
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0961834439674991,
      "learning_rate": 9.452482512485386e-06,
      "loss": 0.6881,
      "step": 1092
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1129848859481177,
      "learning_rate": 9.451294504891824e-06,
      "loss": 0.6388,
      "step": 1093
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0409418072682515,
      "learning_rate": 9.450105284642938e-06,
      "loss": 0.6821,
      "step": 1094
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6940216433613278,
      "learning_rate": 9.448914852062705e-06,
      "loss": 0.4935,
      "step": 1095
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.041109449996568,
      "learning_rate": 9.447723207475432e-06,
      "loss": 0.599,
      "step": 1096
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.071810706869906,
      "learning_rate": 9.446530351205754e-06,
      "loss": 0.6563,
      "step": 1097
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0791535216175594,
      "learning_rate": 9.445336283578639e-06,
      "loss": 0.6994,
      "step": 1098
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.131459703710808,
      "learning_rate": 9.444141004919385e-06,
      "loss": 0.6824,
      "step": 1099
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1109924808320397,
      "learning_rate": 9.442944515553616e-06,
      "loss": 0.6892,
      "step": 1100
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0613221388403484,
      "learning_rate": 9.441746815807292e-06,
      "loss": 0.6862,
      "step": 1101
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1481310697294973,
      "learning_rate": 9.440547906006697e-06,
      "loss": 0.7421,
      "step": 1102
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1421279577171106,
      "learning_rate": 9.439347786478448e-06,
      "loss": 0.6875,
      "step": 1103
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.182388471131474,
      "learning_rate": 9.438146457549491e-06,
      "loss": 0.7018,
      "step": 1104
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1853389093598405,
      "learning_rate": 9.436943919547101e-06,
      "loss": 0.6955,
      "step": 1105
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.14598600020169,
      "learning_rate": 9.435740172798881e-06,
      "loss": 0.6452,
      "step": 1106
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0069612755304904,
      "learning_rate": 9.434535217632768e-06,
      "loss": 0.6274,
      "step": 1107
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.142387996447214,
      "learning_rate": 9.433329054377023e-06,
      "loss": 0.6869,
      "step": 1108
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.117654650314097,
      "learning_rate": 9.432121683360235e-06,
      "loss": 0.736,
      "step": 1109
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1278281277060207,
      "learning_rate": 9.430913104911331e-06,
      "loss": 0.6747,
      "step": 1110
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1349248587271872,
      "learning_rate": 9.429703319359558e-06,
      "loss": 0.672,
      "step": 1111
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.024185517192485,
      "learning_rate": 9.428492327034496e-06,
      "loss": 0.6033,
      "step": 1112
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9511302016841211,
      "learning_rate": 9.427280128266049e-06,
      "loss": 0.6185,
      "step": 1113
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1228438446502222,
      "learning_rate": 9.42606672338446e-06,
      "loss": 0.7208,
      "step": 1114
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.036050911711722,
      "learning_rate": 9.424852112720289e-06,
      "loss": 0.5679,
      "step": 1115
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.129863645277397,
      "learning_rate": 9.42363629660443e-06,
      "loss": 0.6281,
      "step": 1116
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0226645253793512,
      "learning_rate": 9.422419275368107e-06,
      "loss": 0.5991,
      "step": 1117
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.069717462366646,
      "learning_rate": 9.421201049342867e-06,
      "loss": 0.6227,
      "step": 1118
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0827797842288633,
      "learning_rate": 9.41998161886059e-06,
      "loss": 0.6439,
      "step": 1119
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1153927857178165,
      "learning_rate": 9.418760984253485e-06,
      "loss": 0.6253,
      "step": 1120
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0532959580783336,
      "learning_rate": 9.417539145854082e-06,
      "loss": 0.6771,
      "step": 1121
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0562968714954466,
      "learning_rate": 9.416316103995246e-06,
      "loss": 0.6069,
      "step": 1122
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0457506568836676,
      "learning_rate": 9.415091859010165e-06,
      "loss": 0.5926,
      "step": 1123
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0366500652286177,
      "learning_rate": 9.413866411232361e-06,
      "loss": 0.6296,
      "step": 1124
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0985700431079428,
      "learning_rate": 9.41263976099568e-06,
      "loss": 0.6496,
      "step": 1125
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0169290242622118,
      "learning_rate": 9.411411908634289e-06,
      "loss": 0.6111,
      "step": 1126
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0410409589621543,
      "learning_rate": 9.410182854482693e-06,
      "loss": 0.6308,
      "step": 1127
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8426116974971252,
      "learning_rate": 9.40895259887572e-06,
      "loss": 0.5101,
      "step": 1128
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1097992827551972,
      "learning_rate": 9.407721142148528e-06,
      "loss": 0.6285,
      "step": 1129
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9762140060945351,
      "learning_rate": 9.406488484636595e-06,
      "loss": 0.6862,
      "step": 1130
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2300256645964416,
      "learning_rate": 9.405254626675736e-06,
      "loss": 0.7907,
      "step": 1131
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0385376128817734,
      "learning_rate": 9.404019568602081e-06,
      "loss": 0.6059,
      "step": 1132
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0396253594334501,
      "learning_rate": 9.402783310752102e-06,
      "loss": 0.6212,
      "step": 1133
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1077407311219116,
      "learning_rate": 9.401545853462584e-06,
      "loss": 0.6282,
      "step": 1134
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0331647075588999,
      "learning_rate": 9.400307197070646e-06,
      "loss": 0.6491,
      "step": 1135
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0598044177470136,
      "learning_rate": 9.399067341913732e-06,
      "loss": 0.5906,
      "step": 1136
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.080395347863945,
      "learning_rate": 9.397826288329615e-06,
      "loss": 0.6683,
      "step": 1137
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7950180638145538,
      "learning_rate": 9.396584036656388e-06,
      "loss": 0.4966,
      "step": 1138
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2139499886683336,
      "learning_rate": 9.395340587232478e-06,
      "loss": 0.7255,
      "step": 1139
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0494093727175315,
      "learning_rate": 9.394095940396632e-06,
      "loss": 0.6709,
      "step": 1140
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.042252114979571,
      "learning_rate": 9.39285009648793e-06,
      "loss": 0.5401,
      "step": 1141
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2010949383594216,
      "learning_rate": 9.391603055845771e-06,
      "loss": 0.7117,
      "step": 1142
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9972246359789446,
      "learning_rate": 9.390354818809886e-06,
      "loss": 0.7257,
      "step": 1143
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9553562407962161,
      "learning_rate": 9.389105385720325e-06,
      "loss": 0.5333,
      "step": 1144
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0298252606824072,
      "learning_rate": 9.387854756917472e-06,
      "loss": 0.5917,
      "step": 1145
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0586860634228825,
      "learning_rate": 9.386602932742032e-06,
      "loss": 0.6591,
      "step": 1146
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.119388948541729,
      "learning_rate": 9.385349913535034e-06,
      "loss": 0.6465,
      "step": 1147
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0953107410902572,
      "learning_rate": 9.384095699637839e-06,
      "loss": 0.6314,
      "step": 1148
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6623052598265775,
      "learning_rate": 9.382840291392124e-06,
      "loss": 0.4491,
      "step": 1149
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0634631297463455,
      "learning_rate": 9.3815836891399e-06,
      "loss": 0.6581,
      "step": 1150
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0812505534269203,
      "learning_rate": 9.380325893223503e-06,
      "loss": 0.7094,
      "step": 1151
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6011310255268045,
      "learning_rate": 9.379066903985588e-06,
      "loss": 0.4815,
      "step": 1152
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0593325736721697,
      "learning_rate": 9.377806721769138e-06,
      "loss": 0.616,
      "step": 1153
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0850398478736807,
      "learning_rate": 9.37654534691746e-06,
      "loss": 0.6381,
      "step": 1154
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0907408361667887,
      "learning_rate": 9.37528277977419e-06,
      "loss": 0.5573,
      "step": 1155
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0922646132955156,
      "learning_rate": 9.374019020683287e-06,
      "loss": 0.6532,
      "step": 1156
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1344615053723808,
      "learning_rate": 9.372754069989032e-06,
      "loss": 0.6822,
      "step": 1157
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0555579156587251,
      "learning_rate": 9.371487928036032e-06,
      "loss": 0.6476,
      "step": 1158
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0259261792104715,
      "learning_rate": 9.37022059516922e-06,
      "loss": 0.6442,
      "step": 1159
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8984732101302717,
      "learning_rate": 9.368952071733851e-06,
      "loss": 0.604,
      "step": 1160
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1010232792375327,
      "learning_rate": 9.367682358075509e-06,
      "loss": 0.7074,
      "step": 1161
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0814698431766612,
      "learning_rate": 9.366411454540095e-06,
      "loss": 0.6852,
      "step": 1162
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9417378999784696,
      "learning_rate": 9.36513936147384e-06,
      "loss": 0.6336,
      "step": 1163
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9926096428894494,
      "learning_rate": 9.363866079223299e-06,
      "loss": 0.6062,
      "step": 1164
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2087772481159154,
      "learning_rate": 9.362591608135346e-06,
      "loss": 0.7518,
      "step": 1165
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4209728972648705,
      "learning_rate": 9.361315948557182e-06,
      "loss": 0.6574,
      "step": 1166
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0622213307593036,
      "learning_rate": 9.360039100836335e-06,
      "loss": 0.6411,
      "step": 1167
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1067636537158854,
      "learning_rate": 9.358761065320653e-06,
      "loss": 0.6431,
      "step": 1168
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1398463939948387,
      "learning_rate": 9.357481842358302e-06,
      "loss": 0.7377,
      "step": 1169
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9815774673164988,
      "learning_rate": 9.356201432297788e-06,
      "loss": 0.6614,
      "step": 1170
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8941479688054781,
      "learning_rate": 9.354919835487922e-06,
      "loss": 0.4741,
      "step": 1171
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.044632672835645,
      "learning_rate": 9.353637052277848e-06,
      "loss": 0.656,
      "step": 1172
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.052307133992648,
      "learning_rate": 9.352353083017034e-06,
      "loss": 0.6429,
      "step": 1173
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2004441664637213,
      "learning_rate": 9.351067928055265e-06,
      "loss": 0.7377,
      "step": 1174
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.127592989401024,
      "learning_rate": 9.349781587742655e-06,
      "loss": 0.6029,
      "step": 1175
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1248373783626897,
      "learning_rate": 9.348494062429639e-06,
      "loss": 0.6627,
      "step": 1176
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0168885892101591,
      "learning_rate": 9.347205352466972e-06,
      "loss": 0.5711,
      "step": 1177
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1507393033298814,
      "learning_rate": 9.345915458205736e-06,
      "loss": 0.6199,
      "step": 1178
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0560258294320832,
      "learning_rate": 9.344624379997335e-06,
      "loss": 0.6065,
      "step": 1179
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1027335552100108,
      "learning_rate": 9.343332118193492e-06,
      "loss": 0.66,
      "step": 1180
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0216012701447972,
      "learning_rate": 9.342038673146253e-06,
      "loss": 0.6253,
      "step": 1181
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0188424914554506,
      "learning_rate": 9.340744045207992e-06,
      "loss": 0.6847,
      "step": 1182
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.044803461282638,
      "learning_rate": 9.3394482347314e-06,
      "loss": 0.5991,
      "step": 1183
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1602146469722492,
      "learning_rate": 9.338151242069491e-06,
      "loss": 0.6096,
      "step": 1184
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0128035216136144,
      "learning_rate": 9.336853067575603e-06,
      "loss": 0.6396,
      "step": 1185
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1933892623226698,
      "learning_rate": 9.335553711603393e-06,
      "loss": 0.6524,
      "step": 1186
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0013575123851317,
      "learning_rate": 9.33425317450684e-06,
      "loss": 0.573,
      "step": 1187
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1586743928279306,
      "learning_rate": 9.332951456640249e-06,
      "loss": 0.6981,
      "step": 1188
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9976559520022303,
      "learning_rate": 9.331648558358244e-06,
      "loss": 0.6135,
      "step": 1189
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0464451762886755,
      "learning_rate": 9.330344480015767e-06,
      "loss": 0.6205,
      "step": 1190
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1256798629405222,
      "learning_rate": 9.329039221968087e-06,
      "loss": 0.6796,
      "step": 1191
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7693710804160085,
      "learning_rate": 9.327732784570794e-06,
      "loss": 0.5184,
      "step": 1192
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9727490541022548,
      "learning_rate": 9.326425168179795e-06,
      "loss": 0.6014,
      "step": 1193
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0697413606763762,
      "learning_rate": 9.325116373151322e-06,
      "loss": 0.673,
      "step": 1194
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0501332540088344,
      "learning_rate": 9.323806399841927e-06,
      "loss": 0.6269,
      "step": 1195
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0820759551201784,
      "learning_rate": 9.32249524860848e-06,
      "loss": 0.6542,
      "step": 1196
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0323939610653803,
      "learning_rate": 9.321182919808179e-06,
      "loss": 0.6186,
      "step": 1197
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.156914332748338,
      "learning_rate": 9.319869413798535e-06,
      "loss": 0.6454,
      "step": 1198
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.16910904059714,
      "learning_rate": 9.318554730937385e-06,
      "loss": 0.6349,
      "step": 1199
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6277148539258679,
      "learning_rate": 9.317238871582886e-06,
      "loss": 0.4597,
      "step": 1200
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9519685426947867,
      "learning_rate": 9.31592183609351e-06,
      "loss": 0.566,
      "step": 1201
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1131236362461228,
      "learning_rate": 9.31460362482806e-06,
      "loss": 0.5806,
      "step": 1202
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9914960013657294,
      "learning_rate": 9.313284238145648e-06,
      "loss": 0.5832,
      "step": 1203
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1394984204933547,
      "learning_rate": 9.311963676405716e-06,
      "loss": 0.6665,
      "step": 1204
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.141113550590736,
      "learning_rate": 9.310641939968016e-06,
      "loss": 0.6722,
      "step": 1205
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0899706323857257,
      "learning_rate": 9.309319029192627e-06,
      "loss": 0.6116,
      "step": 1206
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1064720127728438,
      "learning_rate": 9.307994944439949e-06,
      "loss": 0.6702,
      "step": 1207
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1719854279700321,
      "learning_rate": 9.306669686070697e-06,
      "loss": 0.6232,
      "step": 1208
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0968729728740734,
      "learning_rate": 9.30534325444591e-06,
      "loss": 0.7028,
      "step": 1209
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6894523783298816,
      "learning_rate": 9.304015649926941e-06,
      "loss": 0.4883,
      "step": 1210
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.086859867690113,
      "learning_rate": 9.30268687287547e-06,
      "loss": 0.683,
      "step": 1211
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0680938255003465,
      "learning_rate": 9.30135692365349e-06,
      "loss": 0.6597,
      "step": 1212
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.012338176949859,
      "learning_rate": 9.300025802623316e-06,
      "loss": 0.6656,
      "step": 1213
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0306906641337574,
      "learning_rate": 9.29869351014758e-06,
      "loss": 0.5925,
      "step": 1214
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.719161842505433,
      "learning_rate": 9.29736004658924e-06,
      "loss": 0.4964,
      "step": 1215
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1849584994273956,
      "learning_rate": 9.296025412311567e-06,
      "loss": 0.7035,
      "step": 1216
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0587016842136787,
      "learning_rate": 9.294689607678151e-06,
      "loss": 0.656,
      "step": 1217
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1043385773877985,
      "learning_rate": 9.293352633052901e-06,
      "loss": 0.5673,
      "step": 1218
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0161825194031135,
      "learning_rate": 9.292014488800046e-06,
      "loss": 0.6026,
      "step": 1219
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6190578711695649,
      "learning_rate": 9.290675175284135e-06,
      "loss": 0.4907,
      "step": 1220
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.125714700914956,
      "learning_rate": 9.289334692870033e-06,
      "loss": 0.7165,
      "step": 1221
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2875941445325274,
      "learning_rate": 9.287993041922924e-06,
      "loss": 0.7112,
      "step": 1222
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1078222309915893,
      "learning_rate": 9.286650222808314e-06,
      "loss": 0.7601,
      "step": 1223
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.999944304866867,
      "learning_rate": 9.285306235892017e-06,
      "loss": 0.5661,
      "step": 1224
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9952463958134803,
      "learning_rate": 9.283961081540178e-06,
      "loss": 0.6069,
      "step": 1225
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1478291170578638,
      "learning_rate": 9.282614760119252e-06,
      "loss": 0.6923,
      "step": 1226
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.024788043357834,
      "learning_rate": 9.281267271996015e-06,
      "loss": 0.7213,
      "step": 1227
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.048442578485777,
      "learning_rate": 9.27991861753756e-06,
      "loss": 0.6394,
      "step": 1228
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.022485456447724,
      "learning_rate": 9.278568797111294e-06,
      "loss": 0.6112,
      "step": 1229
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0888123572603883,
      "learning_rate": 9.277217811084952e-06,
      "loss": 0.5784,
      "step": 1230
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0939112564306201,
      "learning_rate": 9.275865659826573e-06,
      "loss": 0.6281,
      "step": 1231
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.915737101049708,
      "learning_rate": 9.274512343704525e-06,
      "loss": 0.598,
      "step": 1232
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9767689527332065,
      "learning_rate": 9.273157863087486e-06,
      "loss": 0.5339,
      "step": 1233
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0785580622773865,
      "learning_rate": 9.271802218344455e-06,
      "loss": 0.6258,
      "step": 1234
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9939454227184851,
      "learning_rate": 9.270445409844749e-06,
      "loss": 0.6299,
      "step": 1235
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0079941254433253,
      "learning_rate": 9.269087437957996e-06,
      "loss": 0.6046,
      "step": 1236
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1072551085362412,
      "learning_rate": 9.267728303054146e-06,
      "loss": 0.6721,
      "step": 1237
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1053569057984955,
      "learning_rate": 9.266368005503465e-06,
      "loss": 0.6123,
      "step": 1238
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0139210005755706,
      "learning_rate": 9.265006545676538e-06,
      "loss": 0.6156,
      "step": 1239
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0743280202582823,
      "learning_rate": 9.263643923944262e-06,
      "loss": 0.6568,
      "step": 1240
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.198856314599014,
      "learning_rate": 9.262280140677852e-06,
      "loss": 0.7356,
      "step": 1241
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9865904398756403,
      "learning_rate": 9.260915196248842e-06,
      "loss": 0.5548,
      "step": 1242
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6984235997615423,
      "learning_rate": 9.259549091029082e-06,
      "loss": 0.4859,
      "step": 1243
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0443175137011091,
      "learning_rate": 9.258181825390732e-06,
      "loss": 0.6019,
      "step": 1244
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1430195282177522,
      "learning_rate": 9.256813399706276e-06,
      "loss": 0.6485,
      "step": 1245
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0702155304377488,
      "learning_rate": 9.25544381434851e-06,
      "loss": 0.6268,
      "step": 1246
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1258650641586976,
      "learning_rate": 9.254073069690545e-06,
      "loss": 0.6028,
      "step": 1247
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9979885635295688,
      "learning_rate": 9.252701166105815e-06,
      "loss": 0.5231,
      "step": 1248
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1225800216660617,
      "learning_rate": 9.251328103968059e-06,
      "loss": 0.695,
      "step": 1249
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0150053148432505,
      "learning_rate": 9.24995388365134e-06,
      "loss": 0.5764,
      "step": 1250
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0567664311953584,
      "learning_rate": 9.24857850553003e-06,
      "loss": 0.6376,
      "step": 1251
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0209686185559421,
      "learning_rate": 9.247201969978825e-06,
      "loss": 0.6542,
      "step": 1252
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0707225531437685,
      "learning_rate": 9.245824277372729e-06,
      "loss": 0.6318,
      "step": 1253
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.220432558688987,
      "learning_rate": 9.24444542808706e-06,
      "loss": 0.6879,
      "step": 1254
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1569060907214943,
      "learning_rate": 9.24306542249746e-06,
      "loss": 0.674,
      "step": 1255
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.069302019008103,
      "learning_rate": 9.241684260979877e-06,
      "loss": 0.6861,
      "step": 1256
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0313837833709343,
      "learning_rate": 9.240301943910578e-06,
      "loss": 0.6203,
      "step": 1257
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1356470117218578,
      "learning_rate": 9.238918471666148e-06,
      "loss": 0.6329,
      "step": 1258
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0759068997696632,
      "learning_rate": 9.237533844623477e-06,
      "loss": 0.583,
      "step": 1259
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.072656661028882,
      "learning_rate": 9.236148063159778e-06,
      "loss": 0.6853,
      "step": 1260
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0307163773942474,
      "learning_rate": 9.234761127652578e-06,
      "loss": 0.5126,
      "step": 1261
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0948171100839685,
      "learning_rate": 9.233373038479716e-06,
      "loss": 0.7146,
      "step": 1262
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1143406928943824,
      "learning_rate": 9.231983796019342e-06,
      "loss": 0.6514,
      "step": 1263
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0767395438149383,
      "learning_rate": 9.230593400649928e-06,
      "loss": 0.6284,
      "step": 1264
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0141466953354,
      "learning_rate": 9.229201852750254e-06,
      "loss": 0.6817,
      "step": 1265
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0049429474092488,
      "learning_rate": 9.227809152699418e-06,
      "loss": 0.6424,
      "step": 1266
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.101760671320198,
      "learning_rate": 9.226415300876828e-06,
      "loss": 0.6909,
      "step": 1267
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0700946452574387,
      "learning_rate": 9.225020297662208e-06,
      "loss": 0.6795,
      "step": 1268
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7404293522852861,
      "learning_rate": 9.223624143435595e-06,
      "loss": 0.4872,
      "step": 1269
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0312883836740334,
      "learning_rate": 9.222226838577342e-06,
      "loss": 0.5905,
      "step": 1270
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2077457820550166,
      "learning_rate": 9.22082838346811e-06,
      "loss": 0.6995,
      "step": 1271
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0529698040361815,
      "learning_rate": 9.21942877848888e-06,
      "loss": 0.5983,
      "step": 1272
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0022401754811046,
      "learning_rate": 9.21802802402094e-06,
      "loss": 0.6284,
      "step": 1273
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9833586086698595,
      "learning_rate": 9.216626120445897e-06,
      "loss": 0.5923,
      "step": 1274
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0530649725869778,
      "learning_rate": 9.215223068145666e-06,
      "loss": 0.677,
      "step": 1275
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0464753437976047,
      "learning_rate": 9.213818867502478e-06,
      "loss": 0.6045,
      "step": 1276
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9926183539978418,
      "learning_rate": 9.212413518898878e-06,
      "loss": 0.5699,
      "step": 1277
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0116267351415305,
      "learning_rate": 9.211007022717716e-06,
      "loss": 0.5614,
      "step": 1278
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.076793931414792,
      "learning_rate": 9.209599379342165e-06,
      "loss": 0.6384,
      "step": 1279
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1052948235191529,
      "learning_rate": 9.208190589155705e-06,
      "loss": 0.6909,
      "step": 1280
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0363158883956463,
      "learning_rate": 9.206780652542131e-06,
      "loss": 0.6198,
      "step": 1281
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0293878264882128,
      "learning_rate": 9.205369569885544e-06,
      "loss": 0.6484,
      "step": 1282
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.016716085100523,
      "learning_rate": 9.203957341570367e-06,
      "loss": 0.5898,
      "step": 1283
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6948168492948111,
      "learning_rate": 9.202543967981325e-06,
      "loss": 0.499,
      "step": 1284
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2041013176322988,
      "learning_rate": 9.201129449503463e-06,
      "loss": 0.6232,
      "step": 1285
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.052663557529718,
      "learning_rate": 9.199713786522135e-06,
      "loss": 0.6386,
      "step": 1286
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1784149455736117,
      "learning_rate": 9.198296979423006e-06,
      "loss": 0.6909,
      "step": 1287
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9868320250156666,
      "learning_rate": 9.196879028592052e-06,
      "loss": 0.5193,
      "step": 1288
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0784236032227814,
      "learning_rate": 9.195459934415563e-06,
      "loss": 0.6336,
      "step": 1289
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0607256825807188,
      "learning_rate": 9.19403969728014e-06,
      "loss": 0.5959,
      "step": 1290
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0201409310772067,
      "learning_rate": 9.192618317572693e-06,
      "loss": 0.5715,
      "step": 1291
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0967435752614172,
      "learning_rate": 9.191195795680447e-06,
      "loss": 0.5446,
      "step": 1292
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0954440188196075,
      "learning_rate": 9.189772131990935e-06,
      "loss": 0.5848,
      "step": 1293
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0609636422321207,
      "learning_rate": 9.188347326892002e-06,
      "loss": 0.635,
      "step": 1294
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0161170852666188,
      "learning_rate": 9.186921380771806e-06,
      "loss": 0.6672,
      "step": 1295
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.000184592347299,
      "learning_rate": 9.185494294018811e-06,
      "loss": 0.5943,
      "step": 1296
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9876484186694694,
      "learning_rate": 9.184066067021798e-06,
      "loss": 0.6028,
      "step": 1297
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0411622139889567,
      "learning_rate": 9.182636700169854e-06,
      "loss": 0.6229,
      "step": 1298
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6684721939962599,
      "learning_rate": 9.181206193852376e-06,
      "loss": 0.448,
      "step": 1299
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0460170258968344,
      "learning_rate": 9.179774548459077e-06,
      "loss": 0.6166,
      "step": 1300
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0606393057387997,
      "learning_rate": 9.178341764379974e-06,
      "loss": 0.6426,
      "step": 1301
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.156393292749792,
      "learning_rate": 9.176907842005398e-06,
      "loss": 0.6889,
      "step": 1302
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.027990199636308,
      "learning_rate": 9.17547278172599e-06,
      "loss": 0.5626,
      "step": 1303
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0138277480040518,
      "learning_rate": 9.174036583932695e-06,
      "loss": 0.6696,
      "step": 1304
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6525229324631442,
      "learning_rate": 9.17259924901678e-06,
      "loss": 0.5144,
      "step": 1305
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1085156960711897,
      "learning_rate": 9.17116077736981e-06,
      "loss": 0.7345,
      "step": 1306
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0816571505859058,
      "learning_rate": 9.169721169383666e-06,
      "loss": 0.6738,
      "step": 1307
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9889485788779966,
      "learning_rate": 9.168280425450535e-06,
      "loss": 0.596,
      "step": 1308
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0800951920606323,
      "learning_rate": 9.166838545962917e-06,
      "loss": 0.6372,
      "step": 1309
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0834988246180475,
      "learning_rate": 9.165395531313622e-06,
      "loss": 0.6318,
      "step": 1310
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.03066141556626,
      "learning_rate": 9.163951381895761e-06,
      "loss": 0.6419,
      "step": 1311
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0876845816630245,
      "learning_rate": 9.162506098102765e-06,
      "loss": 0.6852,
      "step": 1312
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.058458063999532,
      "learning_rate": 9.161059680328368e-06,
      "loss": 0.5862,
      "step": 1313
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0019865134231243,
      "learning_rate": 9.159612128966615e-06,
      "loss": 0.5792,
      "step": 1314
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0547634676928261,
      "learning_rate": 9.158163444411857e-06,
      "loss": 0.6837,
      "step": 1315
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0883117640684918,
      "learning_rate": 9.156713627058754e-06,
      "loss": 0.605,
      "step": 1316
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0185022650609008,
      "learning_rate": 9.155262677302281e-06,
      "loss": 0.6413,
      "step": 1317
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0170124100150486,
      "learning_rate": 9.153810595537713e-06,
      "loss": 0.5307,
      "step": 1318
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1311136669530881,
      "learning_rate": 9.152357382160639e-06,
      "loss": 0.6308,
      "step": 1319
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6541292516969023,
      "learning_rate": 9.150903037566954e-06,
      "loss": 0.4329,
      "step": 1320
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1487889041829114,
      "learning_rate": 9.14944756215286e-06,
      "loss": 0.666,
      "step": 1321
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.024128203359301,
      "learning_rate": 9.14799095631487e-06,
      "loss": 0.6238,
      "step": 1322
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9882189189352942,
      "learning_rate": 9.146533220449804e-06,
      "loss": 0.5338,
      "step": 1323
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0898690130374142,
      "learning_rate": 9.14507435495479e-06,
      "loss": 0.6479,
      "step": 1324
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0326716813996526,
      "learning_rate": 9.143614360227261e-06,
      "loss": 0.6237,
      "step": 1325
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0577939091282096,
      "learning_rate": 9.142153236664961e-06,
      "loss": 0.6411,
      "step": 1326
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9604742133786004,
      "learning_rate": 9.14069098466594e-06,
      "loss": 0.5531,
      "step": 1327
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7125360178080434,
      "learning_rate": 9.139227604628556e-06,
      "loss": 0.5467,
      "step": 1328
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0423687649540945,
      "learning_rate": 9.137763096951474e-06,
      "loss": 0.6056,
      "step": 1329
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0834471726699733,
      "learning_rate": 9.136297462033667e-06,
      "loss": 0.7023,
      "step": 1330
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1300533974132025,
      "learning_rate": 9.134830700274413e-06,
      "loss": 0.6385,
      "step": 1331
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0765766405327657,
      "learning_rate": 9.133362812073298e-06,
      "loss": 0.596,
      "step": 1332
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0986521761479784,
      "learning_rate": 9.131893797830219e-06,
      "loss": 0.6769,
      "step": 1333
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1294104115009782,
      "learning_rate": 9.130423657945369e-06,
      "loss": 0.5503,
      "step": 1334
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0150987693400615,
      "learning_rate": 9.128952392819262e-06,
      "loss": 0.6175,
      "step": 1335
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1455229913482834,
      "learning_rate": 9.127480002852706e-06,
      "loss": 0.6709,
      "step": 1336
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.113249858812364,
      "learning_rate": 9.126006488446824e-06,
      "loss": 0.673,
      "step": 1337
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.081055884443081,
      "learning_rate": 9.12453185000304e-06,
      "loss": 0.6924,
      "step": 1338
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1666976481591584,
      "learning_rate": 9.123056087923088e-06,
      "loss": 0.7032,
      "step": 1339
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0228319307836449,
      "learning_rate": 9.121579202609004e-06,
      "loss": 0.6525,
      "step": 1340
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6469079481282709,
      "learning_rate": 9.12010119446313e-06,
      "loss": 0.485,
      "step": 1341
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7042101347952994,
      "learning_rate": 9.118622063888124e-06,
      "loss": 0.5174,
      "step": 1342
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0981131784492781,
      "learning_rate": 9.117141811286935e-06,
      "loss": 0.6659,
      "step": 1343
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1283918435241405,
      "learning_rate": 9.115660437062828e-06,
      "loss": 0.6193,
      "step": 1344
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0325628029589136,
      "learning_rate": 9.114177941619369e-06,
      "loss": 0.6506,
      "step": 1345
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1704250599488917,
      "learning_rate": 9.11269432536043e-06,
      "loss": 0.635,
      "step": 1346
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0487258624444653,
      "learning_rate": 9.11120958869019e-06,
      "loss": 0.6473,
      "step": 1347
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6630797570640432,
      "learning_rate": 9.109723732013132e-06,
      "loss": 0.492,
      "step": 1348
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0631836038739766,
      "learning_rate": 9.108236755734045e-06,
      "loss": 0.6399,
      "step": 1349
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0416000412977222,
      "learning_rate": 9.10674866025802e-06,
      "loss": 0.6401,
      "step": 1350
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0476568157916513,
      "learning_rate": 9.105259445990457e-06,
      "loss": 0.7022,
      "step": 1351
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.993864636778809,
      "learning_rate": 9.10376911333706e-06,
      "loss": 0.5898,
      "step": 1352
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0413112967865412,
      "learning_rate": 9.102277662703834e-06,
      "loss": 0.6153,
      "step": 1353
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.010006342010514,
      "learning_rate": 9.100785094497093e-06,
      "loss": 0.6296,
      "step": 1354
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0096545116284337,
      "learning_rate": 9.099291409123454e-06,
      "loss": 0.7163,
      "step": 1355
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0439025396194834,
      "learning_rate": 9.097796606989838e-06,
      "loss": 0.6266,
      "step": 1356
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0448381189778546,
      "learning_rate": 9.09630068850347e-06,
      "loss": 0.6362,
      "step": 1357
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.08568752755913,
      "learning_rate": 9.094803654071877e-06,
      "loss": 0.6969,
      "step": 1358
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.049579526620219,
      "learning_rate": 9.093305504102897e-06,
      "loss": 0.5182,
      "step": 1359
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0359177107256712,
      "learning_rate": 9.091806239004664e-06,
      "loss": 0.684,
      "step": 1360
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6931210113715948,
      "learning_rate": 9.090305859185619e-06,
      "loss": 0.4926,
      "step": 1361
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9921018624879908,
      "learning_rate": 9.088804365054511e-06,
      "loss": 0.6219,
      "step": 1362
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6239639129793633,
      "learning_rate": 9.087301757020384e-06,
      "loss": 0.6003,
      "step": 1363
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9825841550040483,
      "learning_rate": 9.08579803549259e-06,
      "loss": 0.5938,
      "step": 1364
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0680435943703834,
      "learning_rate": 9.084293200880787e-06,
      "loss": 0.6651,
      "step": 1365
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9932673877387073,
      "learning_rate": 9.082787253594932e-06,
      "loss": 0.6031,
      "step": 1366
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9420629925339018,
      "learning_rate": 9.081280194045286e-06,
      "loss": 0.5354,
      "step": 1367
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0573858054704095,
      "learning_rate": 9.079772022642413e-06,
      "loss": 0.6701,
      "step": 1368
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0444577130293116,
      "learning_rate": 9.078262739797184e-06,
      "loss": 0.6534,
      "step": 1369
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0065191470541899,
      "learning_rate": 9.076752345920764e-06,
      "loss": 0.6137,
      "step": 1370
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0129837858779838,
      "learning_rate": 9.075240841424629e-06,
      "loss": 0.5639,
      "step": 1371
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1543883971204312,
      "learning_rate": 9.073728226720555e-06,
      "loss": 0.6625,
      "step": 1372
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0972377559857236,
      "learning_rate": 9.07221450222062e-06,
      "loss": 0.7125,
      "step": 1373
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.956946091256148,
      "learning_rate": 9.070699668337202e-06,
      "loss": 0.6118,
      "step": 1374
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0322466730180362,
      "learning_rate": 9.069183725482984e-06,
      "loss": 0.5909,
      "step": 1375
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0817020230398517,
      "learning_rate": 9.067666674070952e-06,
      "loss": 0.6058,
      "step": 1376
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9129587604116987,
      "learning_rate": 9.066148514514395e-06,
      "loss": 0.603,
      "step": 1377
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0166212285215896,
      "learning_rate": 9.064629247226895e-06,
      "loss": 0.6555,
      "step": 1378
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.054780511818078,
      "learning_rate": 9.063108872622348e-06,
      "loss": 0.644,
      "step": 1379
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1700083035659647,
      "learning_rate": 9.061587391114942e-06,
      "loss": 0.7466,
      "step": 1380
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0505106069264116,
      "learning_rate": 9.060064803119173e-06,
      "loss": 0.72,
      "step": 1381
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0551900581640228,
      "learning_rate": 9.058541109049836e-06,
      "loss": 0.6341,
      "step": 1382
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1194935385821996,
      "learning_rate": 9.057016309322026e-06,
      "loss": 0.817,
      "step": 1383
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0668787387305103,
      "learning_rate": 9.055490404351141e-06,
      "loss": 0.6579,
      "step": 1384
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0637976882370652,
      "learning_rate": 9.05396339455288e-06,
      "loss": 0.6477,
      "step": 1385
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.161854519320283,
      "learning_rate": 9.05243528034324e-06,
      "loss": 0.6956,
      "step": 1386
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9453457969355008,
      "learning_rate": 9.050906062138527e-06,
      "loss": 0.693,
      "step": 1387
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0505183870051222,
      "learning_rate": 9.049375740355338e-06,
      "loss": 0.7126,
      "step": 1388
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0856608096892482,
      "learning_rate": 9.047844315410574e-06,
      "loss": 0.6582,
      "step": 1389
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.093157385772696,
      "learning_rate": 9.04631178772144e-06,
      "loss": 0.6326,
      "step": 1390
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0677806164608241,
      "learning_rate": 9.044778157705437e-06,
      "loss": 0.6593,
      "step": 1391
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0416462731597982,
      "learning_rate": 9.04324342578037e-06,
      "loss": 0.6476,
      "step": 1392
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0405003588410895,
      "learning_rate": 9.04170759236434e-06,
      "loss": 0.636,
      "step": 1393
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.109274215901161,
      "learning_rate": 9.040170657875753e-06,
      "loss": 0.5808,
      "step": 1394
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1082056816133068,
      "learning_rate": 9.038632622733311e-06,
      "loss": 0.63,
      "step": 1395
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1026516262977042,
      "learning_rate": 9.037093487356016e-06,
      "loss": 0.7177,
      "step": 1396
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0644638696196977,
      "learning_rate": 9.035553252163172e-06,
      "loss": 0.6869,
      "step": 1397
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9806611002038922,
      "learning_rate": 9.034011917574382e-06,
      "loss": 0.6167,
      "step": 1398
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0878974003679027,
      "learning_rate": 9.032469484009548e-06,
      "loss": 0.6647,
      "step": 1399
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0801494168326196,
      "learning_rate": 9.030925951888869e-06,
      "loss": 0.6665,
      "step": 1400
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0497997377947759,
      "learning_rate": 9.029381321632849e-06,
      "loss": 0.6555,
      "step": 1401
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0588087676173972,
      "learning_rate": 9.027835593662285e-06,
      "loss": 0.6029,
      "step": 1402
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0365680610901784,
      "learning_rate": 9.026288768398278e-06,
      "loss": 0.6319,
      "step": 1403
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8906181517900568,
      "learning_rate": 9.024740846262225e-06,
      "loss": 0.5554,
      "step": 1404
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0639166313317698,
      "learning_rate": 9.023191827675824e-06,
      "loss": 0.6312,
      "step": 1405
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0778823460216507,
      "learning_rate": 9.021641713061069e-06,
      "loss": 0.6586,
      "step": 1406
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0422156539703293,
      "learning_rate": 9.020090502840253e-06,
      "loss": 0.6578,
      "step": 1407
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1229123110341979,
      "learning_rate": 9.01853819743597e-06,
      "loss": 0.6788,
      "step": 1408
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2101381260602155,
      "learning_rate": 9.016984797271112e-06,
      "loss": 0.6664,
      "step": 1409
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0566900395722054,
      "learning_rate": 9.015430302768865e-06,
      "loss": 0.5809,
      "step": 1410
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1469800123289702,
      "learning_rate": 9.013874714352716e-06,
      "loss": 0.7116,
      "step": 1411
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9971819448767034,
      "learning_rate": 9.012318032446454e-06,
      "loss": 0.6547,
      "step": 1412
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0073186291041025,
      "learning_rate": 9.010760257474158e-06,
      "loss": 0.6181,
      "step": 1413
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0571180880945343,
      "learning_rate": 9.009201389860212e-06,
      "loss": 0.6798,
      "step": 1414
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.088820057183002,
      "learning_rate": 9.007641430029292e-06,
      "loss": 0.6245,
      "step": 1415
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0307297331627165,
      "learning_rate": 9.006080378406377e-06,
      "loss": 0.6237,
      "step": 1416
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.106008756404052,
      "learning_rate": 9.004518235416737e-06,
      "loss": 0.7473,
      "step": 1417
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0761915208260295,
      "learning_rate": 9.002955001485945e-06,
      "loss": 0.6418,
      "step": 1418
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0519003102911193,
      "learning_rate": 9.001390677039868e-06,
      "loss": 0.7114,
      "step": 1419
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.06933104192568,
      "learning_rate": 8.999825262504672e-06,
      "loss": 0.6671,
      "step": 1420
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0024346237329713,
      "learning_rate": 8.998258758306819e-06,
      "loss": 0.5722,
      "step": 1421
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0884012946637913,
      "learning_rate": 8.996691164873068e-06,
      "loss": 0.6639,
      "step": 1422
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0792796001461855,
      "learning_rate": 8.995122482630473e-06,
      "loss": 0.5921,
      "step": 1423
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0253928774530683,
      "learning_rate": 8.993552712006388e-06,
      "loss": 0.6353,
      "step": 1424
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0611814316767676,
      "learning_rate": 8.99198185342846e-06,
      "loss": 0.6448,
      "step": 1425
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9729411275824148,
      "learning_rate": 8.990409907324638e-06,
      "loss": 0.6037,
      "step": 1426
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9881382168648065,
      "learning_rate": 8.98883687412316e-06,
      "loss": 0.6376,
      "step": 1427
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0155742607429574,
      "learning_rate": 8.987262754252565e-06,
      "loss": 0.6051,
      "step": 1428
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0399315985565132,
      "learning_rate": 8.985687548141685e-06,
      "loss": 0.6584,
      "step": 1429
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0026553178023434,
      "learning_rate": 8.984111256219651e-06,
      "loss": 0.6098,
      "step": 1430
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9832513074201062,
      "learning_rate": 8.982533878915889e-06,
      "loss": 0.6352,
      "step": 1431
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0579708074182446,
      "learning_rate": 8.98095541666012e-06,
      "loss": 0.6537,
      "step": 1432
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1130460136431584,
      "learning_rate": 8.979375869882358e-06,
      "loss": 0.6532,
      "step": 1433
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0002837188096758,
      "learning_rate": 8.977795239012916e-06,
      "loss": 0.6066,
      "step": 1434
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9808336244622646,
      "learning_rate": 8.976213524482404e-06,
      "loss": 0.6766,
      "step": 1435
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.096713230384531,
      "learning_rate": 8.974630726721723e-06,
      "loss": 0.6626,
      "step": 1436
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0578575299460342,
      "learning_rate": 8.97304684616207e-06,
      "loss": 0.663,
      "step": 1437
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9683831751834716,
      "learning_rate": 8.97146188323494e-06,
      "loss": 0.5683,
      "step": 1438
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0589364304161788,
      "learning_rate": 8.969875838372117e-06,
      "loss": 0.6491,
      "step": 1439
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0311570189096695,
      "learning_rate": 8.968288712005688e-06,
      "loss": 0.595,
      "step": 1440
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9975930813864733,
      "learning_rate": 8.966700504568025e-06,
      "loss": 0.6322,
      "step": 1441
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.067823484224228,
      "learning_rate": 8.965111216491803e-06,
      "loss": 0.6464,
      "step": 1442
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0283474615731316,
      "learning_rate": 8.963520848209985e-06,
      "loss": 0.576,
      "step": 1443
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0121956374436312,
      "learning_rate": 8.961929400155833e-06,
      "loss": 0.6495,
      "step": 1444
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9947993733094849,
      "learning_rate": 8.960336872762903e-06,
      "loss": 0.668,
      "step": 1445
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1018081944338125,
      "learning_rate": 8.958743266465041e-06,
      "loss": 0.6781,
      "step": 1446
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.032205939594897,
      "learning_rate": 8.957148581696389e-06,
      "loss": 0.614,
      "step": 1447
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0018226932024632,
      "learning_rate": 8.955552818891384e-06,
      "loss": 0.5993,
      "step": 1448
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9487645848948856,
      "learning_rate": 8.953955978484756e-06,
      "loss": 0.6174,
      "step": 1449
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.054663585165408,
      "learning_rate": 8.95235806091153e-06,
      "loss": 0.5258,
      "step": 1450
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1508342107909812,
      "learning_rate": 8.950759066607017e-06,
      "loss": 0.6782,
      "step": 1451
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0676223992114562,
      "learning_rate": 8.949158996006834e-06,
      "loss": 0.6692,
      "step": 1452
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0260256617593244,
      "learning_rate": 8.94755784954688e-06,
      "loss": 0.6633,
      "step": 1453
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0187619959840217,
      "learning_rate": 8.945955627663353e-06,
      "loss": 0.5627,
      "step": 1454
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0518968378568396,
      "learning_rate": 8.944352330792741e-06,
      "loss": 0.5802,
      "step": 1455
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0130467867865363,
      "learning_rate": 8.942747959371829e-06,
      "loss": 0.5963,
      "step": 1456
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1149757650642895,
      "learning_rate": 8.941142513837689e-06,
      "loss": 0.6366,
      "step": 1457
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0726075141671725,
      "learning_rate": 8.939535994627692e-06,
      "loss": 0.6282,
      "step": 1458
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0537008522251274,
      "learning_rate": 8.937928402179495e-06,
      "loss": 0.6026,
      "step": 1459
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.975524913513794,
      "learning_rate": 8.936319736931051e-06,
      "loss": 0.5593,
      "step": 1460
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0360998035398263,
      "learning_rate": 8.934709999320605e-06,
      "loss": 0.6961,
      "step": 1461
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.034605856925702,
      "learning_rate": 8.933099189786697e-06,
      "loss": 0.6852,
      "step": 1462
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0870992225092975,
      "learning_rate": 8.93148730876815e-06,
      "loss": 0.5848,
      "step": 1463
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1493617672520888,
      "learning_rate": 8.92987435670409e-06,
      "loss": 0.6141,
      "step": 1464
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0063412205856306,
      "learning_rate": 8.928260334033927e-06,
      "loss": 0.6286,
      "step": 1465
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.037769738541675,
      "learning_rate": 8.926645241197365e-06,
      "loss": 0.6547,
      "step": 1466
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0365758173384156,
      "learning_rate": 8.925029078634401e-06,
      "loss": 0.6165,
      "step": 1467
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0080011580160892,
      "learning_rate": 8.923411846785322e-06,
      "loss": 0.5648,
      "step": 1468
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0489487831016127,
      "learning_rate": 8.921793546090709e-06,
      "loss": 0.6436,
      "step": 1469
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0176680361301425,
      "learning_rate": 8.920174176991426e-06,
      "loss": 0.5276,
      "step": 1470
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.086557866647455,
      "learning_rate": 8.918553739928637e-06,
      "loss": 0.6442,
      "step": 1471
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0360629532542711,
      "learning_rate": 8.916932235343797e-06,
      "loss": 0.5219,
      "step": 1472
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9150853989413839,
      "learning_rate": 8.915309663678641e-06,
      "loss": 0.4808,
      "step": 1473
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.738454409470134,
      "learning_rate": 8.913686025375207e-06,
      "loss": 0.4826,
      "step": 1474
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0164443478842413,
      "learning_rate": 8.91206132087582e-06,
      "loss": 0.5974,
      "step": 1475
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.078104225078703,
      "learning_rate": 8.910435550623093e-06,
      "loss": 0.679,
      "step": 1476
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0343444515289968,
      "learning_rate": 8.908808715059929e-06,
      "loss": 0.6405,
      "step": 1477
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0254828394527824,
      "learning_rate": 8.907180814629526e-06,
      "loss": 0.5603,
      "step": 1478
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0413531257034343,
      "learning_rate": 8.905551849775365e-06,
      "loss": 0.5764,
      "step": 1479
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0934818756944118,
      "learning_rate": 8.903921820941224e-06,
      "loss": 0.6805,
      "step": 1480
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0511557236364124,
      "learning_rate": 8.902290728571165e-06,
      "loss": 0.5895,
      "step": 1481
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0657948372414696,
      "learning_rate": 8.900658573109546e-06,
      "loss": 0.6218,
      "step": 1482
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0024071700845156,
      "learning_rate": 8.89902535500101e-06,
      "loss": 0.5925,
      "step": 1483
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0387619596933229,
      "learning_rate": 8.897391074690489e-06,
      "loss": 0.6822,
      "step": 1484
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0312979038838492,
      "learning_rate": 8.895755732623207e-06,
      "loss": 0.6449,
      "step": 1485
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0790057677602334,
      "learning_rate": 8.894119329244675e-06,
      "loss": 0.6883,
      "step": 1486
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9995900236789885,
      "learning_rate": 8.892481865000698e-06,
      "loss": 0.5953,
      "step": 1487
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0130139338640964,
      "learning_rate": 8.890843340337363e-06,
      "loss": 0.6232,
      "step": 1488
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0122041981422187,
      "learning_rate": 8.88920375570105e-06,
      "loss": 0.5069,
      "step": 1489
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1317800510763913,
      "learning_rate": 8.887563111538428e-06,
      "loss": 0.5805,
      "step": 1490
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0630407526416232,
      "learning_rate": 8.885921408296454e-06,
      "loss": 0.616,
      "step": 1491
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0687415970615437,
      "learning_rate": 8.88427864642237e-06,
      "loss": 0.6205,
      "step": 1492
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.032691954140164,
      "learning_rate": 8.882634826363714e-06,
      "loss": 0.5602,
      "step": 1493
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0151872248202518,
      "learning_rate": 8.880989948568303e-06,
      "loss": 0.6209,
      "step": 1494
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9605974662378616,
      "learning_rate": 8.879344013484253e-06,
      "loss": 0.5996,
      "step": 1495
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1246166032300906,
      "learning_rate": 8.877697021559958e-06,
      "loss": 0.7271,
      "step": 1496
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0988429410556217,
      "learning_rate": 8.876048973244105e-06,
      "loss": 0.6674,
      "step": 1497
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0540556185526617,
      "learning_rate": 8.874399868985668e-06,
      "loss": 0.5741,
      "step": 1498
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.154351969784402,
      "learning_rate": 8.872749709233907e-06,
      "loss": 0.6944,
      "step": 1499
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0479973778098817,
      "learning_rate": 8.871098494438375e-06,
      "loss": 0.64,
      "step": 1500
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1089315627007361,
      "learning_rate": 8.869446225048903e-06,
      "loss": 0.6268,
      "step": 1501
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.021501253264102,
      "learning_rate": 8.867792901515617e-06,
      "loss": 0.5781,
      "step": 1502
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.09310067762115,
      "learning_rate": 8.866138524288929e-06,
      "loss": 0.6449,
      "step": 1503
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.01618572080514,
      "learning_rate": 8.864483093819537e-06,
      "loss": 0.7068,
      "step": 1504
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0700028006954734,
      "learning_rate": 8.862826610558427e-06,
      "loss": 0.624,
      "step": 1505
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0417879877839549,
      "learning_rate": 8.861169074956865e-06,
      "loss": 0.5715,
      "step": 1506
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0941366897768456,
      "learning_rate": 8.859510487466415e-06,
      "loss": 0.6658,
      "step": 1507
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0894715757770213,
      "learning_rate": 8.85785084853892e-06,
      "loss": 0.6816,
      "step": 1508
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0440828024241646,
      "learning_rate": 8.856190158626512e-06,
      "loss": 0.6645,
      "step": 1509
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9800280676379383,
      "learning_rate": 8.854528418181609e-06,
      "loss": 0.6314,
      "step": 1510
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0346539432910857,
      "learning_rate": 8.852865627656912e-06,
      "loss": 0.6212,
      "step": 1511
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1031783112781088,
      "learning_rate": 8.851201787505415e-06,
      "loss": 0.6458,
      "step": 1512
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0778708148484915,
      "learning_rate": 8.84953689818039e-06,
      "loss": 0.673,
      "step": 1513
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9833523491723598,
      "learning_rate": 8.847870960135403e-06,
      "loss": 0.5401,
      "step": 1514
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9947151505723325,
      "learning_rate": 8.8462039738243e-06,
      "loss": 0.6156,
      "step": 1515
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0185927764063212,
      "learning_rate": 8.84453593970121e-06,
      "loss": 0.6498,
      "step": 1516
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0269473423982132,
      "learning_rate": 8.842866858220558e-06,
      "loss": 0.6381,
      "step": 1517
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0199822959916656,
      "learning_rate": 8.841196729837044e-06,
      "loss": 0.6154,
      "step": 1518
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9765012900082941,
      "learning_rate": 8.839525555005656e-06,
      "loss": 0.7073,
      "step": 1519
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1495053197757263,
      "learning_rate": 8.837853334181669e-06,
      "loss": 0.7184,
      "step": 1520
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0555606447747357,
      "learning_rate": 8.836180067820646e-06,
      "loss": 0.5939,
      "step": 1521
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0782147776476414,
      "learning_rate": 8.834505756378425e-06,
      "loss": 0.6539,
      "step": 1522
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0669674198660808,
      "learning_rate": 8.832830400311137e-06,
      "loss": 0.6574,
      "step": 1523
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0555694178294315,
      "learning_rate": 8.831154000075196e-06,
      "loss": 0.6395,
      "step": 1524
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0207620301246727,
      "learning_rate": 8.829476556127301e-06,
      "loss": 0.601,
      "step": 1525
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.032999601432603,
      "learning_rate": 8.82779806892443e-06,
      "loss": 0.6212,
      "step": 1526
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0697205731503756,
      "learning_rate": 8.826118538923851e-06,
      "loss": 0.652,
      "step": 1527
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0181741600530463,
      "learning_rate": 8.824437966583114e-06,
      "loss": 0.711,
      "step": 1528
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0691253459695536,
      "learning_rate": 8.822756352360056e-06,
      "loss": 0.6733,
      "step": 1529
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2300538157109882,
      "learning_rate": 8.82107369671279e-06,
      "loss": 0.6741,
      "step": 1530
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.032340800337491,
      "learning_rate": 8.819390000099723e-06,
      "loss": 0.632,
      "step": 1531
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0585166360182745,
      "learning_rate": 8.817705262979536e-06,
      "loss": 0.564,
      "step": 1532
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0807736682342988,
      "learning_rate": 8.8160194858112e-06,
      "loss": 0.6535,
      "step": 1533
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0057269110316291,
      "learning_rate": 8.814332669053968e-06,
      "loss": 0.6781,
      "step": 1534
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0690694637543459,
      "learning_rate": 8.812644813167372e-06,
      "loss": 0.6127,
      "step": 1535
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.95297497030005,
      "learning_rate": 8.810955918611235e-06,
      "loss": 0.5491,
      "step": 1536
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0643776281675157,
      "learning_rate": 8.809265985845655e-06,
      "loss": 0.6269,
      "step": 1537
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7657711567997146,
      "learning_rate": 8.807575015331019e-06,
      "loss": 0.5138,
      "step": 1538
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0551066490204484,
      "learning_rate": 8.805883007527992e-06,
      "loss": 0.6505,
      "step": 1539
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.129762087885999,
      "learning_rate": 8.804189962897521e-06,
      "loss": 0.7038,
      "step": 1540
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1004353013799344,
      "learning_rate": 8.802495881900844e-06,
      "loss": 0.5983,
      "step": 1541
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9400236819690566,
      "learning_rate": 8.80080076499947e-06,
      "loss": 0.6222,
      "step": 1542
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0662222392052771,
      "learning_rate": 8.7991046126552e-06,
      "loss": 0.6768,
      "step": 1543
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9964479661598958,
      "learning_rate": 8.79740742533011e-06,
      "loss": 0.624,
      "step": 1544
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1149768599964365,
      "learning_rate": 8.795709203486563e-06,
      "loss": 0.6699,
      "step": 1545
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9798992543153964,
      "learning_rate": 8.794009947587197e-06,
      "loss": 0.6298,
      "step": 1546
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1543236725068406,
      "learning_rate": 8.79230965809494e-06,
      "loss": 0.6395,
      "step": 1547
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0827574631517152,
      "learning_rate": 8.790608335472995e-06,
      "loss": 0.6594,
      "step": 1548
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0675398408332948,
      "learning_rate": 8.788905980184851e-06,
      "loss": 0.6311,
      "step": 1549
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9442519024763887,
      "learning_rate": 8.787202592694278e-06,
      "loss": 0.5528,
      "step": 1550
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.812937967542351,
      "learning_rate": 8.785498173465323e-06,
      "loss": 0.5466,
      "step": 1551
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.123070633027778,
      "learning_rate": 8.783792722962316e-06,
      "loss": 0.7048,
      "step": 1552
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0695087180612246,
      "learning_rate": 8.782086241649874e-06,
      "loss": 0.5786,
      "step": 1553
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1103592582331396,
      "learning_rate": 8.780378729992884e-06,
      "loss": 0.6329,
      "step": 1554
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0819087202157869,
      "learning_rate": 8.778670188456519e-06,
      "loss": 0.6534,
      "step": 1555
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6377229052319721,
      "learning_rate": 8.776960617506237e-06,
      "loss": 0.4977,
      "step": 1556
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9943102584229293,
      "learning_rate": 8.77525001760777e-06,
      "loss": 0.5542,
      "step": 1557
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0000370786761754,
      "learning_rate": 8.773538389227134e-06,
      "loss": 0.7041,
      "step": 1558
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0389936307919212,
      "learning_rate": 8.771825732830622e-06,
      "loss": 0.6764,
      "step": 1559
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0162407585068645,
      "learning_rate": 8.77011204888481e-06,
      "loss": 0.607,
      "step": 1560
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9675094640830435,
      "learning_rate": 8.76839733785655e-06,
      "loss": 0.6181,
      "step": 1561
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.099032648864559,
      "learning_rate": 8.766681600212981e-06,
      "loss": 0.6055,
      "step": 1562
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0353773626494007,
      "learning_rate": 8.764964836421515e-06,
      "loss": 0.6413,
      "step": 1563
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0506688255339323,
      "learning_rate": 8.763247046949843e-06,
      "loss": 0.6327,
      "step": 1564
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0335132240676155,
      "learning_rate": 8.761528232265944e-06,
      "loss": 0.6418,
      "step": 1565
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9778028796006232,
      "learning_rate": 8.759808392838066e-06,
      "loss": 0.5767,
      "step": 1566
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0158305395107219,
      "learning_rate": 8.75808752913474e-06,
      "loss": 0.6211,
      "step": 1567
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9416086540391815,
      "learning_rate": 8.756365641624782e-06,
      "loss": 0.653,
      "step": 1568
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1188704018876807,
      "learning_rate": 8.754642730777276e-06,
      "loss": 0.721,
      "step": 1569
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0925725161919906,
      "learning_rate": 8.752918797061593e-06,
      "loss": 0.5907,
      "step": 1570
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.07778335632773,
      "learning_rate": 8.751193840947382e-06,
      "loss": 0.624,
      "step": 1571
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.002404163704054,
      "learning_rate": 8.749467862904565e-06,
      "loss": 0.6001,
      "step": 1572
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0167764354228188,
      "learning_rate": 8.747740863403348e-06,
      "loss": 0.5923,
      "step": 1573
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0699150666345603,
      "learning_rate": 8.746012842914214e-06,
      "loss": 0.5538,
      "step": 1574
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9707404125730749,
      "learning_rate": 8.74428380190792e-06,
      "loss": 0.6715,
      "step": 1575
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9992683726708323,
      "learning_rate": 8.742553740855507e-06,
      "loss": 0.5782,
      "step": 1576
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.988104761649749,
      "learning_rate": 8.74082266022829e-06,
      "loss": 0.6677,
      "step": 1577
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9748303766760963,
      "learning_rate": 8.739090560497864e-06,
      "loss": 0.6467,
      "step": 1578
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0594982513880946,
      "learning_rate": 8.737357442136104e-06,
      "loss": 0.6756,
      "step": 1579
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0402860973172223,
      "learning_rate": 8.735623305615153e-06,
      "loss": 0.6505,
      "step": 1580
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.982373754870477,
      "learning_rate": 8.733888151407441e-06,
      "loss": 0.6607,
      "step": 1581
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0317524778602136,
      "learning_rate": 8.732151979985671e-06,
      "loss": 0.6938,
      "step": 1582
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0761223281040657,
      "learning_rate": 8.730414791822825e-06,
      "loss": 0.6617,
      "step": 1583
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.005720666490807,
      "learning_rate": 8.72867658739216e-06,
      "loss": 0.5116,
      "step": 1584
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0867411532114422,
      "learning_rate": 8.726937367167211e-06,
      "loss": 0.6918,
      "step": 1585
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9964653117831407,
      "learning_rate": 8.72519713162179e-06,
      "loss": 0.5608,
      "step": 1586
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0870998554991365,
      "learning_rate": 8.723455881229984e-06,
      "loss": 0.6413,
      "step": 1587
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0685465894499973,
      "learning_rate": 8.721713616466158e-06,
      "loss": 0.6507,
      "step": 1588
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0556099452071792,
      "learning_rate": 8.719970337804953e-06,
      "loss": 0.5845,
      "step": 1589
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.102439536532582,
      "learning_rate": 8.718226045721287e-06,
      "loss": 0.6866,
      "step": 1590
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0903285475999822,
      "learning_rate": 8.716480740690353e-06,
      "loss": 0.6781,
      "step": 1591
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.048903353082005,
      "learning_rate": 8.71473442318762e-06,
      "loss": 0.6703,
      "step": 1592
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.056870976515927,
      "learning_rate": 8.712987093688833e-06,
      "loss": 0.6836,
      "step": 1593
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9626053247305686,
      "learning_rate": 8.711238752670012e-06,
      "loss": 0.5208,
      "step": 1594
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1344262678857058,
      "learning_rate": 8.709489400607453e-06,
      "loss": 0.6873,
      "step": 1595
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1133076118515337,
      "learning_rate": 8.70773903797773e-06,
      "loss": 0.709,
      "step": 1596
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0883570248936871,
      "learning_rate": 8.705987665257688e-06,
      "loss": 0.6366,
      "step": 1597
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0183267678822412,
      "learning_rate": 8.704235282924449e-06,
      "loss": 0.6171,
      "step": 1598
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0946328692999128,
      "learning_rate": 8.702481891455414e-06,
      "loss": 0.6611,
      "step": 1599
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9679801197678418,
      "learning_rate": 8.70072749132825e-06,
      "loss": 0.5326,
      "step": 1600
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1603361555886775,
      "learning_rate": 8.698972083020905e-06,
      "loss": 0.6206,
      "step": 1601
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6978951406280792,
      "learning_rate": 8.697215667011605e-06,
      "loss": 0.487,
      "step": 1602
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9392583365461291,
      "learning_rate": 8.69545824377884e-06,
      "loss": 0.5962,
      "step": 1603
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1042017911288597,
      "learning_rate": 8.693699813801387e-06,
      "loss": 0.6476,
      "step": 1604
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0563581847605283,
      "learning_rate": 8.691940377558284e-06,
      "loss": 0.6606,
      "step": 1605
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.025950882237531,
      "learning_rate": 8.690179935528853e-06,
      "loss": 0.6165,
      "step": 1606
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9559777331165553,
      "learning_rate": 8.68841848819269e-06,
      "loss": 0.6004,
      "step": 1607
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9766649953650622,
      "learning_rate": 8.686656036029657e-06,
      "loss": 0.5941,
      "step": 1608
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9891151804843429,
      "learning_rate": 8.684892579519897e-06,
      "loss": 0.5936,
      "step": 1609
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0928074277795945,
      "learning_rate": 8.683128119143824e-06,
      "loss": 0.663,
      "step": 1610
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0318830969797916,
      "learning_rate": 8.681362655382125e-06,
      "loss": 0.6325,
      "step": 1611
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1723120465259382,
      "learning_rate": 8.67959618871576e-06,
      "loss": 0.6772,
      "step": 1612
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9998221958513286,
      "learning_rate": 8.677828719625967e-06,
      "loss": 0.6309,
      "step": 1613
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0812882573047393,
      "learning_rate": 8.676060248594248e-06,
      "loss": 0.6204,
      "step": 1614
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1269224636145994,
      "learning_rate": 8.674290776102388e-06,
      "loss": 0.6721,
      "step": 1615
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0365051895780908,
      "learning_rate": 8.67252030263244e-06,
      "loss": 0.5391,
      "step": 1616
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0056816044417796,
      "learning_rate": 8.670748828666725e-06,
      "loss": 0.6696,
      "step": 1617
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9627779876234087,
      "learning_rate": 8.668976354687844e-06,
      "loss": 0.5621,
      "step": 1618
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1217891486778544,
      "learning_rate": 8.66720288117867e-06,
      "loss": 0.7468,
      "step": 1619
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0085984162496406,
      "learning_rate": 8.665428408622343e-06,
      "loss": 0.6715,
      "step": 1620
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7104010120169026,
      "learning_rate": 8.66365293750228e-06,
      "loss": 0.5029,
      "step": 1621
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8671782712952599,
      "learning_rate": 8.661876468302167e-06,
      "loss": 0.5123,
      "step": 1622
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0199256782742885,
      "learning_rate": 8.660099001505965e-06,
      "loss": 0.6491,
      "step": 1623
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.002738068993366,
      "learning_rate": 8.658320537597901e-06,
      "loss": 0.6108,
      "step": 1624
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9719023566018004,
      "learning_rate": 8.656541077062483e-06,
      "loss": 0.515,
      "step": 1625
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0422212483749724,
      "learning_rate": 8.654760620384482e-06,
      "loss": 0.6412,
      "step": 1626
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0728643077025237,
      "learning_rate": 8.652979168048944e-06,
      "loss": 0.5881,
      "step": 1627
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1472426260470576,
      "learning_rate": 8.651196720541186e-06,
      "loss": 0.7363,
      "step": 1628
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0274908537313772,
      "learning_rate": 8.649413278346795e-06,
      "loss": 0.6964,
      "step": 1629
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0645912151847958,
      "learning_rate": 8.64762884195163e-06,
      "loss": 0.644,
      "step": 1630
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.047408954340107,
      "learning_rate": 8.64584341184182e-06,
      "loss": 0.6368,
      "step": 1631
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1059072759282695,
      "learning_rate": 8.644056988503769e-06,
      "loss": 0.6356,
      "step": 1632
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.032691888462967,
      "learning_rate": 8.642269572424143e-06,
      "loss": 0.6946,
      "step": 1633
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.120360130844582,
      "learning_rate": 8.640481164089887e-06,
      "loss": 0.6493,
      "step": 1634
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0373968404061225,
      "learning_rate": 8.63869176398821e-06,
      "loss": 0.6239,
      "step": 1635
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.11191931049576,
      "learning_rate": 8.636901372606596e-06,
      "loss": 0.6425,
      "step": 1636
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0032186564743637,
      "learning_rate": 8.635109990432797e-06,
      "loss": 0.6234,
      "step": 1637
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0633552308228533,
      "learning_rate": 8.633317617954832e-06,
      "loss": 0.5954,
      "step": 1638
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.002080694788509,
      "learning_rate": 8.631524255660997e-06,
      "loss": 0.5875,
      "step": 1639
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9908341621652779,
      "learning_rate": 8.629729904039853e-06,
      "loss": 0.567,
      "step": 1640
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6993224514719861,
      "learning_rate": 8.627934563580226e-06,
      "loss": 0.5199,
      "step": 1641
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0258685643234537,
      "learning_rate": 8.62613823477122e-06,
      "loss": 0.6492,
      "step": 1642
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0418792506748107,
      "learning_rate": 8.624340918102206e-06,
      "loss": 0.5835,
      "step": 1643
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0721806467380202,
      "learning_rate": 8.622542614062816e-06,
      "loss": 0.5924,
      "step": 1644
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0507311790555824,
      "learning_rate": 8.620743323142966e-06,
      "loss": 0.6493,
      "step": 1645
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6724881926513512,
      "learning_rate": 8.618943045832826e-06,
      "loss": 0.4854,
      "step": 1646
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.104824565932021,
      "learning_rate": 8.617141782622844e-06,
      "loss": 0.6777,
      "step": 1647
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0486965750493942,
      "learning_rate": 8.615339534003735e-06,
      "loss": 0.6008,
      "step": 1648
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.161208417107663,
      "learning_rate": 8.613536300466476e-06,
      "loss": 0.6905,
      "step": 1649
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0695064463705504,
      "learning_rate": 8.611732082502324e-06,
      "loss": 0.6688,
      "step": 1650
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0856943486246142,
      "learning_rate": 8.609926880602794e-06,
      "loss": 0.5956,
      "step": 1651
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9986924330537926,
      "learning_rate": 8.608120695259674e-06,
      "loss": 0.6165,
      "step": 1652
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0726488388280122,
      "learning_rate": 8.606313526965017e-06,
      "loss": 0.6646,
      "step": 1653
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1011202599420091,
      "learning_rate": 8.604505376211148e-06,
      "loss": 0.7196,
      "step": 1654
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0752769150535895,
      "learning_rate": 8.602696243490653e-06,
      "loss": 0.5833,
      "step": 1655
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0512314245295338,
      "learning_rate": 8.600886129296396e-06,
      "loss": 0.5489,
      "step": 1656
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0134722990641758,
      "learning_rate": 8.599075034121496e-06,
      "loss": 0.5853,
      "step": 1657
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0275243492764028,
      "learning_rate": 8.59726295845935e-06,
      "loss": 0.6627,
      "step": 1658
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9461698106458887,
      "learning_rate": 8.595449902803612e-06,
      "loss": 0.5607,
      "step": 1659
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9762099649357988,
      "learning_rate": 8.593635867648214e-06,
      "loss": 0.5623,
      "step": 1660
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.140689772672225,
      "learning_rate": 8.591820853487344e-06,
      "loss": 0.7361,
      "step": 1661
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9985142548871073,
      "learning_rate": 8.590004860815466e-06,
      "loss": 0.5413,
      "step": 1662
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8956616354421242,
      "learning_rate": 8.588187890127305e-06,
      "loss": 0.5584,
      "step": 1663
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.071073436389438,
      "learning_rate": 8.586369941917852e-06,
      "loss": 0.6574,
      "step": 1664
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0729321275082626,
      "learning_rate": 8.58455101668237e-06,
      "loss": 0.6423,
      "step": 1665
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9547901415558567,
      "learning_rate": 8.58273111491638e-06,
      "loss": 0.6275,
      "step": 1666
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.010845465546539,
      "learning_rate": 8.580910237115678e-06,
      "loss": 0.6286,
      "step": 1667
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1076513713081735,
      "learning_rate": 8.579088383776318e-06,
      "loss": 0.7099,
      "step": 1668
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9815780758241532,
      "learning_rate": 8.577265555394626e-06,
      "loss": 0.5853,
      "step": 1669
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.000518277494108,
      "learning_rate": 8.575441752467185e-06,
      "loss": 0.6254,
      "step": 1670
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.000172503910287,
      "learning_rate": 8.573616975490855e-06,
      "loss": 0.6888,
      "step": 1671
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9803484356550121,
      "learning_rate": 8.571791224962754e-06,
      "loss": 0.6145,
      "step": 1672
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0170471121727223,
      "learning_rate": 8.569964501380266e-06,
      "loss": 0.649,
      "step": 1673
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0161820053555706,
      "learning_rate": 8.56813680524104e-06,
      "loss": 0.5567,
      "step": 1674
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0937265479865304,
      "learning_rate": 8.566308137042995e-06,
      "loss": 0.6646,
      "step": 1675
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0252371511959673,
      "learning_rate": 8.564478497284306e-06,
      "loss": 0.665,
      "step": 1676
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0133537498431278,
      "learning_rate": 8.562647886463417e-06,
      "loss": 0.6445,
      "step": 1677
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9002123617800141,
      "learning_rate": 8.560816305079041e-06,
      "loss": 0.5146,
      "step": 1678
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0368346283617285,
      "learning_rate": 8.558983753630149e-06,
      "loss": 0.5916,
      "step": 1679
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.642759485710064,
      "learning_rate": 8.557150232615977e-06,
      "loss": 0.4956,
      "step": 1680
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1801674712653232,
      "learning_rate": 8.55531574253603e-06,
      "loss": 0.7094,
      "step": 1681
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.045259928674509,
      "learning_rate": 8.55348028389007e-06,
      "loss": 0.651,
      "step": 1682
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0716332254571803,
      "learning_rate": 8.55164385717813e-06,
      "loss": 0.6636,
      "step": 1683
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0143646475325296,
      "learning_rate": 8.549806462900503e-06,
      "loss": 0.6934,
      "step": 1684
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9946778407753984,
      "learning_rate": 8.547968101557742e-06,
      "loss": 0.5928,
      "step": 1685
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.020327734981545,
      "learning_rate": 8.54612877365067e-06,
      "loss": 0.6334,
      "step": 1686
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9904809554457953,
      "learning_rate": 8.544288479680371e-06,
      "loss": 0.5466,
      "step": 1687
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9931222332379382,
      "learning_rate": 8.542447220148191e-06,
      "loss": 0.6376,
      "step": 1688
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1352056130280164,
      "learning_rate": 8.540604995555741e-06,
      "loss": 0.6504,
      "step": 1689
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9748514467536232,
      "learning_rate": 8.538761806404892e-06,
      "loss": 0.5934,
      "step": 1690
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7040353558143975,
      "learning_rate": 8.53691765319778e-06,
      "loss": 0.4651,
      "step": 1691
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9641000802361597,
      "learning_rate": 8.535072536436805e-06,
      "loss": 0.6185,
      "step": 1692
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0675220196805224,
      "learning_rate": 8.533226456624624e-06,
      "loss": 0.5808,
      "step": 1693
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.045049007010878,
      "learning_rate": 8.531379414264165e-06,
      "loss": 0.5948,
      "step": 1694
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1690712166726422,
      "learning_rate": 8.52953140985861e-06,
      "loss": 0.7885,
      "step": 1695
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9691304148377573,
      "learning_rate": 8.527682443911405e-06,
      "loss": 0.5909,
      "step": 1696
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.051325752753385,
      "learning_rate": 8.525832516926262e-06,
      "loss": 0.6795,
      "step": 1697
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9874207573382526,
      "learning_rate": 8.52398162940715e-06,
      "loss": 0.5093,
      "step": 1698
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0880948909844912,
      "learning_rate": 8.522129781858306e-06,
      "loss": 0.6809,
      "step": 1699
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9408074018609945,
      "learning_rate": 8.52027697478422e-06,
      "loss": 0.5174,
      "step": 1700
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0706556977446764,
      "learning_rate": 8.518423208689647e-06,
      "loss": 0.6269,
      "step": 1701
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0385456106117927,
      "learning_rate": 8.516568484079609e-06,
      "loss": 0.6644,
      "step": 1702
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.953706524864836,
      "learning_rate": 8.514712801459379e-06,
      "loss": 0.6572,
      "step": 1703
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1725342513906345,
      "learning_rate": 8.5128561613345e-06,
      "loss": 0.6581,
      "step": 1704
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0263882645496227,
      "learning_rate": 8.510998564210769e-06,
      "loss": 0.5504,
      "step": 1705
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9864510468345877,
      "learning_rate": 8.509140010594248e-06,
      "loss": 0.6148,
      "step": 1706
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9987834851991878,
      "learning_rate": 8.50728050099126e-06,
      "loss": 0.5584,
      "step": 1707
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9878352585628231,
      "learning_rate": 8.505420035908383e-06,
      "loss": 0.6151,
      "step": 1708
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0196462045112449,
      "learning_rate": 8.503558615852461e-06,
      "loss": 0.6803,
      "step": 1709
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0342314761616909,
      "learning_rate": 8.501696241330594e-06,
      "loss": 0.6665,
      "step": 1710
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9230466365766429,
      "learning_rate": 8.49983291285015e-06,
      "loss": 0.6374,
      "step": 1711
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9896275839919103,
      "learning_rate": 8.497968630918743e-06,
      "loss": 0.6063,
      "step": 1712
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0594584151383968,
      "learning_rate": 8.496103396044262e-06,
      "loss": 0.6331,
      "step": 1713
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0768817228954901,
      "learning_rate": 8.494237208734843e-06,
      "loss": 0.6032,
      "step": 1714
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9779559942411508,
      "learning_rate": 8.492370069498892e-06,
      "loss": 0.6105,
      "step": 1715
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9721636982428149,
      "learning_rate": 8.490501978845064e-06,
      "loss": 0.5546,
      "step": 1716
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9917190224004617,
      "learning_rate": 8.488632937282281e-06,
      "loss": 0.6276,
      "step": 1717
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0606321311516793,
      "learning_rate": 8.486762945319722e-06,
      "loss": 0.6596,
      "step": 1718
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0568207882543708,
      "learning_rate": 8.484892003466823e-06,
      "loss": 0.6061,
      "step": 1719
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9422961481839436,
      "learning_rate": 8.48302011223328e-06,
      "loss": 0.5013,
      "step": 1720
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1225999886105238,
      "learning_rate": 8.48114727212905e-06,
      "loss": 0.6369,
      "step": 1721
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0282463592664843,
      "learning_rate": 8.479273483664344e-06,
      "loss": 0.6249,
      "step": 1722
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.056128730276417,
      "learning_rate": 8.477398747349632e-06,
      "loss": 0.628,
      "step": 1723
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.057671165168062,
      "learning_rate": 8.47552306369565e-06,
      "loss": 0.6328,
      "step": 1724
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.022173977872148,
      "learning_rate": 8.473646433213378e-06,
      "loss": 0.5946,
      "step": 1725
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.939260753021954,
      "learning_rate": 8.471768856414069e-06,
      "loss": 0.5845,
      "step": 1726
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9894676407032139,
      "learning_rate": 8.469890333809223e-06,
      "loss": 0.607,
      "step": 1727
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.961056712135085,
      "learning_rate": 8.468010865910601e-06,
      "loss": 0.6008,
      "step": 1728
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1253906678005074,
      "learning_rate": 8.466130453230224e-06,
      "loss": 0.6585,
      "step": 1729
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.903592969972972,
      "learning_rate": 8.464249096280368e-06,
      "loss": 0.5695,
      "step": 1730
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9605140262275187,
      "learning_rate": 8.462366795573564e-06,
      "loss": 0.5731,
      "step": 1731
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.068246171383511,
      "learning_rate": 8.460483551622606e-06,
      "loss": 0.66,
      "step": 1732
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.03825456378861,
      "learning_rate": 8.458599364940537e-06,
      "loss": 0.6649,
      "step": 1733
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0446250644890465,
      "learning_rate": 8.456714236040664e-06,
      "loss": 0.6022,
      "step": 1734
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9998660039857896,
      "learning_rate": 8.45482816543655e-06,
      "loss": 0.7309,
      "step": 1735
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9995765188451078,
      "learning_rate": 8.45294115364201e-06,
      "loss": 0.6012,
      "step": 1736
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9538894410633489,
      "learning_rate": 8.45105320117112e-06,
      "loss": 0.6484,
      "step": 1737
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0895376379022634,
      "learning_rate": 8.449164308538209e-06,
      "loss": 0.675,
      "step": 1738
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0196965965900842,
      "learning_rate": 8.447274476257863e-06,
      "loss": 0.6398,
      "step": 1739
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6720800490790071,
      "learning_rate": 8.445383704844925e-06,
      "loss": 0.4733,
      "step": 1740
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0028822644163902,
      "learning_rate": 8.443491994814493e-06,
      "loss": 0.7383,
      "step": 1741
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1008273233074435,
      "learning_rate": 8.441599346681921e-06,
      "loss": 0.6774,
      "step": 1742
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9577838293406524,
      "learning_rate": 8.43970576096282e-06,
      "loss": 0.5457,
      "step": 1743
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0392775422049816,
      "learning_rate": 8.437811238173053e-06,
      "loss": 0.665,
      "step": 1744
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9314100910470888,
      "learning_rate": 8.43591577882874e-06,
      "loss": 0.5046,
      "step": 1745
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0213911085769738,
      "learning_rate": 8.43401938344626e-06,
      "loss": 0.5563,
      "step": 1746
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0215389870134195,
      "learning_rate": 8.432122052542238e-06,
      "loss": 0.5047,
      "step": 1747
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0332692333333713,
      "learning_rate": 8.430223786633563e-06,
      "loss": 0.5708,
      "step": 1748
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0382015802913187,
      "learning_rate": 8.428324586237374e-06,
      "loss": 0.6369,
      "step": 1749
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0182815950317061,
      "learning_rate": 8.426424451871063e-06,
      "loss": 0.6505,
      "step": 1750
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0545239815752525,
      "learning_rate": 8.424523384052284e-06,
      "loss": 0.6761,
      "step": 1751
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9844930747306614,
      "learning_rate": 8.422621383298936e-06,
      "loss": 0.6457,
      "step": 1752
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2458380475432014,
      "learning_rate": 8.42071845012918e-06,
      "loss": 0.6377,
      "step": 1753
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0278551000822955,
      "learning_rate": 8.418814585061423e-06,
      "loss": 0.5611,
      "step": 1754
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.969035816844534,
      "learning_rate": 8.416909788614335e-06,
      "loss": 0.6018,
      "step": 1755
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0294185081507161,
      "learning_rate": 8.415004061306833e-06,
      "loss": 0.6845,
      "step": 1756
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.049654602466264,
      "learning_rate": 8.413097403658089e-06,
      "loss": 0.6566,
      "step": 1757
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0863304861635876,
      "learning_rate": 8.411189816187528e-06,
      "loss": 0.6343,
      "step": 1758
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1089916865262657,
      "learning_rate": 8.409281299414833e-06,
      "loss": 0.5735,
      "step": 1759
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0471448937479653,
      "learning_rate": 8.407371853859935e-06,
      "loss": 0.6563,
      "step": 1760
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0176193029037608,
      "learning_rate": 8.405461480043019e-06,
      "loss": 0.5793,
      "step": 1761
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0747650784827685,
      "learning_rate": 8.403550178484521e-06,
      "loss": 0.6258,
      "step": 1762
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.737891340712526,
      "learning_rate": 8.401637949705138e-06,
      "loss": 0.5266,
      "step": 1763
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9987039254659821,
      "learning_rate": 8.399724794225809e-06,
      "loss": 0.5305,
      "step": 1764
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1605615911053861,
      "learning_rate": 8.397810712567732e-06,
      "loss": 0.6274,
      "step": 1765
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0138026620404887,
      "learning_rate": 8.39589570525236e-06,
      "loss": 0.6854,
      "step": 1766
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1234830093166164,
      "learning_rate": 8.393979772801386e-06,
      "loss": 0.6971,
      "step": 1767
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9897540411751579,
      "learning_rate": 8.392062915736765e-06,
      "loss": 0.5885,
      "step": 1768
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0395495039921814,
      "learning_rate": 8.390145134580705e-06,
      "loss": 0.5927,
      "step": 1769
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0345495825272193,
      "learning_rate": 8.38822642985566e-06,
      "loss": 0.6257,
      "step": 1770
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9850108264967862,
      "learning_rate": 8.386306802084339e-06,
      "loss": 0.6265,
      "step": 1771
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9567888483302034,
      "learning_rate": 8.3843862517897e-06,
      "loss": 0.621,
      "step": 1772
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0648065753018972,
      "learning_rate": 8.382464779494954e-06,
      "loss": 0.534,
      "step": 1773
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9975528294466697,
      "learning_rate": 8.380542385723566e-06,
      "loss": 0.591,
      "step": 1774
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0405999846229763,
      "learning_rate": 8.378619070999245e-06,
      "loss": 0.6167,
      "step": 1775
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7944337860919333,
      "learning_rate": 8.37669483584596e-06,
      "loss": 0.4972,
      "step": 1776
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1668380656422028,
      "learning_rate": 8.37476968078792e-06,
      "loss": 0.751,
      "step": 1777
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0541695218557057,
      "learning_rate": 8.372843606349594e-06,
      "loss": 0.5835,
      "step": 1778
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1004276898319358,
      "learning_rate": 8.370916613055695e-06,
      "loss": 0.6074,
      "step": 1779
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0838407787688848,
      "learning_rate": 8.368988701431192e-06,
      "loss": 0.6447,
      "step": 1780
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0753627423737728,
      "learning_rate": 8.3670598720013e-06,
      "loss": 0.6613,
      "step": 1781
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.080036292743852,
      "learning_rate": 8.365130125291485e-06,
      "loss": 0.6769,
      "step": 1782
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0854087123461564,
      "learning_rate": 8.363199461827464e-06,
      "loss": 0.7298,
      "step": 1783
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9766129535654418,
      "learning_rate": 8.361267882135203e-06,
      "loss": 0.5707,
      "step": 1784
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.100336670122762,
      "learning_rate": 8.359335386740916e-06,
      "loss": 0.5962,
      "step": 1785
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9589186998604762,
      "learning_rate": 8.357401976171069e-06,
      "loss": 0.6151,
      "step": 1786
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9308742840867967,
      "learning_rate": 8.355467650952375e-06,
      "loss": 0.5709,
      "step": 1787
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0525982221057608,
      "learning_rate": 8.353532411611801e-06,
      "loss": 0.7567,
      "step": 1788
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9844658255154851,
      "learning_rate": 8.351596258676558e-06,
      "loss": 0.5985,
      "step": 1789
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0626351270358532,
      "learning_rate": 8.349659192674104e-06,
      "loss": 0.6399,
      "step": 1790
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9713130770029553,
      "learning_rate": 8.347721214132154e-06,
      "loss": 0.6431,
      "step": 1791
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0879141458714308,
      "learning_rate": 8.345782323578664e-06,
      "loss": 0.5821,
      "step": 1792
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0363899631405824,
      "learning_rate": 8.343842521541844e-06,
      "loss": 0.6696,
      "step": 1793
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0570492613741975,
      "learning_rate": 8.341901808550147e-06,
      "loss": 0.7025,
      "step": 1794
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.096954505510325,
      "learning_rate": 8.339960185132275e-06,
      "loss": 0.6787,
      "step": 1795
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1108423117853232,
      "learning_rate": 8.338017651817183e-06,
      "loss": 0.6773,
      "step": 1796
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8848227765446441,
      "learning_rate": 8.336074209134071e-06,
      "loss": 0.4836,
      "step": 1797
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0482380217025193,
      "learning_rate": 8.334129857612383e-06,
      "loss": 0.6715,
      "step": 1798
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8782054633558585,
      "learning_rate": 8.332184597781818e-06,
      "loss": 0.5491,
      "step": 1799
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0179283223169913,
      "learning_rate": 8.330238430172315e-06,
      "loss": 0.5877,
      "step": 1800
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.061746729547745,
      "learning_rate": 8.328291355314067e-06,
      "loss": 0.6625,
      "step": 1801
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0650736576041888,
      "learning_rate": 8.326343373737506e-06,
      "loss": 0.581,
      "step": 1802
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0370813272018884,
      "learning_rate": 8.32439448597332e-06,
      "loss": 0.5894,
      "step": 1803
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0550795502537282,
      "learning_rate": 8.322444692552437e-06,
      "loss": 0.6963,
      "step": 1804
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0230569925726605,
      "learning_rate": 8.320493994006039e-06,
      "loss": 0.6726,
      "step": 1805
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0440828654983443,
      "learning_rate": 8.318542390865546e-06,
      "loss": 0.624,
      "step": 1806
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0485655865343806,
      "learning_rate": 8.316589883662629e-06,
      "loss": 0.6669,
      "step": 1807
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9826038180876443,
      "learning_rate": 8.314636472929206e-06,
      "loss": 0.5886,
      "step": 1808
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.124641565398034,
      "learning_rate": 8.31268215919744e-06,
      "loss": 0.6869,
      "step": 1809
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0359551621696965,
      "learning_rate": 8.310726942999736e-06,
      "loss": 0.67,
      "step": 1810
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.724045565802701,
      "learning_rate": 8.308770824868757e-06,
      "loss": 0.5101,
      "step": 1811
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0422446369189036,
      "learning_rate": 8.306813805337395e-06,
      "loss": 0.6936,
      "step": 1812
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9844521046984838,
      "learning_rate": 8.3048558849388e-06,
      "loss": 0.6305,
      "step": 1813
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9033114546958253,
      "learning_rate": 8.302897064206363e-06,
      "loss": 0.6331,
      "step": 1814
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9933380467035439,
      "learning_rate": 8.300937343673722e-06,
      "loss": 0.6537,
      "step": 1815
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.979713352894971,
      "learning_rate": 8.298976723874757e-06,
      "loss": 0.5506,
      "step": 1816
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1059823706623375,
      "learning_rate": 8.297015205343595e-06,
      "loss": 0.5433,
      "step": 1817
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0057659927592761,
      "learning_rate": 8.295052788614608e-06,
      "loss": 0.6048,
      "step": 1818
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0369388908581687,
      "learning_rate": 8.293089474222414e-06,
      "loss": 0.5785,
      "step": 1819
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1722359854213027,
      "learning_rate": 8.291125262701874e-06,
      "loss": 0.7334,
      "step": 1820
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.062333157494112,
      "learning_rate": 8.289160154588088e-06,
      "loss": 0.7139,
      "step": 1821
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0873619579903715,
      "learning_rate": 8.287194150416413e-06,
      "loss": 0.6306,
      "step": 1822
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0539449763129014,
      "learning_rate": 8.285227250722439e-06,
      "loss": 0.5959,
      "step": 1823
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0232104075733648,
      "learning_rate": 8.283259456042e-06,
      "loss": 0.6106,
      "step": 1824
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0194748021325402,
      "learning_rate": 8.281290766911187e-06,
      "loss": 0.5285,
      "step": 1825
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.984156354882325,
      "learning_rate": 8.279321183866317e-06,
      "loss": 0.6307,
      "step": 1826
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0774045140665722,
      "learning_rate": 8.277350707443964e-06,
      "loss": 0.6489,
      "step": 1827
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0531625800835027,
      "learning_rate": 8.275379338180936e-06,
      "loss": 0.5808,
      "step": 1828
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0644924304710526,
      "learning_rate": 8.273407076614288e-06,
      "loss": 0.6346,
      "step": 1829
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.009802786104154,
      "learning_rate": 8.271433923281322e-06,
      "loss": 0.6556,
      "step": 1830
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.028178110883418,
      "learning_rate": 8.26945987871958e-06,
      "loss": 0.6357,
      "step": 1831
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0426849063270975,
      "learning_rate": 8.26748494346684e-06,
      "loss": 0.5869,
      "step": 1832
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1012827784984658,
      "learning_rate": 8.265509118061135e-06,
      "loss": 0.6044,
      "step": 1833
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0215863309491484,
      "learning_rate": 8.26353240304073e-06,
      "loss": 0.6145,
      "step": 1834
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0822053368983189,
      "learning_rate": 8.26155479894414e-06,
      "loss": 0.6557,
      "step": 1835
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1772159547213141,
      "learning_rate": 8.259576306310118e-06,
      "loss": 0.6654,
      "step": 1836
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0556004590722161,
      "learning_rate": 8.257596925677657e-06,
      "loss": 0.5879,
      "step": 1837
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9753247931823891,
      "learning_rate": 8.255616657586e-06,
      "loss": 0.5754,
      "step": 1838
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0180753187469045,
      "learning_rate": 8.253635502574623e-06,
      "loss": 0.6365,
      "step": 1839
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.985576710942524,
      "learning_rate": 8.251653461183249e-06,
      "loss": 0.5927,
      "step": 1840
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0959951788768731,
      "learning_rate": 8.249670533951838e-06,
      "loss": 0.6575,
      "step": 1841
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.008307485146341,
      "learning_rate": 8.247686721420596e-06,
      "loss": 0.6734,
      "step": 1842
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0330964672161271,
      "learning_rate": 8.24570202412997e-06,
      "loss": 0.6028,
      "step": 1843
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7273470216488935,
      "learning_rate": 8.243716442620644e-06,
      "loss": 0.5474,
      "step": 1844
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0621734748208893,
      "learning_rate": 8.241729977433543e-06,
      "loss": 0.6228,
      "step": 1845
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0855446138408735,
      "learning_rate": 8.239742629109839e-06,
      "loss": 0.6811,
      "step": 1846
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0896725078555813,
      "learning_rate": 8.237754398190937e-06,
      "loss": 0.6174,
      "step": 1847
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1044736210340818,
      "learning_rate": 8.235765285218491e-06,
      "loss": 0.6563,
      "step": 1848
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0702517521318489,
      "learning_rate": 8.233775290734385e-06,
      "loss": 0.6275,
      "step": 1849
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6144811204624225,
      "learning_rate": 8.23178441528075e-06,
      "loss": 0.4909,
      "step": 1850
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9586766604210359,
      "learning_rate": 8.229792659399957e-06,
      "loss": 0.5827,
      "step": 1851
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0332765274425348,
      "learning_rate": 8.227800023634614e-06,
      "loss": 0.6434,
      "step": 1852
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6655779450340815,
      "learning_rate": 8.225806508527568e-06,
      "loss": 0.5153,
      "step": 1853
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1578013782441694,
      "learning_rate": 8.22381211462191e-06,
      "loss": 0.7505,
      "step": 1854
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7126967873007504,
      "learning_rate": 8.221816842460967e-06,
      "loss": 0.4443,
      "step": 1855
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1509064268853675,
      "learning_rate": 8.219820692588306e-06,
      "loss": 0.6913,
      "step": 1856
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0581320978820739,
      "learning_rate": 8.217823665547733e-06,
      "loss": 0.6221,
      "step": 1857
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.048072805009424,
      "learning_rate": 8.215825761883295e-06,
      "loss": 0.6806,
      "step": 1858
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0464075588370685,
      "learning_rate": 8.213826982139273e-06,
      "loss": 0.6497,
      "step": 1859
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.081698835502318,
      "learning_rate": 8.211827326860193e-06,
      "loss": 0.6691,
      "step": 1860
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.970227583613651,
      "learning_rate": 8.209826796590814e-06,
      "loss": 0.5103,
      "step": 1861
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1050004756068814,
      "learning_rate": 8.207825391876137e-06,
      "loss": 0.6524,
      "step": 1862
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0669457282545567,
      "learning_rate": 8.205823113261397e-06,
      "loss": 0.6666,
      "step": 1863
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0727784349856202,
      "learning_rate": 8.203819961292074e-06,
      "loss": 0.6157,
      "step": 1864
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1102033134509754,
      "learning_rate": 8.20181593651388e-06,
      "loss": 0.6717,
      "step": 1865
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1644967565480249,
      "learning_rate": 8.199811039472764e-06,
      "loss": 0.6919,
      "step": 1866
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9733643986624221,
      "learning_rate": 8.197805270714918e-06,
      "loss": 0.595,
      "step": 1867
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9757559127369736,
      "learning_rate": 8.19579863078677e-06,
      "loss": 0.5169,
      "step": 1868
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0823891491321702,
      "learning_rate": 8.193791120234982e-06,
      "loss": 0.6022,
      "step": 1869
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0136407057982078,
      "learning_rate": 8.191782739606455e-06,
      "loss": 0.6133,
      "step": 1870
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8469244577806043,
      "learning_rate": 8.189773489448328e-06,
      "loss": 0.5035,
      "step": 1871
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.996652341703191,
      "learning_rate": 8.187763370307975e-06,
      "loss": 0.5763,
      "step": 1872
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7852253493185378,
      "learning_rate": 8.185752382733007e-06,
      "loss": 0.4269,
      "step": 1873
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0916002356645216,
      "learning_rate": 8.183740527271278e-06,
      "loss": 0.669,
      "step": 1874
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.136826792078389,
      "learning_rate": 8.181727804470867e-06,
      "loss": 0.6673,
      "step": 1875
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9991595438422527,
      "learning_rate": 8.179714214880096e-06,
      "loss": 0.6238,
      "step": 1876
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0484510568340117,
      "learning_rate": 8.177699759047525e-06,
      "loss": 0.6568,
      "step": 1877
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9547526290395373,
      "learning_rate": 8.175684437521946e-06,
      "loss": 0.6039,
      "step": 1878
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0683359057504307,
      "learning_rate": 8.173668250852389e-06,
      "loss": 0.6622,
      "step": 1879
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0374471957933837,
      "learning_rate": 8.171651199588118e-06,
      "loss": 0.6153,
      "step": 1880
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7482258719262316,
      "learning_rate": 8.16963328427863e-06,
      "loss": 0.4951,
      "step": 1881
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0673835289948106,
      "learning_rate": 8.167614505473667e-06,
      "loss": 0.648,
      "step": 1882
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.031267899413725,
      "learning_rate": 8.165594863723197e-06,
      "loss": 0.6337,
      "step": 1883
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9887973193361107,
      "learning_rate": 8.163574359577422e-06,
      "loss": 0.6518,
      "step": 1884
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0216816123151,
      "learning_rate": 8.16155299358679e-06,
      "loss": 0.6523,
      "step": 1885
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9361258937232282,
      "learning_rate": 8.159530766301974e-06,
      "loss": 0.523,
      "step": 1886
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0822460140835453,
      "learning_rate": 8.157507678273884e-06,
      "loss": 0.6098,
      "step": 1887
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9846889417721696,
      "learning_rate": 8.155483730053664e-06,
      "loss": 0.6812,
      "step": 1888
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0472018705696453,
      "learning_rate": 8.153458922192696e-06,
      "loss": 0.5772,
      "step": 1889
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.681976135012787,
      "learning_rate": 8.15143325524259e-06,
      "loss": 0.4605,
      "step": 1890
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1251818949576624,
      "learning_rate": 8.149406729755198e-06,
      "loss": 0.7063,
      "step": 1891
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0859060596255317,
      "learning_rate": 8.147379346282599e-06,
      "loss": 0.7091,
      "step": 1892
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9970457265637536,
      "learning_rate": 8.145351105377107e-06,
      "loss": 0.5934,
      "step": 1893
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9487372532225709,
      "learning_rate": 8.14332200759127e-06,
      "loss": 0.6104,
      "step": 1894
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.038491933667674,
      "learning_rate": 8.141292053477873e-06,
      "loss": 0.6704,
      "step": 1895
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9900469637871521,
      "learning_rate": 8.139261243589933e-06,
      "loss": 0.676,
      "step": 1896
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.925498033304616,
      "learning_rate": 8.137229578480694e-06,
      "loss": 0.5594,
      "step": 1897
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0244901228059946,
      "learning_rate": 8.135197058703638e-06,
      "loss": 0.5916,
      "step": 1898
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0213745065267665,
      "learning_rate": 8.133163684812484e-06,
      "loss": 0.6685,
      "step": 1899
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9936728441669446,
      "learning_rate": 8.131129457361176e-06,
      "loss": 0.6565,
      "step": 1900
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0348157107047065,
      "learning_rate": 8.129094376903891e-06,
      "loss": 0.6504,
      "step": 1901
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0768488551822888,
      "learning_rate": 8.127058443995046e-06,
      "loss": 0.6317,
      "step": 1902
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9281023505153545,
      "learning_rate": 8.125021659189281e-06,
      "loss": 0.5736,
      "step": 1903
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0263393171481847,
      "learning_rate": 8.122984023041476e-06,
      "loss": 0.5959,
      "step": 1904
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7683028202767916,
      "learning_rate": 8.120945536106738e-06,
      "loss": 0.5255,
      "step": 1905
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0975191352655638,
      "learning_rate": 8.118906198940403e-06,
      "loss": 0.7127,
      "step": 1906
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9932971526634823,
      "learning_rate": 8.11686601209805e-06,
      "loss": 0.5287,
      "step": 1907
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6412213084705646,
      "learning_rate": 8.114824976135478e-06,
      "loss": 0.5015,
      "step": 1908
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.985160308454403,
      "learning_rate": 8.11278309160872e-06,
      "loss": 0.6109,
      "step": 1909
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1657540101316224,
      "learning_rate": 8.110740359074046e-06,
      "loss": 0.7356,
      "step": 1910
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.968709572572293,
      "learning_rate": 8.108696779087949e-06,
      "loss": 0.6632,
      "step": 1911
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9303354822475052,
      "learning_rate": 8.106652352207157e-06,
      "loss": 0.6414,
      "step": 1912
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9523928651950057,
      "learning_rate": 8.10460707898863e-06,
      "loss": 0.511,
      "step": 1913
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.085256702337992,
      "learning_rate": 8.102560959989554e-06,
      "loss": 0.6768,
      "step": 1914
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9278595028727968,
      "learning_rate": 8.100513995767352e-06,
      "loss": 0.5906,
      "step": 1915
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.100670424495352,
      "learning_rate": 8.098466186879669e-06,
      "loss": 0.7041,
      "step": 1916
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0863607266027058,
      "learning_rate": 8.096417533884388e-06,
      "loss": 0.6143,
      "step": 1917
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9766862867669861,
      "learning_rate": 8.094368037339619e-06,
      "loss": 0.6118,
      "step": 1918
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6595381921918712,
      "learning_rate": 8.092317697803698e-06,
      "loss": 0.5112,
      "step": 1919
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9613728465613319,
      "learning_rate": 8.090266515835197e-06,
      "loss": 0.5977,
      "step": 1920
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0375818113270032,
      "learning_rate": 8.088214491992912e-06,
      "loss": 0.6795,
      "step": 1921
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9221077039518808,
      "learning_rate": 8.086161626835873e-06,
      "loss": 0.5526,
      "step": 1922
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9923873972020701,
      "learning_rate": 8.084107920923336e-06,
      "loss": 0.6165,
      "step": 1923
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0272871893814814,
      "learning_rate": 8.08205337481479e-06,
      "loss": 0.669,
      "step": 1924
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0074420599253882,
      "learning_rate": 8.079997989069945e-06,
      "loss": 0.6777,
      "step": 1925
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0334715874509215,
      "learning_rate": 8.077941764248746e-06,
      "loss": 0.5763,
      "step": 1926
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0684268300674156,
      "learning_rate": 8.075884700911368e-06,
      "loss": 0.6411,
      "step": 1927
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0001782158510906,
      "learning_rate": 8.07382679961821e-06,
      "loss": 0.6011,
      "step": 1928
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1242012273356614,
      "learning_rate": 8.071768060929903e-06,
      "loss": 0.6344,
      "step": 1929
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9645809750550852,
      "learning_rate": 8.0697084854073e-06,
      "loss": 0.5707,
      "step": 1930
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0689899672043581,
      "learning_rate": 8.06764807361149e-06,
      "loss": 0.6,
      "step": 1931
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.992122881121202,
      "learning_rate": 8.065586826103782e-06,
      "loss": 0.6058,
      "step": 1932
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9874474650573509,
      "learning_rate": 8.063524743445721e-06,
      "loss": 0.5869,
      "step": 1933
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0046332963029097,
      "learning_rate": 8.061461826199075e-06,
      "loss": 0.5953,
      "step": 1934
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0613108066732035,
      "learning_rate": 8.059398074925835e-06,
      "loss": 0.6438,
      "step": 1935
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.043381392212319,
      "learning_rate": 8.057333490188228e-06,
      "loss": 0.5762,
      "step": 1936
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0292761210038501,
      "learning_rate": 8.055268072548704e-06,
      "loss": 0.5756,
      "step": 1937
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0777478814735153,
      "learning_rate": 8.053201822569933e-06,
      "loss": 0.6444,
      "step": 1938
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0497761945390296,
      "learning_rate": 8.051134740814827e-06,
      "loss": 0.6165,
      "step": 1939
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.111693470563027,
      "learning_rate": 8.049066827846513e-06,
      "loss": 0.6035,
      "step": 1940
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0284638581571135,
      "learning_rate": 8.046998084228347e-06,
      "loss": 0.6671,
      "step": 1941
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0507331996853633,
      "learning_rate": 8.044928510523911e-06,
      "loss": 0.6332,
      "step": 1942
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.07407758394666,
      "learning_rate": 8.042858107297015e-06,
      "loss": 0.65,
      "step": 1943
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9884761789744048,
      "learning_rate": 8.040786875111694e-06,
      "loss": 0.6502,
      "step": 1944
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0627479018315698,
      "learning_rate": 8.03871481453221e-06,
      "loss": 0.6663,
      "step": 1945
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1768227769577873,
      "learning_rate": 8.036641926123043e-06,
      "loss": 0.695,
      "step": 1946
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9830213721304991,
      "learning_rate": 8.034568210448914e-06,
      "loss": 0.5873,
      "step": 1947
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9411894878501208,
      "learning_rate": 8.032493668074756e-06,
      "loss": 0.6515,
      "step": 1948
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0227226684578699,
      "learning_rate": 8.03041829956573e-06,
      "loss": 0.6406,
      "step": 1949
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0784308805027176,
      "learning_rate": 8.028342105487226e-06,
      "loss": 0.7069,
      "step": 1950
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0665093746620757,
      "learning_rate": 8.026265086404856e-06,
      "loss": 0.6029,
      "step": 1951
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0665936853303892,
      "learning_rate": 8.024187242884456e-06,
      "loss": 0.6731,
      "step": 1952
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0500557340609795,
      "learning_rate": 8.02210857549209e-06,
      "loss": 0.6038,
      "step": 1953
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0003403707183904,
      "learning_rate": 8.020029084794043e-06,
      "loss": 0.6547,
      "step": 1954
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0763977392867565,
      "learning_rate": 8.017948771356824e-06,
      "loss": 0.6499,
      "step": 1955
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9482502217798888,
      "learning_rate": 8.01586763574717e-06,
      "loss": 0.5633,
      "step": 1956
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0072454045966897,
      "learning_rate": 8.01378567853204e-06,
      "loss": 0.6581,
      "step": 1957
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1158516183273934,
      "learning_rate": 8.011702900278614e-06,
      "loss": 0.6772,
      "step": 1958
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9069256411697026,
      "learning_rate": 8.009619301554303e-06,
      "loss": 0.5151,
      "step": 1959
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9975652633698152,
      "learning_rate": 8.007534882926731e-06,
      "loss": 0.526,
      "step": 1960
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0210895217525349,
      "learning_rate": 8.005449644963756e-06,
      "loss": 0.6013,
      "step": 1961
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0261806505993152,
      "learning_rate": 8.00336358823345e-06,
      "loss": 0.6103,
      "step": 1962
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9369903753274512,
      "learning_rate": 8.001276713304116e-06,
      "loss": 0.6392,
      "step": 1963
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0411867265022243,
      "learning_rate": 7.999189020744273e-06,
      "loss": 0.5983,
      "step": 1964
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9235641784399136,
      "learning_rate": 7.997100511122669e-06,
      "loss": 0.6123,
      "step": 1965
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.01542544732557,
      "learning_rate": 7.99501118500827e-06,
      "loss": 0.6525,
      "step": 1966
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0057105494489462,
      "learning_rate": 7.992921042970264e-06,
      "loss": 0.5757,
      "step": 1967
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0477621982696093,
      "learning_rate": 7.990830085578068e-06,
      "loss": 0.6993,
      "step": 1968
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0026430858701145,
      "learning_rate": 7.988738313401312e-06,
      "loss": 0.6398,
      "step": 1969
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0620971683207427,
      "learning_rate": 7.986645727009856e-06,
      "loss": 0.6552,
      "step": 1970
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.031824942138432,
      "learning_rate": 7.984552326973776e-06,
      "loss": 0.6628,
      "step": 1971
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0665716750502996,
      "learning_rate": 7.982458113863373e-06,
      "loss": 0.6487,
      "step": 1972
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0029031995074955,
      "learning_rate": 7.980363088249167e-06,
      "loss": 0.6196,
      "step": 1973
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9486493328205855,
      "learning_rate": 7.978267250701904e-06,
      "loss": 0.5783,
      "step": 1974
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9949223714268802,
      "learning_rate": 7.976170601792543e-06,
      "loss": 0.7236,
      "step": 1975
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.016433832809466,
      "learning_rate": 7.97407314209227e-06,
      "loss": 0.6337,
      "step": 1976
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9402565535052956,
      "learning_rate": 7.971974872172497e-06,
      "loss": 0.5083,
      "step": 1977
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0132233493236422,
      "learning_rate": 7.969875792604842e-06,
      "loss": 0.6626,
      "step": 1978
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9951281064603269,
      "learning_rate": 7.967775903961158e-06,
      "loss": 0.6049,
      "step": 1979
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0167015209018815,
      "learning_rate": 7.96567520681351e-06,
      "loss": 0.6799,
      "step": 1980
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0533996850244487,
      "learning_rate": 7.963573701734185e-06,
      "loss": 0.5744,
      "step": 1981
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0090592024541625,
      "learning_rate": 7.961471389295694e-06,
      "loss": 0.5835,
      "step": 1982
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1383223156297746,
      "learning_rate": 7.959368270070763e-06,
      "loss": 0.6271,
      "step": 1983
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1116792050877726,
      "learning_rate": 7.957264344632338e-06,
      "loss": 0.6601,
      "step": 1984
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.018476930765761,
      "learning_rate": 7.95515961355359e-06,
      "loss": 0.6604,
      "step": 1985
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.081145993325459,
      "learning_rate": 7.953054077407903e-06,
      "loss": 0.6891,
      "step": 1986
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9264933505418262,
      "learning_rate": 7.950947736768884e-06,
      "loss": 0.5825,
      "step": 1987
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0420210854137582,
      "learning_rate": 7.948840592210358e-06,
      "loss": 0.7006,
      "step": 1988
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9731076134852483,
      "learning_rate": 7.94673264430637e-06,
      "loss": 0.6131,
      "step": 1989
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9708495311147903,
      "learning_rate": 7.94462389363118e-06,
      "loss": 0.5896,
      "step": 1990
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9956630272461023,
      "learning_rate": 7.942514340759275e-06,
      "loss": 0.6952,
      "step": 1991
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7246089343983644,
      "learning_rate": 7.940403986265353e-06,
      "loss": 0.5023,
      "step": 1992
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.068740135715354,
      "learning_rate": 7.93829283072433e-06,
      "loss": 0.6365,
      "step": 1993
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9761955533982813,
      "learning_rate": 7.936180874711347e-06,
      "loss": 0.5576,
      "step": 1994
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0197491590903691,
      "learning_rate": 7.934068118801758e-06,
      "loss": 0.6007,
      "step": 1995
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.013614928651729,
      "learning_rate": 7.931954563571134e-06,
      "loss": 0.6485,
      "step": 1996
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0366744648945945,
      "learning_rate": 7.92984020959527e-06,
      "loss": 0.6081,
      "step": 1997
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0095285612348635,
      "learning_rate": 7.92772505745017e-06,
      "loss": 0.5701,
      "step": 1998
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9417884432762371,
      "learning_rate": 7.925609107712057e-06,
      "loss": 0.5678,
      "step": 1999
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0019134036388098,
      "learning_rate": 7.923492360957383e-06,
      "loss": 0.6491,
      "step": 2000
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.018472058290425,
      "learning_rate": 7.921374817762801e-06,
      "loss": 0.6862,
      "step": 2001
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.967819687624286,
      "learning_rate": 7.919256478705192e-06,
      "loss": 0.6792,
      "step": 2002
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9329606419877908,
      "learning_rate": 7.917137344361647e-06,
      "loss": 0.5299,
      "step": 2003
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.007616233161508,
      "learning_rate": 7.91501741530948e-06,
      "loss": 0.6973,
      "step": 2004
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0038678740391376,
      "learning_rate": 7.912896692126216e-06,
      "loss": 0.6127,
      "step": 2005
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.980762374008956,
      "learning_rate": 7.910775175389595e-06,
      "loss": 0.6242,
      "step": 2006
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0284556986146682,
      "learning_rate": 7.908652865677584e-06,
      "loss": 0.6633,
      "step": 2007
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0028130598791773,
      "learning_rate": 7.906529763568354e-06,
      "loss": 0.5724,
      "step": 2008
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0505062226778203,
      "learning_rate": 7.904405869640296e-06,
      "loss": 0.597,
      "step": 2009
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.041478169750078,
      "learning_rate": 7.902281184472021e-06,
      "loss": 0.6307,
      "step": 2010
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1258947799335886,
      "learning_rate": 7.900155708642347e-06,
      "loss": 0.6875,
      "step": 2011
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0925877922762228,
      "learning_rate": 7.898029442730316e-06,
      "loss": 0.6599,
      "step": 2012
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1121577938042486,
      "learning_rate": 7.89590238731518e-06,
      "loss": 0.6447,
      "step": 2013
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9891122462044993,
      "learning_rate": 7.893774542976408e-06,
      "loss": 0.6444,
      "step": 2014
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0178244244657477,
      "learning_rate": 7.891645910293683e-06,
      "loss": 0.6464,
      "step": 2015
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1190640597755588,
      "learning_rate": 7.889516489846904e-06,
      "loss": 0.7081,
      "step": 2016
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9376282934354488,
      "learning_rate": 7.887386282216182e-06,
      "loss": 0.5193,
      "step": 2017
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9315979567364392,
      "learning_rate": 7.885255287981845e-06,
      "loss": 0.4983,
      "step": 2018
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9360925707387926,
      "learning_rate": 7.883123507724437e-06,
      "loss": 0.6793,
      "step": 2019
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8240866564978023,
      "learning_rate": 7.88099094202471e-06,
      "loss": 0.553,
      "step": 2020
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0552387252262851,
      "learning_rate": 7.878857591463636e-06,
      "loss": 0.6835,
      "step": 2021
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0532354399847978,
      "learning_rate": 7.876723456622399e-06,
      "loss": 0.6576,
      "step": 2022
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.975130802775477,
      "learning_rate": 7.874588538082394e-06,
      "loss": 0.566,
      "step": 2023
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9266834455280066,
      "learning_rate": 7.872452836425233e-06,
      "loss": 0.6082,
      "step": 2024
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.001348759176678,
      "learning_rate": 7.870316352232739e-06,
      "loss": 0.6213,
      "step": 2025
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1022720059019917,
      "learning_rate": 7.868179086086951e-06,
      "loss": 0.6977,
      "step": 2026
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0891228964044355,
      "learning_rate": 7.866041038570117e-06,
      "loss": 0.6576,
      "step": 2027
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7033120574863014,
      "learning_rate": 7.863902210264702e-06,
      "loss": 0.4756,
      "step": 2028
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0301721011395628,
      "learning_rate": 7.861762601753379e-06,
      "loss": 0.5798,
      "step": 2029
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9713122289800508,
      "learning_rate": 7.859622213619043e-06,
      "loss": 0.6064,
      "step": 2030
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9166861836572292,
      "learning_rate": 7.857481046444785e-06,
      "loss": 0.5651,
      "step": 2031
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.088076409534945,
      "learning_rate": 7.855339100813925e-06,
      "loss": 0.6263,
      "step": 2032
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0616452849750926,
      "learning_rate": 7.853196377309986e-06,
      "loss": 0.6214,
      "step": 2033
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9562817434125473,
      "learning_rate": 7.851052876516708e-06,
      "loss": 0.5707,
      "step": 2034
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.958410897538622,
      "learning_rate": 7.848908599018033e-06,
      "loss": 0.5643,
      "step": 2035
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0219612308357566,
      "learning_rate": 7.846763545398124e-06,
      "loss": 0.6315,
      "step": 2036
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0509593492374971,
      "learning_rate": 7.844617716241358e-06,
      "loss": 0.658,
      "step": 2037
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0089412867215362,
      "learning_rate": 7.842471112132311e-06,
      "loss": 0.5791,
      "step": 2038
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.114245945367318,
      "learning_rate": 7.84032373365578e-06,
      "loss": 0.6294,
      "step": 2039
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.941736617895979,
      "learning_rate": 7.83817558139677e-06,
      "loss": 0.5635,
      "step": 2040
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9983443866436168,
      "learning_rate": 7.836026655940497e-06,
      "loss": 0.597,
      "step": 2041
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0645662242907994,
      "learning_rate": 7.833876957872388e-06,
      "loss": 0.6414,
      "step": 2042
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9629343154807554,
      "learning_rate": 7.83172648777808e-06,
      "loss": 0.5486,
      "step": 2043
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0118379257709842,
      "learning_rate": 7.829575246243417e-06,
      "loss": 0.5936,
      "step": 2044
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9721244453186837,
      "learning_rate": 7.82742323385446e-06,
      "loss": 0.5174,
      "step": 2045
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0398470026578153,
      "learning_rate": 7.825270451197477e-06,
      "loss": 0.6944,
      "step": 2046
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3068989131552775,
      "learning_rate": 7.823116898858945e-06,
      "loss": 0.5915,
      "step": 2047
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0106897674424422,
      "learning_rate": 7.820962577425548e-06,
      "loss": 0.5205,
      "step": 2048
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1883492742811927,
      "learning_rate": 7.818807487484186e-06,
      "loss": 0.7669,
      "step": 2049
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1041324044092566,
      "learning_rate": 7.816651629621963e-06,
      "loss": 0.6771,
      "step": 2050
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9721792400151228,
      "learning_rate": 7.814495004426195e-06,
      "loss": 0.6271,
      "step": 2051
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9849060740235217,
      "learning_rate": 7.812337612484404e-06,
      "loss": 0.5689,
      "step": 2052
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0965701090832938,
      "learning_rate": 7.810179454384326e-06,
      "loss": 0.6196,
      "step": 2053
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0662649921282041,
      "learning_rate": 7.808020530713902e-06,
      "loss": 0.6487,
      "step": 2054
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9779694843125751,
      "learning_rate": 7.805860842061282e-06,
      "loss": 0.6074,
      "step": 2055
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0371165085767355,
      "learning_rate": 7.803700389014824e-06,
      "loss": 0.5966,
      "step": 2056
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9656577001546961,
      "learning_rate": 7.801539172163097e-06,
      "loss": 0.6343,
      "step": 2057
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0907897309078987,
      "learning_rate": 7.799377192094872e-06,
      "loss": 0.5901,
      "step": 2058
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0391155945975157,
      "learning_rate": 7.797214449399136e-06,
      "loss": 0.6718,
      "step": 2059
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.970802655665947,
      "learning_rate": 7.79505094466508e-06,
      "loss": 0.6106,
      "step": 2060
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4975500085063516,
      "learning_rate": 7.792886678482096e-06,
      "loss": 0.5436,
      "step": 2061
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.022620941330031,
      "learning_rate": 7.7907216514398e-06,
      "loss": 0.6167,
      "step": 2062
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.022589008246891,
      "learning_rate": 7.788555864127995e-06,
      "loss": 0.6031,
      "step": 2063
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9488340893757712,
      "learning_rate": 7.786389317136708e-06,
      "loss": 0.6,
      "step": 2064
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1107757558635087,
      "learning_rate": 7.784222011056164e-06,
      "loss": 0.7038,
      "step": 2065
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1408904178479418,
      "learning_rate": 7.782053946476795e-06,
      "loss": 0.6719,
      "step": 2066
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.976330884150759,
      "learning_rate": 7.779885123989244e-06,
      "loss": 0.5981,
      "step": 2067
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0118914254326963,
      "learning_rate": 7.777715544184358e-06,
      "loss": 0.5866,
      "step": 2068
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0463059068660445,
      "learning_rate": 7.775545207653188e-06,
      "loss": 0.7049,
      "step": 2069
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0140843075338823,
      "learning_rate": 7.773374114986996e-06,
      "loss": 0.6366,
      "step": 2070
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0758735668897645,
      "learning_rate": 7.771202266777247e-06,
      "loss": 0.6564,
      "step": 2071
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0084112592974492,
      "learning_rate": 7.76902966361561e-06,
      "loss": 0.5532,
      "step": 2072
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0369849279025734,
      "learning_rate": 7.766856306093966e-06,
      "loss": 0.6544,
      "step": 2073
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0062071651625457,
      "learning_rate": 7.764682194804394e-06,
      "loss": 0.5698,
      "step": 2074
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0565988988799921,
      "learning_rate": 7.762507330339185e-06,
      "loss": 0.5822,
      "step": 2075
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9220190328693751,
      "learning_rate": 7.76033171329083e-06,
      "loss": 0.5636,
      "step": 2076
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.012920946453687,
      "learning_rate": 7.758155344252025e-06,
      "loss": 0.558,
      "step": 2077
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9366909549974815,
      "learning_rate": 7.755978223815678e-06,
      "loss": 0.5931,
      "step": 2078
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.955611734012265,
      "learning_rate": 7.753800352574891e-06,
      "loss": 0.5458,
      "step": 2079
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0149920410921363,
      "learning_rate": 7.751621731122981e-06,
      "loss": 0.5602,
      "step": 2080
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9868284832705365,
      "learning_rate": 7.749442360053462e-06,
      "loss": 0.5291,
      "step": 2081
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0135731372838592,
      "learning_rate": 7.747262239960055e-06,
      "loss": 0.6495,
      "step": 2082
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0063418158904514,
      "learning_rate": 7.745081371436686e-06,
      "loss": 0.6039,
      "step": 2083
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9989108272103109,
      "learning_rate": 7.742899755077482e-06,
      "loss": 0.615,
      "step": 2084
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9760342977118495,
      "learning_rate": 7.740717391476778e-06,
      "loss": 0.5688,
      "step": 2085
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6766906224146437,
      "learning_rate": 7.738534281229106e-06,
      "loss": 0.4681,
      "step": 2086
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9951447643457285,
      "learning_rate": 7.736350424929209e-06,
      "loss": 0.5706,
      "step": 2087
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1770440409555176,
      "learning_rate": 7.734165823172028e-06,
      "loss": 0.6948,
      "step": 2088
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.969816041287878,
      "learning_rate": 7.731980476552708e-06,
      "loss": 0.5987,
      "step": 2089
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9880512918059993,
      "learning_rate": 7.7297943856666e-06,
      "loss": 0.6257,
      "step": 2090
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.667036197838884,
      "learning_rate": 7.727607551109252e-06,
      "loss": 0.4954,
      "step": 2091
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0545500030018207,
      "learning_rate": 7.725419973476422e-06,
      "loss": 0.7486,
      "step": 2092
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1410138392031859,
      "learning_rate": 7.723231653364065e-06,
      "loss": 0.5423,
      "step": 2093
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0689512480568013,
      "learning_rate": 7.72104259136834e-06,
      "loss": 0.6598,
      "step": 2094
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0272364091535449,
      "learning_rate": 7.718852788085604e-06,
      "loss": 0.6589,
      "step": 2095
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0639373981161628,
      "learning_rate": 7.716662244112425e-06,
      "loss": 0.648,
      "step": 2096
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9929762039864671,
      "learning_rate": 7.714470960045565e-06,
      "loss": 0.6687,
      "step": 2097
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.61768735296764,
      "learning_rate": 7.712278936481992e-06,
      "loss": 0.4606,
      "step": 2098
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0090334206433311,
      "learning_rate": 7.710086174018871e-06,
      "loss": 0.6218,
      "step": 2099
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1210131154810046,
      "learning_rate": 7.707892673253572e-06,
      "loss": 0.6655,
      "step": 2100
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9515248636097394,
      "learning_rate": 7.705698434783666e-06,
      "loss": 0.5694,
      "step": 2101
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9654665071364131,
      "learning_rate": 7.703503459206922e-06,
      "loss": 0.7025,
      "step": 2102
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0099391509418814,
      "learning_rate": 7.701307747121314e-06,
      "loss": 0.6657,
      "step": 2103
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0808159811481284,
      "learning_rate": 7.699111299125015e-06,
      "loss": 0.6455,
      "step": 2104
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1068453224189645,
      "learning_rate": 7.696914115816395e-06,
      "loss": 0.7246,
      "step": 2105
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9860676801074832,
      "learning_rate": 7.69471619779403e-06,
      "loss": 0.6059,
      "step": 2106
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0564111640366705,
      "learning_rate": 7.692517545656691e-06,
      "loss": 0.6151,
      "step": 2107
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1464401318846351,
      "learning_rate": 7.690318160003356e-06,
      "loss": 0.6211,
      "step": 2108
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9964937363368471,
      "learning_rate": 7.688118041433192e-06,
      "loss": 0.5264,
      "step": 2109
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0636706486692744,
      "learning_rate": 7.685917190545576e-06,
      "loss": 0.5781,
      "step": 2110
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1369028503260206,
      "learning_rate": 7.683715607940078e-06,
      "loss": 0.6927,
      "step": 2111
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0010836095958955,
      "learning_rate": 7.681513294216476e-06,
      "loss": 0.5885,
      "step": 2112
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9882752742069714,
      "learning_rate": 7.679310249974732e-06,
      "loss": 0.6154,
      "step": 2113
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0545786223824967,
      "learning_rate": 7.677106475815021e-06,
      "loss": 0.6523,
      "step": 2114
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0700322278876584,
      "learning_rate": 7.674901972337712e-06,
      "loss": 0.626,
      "step": 2115
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.035182276123007,
      "learning_rate": 7.672696740143372e-06,
      "loss": 0.6767,
      "step": 2116
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9665535157453695,
      "learning_rate": 7.670490779832767e-06,
      "loss": 0.591,
      "step": 2117
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9659227906536961,
      "learning_rate": 7.668284092006859e-06,
      "loss": 0.6228,
      "step": 2118
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.047978084950697,
      "learning_rate": 7.666076677266813e-06,
      "loss": 0.6053,
      "step": 2119
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0133906562186499,
      "learning_rate": 7.66386853621399e-06,
      "loss": 0.6341,
      "step": 2120
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0814288767218916,
      "learning_rate": 7.661659669449948e-06,
      "loss": 0.6563,
      "step": 2121
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0052283508300457,
      "learning_rate": 7.659450077576444e-06,
      "loss": 0.5873,
      "step": 2122
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0509371560038792,
      "learning_rate": 7.657239761195428e-06,
      "loss": 0.6882,
      "step": 2123
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9998048570627717,
      "learning_rate": 7.655028720909057e-06,
      "loss": 0.5735,
      "step": 2124
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0348978674299782,
      "learning_rate": 7.652816957319674e-06,
      "loss": 0.6686,
      "step": 2125
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0167365285033512,
      "learning_rate": 7.650604471029825e-06,
      "loss": 0.5765,
      "step": 2126
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9954668110172916,
      "learning_rate": 7.648391262642257e-06,
      "loss": 0.6018,
      "step": 2127
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.037648377035903,
      "learning_rate": 7.646177332759906e-06,
      "loss": 0.5853,
      "step": 2128
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.996204503450735,
      "learning_rate": 7.643962681985904e-06,
      "loss": 0.6715,
      "step": 2129
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.050585723821499,
      "learning_rate": 7.641747310923588e-06,
      "loss": 0.6378,
      "step": 2130
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0216530161655883,
      "learning_rate": 7.639531220176484e-06,
      "loss": 0.6345,
      "step": 2131
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.012455957258552,
      "learning_rate": 7.637314410348315e-06,
      "loss": 0.5577,
      "step": 2132
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0420492122468314,
      "learning_rate": 7.635096882043006e-06,
      "loss": 0.6699,
      "step": 2133
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9964637629612332,
      "learning_rate": 7.632878635864666e-06,
      "loss": 0.6171,
      "step": 2134
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0273271574807576,
      "learning_rate": 7.630659672417613e-06,
      "loss": 0.6096,
      "step": 2135
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.025247235444467,
      "learning_rate": 7.628439992306349e-06,
      "loss": 0.6155,
      "step": 2136
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.028249892097847,
      "learning_rate": 7.626219596135578e-06,
      "loss": 0.5321,
      "step": 2137
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0311433362703426,
      "learning_rate": 7.623998484510197e-06,
      "loss": 0.6398,
      "step": 2138
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.024511871922928,
      "learning_rate": 7.621776658035298e-06,
      "loss": 0.6772,
      "step": 2139
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0152003532559413,
      "learning_rate": 7.619554117316165e-06,
      "loss": 0.6664,
      "step": 2140
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0245939333101743,
      "learning_rate": 7.617330862958287e-06,
      "loss": 0.6408,
      "step": 2141
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0574103834858695,
      "learning_rate": 7.615106895567331e-06,
      "loss": 0.6674,
      "step": 2142
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9822803377967566,
      "learning_rate": 7.612882215749172e-06,
      "loss": 0.5971,
      "step": 2143
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9864689838284217,
      "learning_rate": 7.6106568241098745e-06,
      "loss": 0.6234,
      "step": 2144
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9277234406275513,
      "learning_rate": 7.608430721255691e-06,
      "loss": 0.5964,
      "step": 2145
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.021101640579584,
      "learning_rate": 7.606203907793081e-06,
      "loss": 0.6495,
      "step": 2146
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.118582162934433,
      "learning_rate": 7.603976384328684e-06,
      "loss": 0.6569,
      "step": 2147
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.789696472848177,
      "learning_rate": 7.601748151469341e-06,
      "loss": 0.5066,
      "step": 2148
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0313553521648877,
      "learning_rate": 7.599519209822085e-06,
      "loss": 0.5704,
      "step": 2149
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0248594993172224,
      "learning_rate": 7.59728955999414e-06,
      "loss": 0.5842,
      "step": 2150
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.987348512270206,
      "learning_rate": 7.595059202592923e-06,
      "loss": 0.6458,
      "step": 2151
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9094226558474473,
      "learning_rate": 7.5928281382260474e-06,
      "loss": 0.5912,
      "step": 2152
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0300689109469887,
      "learning_rate": 7.590596367501314e-06,
      "loss": 0.6684,
      "step": 2153
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.029470695436444,
      "learning_rate": 7.58836389102672e-06,
      "loss": 0.57,
      "step": 2154
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0607020540061805,
      "learning_rate": 7.586130709410454e-06,
      "loss": 0.7084,
      "step": 2155
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9373987152064778,
      "learning_rate": 7.583896823260894e-06,
      "loss": 0.5209,
      "step": 2156
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.053382756930419,
      "learning_rate": 7.581662233186618e-06,
      "loss": 0.6058,
      "step": 2157
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0074022017538398,
      "learning_rate": 7.5794269397963814e-06,
      "loss": 0.658,
      "step": 2158
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0794665647891577,
      "learning_rate": 7.577190943699145e-06,
      "loss": 0.6272,
      "step": 2159
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.009622520294226,
      "learning_rate": 7.574954245504056e-06,
      "loss": 0.5897,
      "step": 2160
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9547244302119534,
      "learning_rate": 7.572716845820452e-06,
      "loss": 0.5928,
      "step": 2161
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9697951688845026,
      "learning_rate": 7.57047874525786e-06,
      "loss": 0.6137,
      "step": 2162
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.035037533934311,
      "learning_rate": 7.568239944426003e-06,
      "loss": 0.5882,
      "step": 2163
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9502285146991711,
      "learning_rate": 7.5660004439347916e-06,
      "loss": 0.6044,
      "step": 2164
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1698309555959372,
      "learning_rate": 7.563760244394325e-06,
      "loss": 0.6429,
      "step": 2165
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0025932875626726,
      "learning_rate": 7.5615193464149005e-06,
      "loss": 0.5801,
      "step": 2166
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.04058948095227,
      "learning_rate": 7.5592777506069946e-06,
      "loss": 0.5861,
      "step": 2167
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0129456428934478,
      "learning_rate": 7.557035457581284e-06,
      "loss": 0.6721,
      "step": 2168
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9555877351451456,
      "learning_rate": 7.5547924679486294e-06,
      "loss": 0.6082,
      "step": 2169
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0656725285070938,
      "learning_rate": 7.552548782320084e-06,
      "loss": 0.6366,
      "step": 2170
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0544649090672507,
      "learning_rate": 7.550304401306887e-06,
      "loss": 0.6519,
      "step": 2171
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0730712108810618,
      "learning_rate": 7.5480593255204725e-06,
      "loss": 0.5814,
      "step": 2172
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0240247141529004,
      "learning_rate": 7.545813555572461e-06,
      "loss": 0.6339,
      "step": 2173
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0772674605091832,
      "learning_rate": 7.54356709207466e-06,
      "loss": 0.6499,
      "step": 2174
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9723631317521624,
      "learning_rate": 7.5413199356390695e-06,
      "loss": 0.5482,
      "step": 2175
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.058529015675711,
      "learning_rate": 7.539072086877877e-06,
      "loss": 0.5695,
      "step": 2176
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9894891115211268,
      "learning_rate": 7.536823546403458e-06,
      "loss": 0.6019,
      "step": 2177
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0419331223481323,
      "learning_rate": 7.534574314828376e-06,
      "loss": 0.6225,
      "step": 2178
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0840864420534446,
      "learning_rate": 7.532324392765387e-06,
      "loss": 0.6268,
      "step": 2179
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0200666520841024,
      "learning_rate": 7.530073780827427e-06,
      "loss": 0.6167,
      "step": 2180
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0658063472007522,
      "learning_rate": 7.527822479627629e-06,
      "loss": 0.5921,
      "step": 2181
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0937266938289099,
      "learning_rate": 7.525570489779307e-06,
      "loss": 0.5996,
      "step": 2182
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9916347833117392,
      "learning_rate": 7.523317811895965e-06,
      "loss": 0.5871,
      "step": 2183
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1270732830847814,
      "learning_rate": 7.5210644465912975e-06,
      "loss": 0.645,
      "step": 2184
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.042014238832267,
      "learning_rate": 7.518810394479179e-06,
      "loss": 0.659,
      "step": 2185
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9909956008998193,
      "learning_rate": 7.516555656173678e-06,
      "loss": 0.5965,
      "step": 2186
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7324249441449703,
      "learning_rate": 7.514300232289049e-06,
      "loss": 0.5322,
      "step": 2187
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9927544419725202,
      "learning_rate": 7.512044123439728e-06,
      "loss": 0.6095,
      "step": 2188
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0407282337140056,
      "learning_rate": 7.509787330240342e-06,
      "loss": 0.5544,
      "step": 2189
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9731759571235095,
      "learning_rate": 7.507529853305706e-06,
      "loss": 0.5713,
      "step": 2190
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.937500977295957,
      "learning_rate": 7.505271693250817e-06,
      "loss": 0.6045,
      "step": 2191
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9797890327990393,
      "learning_rate": 7.50301285069086e-06,
      "loss": 0.6318,
      "step": 2192
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.132217045604059,
      "learning_rate": 7.500753326241208e-06,
      "loss": 0.6241,
      "step": 2193
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2278289441722754,
      "learning_rate": 7.498493120517415e-06,
      "loss": 0.6196,
      "step": 2194
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9990125360783698,
      "learning_rate": 7.4962322341352256e-06,
      "loss": 0.6263,
      "step": 2195
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9806168080037504,
      "learning_rate": 7.493970667710566e-06,
      "loss": 0.6257,
      "step": 2196
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9699999360034107,
      "learning_rate": 7.491708421859549e-06,
      "loss": 0.614,
      "step": 2197
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9865805033343138,
      "learning_rate": 7.489445497198475e-06,
      "loss": 0.6396,
      "step": 2198
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9982575792646158,
      "learning_rate": 7.487181894343826e-06,
      "loss": 0.6293,
      "step": 2199
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0809635183991693,
      "learning_rate": 7.484917613912267e-06,
      "loss": 0.6394,
      "step": 2200
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9647106534349793,
      "learning_rate": 7.482652656520655e-06,
      "loss": 0.6543,
      "step": 2201
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0943693656857505,
      "learning_rate": 7.480387022786023e-06,
      "loss": 0.628,
      "step": 2202
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.035142011098132,
      "learning_rate": 7.478120713325595e-06,
      "loss": 0.6506,
      "step": 2203
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0651517783072433,
      "learning_rate": 7.4758537287567745e-06,
      "loss": 0.604,
      "step": 2204
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9720933860867106,
      "learning_rate": 7.4735860696971505e-06,
      "loss": 0.5954,
      "step": 2205
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0275337000091138,
      "learning_rate": 7.471317736764497e-06,
      "loss": 0.6617,
      "step": 2206
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0305658457450282,
      "learning_rate": 7.46904873057677e-06,
      "loss": 0.6363,
      "step": 2207
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.937168114333817,
      "learning_rate": 7.466779051752107e-06,
      "loss": 0.5311,
      "step": 2208
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0143455204759564,
      "learning_rate": 7.464508700908836e-06,
      "loss": 0.569,
      "step": 2209
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.005320338403505,
      "learning_rate": 7.4622376786654596e-06,
      "loss": 0.5877,
      "step": 2210
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8966566395654978,
      "learning_rate": 7.459965985640665e-06,
      "loss": 0.5638,
      "step": 2211
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0295116094114811,
      "learning_rate": 7.457693622453329e-06,
      "loss": 0.5714,
      "step": 2212
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.992821429488196,
      "learning_rate": 7.455420589722504e-06,
      "loss": 0.6333,
      "step": 2213
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6837528960501983,
      "learning_rate": 7.453146888067424e-06,
      "loss": 0.4696,
      "step": 2214
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0439748577069399,
      "learning_rate": 7.450872518107511e-06,
      "loss": 0.6377,
      "step": 2215
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.028459715489088,
      "learning_rate": 7.448597480462366e-06,
      "loss": 0.6115,
      "step": 2216
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0620909463080204,
      "learning_rate": 7.446321775751772e-06,
      "loss": 0.6115,
      "step": 2217
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.013921568198763,
      "learning_rate": 7.444045404595692e-06,
      "loss": 0.5665,
      "step": 2218
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9817164372032654,
      "learning_rate": 7.441768367614274e-06,
      "loss": 0.5139,
      "step": 2219
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9564655080395958,
      "learning_rate": 7.439490665427844e-06,
      "loss": 0.569,
      "step": 2220
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1035241164050016,
      "learning_rate": 7.437212298656914e-06,
      "loss": 0.6226,
      "step": 2221
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0279100767605824,
      "learning_rate": 7.434933267922168e-06,
      "loss": 0.6381,
      "step": 2222
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0704208348696829,
      "learning_rate": 7.432653573844483e-06,
      "loss": 0.6738,
      "step": 2223
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.008501522545485,
      "learning_rate": 7.430373217044909e-06,
      "loss": 0.6101,
      "step": 2224
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9466910912366296,
      "learning_rate": 7.428092198144674e-06,
      "loss": 0.5269,
      "step": 2225
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.680135375678601,
      "learning_rate": 7.425810517765196e-06,
      "loss": 0.5063,
      "step": 2226
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9697735314082948,
      "learning_rate": 7.423528176528063e-06,
      "loss": 0.634,
      "step": 2227
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1256424298162675,
      "learning_rate": 7.4212451750550515e-06,
      "loss": 0.663,
      "step": 2228
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1263628726647679,
      "learning_rate": 7.418961513968111e-06,
      "loss": 0.6629,
      "step": 2229
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0739052856172258,
      "learning_rate": 7.416677193889376e-06,
      "loss": 0.6265,
      "step": 2230
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0494167277395259,
      "learning_rate": 7.4143922154411576e-06,
      "loss": 0.6321,
      "step": 2231
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9797237111807897,
      "learning_rate": 7.412106579245945e-06,
      "loss": 0.6439,
      "step": 2232
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9536068776665408,
      "learning_rate": 7.409820285926411e-06,
      "loss": 0.6505,
      "step": 2233
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9168718996040589,
      "learning_rate": 7.407533336105404e-06,
      "loss": 0.6062,
      "step": 2234
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0781537061886268,
      "learning_rate": 7.405245730405954e-06,
      "loss": 0.596,
      "step": 2235
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9824668367239546,
      "learning_rate": 7.402957469451263e-06,
      "loss": 0.5922,
      "step": 2236
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0503225741349496,
      "learning_rate": 7.4006685538647214e-06,
      "loss": 0.6386,
      "step": 2237
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9934324674450471,
      "learning_rate": 7.3983789842698894e-06,
      "loss": 0.5646,
      "step": 2238
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0480707553360378,
      "learning_rate": 7.3960887612905116e-06,
      "loss": 0.6626,
      "step": 2239
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9982594008088528,
      "learning_rate": 7.393797885550506e-06,
      "loss": 0.6133,
      "step": 2240
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.023618108743904,
      "learning_rate": 7.391506357673972e-06,
      "loss": 0.655,
      "step": 2241
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.93858795250245,
      "learning_rate": 7.389214178285182e-06,
      "loss": 0.547,
      "step": 2242
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9802790735934123,
      "learning_rate": 7.386921348008592e-06,
      "loss": 0.5308,
      "step": 2243
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1341996182268386,
      "learning_rate": 7.384627867468831e-06,
      "loss": 0.6223,
      "step": 2244
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0932148014706684,
      "learning_rate": 7.382333737290708e-06,
      "loss": 0.5927,
      "step": 2245
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0139605520628223,
      "learning_rate": 7.3800389580992025e-06,
      "loss": 0.6374,
      "step": 2246
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9016158552972428,
      "learning_rate": 7.37774353051948e-06,
      "loss": 0.588,
      "step": 2247
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0972601640743878,
      "learning_rate": 7.375447455176877e-06,
      "loss": 0.6535,
      "step": 2248
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0268348104925817,
      "learning_rate": 7.373150732696907e-06,
      "loss": 0.6244,
      "step": 2249
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9727916070992491,
      "learning_rate": 7.370853363705261e-06,
      "loss": 0.5854,
      "step": 2250
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9953585514048059,
      "learning_rate": 7.3685553488278064e-06,
      "loss": 0.6082,
      "step": 2251
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0536536084673818,
      "learning_rate": 7.366256688690585e-06,
      "loss": 0.582,
      "step": 2252
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0493702495187964,
      "learning_rate": 7.363957383919815e-06,
      "loss": 0.5751,
      "step": 2253
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.174774506365941,
      "learning_rate": 7.361657435141892e-06,
      "loss": 0.6766,
      "step": 2254
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0670703164117528,
      "learning_rate": 7.3593568429833825e-06,
      "loss": 0.6564,
      "step": 2255
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9755258278824886,
      "learning_rate": 7.357055608071034e-06,
      "loss": 0.5662,
      "step": 2256
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0584981662858932,
      "learning_rate": 7.354753731031765e-06,
      "loss": 0.6651,
      "step": 2257
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.121569052863542,
      "learning_rate": 7.352451212492671e-06,
      "loss": 0.7275,
      "step": 2258
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0476673966220134,
      "learning_rate": 7.350148053081021e-06,
      "loss": 0.6499,
      "step": 2259
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1155294444916661,
      "learning_rate": 7.3478442534242565e-06,
      "loss": 0.6515,
      "step": 2260
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0022557002650023,
      "learning_rate": 7.345539814150002e-06,
      "loss": 0.6748,
      "step": 2261
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0028660692009115,
      "learning_rate": 7.3432347358860445e-06,
      "loss": 0.6944,
      "step": 2262
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0020792931902724,
      "learning_rate": 7.340929019260356e-06,
      "loss": 0.6206,
      "step": 2263
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9825732768808498,
      "learning_rate": 7.338622664901073e-06,
      "loss": 0.5799,
      "step": 2264
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1151849099475757,
      "learning_rate": 7.336315673436512e-06,
      "loss": 0.6722,
      "step": 2265
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.015357200135573,
      "learning_rate": 7.33400804549516e-06,
      "loss": 0.5862,
      "step": 2266
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0559706548634678,
      "learning_rate": 7.331699781705679e-06,
      "loss": 0.6778,
      "step": 2267
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0722907487227722,
      "learning_rate": 7.329390882696904e-06,
      "loss": 0.702,
      "step": 2268
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0439197737276733,
      "learning_rate": 7.327081349097842e-06,
      "loss": 0.589,
      "step": 2269
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0110139710432422,
      "learning_rate": 7.324771181537676e-06,
      "loss": 0.5321,
      "step": 2270
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1525438040790654,
      "learning_rate": 7.322460380645755e-06,
      "loss": 0.6721,
      "step": 2271
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.142277853267506,
      "learning_rate": 7.32014894705161e-06,
      "loss": 0.5394,
      "step": 2272
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7335970205689721,
      "learning_rate": 7.317836881384934e-06,
      "loss": 0.5117,
      "step": 2273
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9866075911196817,
      "learning_rate": 7.315524184275602e-06,
      "loss": 0.6447,
      "step": 2274
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.917340397854701,
      "learning_rate": 7.313210856353653e-06,
      "loss": 0.5889,
      "step": 2275
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9569878512424859,
      "learning_rate": 7.310896898249303e-06,
      "loss": 0.613,
      "step": 2276
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.088564601700148,
      "learning_rate": 7.3085823105929385e-06,
      "loss": 0.6819,
      "step": 2277
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9899626356098077,
      "learning_rate": 7.306267094015116e-06,
      "loss": 0.6117,
      "step": 2278
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9039913760438909,
      "learning_rate": 7.303951249146563e-06,
      "loss": 0.51,
      "step": 2279
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0549402674171047,
      "learning_rate": 7.301634776618183e-06,
      "loss": 0.7116,
      "step": 2280
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9035697292294355,
      "learning_rate": 7.299317677061043e-06,
      "loss": 0.5377,
      "step": 2281
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9895724294011012,
      "learning_rate": 7.2969999511063875e-06,
      "loss": 0.5844,
      "step": 2282
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9455445325187327,
      "learning_rate": 7.294681599385629e-06,
      "loss": 0.5867,
      "step": 2283
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.064716577101577,
      "learning_rate": 7.292362622530347e-06,
      "loss": 0.6012,
      "step": 2284
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0295923136773983,
      "learning_rate": 7.2900430211723e-06,
      "loss": 0.6681,
      "step": 2285
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.118860441117518,
      "learning_rate": 7.287722795943407e-06,
      "loss": 0.6798,
      "step": 2286
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0525432332213427,
      "learning_rate": 7.285401947475764e-06,
      "loss": 0.5869,
      "step": 2287
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.975589404865857,
      "learning_rate": 7.283080476401634e-06,
      "loss": 0.5738,
      "step": 2288
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0006130130616144,
      "learning_rate": 7.280758383353447e-06,
      "loss": 0.6825,
      "step": 2289
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0029200734741783,
      "learning_rate": 7.278435668963807e-06,
      "loss": 0.5513,
      "step": 2290
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.043338981883033,
      "learning_rate": 7.276112333865485e-06,
      "loss": 0.6003,
      "step": 2291
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0401236749370717,
      "learning_rate": 7.273788378691422e-06,
      "loss": 0.6626,
      "step": 2292
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6890032070859601,
      "learning_rate": 7.271463804074728e-06,
      "loss": 0.4936,
      "step": 2293
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9756682529932238,
      "learning_rate": 7.26913861064868e-06,
      "loss": 0.5768,
      "step": 2294
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9935930566997946,
      "learning_rate": 7.2668127990467266e-06,
      "loss": 0.6365,
      "step": 2295
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0855618397764537,
      "learning_rate": 7.26448636990248e-06,
      "loss": 0.6834,
      "step": 2296
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.066300092063712,
      "learning_rate": 7.262159323849725e-06,
      "loss": 0.6713,
      "step": 2297
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.029760753288409,
      "learning_rate": 7.259831661522415e-06,
      "loss": 0.6249,
      "step": 2298
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7170706348975647,
      "learning_rate": 7.257503383554668e-06,
      "loss": 0.4782,
      "step": 2299
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.038851938883799,
      "learning_rate": 7.25517449058077e-06,
      "loss": 0.6068,
      "step": 2300
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9479056382578903,
      "learning_rate": 7.252844983235177e-06,
      "loss": 0.547,
      "step": 2301
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.044590042440595,
      "learning_rate": 7.250514862152509e-06,
      "loss": 0.5581,
      "step": 2302
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0397548846106945,
      "learning_rate": 7.248184127967557e-06,
      "loss": 0.6095,
      "step": 2303
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9838879368610903,
      "learning_rate": 7.245852781315278e-06,
      "loss": 0.6542,
      "step": 2304
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9953804012244276,
      "learning_rate": 7.243520822830794e-06,
      "loss": 0.6014,
      "step": 2305
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9967313948299399,
      "learning_rate": 7.241188253149395e-06,
      "loss": 0.5697,
      "step": 2306
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0202032628876836,
      "learning_rate": 7.238855072906537e-06,
      "loss": 0.6795,
      "step": 2307
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6526329759037361,
      "learning_rate": 7.236521282737842e-06,
      "loss": 0.4914,
      "step": 2308
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9928526682224015,
      "learning_rate": 7.2341868832791e-06,
      "loss": 0.5976,
      "step": 2309
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0281525604653499,
      "learning_rate": 7.231851875166266e-06,
      "loss": 0.5862,
      "step": 2310
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9643959959834605,
      "learning_rate": 7.229516259035458e-06,
      "loss": 0.5179,
      "step": 2311
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0602900575715508,
      "learning_rate": 7.227180035522966e-06,
      "loss": 0.6703,
      "step": 2312
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0425834460317396,
      "learning_rate": 7.224843205265239e-06,
      "loss": 0.5748,
      "step": 2313
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0754807284672718,
      "learning_rate": 7.222505768898894e-06,
      "loss": 0.62,
      "step": 2314
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6700798805359844,
      "learning_rate": 7.220167727060714e-06,
      "loss": 0.5045,
      "step": 2315
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0724473126007261,
      "learning_rate": 7.217829080387648e-06,
      "loss": 0.6219,
      "step": 2316
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0305351586745446,
      "learning_rate": 7.2154898295168045e-06,
      "loss": 0.6645,
      "step": 2317
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0803282059123502,
      "learning_rate": 7.21314997508546e-06,
      "loss": 0.645,
      "step": 2318
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9930170025838548,
      "learning_rate": 7.210809517731057e-06,
      "loss": 0.5948,
      "step": 2319
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0191853221584506,
      "learning_rate": 7.2084684580912e-06,
      "loss": 0.5696,
      "step": 2320
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0341977117673886,
      "learning_rate": 7.206126796803659e-06,
      "loss": 0.6438,
      "step": 2321
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.176069048803107,
      "learning_rate": 7.203784534506364e-06,
      "loss": 0.6832,
      "step": 2322
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9895004832667399,
      "learning_rate": 7.201441671837417e-06,
      "loss": 0.6319,
      "step": 2323
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9318644353326777,
      "learning_rate": 7.199098209435073e-06,
      "loss": 0.6052,
      "step": 2324
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6971038331340554,
      "learning_rate": 7.19675414793776e-06,
      "loss": 0.5093,
      "step": 2325
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0241133214838547,
      "learning_rate": 7.194409487984063e-06,
      "loss": 0.5799,
      "step": 2326
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0192936716129128,
      "learning_rate": 7.192064230212733e-06,
      "loss": 0.5784,
      "step": 2327
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.128273949257954,
      "learning_rate": 7.189718375262681e-06,
      "loss": 0.6741,
      "step": 2328
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0169799227839607,
      "learning_rate": 7.1873719237729835e-06,
      "loss": 0.6721,
      "step": 2329
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0970038244967593,
      "learning_rate": 7.18502487638288e-06,
      "loss": 0.6683,
      "step": 2330
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0514332467639362,
      "learning_rate": 7.182677233731768e-06,
      "loss": 0.6985,
      "step": 2331
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9860533829956002,
      "learning_rate": 7.1803289964592115e-06,
      "loss": 0.6123,
      "step": 2332
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9415702344101353,
      "learning_rate": 7.177980165204935e-06,
      "loss": 0.6253,
      "step": 2333
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.981775889655673,
      "learning_rate": 7.1756307406088275e-06,
      "loss": 0.67,
      "step": 2334
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0172659585057755,
      "learning_rate": 7.173280723310932e-06,
      "loss": 0.6384,
      "step": 2335
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0760415073559442,
      "learning_rate": 7.170930113951462e-06,
      "loss": 0.6651,
      "step": 2336
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1222808027226474,
      "learning_rate": 7.168578913170788e-06,
      "loss": 0.6946,
      "step": 2337
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.03926099578737,
      "learning_rate": 7.166227121609439e-06,
      "loss": 0.608,
      "step": 2338
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.951246740193479,
      "learning_rate": 7.16387473990811e-06,
      "loss": 0.5662,
      "step": 2339
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9764330286390912,
      "learning_rate": 7.1615217687076555e-06,
      "loss": 0.5922,
      "step": 2340
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0536564721478936,
      "learning_rate": 7.159168208649086e-06,
      "loss": 0.659,
      "step": 2341
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9477427708930591,
      "learning_rate": 7.15681406037358e-06,
      "loss": 0.4902,
      "step": 2342
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0462160185072122,
      "learning_rate": 7.154459324522474e-06,
      "loss": 0.6021,
      "step": 2343
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0157462260641292,
      "learning_rate": 7.152104001737254e-06,
      "loss": 0.6139,
      "step": 2344
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0722983663725614,
      "learning_rate": 7.149748092659585e-06,
      "loss": 0.7137,
      "step": 2345
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.940472714708764,
      "learning_rate": 7.147391597931277e-06,
      "loss": 0.5548,
      "step": 2346
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7476040474905417,
      "learning_rate": 7.145034518194304e-06,
      "loss": 0.5133,
      "step": 2347
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6878141988429427,
      "learning_rate": 7.1426768540908e-06,
      "loss": 0.4676,
      "step": 2348
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0651734573065943,
      "learning_rate": 7.140318606263058e-06,
      "loss": 0.635,
      "step": 2349
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0337203632659062,
      "learning_rate": 7.137959775353529e-06,
      "loss": 0.5942,
      "step": 2350
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0728951921094223,
      "learning_rate": 7.135600362004824e-06,
      "loss": 0.7703,
      "step": 2351
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1055960475929942,
      "learning_rate": 7.133240366859713e-06,
      "loss": 0.6393,
      "step": 2352
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9667227418057721,
      "learning_rate": 7.130879790561122e-06,
      "loss": 0.6511,
      "step": 2353
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7663740701338713,
      "learning_rate": 7.128518633752139e-06,
      "loss": 0.4795,
      "step": 2354
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8530005604396605,
      "learning_rate": 7.126156897076005e-06,
      "loss": 0.5593,
      "step": 2355
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0781745990924998,
      "learning_rate": 7.123794581176127e-06,
      "loss": 0.6442,
      "step": 2356
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6999974646325233,
      "learning_rate": 7.121431686696061e-06,
      "loss": 0.4698,
      "step": 2357
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0579815294938772,
      "learning_rate": 7.119068214279525e-06,
      "loss": 0.5452,
      "step": 2358
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1276000233987107,
      "learning_rate": 7.116704164570398e-06,
      "loss": 0.6478,
      "step": 2359
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.025832808921346,
      "learning_rate": 7.114339538212707e-06,
      "loss": 0.7092,
      "step": 2360
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9989034423337582,
      "learning_rate": 7.111974335850644e-06,
      "loss": 0.5454,
      "step": 2361
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6376899780931057,
      "learning_rate": 7.1096085581285555e-06,
      "loss": 0.518,
      "step": 2362
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0496611821914672,
      "learning_rate": 7.1072422056909426e-06,
      "loss": 0.5956,
      "step": 2363
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0888408750761576,
      "learning_rate": 7.104875279182468e-06,
      "loss": 0.7206,
      "step": 2364
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.993143639084888,
      "learning_rate": 7.102507779247947e-06,
      "loss": 0.5604,
      "step": 2365
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.056484529223308,
      "learning_rate": 7.100139706532347e-06,
      "loss": 0.7084,
      "step": 2366
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1199644604555272,
      "learning_rate": 7.097771061680805e-06,
      "loss": 0.7009,
      "step": 2367
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0338586201100615,
      "learning_rate": 7.095401845338598e-06,
      "loss": 0.612,
      "step": 2368
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0110404456672186,
      "learning_rate": 7.093032058151168e-06,
      "loss": 0.5755,
      "step": 2369
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7066641201528922,
      "learning_rate": 7.090661700764112e-06,
      "loss": 0.5029,
      "step": 2370
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0042353068904448,
      "learning_rate": 7.088290773823177e-06,
      "loss": 0.6437,
      "step": 2371
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9138896162531205,
      "learning_rate": 7.085919277974274e-06,
      "loss": 0.4887,
      "step": 2372
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9934873959798128,
      "learning_rate": 7.083547213863458e-06,
      "loss": 0.6607,
      "step": 2373
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.087857260483321,
      "learning_rate": 7.0811745821369495e-06,
      "loss": 0.6244,
      "step": 2374
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9438686380348272,
      "learning_rate": 7.0788013834411165e-06,
      "loss": 0.529,
      "step": 2375
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.092078836695521,
      "learning_rate": 7.0764276184224845e-06,
      "loss": 0.6282,
      "step": 2376
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.031629186140058,
      "learning_rate": 7.07405328772773e-06,
      "loss": 0.6555,
      "step": 2377
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9780903942633322,
      "learning_rate": 7.071678392003691e-06,
      "loss": 0.6471,
      "step": 2378
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0456899084735127,
      "learning_rate": 7.069302931897352e-06,
      "loss": 0.6171,
      "step": 2379
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9588691913296155,
      "learning_rate": 7.0669269080558515e-06,
      "loss": 0.5885,
      "step": 2380
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0385791001888502,
      "learning_rate": 7.064550321126489e-06,
      "loss": 0.6038,
      "step": 2381
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6861688099958749,
      "learning_rate": 7.0621731717567055e-06,
      "loss": 0.4653,
      "step": 2382
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0446725209149121,
      "learning_rate": 7.059795460594109e-06,
      "loss": 0.6505,
      "step": 2383
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9220015695075581,
      "learning_rate": 7.057417188286449e-06,
      "loss": 0.5945,
      "step": 2384
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9625529045516624,
      "learning_rate": 7.0550383554816345e-06,
      "loss": 0.6274,
      "step": 2385
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9439610736644684,
      "learning_rate": 7.052658962827724e-06,
      "loss": 0.6151,
      "step": 2386
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0591847510302013,
      "learning_rate": 7.050279010972932e-06,
      "loss": 0.6301,
      "step": 2387
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0462506749042528,
      "learning_rate": 7.047898500565619e-06,
      "loss": 0.5891,
      "step": 2388
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6435080393858074,
      "learning_rate": 7.045517432254304e-06,
      "loss": 0.444,
      "step": 2389
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9358529603801624,
      "learning_rate": 7.043135806687655e-06,
      "loss": 0.6055,
      "step": 2390
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0585678958383646,
      "learning_rate": 7.040753624514494e-06,
      "loss": 0.6812,
      "step": 2391
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9806199373411311,
      "learning_rate": 7.038370886383793e-06,
      "loss": 0.5832,
      "step": 2392
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9960809853376439,
      "learning_rate": 7.035987592944672e-06,
      "loss": 0.5949,
      "step": 2393
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9371568186470512,
      "learning_rate": 7.03360374484641e-06,
      "loss": 0.6317,
      "step": 2394
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0861361535105436,
      "learning_rate": 7.031219342738431e-06,
      "loss": 0.6498,
      "step": 2395
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9834551502830807,
      "learning_rate": 7.028834387270311e-06,
      "loss": 0.5666,
      "step": 2396
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0212889032906192,
      "learning_rate": 7.02644887909178e-06,
      "loss": 0.5985,
      "step": 2397
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.005482585269343,
      "learning_rate": 7.024062818852716e-06,
      "loss": 0.5555,
      "step": 2398
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9473593003588137,
      "learning_rate": 7.021676207203145e-06,
      "loss": 0.5761,
      "step": 2399
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0196190720589509,
      "learning_rate": 7.019289044793247e-06,
      "loss": 0.5582,
      "step": 2400
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.129195407577771,
      "learning_rate": 7.016901332273352e-06,
      "loss": 0.7053,
      "step": 2401
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0365452022713293,
      "learning_rate": 7.014513070293938e-06,
      "loss": 0.6153,
      "step": 2402
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0545770627082933,
      "learning_rate": 7.012124259505633e-06,
      "loss": 0.5732,
      "step": 2403
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9899008600664705,
      "learning_rate": 7.0097349005592145e-06,
      "loss": 0.571,
      "step": 2404
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9701003512959118,
      "learning_rate": 7.007344994105612e-06,
      "loss": 0.5922,
      "step": 2405
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0777204599898036,
      "learning_rate": 7.004954540795899e-06,
      "loss": 0.6939,
      "step": 2406
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9446134143788822,
      "learning_rate": 7.002563541281302e-06,
      "loss": 0.6094,
      "step": 2407
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.970206267487964,
      "learning_rate": 7.000171996213196e-06,
      "loss": 0.5696,
      "step": 2408
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.027960703484031,
      "learning_rate": 6.997779906243103e-06,
      "loss": 0.5221,
      "step": 2409
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0585753660298387,
      "learning_rate": 6.995387272022695e-06,
      "loss": 0.5642,
      "step": 2410
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0286105451501615,
      "learning_rate": 6.99299409420379e-06,
      "loss": 0.6874,
      "step": 2411
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8940586357595608,
      "learning_rate": 6.9906003734383565e-06,
      "loss": 0.5707,
      "step": 2412
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.003678956389507,
      "learning_rate": 6.98820611037851e-06,
      "loss": 0.674,
      "step": 2413
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9909017986769085,
      "learning_rate": 6.985811305676515e-06,
      "loss": 0.6255,
      "step": 2414
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0267431188087748,
      "learning_rate": 6.98341595998478e-06,
      "loss": 0.5667,
      "step": 2415
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9549869812343352,
      "learning_rate": 6.981020073955866e-06,
      "loss": 0.5229,
      "step": 2416
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0405645009970432,
      "learning_rate": 6.978623648242474e-06,
      "loss": 0.6078,
      "step": 2417
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0266299686590368,
      "learning_rate": 6.9762266834974605e-06,
      "loss": 0.6601,
      "step": 2418
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0920923884683573,
      "learning_rate": 6.973829180373823e-06,
      "loss": 0.582,
      "step": 2419
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0210098897673845,
      "learning_rate": 6.971431139524709e-06,
      "loss": 0.6358,
      "step": 2420
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0695916584153915,
      "learning_rate": 6.96903256160341e-06,
      "loss": 0.5862,
      "step": 2421
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9852205104796034,
      "learning_rate": 6.966633447263362e-06,
      "loss": 0.6132,
      "step": 2422
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9778435179856887,
      "learning_rate": 6.964233797158155e-06,
      "loss": 0.5942,
      "step": 2423
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.05774224895171,
      "learning_rate": 6.961833611941515e-06,
      "loss": 0.5974,
      "step": 2424
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1036777854551643,
      "learning_rate": 6.959432892267324e-06,
      "loss": 0.5907,
      "step": 2425
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0125481852489482,
      "learning_rate": 6.957031638789598e-06,
      "loss": 0.5499,
      "step": 2426
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6909967251649658,
      "learning_rate": 6.954629852162509e-06,
      "loss": 0.4617,
      "step": 2427
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0469551907163064,
      "learning_rate": 6.952227533040369e-06,
      "loss": 0.6015,
      "step": 2428
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0748351567085235,
      "learning_rate": 6.949824682077635e-06,
      "loss": 0.6474,
      "step": 2429
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9342974819635594,
      "learning_rate": 6.947421299928909e-06,
      "loss": 0.5626,
      "step": 2430
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0208567590937923,
      "learning_rate": 6.945017387248942e-06,
      "loss": 0.5843,
      "step": 2431
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9491646650236966,
      "learning_rate": 6.942612944692624e-06,
      "loss": 0.5344,
      "step": 2432
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0033886204305744,
      "learning_rate": 6.940207972914989e-06,
      "loss": 0.5614,
      "step": 2433
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0115612725130536,
      "learning_rate": 6.9378024725712225e-06,
      "loss": 0.5595,
      "step": 2434
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9533654936049077,
      "learning_rate": 6.935396444316646e-06,
      "loss": 0.5382,
      "step": 2435
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6829490251946648,
      "learning_rate": 6.9329898888067295e-06,
      "loss": 0.4871,
      "step": 2436
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7897422233734538,
      "learning_rate": 6.930582806697082e-06,
      "loss": 0.4901,
      "step": 2437
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0770851751663981,
      "learning_rate": 6.928175198643463e-06,
      "loss": 0.6282,
      "step": 2438
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9399303558611184,
      "learning_rate": 6.9257670653017674e-06,
      "loss": 0.5918,
      "step": 2439
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9824994897873479,
      "learning_rate": 6.92335840732804e-06,
      "loss": 0.5416,
      "step": 2440
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.008968292114502,
      "learning_rate": 6.9209492253784664e-06,
      "loss": 0.6124,
      "step": 2441
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0217020662128589,
      "learning_rate": 6.91853952010937e-06,
      "loss": 0.6114,
      "step": 2442
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0667125506308615,
      "learning_rate": 6.916129292177225e-06,
      "loss": 0.7157,
      "step": 2443
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.035584760366978,
      "learning_rate": 6.913718542238642e-06,
      "loss": 0.6149,
      "step": 2444
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.073085884584207,
      "learning_rate": 6.911307270950376e-06,
      "loss": 0.5777,
      "step": 2445
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0086918945637111,
      "learning_rate": 6.908895478969324e-06,
      "loss": 0.6354,
      "step": 2446
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0823716969148602,
      "learning_rate": 6.906483166952526e-06,
      "loss": 0.6626,
      "step": 2447
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.996536880399445,
      "learning_rate": 6.904070335557158e-06,
      "loss": 0.6027,
      "step": 2448
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9309828118928946,
      "learning_rate": 6.9016569854405466e-06,
      "loss": 0.5478,
      "step": 2449
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9216296862896817,
      "learning_rate": 6.899243117260153e-06,
      "loss": 0.568,
      "step": 2450
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.003238865979167,
      "learning_rate": 6.896828731673579e-06,
      "loss": 0.6199,
      "step": 2451
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.000618321119245,
      "learning_rate": 6.894413829338576e-06,
      "loss": 0.66,
      "step": 2452
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9693475378799835,
      "learning_rate": 6.891998410913021e-06,
      "loss": 0.6286,
      "step": 2453
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.029799804230898,
      "learning_rate": 6.88958247705495e-06,
      "loss": 0.6492,
      "step": 2454
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.106307507461913,
      "learning_rate": 6.887166028422524e-06,
      "loss": 0.5799,
      "step": 2455
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0998126940588364,
      "learning_rate": 6.884749065674051e-06,
      "loss": 0.6056,
      "step": 2456
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9931039313121502,
      "learning_rate": 6.88233158946798e-06,
      "loss": 0.5835,
      "step": 2457
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0717356054480864,
      "learning_rate": 6.879913600462898e-06,
      "loss": 0.6554,
      "step": 2458
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0016209251132813,
      "learning_rate": 6.87749509931753e-06,
      "loss": 0.5684,
      "step": 2459
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0665726340574677,
      "learning_rate": 6.875076086690744e-06,
      "loss": 0.6602,
      "step": 2460
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9987147061928905,
      "learning_rate": 6.8726565632415445e-06,
      "loss": 0.5686,
      "step": 2461
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9347817546836513,
      "learning_rate": 6.8702365296290775e-06,
      "loss": 0.6116,
      "step": 2462
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9682285847422266,
      "learning_rate": 6.867815986512627e-06,
      "loss": 0.5283,
      "step": 2463
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.041555614335137,
      "learning_rate": 6.865394934551613e-06,
      "loss": 0.6581,
      "step": 2464
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9995520006877903,
      "learning_rate": 6.862973374405601e-06,
      "loss": 0.5565,
      "step": 2465
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9553043661881615,
      "learning_rate": 6.860551306734289e-06,
      "loss": 0.6583,
      "step": 2466
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0315210481722183,
      "learning_rate": 6.858128732197513e-06,
      "loss": 0.7132,
      "step": 2467
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9846072308797008,
      "learning_rate": 6.855705651455252e-06,
      "loss": 0.6106,
      "step": 2468
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0811711598343194,
      "learning_rate": 6.853282065167618e-06,
      "loss": 0.6422,
      "step": 2469
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0907309562695893,
      "learning_rate": 6.850857973994865e-06,
      "loss": 0.6878,
      "step": 2470
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.952879298508714,
      "learning_rate": 6.8484333785973786e-06,
      "loss": 0.61,
      "step": 2471
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.058760380792588,
      "learning_rate": 6.846008279635688e-06,
      "loss": 0.6212,
      "step": 2472
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0080977940179607,
      "learning_rate": 6.8435826777704575e-06,
      "loss": 0.5835,
      "step": 2473
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0899744275175063,
      "learning_rate": 6.841156573662486e-06,
      "loss": 0.6517,
      "step": 2474
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0085032895336206,
      "learning_rate": 6.8387299679727125e-06,
      "loss": 0.6277,
      "step": 2475
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1161474652349777,
      "learning_rate": 6.836302861362211e-06,
      "loss": 0.6935,
      "step": 2476
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9561434316333456,
      "learning_rate": 6.8338752544921915e-06,
      "loss": 0.584,
      "step": 2477
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0369621984465147,
      "learning_rate": 6.831447148024002e-06,
      "loss": 0.6416,
      "step": 2478
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9831594166896802,
      "learning_rate": 6.829018542619125e-06,
      "loss": 0.6787,
      "step": 2479
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8515645803453231,
      "learning_rate": 6.82658943893918e-06,
      "loss": 0.4987,
      "step": 2480
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0602170636627577,
      "learning_rate": 6.824159837645921e-06,
      "loss": 0.5948,
      "step": 2481
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1235667182021651,
      "learning_rate": 6.821729739401239e-06,
      "loss": 0.6898,
      "step": 2482
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9200206686011301,
      "learning_rate": 6.8192991448671605e-06,
      "loss": 0.538,
      "step": 2483
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0213212703751973,
      "learning_rate": 6.8168680547058455e-06,
      "loss": 0.5104,
      "step": 2484
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7255726756540283,
      "learning_rate": 6.81443646957959e-06,
      "loss": 0.4639,
      "step": 2485
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0265385465932637,
      "learning_rate": 6.812004390150825e-06,
      "loss": 0.6828,
      "step": 2486
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9685166294120701,
      "learning_rate": 6.809571817082117e-06,
      "loss": 0.6158,
      "step": 2487
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9651939791120863,
      "learning_rate": 6.807138751036163e-06,
      "loss": 0.6089,
      "step": 2488
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0473639433561777,
      "learning_rate": 6.804705192675799e-06,
      "loss": 0.6069,
      "step": 2489
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9550060941541216,
      "learning_rate": 6.802271142663994e-06,
      "loss": 0.5115,
      "step": 2490
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9774494673745749,
      "learning_rate": 6.799836601663851e-06,
      "loss": 0.6043,
      "step": 2491
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9417199926922714,
      "learning_rate": 6.797401570338604e-06,
      "loss": 0.5986,
      "step": 2492
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9403230727109817,
      "learning_rate": 6.794966049351625e-06,
      "loss": 0.5489,
      "step": 2493
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0420091789920436,
      "learning_rate": 6.792530039366414e-06,
      "loss": 0.7169,
      "step": 2494
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9880195015039127,
      "learning_rate": 6.790093541046609e-06,
      "loss": 0.5736,
      "step": 2495
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9606500870149525,
      "learning_rate": 6.787656555055979e-06,
      "loss": 0.5969,
      "step": 2496
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9948106035363221,
      "learning_rate": 6.785219082058426e-06,
      "loss": 0.6565,
      "step": 2497
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9872436683254563,
      "learning_rate": 6.782781122717987e-06,
      "loss": 0.5967,
      "step": 2498
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1118934956516229,
      "learning_rate": 6.780342677698826e-06,
      "loss": 0.6268,
      "step": 2499
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0499605169795165,
      "learning_rate": 6.777903747665245e-06,
      "loss": 0.5868,
      "step": 2500
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0579897809382857,
      "learning_rate": 6.775464333281674e-06,
      "loss": 0.6756,
      "step": 2501
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.063498562823953,
      "learning_rate": 6.773024435212678e-06,
      "loss": 0.7748,
      "step": 2502
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9499005028851514,
      "learning_rate": 6.770584054122954e-06,
      "loss": 0.576,
      "step": 2503
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9903775626135637,
      "learning_rate": 6.7681431906773255e-06,
      "loss": 0.641,
      "step": 2504
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0776084816022675,
      "learning_rate": 6.765701845540753e-06,
      "loss": 0.6238,
      "step": 2505
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0119200822677583,
      "learning_rate": 6.763260019378325e-06,
      "loss": 0.6679,
      "step": 2506
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9799769655119922,
      "learning_rate": 6.760817712855266e-06,
      "loss": 0.5822,
      "step": 2507
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0051958202205935,
      "learning_rate": 6.75837492663692e-06,
      "loss": 0.5734,
      "step": 2508
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9384966957369636,
      "learning_rate": 6.755931661388778e-06,
      "loss": 0.4679,
      "step": 2509
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.007386600122028,
      "learning_rate": 6.753487917776447e-06,
      "loss": 0.5655,
      "step": 2510
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0806513266957567,
      "learning_rate": 6.751043696465674e-06,
      "loss": 0.6519,
      "step": 2511
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0542621529396017,
      "learning_rate": 6.748598998122328e-06,
      "loss": 0.6419,
      "step": 2512
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6961910299745654,
      "learning_rate": 6.746153823412416e-06,
      "loss": 0.504,
      "step": 2513
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.000283417248506,
      "learning_rate": 6.7437081730020695e-06,
      "loss": 0.592,
      "step": 2514
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1235580595426367,
      "learning_rate": 6.7412620475575495e-06,
      "loss": 0.6953,
      "step": 2515
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0048097947468948,
      "learning_rate": 6.7388154477452505e-06,
      "loss": 0.6361,
      "step": 2516
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.066195123774335,
      "learning_rate": 6.736368374231693e-06,
      "loss": 0.5936,
      "step": 2517
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9998213160074564,
      "learning_rate": 6.733920827683529e-06,
      "loss": 0.5973,
      "step": 2518
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9506109324333402,
      "learning_rate": 6.731472808767532e-06,
      "loss": 0.6179,
      "step": 2519
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9535603780080965,
      "learning_rate": 6.729024318150617e-06,
      "loss": 0.6172,
      "step": 2520
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0103086100808496,
      "learning_rate": 6.726575356499814e-06,
      "loss": 0.6309,
      "step": 2521
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9317522355461826,
      "learning_rate": 6.724125924482292e-06,
      "loss": 0.4668,
      "step": 2522
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.522798330174133,
      "learning_rate": 6.7216760227653426e-06,
      "loss": 0.5464,
      "step": 2523
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.969246569606102,
      "learning_rate": 6.7192256520163844e-06,
      "loss": 0.5826,
      "step": 2524
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0052706170470391,
      "learning_rate": 6.7167748129029705e-06,
      "loss": 0.5466,
      "step": 2525
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0227874255285323,
      "learning_rate": 6.714323506092773e-06,
      "loss": 0.6505,
      "step": 2526
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9505392323519235,
      "learning_rate": 6.711871732253596e-06,
      "loss": 0.5293,
      "step": 2527
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9377052049095732,
      "learning_rate": 6.709419492053373e-06,
      "loss": 0.5919,
      "step": 2528
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9932410292132299,
      "learning_rate": 6.706966786160159e-06,
      "loss": 0.6502,
      "step": 2529
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0480315900430064,
      "learning_rate": 6.7045136152421395e-06,
      "loss": 0.6338,
      "step": 2530
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9254879553161067,
      "learning_rate": 6.702059979967627e-06,
      "loss": 0.5205,
      "step": 2531
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9588210869232551,
      "learning_rate": 6.699605881005058e-06,
      "loss": 0.5756,
      "step": 2532
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0655037342255094,
      "learning_rate": 6.697151319022996e-06,
      "loss": 0.6279,
      "step": 2533
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9598108868864681,
      "learning_rate": 6.694696294690133e-06,
      "loss": 0.6104,
      "step": 2534
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8996195883080861,
      "learning_rate": 6.692240808675286e-06,
      "loss": 0.6314,
      "step": 2535
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9669811778226647,
      "learning_rate": 6.689784861647395e-06,
      "loss": 0.6018,
      "step": 2536
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6518424321418883,
      "learning_rate": 6.6873284542755275e-06,
      "loss": 0.4843,
      "step": 2537
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9622863633926381,
      "learning_rate": 6.684871587228878e-06,
      "loss": 0.5248,
      "step": 2538
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0711655979713472,
      "learning_rate": 6.682414261176765e-06,
      "loss": 0.6233,
      "step": 2539
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.012294791062189,
      "learning_rate": 6.6799564767886305e-06,
      "loss": 0.6004,
      "step": 2540
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0310829934042711,
      "learning_rate": 6.677498234734045e-06,
      "loss": 0.6348,
      "step": 2541
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9737609522783973,
      "learning_rate": 6.675039535682699e-06,
      "loss": 0.5559,
      "step": 2542
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.954720512338294,
      "learning_rate": 6.67258038030441e-06,
      "loss": 0.5958,
      "step": 2543
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9531872991714292,
      "learning_rate": 6.67012076926912e-06,
      "loss": 0.5661,
      "step": 2544
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9941276543918149,
      "learning_rate": 6.667660703246897e-06,
      "loss": 0.6232,
      "step": 2545
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0130999863311994,
      "learning_rate": 6.665200182907928e-06,
      "loss": 0.5833,
      "step": 2546
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0434174066359805,
      "learning_rate": 6.662739208922529e-06,
      "loss": 0.6335,
      "step": 2547
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0024392468491032,
      "learning_rate": 6.660277781961135e-06,
      "loss": 0.6354,
      "step": 2548
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9948464686682033,
      "learning_rate": 6.6578159026943064e-06,
      "loss": 0.6188,
      "step": 2549
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9856599808866728,
      "learning_rate": 6.655353571792729e-06,
      "loss": 0.6017,
      "step": 2550
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0947400114544834,
      "learning_rate": 6.652890789927209e-06,
      "loss": 0.6769,
      "step": 2551
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9821868481112924,
      "learning_rate": 6.650427557768674e-06,
      "loss": 0.5932,
      "step": 2552
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0362479062746677,
      "learning_rate": 6.647963875988179e-06,
      "loss": 0.6618,
      "step": 2553
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0045773425075395,
      "learning_rate": 6.645499745256898e-06,
      "loss": 0.6368,
      "step": 2554
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0449898981481693,
      "learning_rate": 6.643035166246128e-06,
      "loss": 0.7077,
      "step": 2555
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9507740410875614,
      "learning_rate": 6.640570139627288e-06,
      "loss": 0.5956,
      "step": 2556
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9738974634356756,
      "learning_rate": 6.638104666071918e-06,
      "loss": 0.5509,
      "step": 2557
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0635617442122571,
      "learning_rate": 6.635638746251685e-06,
      "loss": 0.632,
      "step": 2558
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9885724221795573,
      "learning_rate": 6.6331723808383674e-06,
      "loss": 0.5276,
      "step": 2559
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0193579183091526,
      "learning_rate": 6.630705570503878e-06,
      "loss": 0.5558,
      "step": 2560
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9991712228106975,
      "learning_rate": 6.628238315920239e-06,
      "loss": 0.6076,
      "step": 2561
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1167738481899345,
      "learning_rate": 6.6257706177595994e-06,
      "loss": 0.6637,
      "step": 2562
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6284012407550957,
      "learning_rate": 6.62330247669423e-06,
      "loss": 0.4411,
      "step": 2563
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9905162474639524,
      "learning_rate": 6.62083389339652e-06,
      "loss": 0.5821,
      "step": 2564
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9685951226184077,
      "learning_rate": 6.618364868538978e-06,
      "loss": 0.5848,
      "step": 2565
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0103096425202092,
      "learning_rate": 6.6158954027942345e-06,
      "loss": 0.6609,
      "step": 2566
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.994775628196965,
      "learning_rate": 6.6134254968350434e-06,
      "loss": 0.6037,
      "step": 2567
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.023326086655465,
      "learning_rate": 6.610955151334269e-06,
      "loss": 0.6006,
      "step": 2568
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8936437066842363,
      "learning_rate": 6.608484366964908e-06,
      "loss": 0.4609,
      "step": 2569
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.972262763337525,
      "learning_rate": 6.606013144400065e-06,
      "loss": 0.5336,
      "step": 2570
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9988413884974223,
      "learning_rate": 6.603541484312974e-06,
      "loss": 0.6078,
      "step": 2571
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0279655296212966,
      "learning_rate": 6.601069387376979e-06,
      "loss": 0.5857,
      "step": 2572
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9352353970546785,
      "learning_rate": 6.5985968542655495e-06,
      "loss": 0.5483,
      "step": 2573
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0647205123561074,
      "learning_rate": 6.596123885652272e-06,
      "loss": 0.615,
      "step": 2574
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9286199916916759,
      "learning_rate": 6.593650482210851e-06,
      "loss": 0.5893,
      "step": 2575
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.022550424170138,
      "learning_rate": 6.591176644615108e-06,
      "loss": 0.5827,
      "step": 2576
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9020692570562443,
      "learning_rate": 6.588702373538987e-06,
      "loss": 0.5308,
      "step": 2577
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9567718875966726,
      "learning_rate": 6.5862276696565454e-06,
      "loss": 0.5948,
      "step": 2578
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8917139516992794,
      "learning_rate": 6.583752533641963e-06,
      "loss": 0.6136,
      "step": 2579
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9404116792845437,
      "learning_rate": 6.581276966169534e-06,
      "loss": 0.6126,
      "step": 2580
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9825803583085433,
      "learning_rate": 6.57880096791367e-06,
      "loss": 0.5271,
      "step": 2581
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.064116122590369,
      "learning_rate": 6.576324539548904e-06,
      "loss": 0.6496,
      "step": 2582
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0116707884223723,
      "learning_rate": 6.573847681749881e-06,
      "loss": 0.5438,
      "step": 2583
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0705515854369636,
      "learning_rate": 6.5713703951913665e-06,
      "loss": 0.5716,
      "step": 2584
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9754768091583119,
      "learning_rate": 6.5688926805482425e-06,
      "loss": 0.5629,
      "step": 2585
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0264659596369923,
      "learning_rate": 6.566414538495504e-06,
      "loss": 0.6195,
      "step": 2586
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7355078677122344,
      "learning_rate": 6.563935969708266e-06,
      "loss": 0.484,
      "step": 2587
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0828541334774135,
      "learning_rate": 6.561456974861761e-06,
      "loss": 0.6887,
      "step": 2588
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0433305776996245,
      "learning_rate": 6.558977554631334e-06,
      "loss": 0.6548,
      "step": 2589
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.959756042998032,
      "learning_rate": 6.556497709692447e-06,
      "loss": 0.6113,
      "step": 2590
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0452700204897056,
      "learning_rate": 6.554017440720679e-06,
      "loss": 0.6809,
      "step": 2591
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0586517651126965,
      "learning_rate": 6.551536748391724e-06,
      "loss": 0.5937,
      "step": 2592
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9699683269666205,
      "learning_rate": 6.54905563338139e-06,
      "loss": 0.596,
      "step": 2593
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0143080749930315,
      "learning_rate": 6.546574096365601e-06,
      "loss": 0.6051,
      "step": 2594
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.041784056222085,
      "learning_rate": 6.544092138020397e-06,
      "loss": 0.6251,
      "step": 2595
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9288845561019567,
      "learning_rate": 6.541609759021933e-06,
      "loss": 0.6075,
      "step": 2596
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9118238654049805,
      "learning_rate": 6.539126960046474e-06,
      "loss": 0.5901,
      "step": 2597
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0166587787271026,
      "learning_rate": 6.536643741770406e-06,
      "loss": 0.6267,
      "step": 2598
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9926945114865702,
      "learning_rate": 6.534160104870224e-06,
      "loss": 0.6482,
      "step": 2599
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0759394956670143,
      "learning_rate": 6.531676050022539e-06,
      "loss": 0.6768,
      "step": 2600
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.058683229891837,
      "learning_rate": 6.529191577904079e-06,
      "loss": 0.7123,
      "step": 2601
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0133936734319813,
      "learning_rate": 6.526706689191681e-06,
      "loss": 0.6449,
      "step": 2602
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0187962886244175,
      "learning_rate": 6.524221384562295e-06,
      "loss": 0.6246,
      "step": 2603
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6729077622455986,
      "learning_rate": 6.521735664692989e-06,
      "loss": 0.4895,
      "step": 2604
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9122792620685437,
      "learning_rate": 6.519249530260943e-06,
      "loss": 0.5501,
      "step": 2605
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.111684736362179,
      "learning_rate": 6.516762981943444e-06,
      "loss": 0.7472,
      "step": 2606
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.020036355096479,
      "learning_rate": 6.514276020417901e-06,
      "loss": 0.5665,
      "step": 2607
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0002754869507022,
      "learning_rate": 6.511788646361828e-06,
      "loss": 0.6063,
      "step": 2608
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1300316086703024,
      "learning_rate": 6.509300860452854e-06,
      "loss": 0.6924,
      "step": 2609
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.030616281329045,
      "learning_rate": 6.506812663368722e-06,
      "loss": 0.6126,
      "step": 2610
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1052548555258808,
      "learning_rate": 6.504324055787285e-06,
      "loss": 0.6465,
      "step": 2611
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9662603377084739,
      "learning_rate": 6.501835038386509e-06,
      "loss": 0.6505,
      "step": 2612
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9047798069380502,
      "learning_rate": 6.499345611844471e-06,
      "loss": 0.6108,
      "step": 2613
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8802146366517508,
      "learning_rate": 6.496855776839357e-06,
      "loss": 0.5185,
      "step": 2614
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9208894234839756,
      "learning_rate": 6.494365534049469e-06,
      "loss": 0.5088,
      "step": 2615
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9267487781505049,
      "learning_rate": 6.491874884153217e-06,
      "loss": 0.5364,
      "step": 2616
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0066387163399229,
      "learning_rate": 6.489383827829122e-06,
      "loss": 0.7278,
      "step": 2617
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0451028431855591,
      "learning_rate": 6.486892365755819e-06,
      "loss": 0.6354,
      "step": 2618
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.90627713695524,
      "learning_rate": 6.4844004986120465e-06,
      "loss": 0.5266,
      "step": 2619
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.042139075169554,
      "learning_rate": 6.481908227076663e-06,
      "loss": 0.6029,
      "step": 2620
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9213864250898641,
      "learning_rate": 6.479415551828627e-06,
      "loss": 0.5656,
      "step": 2621
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0568248380324632,
      "learning_rate": 6.476922473547016e-06,
      "loss": 0.5514,
      "step": 2622
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0729407017376156,
      "learning_rate": 6.474428992911011e-06,
      "loss": 0.5717,
      "step": 2623
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.051282653158837,
      "learning_rate": 6.471935110599907e-06,
      "loss": 0.5708,
      "step": 2624
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9679742272764066,
      "learning_rate": 6.469440827293103e-06,
      "loss": 0.5925,
      "step": 2625
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0209790275830017,
      "learning_rate": 6.466946143670113e-06,
      "loss": 0.6208,
      "step": 2626
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0493256163108942,
      "learning_rate": 6.464451060410556e-06,
      "loss": 0.5772,
      "step": 2627
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1355684616864483,
      "learning_rate": 6.461955578194163e-06,
      "loss": 0.6956,
      "step": 2628
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.982779263033437,
      "learning_rate": 6.459459697700772e-06,
      "loss": 0.5499,
      "step": 2629
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9708244861643506,
      "learning_rate": 6.456963419610327e-06,
      "loss": 0.6408,
      "step": 2630
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9301758824880064,
      "learning_rate": 6.454466744602888e-06,
      "loss": 0.496,
      "step": 2631
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9881125578914289,
      "learning_rate": 6.451969673358613e-06,
      "loss": 0.595,
      "step": 2632
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.019576817251507,
      "learning_rate": 6.449472206557776e-06,
      "loss": 0.5994,
      "step": 2633
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1079324193948537,
      "learning_rate": 6.4469743448807546e-06,
      "loss": 0.6594,
      "step": 2634
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.019981864460688,
      "learning_rate": 6.444476089008037e-06,
      "loss": 0.615,
      "step": 2635
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0170709463386776,
      "learning_rate": 6.441977439620214e-06,
      "loss": 0.5923,
      "step": 2636
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0698401477799844,
      "learning_rate": 6.439478397397989e-06,
      "loss": 0.6591,
      "step": 2637
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0712602927839843,
      "learning_rate": 6.436978963022168e-06,
      "loss": 0.669,
      "step": 2638
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6810625557873233,
      "learning_rate": 6.434479137173667e-06,
      "loss": 0.5347,
      "step": 2639
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0406801595627877,
      "learning_rate": 6.431978920533509e-06,
      "loss": 0.6419,
      "step": 2640
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9644160598017195,
      "learning_rate": 6.429478313782817e-06,
      "loss": 0.6425,
      "step": 2641
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9899673251130092,
      "learning_rate": 6.426977317602833e-06,
      "loss": 0.5918,
      "step": 2642
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.033403924011447,
      "learning_rate": 6.424475932674889e-06,
      "loss": 0.5858,
      "step": 2643
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0347320925473982,
      "learning_rate": 6.421974159680437e-06,
      "loss": 0.6935,
      "step": 2644
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9532589680480318,
      "learning_rate": 6.419471999301027e-06,
      "loss": 0.5837,
      "step": 2645
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9500626662746939,
      "learning_rate": 6.416969452218317e-06,
      "loss": 0.6478,
      "step": 2646
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9523691191776804,
      "learning_rate": 6.414466519114068e-06,
      "loss": 0.5708,
      "step": 2647
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9705145228209704,
      "learning_rate": 6.411963200670148e-06,
      "loss": 0.6089,
      "step": 2648
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0906113182763968,
      "learning_rate": 6.409459497568533e-06,
      "loss": 0.7288,
      "step": 2649
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9087301013461447,
      "learning_rate": 6.406955410491298e-06,
      "loss": 0.5679,
      "step": 2650
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9129516490165551,
      "learning_rate": 6.404450940120628e-06,
      "loss": 0.3973,
      "step": 2651
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9290740070857407,
      "learning_rate": 6.4019460871388055e-06,
      "loss": 0.5428,
      "step": 2652
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.033949366708766,
      "learning_rate": 6.399440852228226e-06,
      "loss": 0.5746,
      "step": 2653
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8870683885883719,
      "learning_rate": 6.396935236071381e-06,
      "loss": 0.5561,
      "step": 2654
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9420004205588067,
      "learning_rate": 6.394429239350872e-06,
      "loss": 0.5819,
      "step": 2655
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9837763290162224,
      "learning_rate": 6.3919228627494e-06,
      "loss": 0.5508,
      "step": 2656
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0103917971952394,
      "learning_rate": 6.38941610694977e-06,
      "loss": 0.5911,
      "step": 2657
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9576469251702779,
      "learning_rate": 6.386908972634897e-06,
      "loss": 0.5567,
      "step": 2658
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1027578881629692,
      "learning_rate": 6.384401460487787e-06,
      "loss": 0.6401,
      "step": 2659
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1152135785755155,
      "learning_rate": 6.381893571191558e-06,
      "loss": 0.6281,
      "step": 2660
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9219426771119541,
      "learning_rate": 6.379385305429428e-06,
      "loss": 0.517,
      "step": 2661
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0318005231477991,
      "learning_rate": 6.376876663884719e-06,
      "loss": 0.6773,
      "step": 2662
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9814015852072634,
      "learning_rate": 6.374367647240851e-06,
      "loss": 0.5852,
      "step": 2663
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0946873461959152,
      "learning_rate": 6.371858256181352e-06,
      "loss": 0.7085,
      "step": 2664
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1257790798105276,
      "learning_rate": 6.3693484913898494e-06,
      "loss": 0.6383,
      "step": 2665
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0230386951645156,
      "learning_rate": 6.36683835355007e-06,
      "loss": 0.5802,
      "step": 2666
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9928835525742654,
      "learning_rate": 6.364327843345847e-06,
      "loss": 0.6074,
      "step": 2667
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0816457454041124,
      "learning_rate": 6.361816961461111e-06,
      "loss": 0.5939,
      "step": 2668
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0035418526957072,
      "learning_rate": 6.359305708579897e-06,
      "loss": 0.5019,
      "step": 2669
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9959592208618275,
      "learning_rate": 6.356794085386337e-06,
      "loss": 0.6783,
      "step": 2670
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9481317734574466,
      "learning_rate": 6.3542820925646696e-06,
      "loss": 0.484,
      "step": 2671
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0455324909362815,
      "learning_rate": 6.351769730799227e-06,
      "loss": 0.6176,
      "step": 2672
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9619067550739812,
      "learning_rate": 6.349257000774452e-06,
      "loss": 0.5764,
      "step": 2673
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0852462274541015,
      "learning_rate": 6.346743903174872e-06,
      "loss": 0.6207,
      "step": 2674
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9494570703108391,
      "learning_rate": 6.344230438685134e-06,
      "loss": 0.5964,
      "step": 2675
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9871075296923102,
      "learning_rate": 6.3417166079899685e-06,
      "loss": 0.6679,
      "step": 2676
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9918370749407301,
      "learning_rate": 6.339202411774215e-06,
      "loss": 0.641,
      "step": 2677
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9971048416139556,
      "learning_rate": 6.336687850722809e-06,
      "loss": 0.5954,
      "step": 2678
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9833572599683785,
      "learning_rate": 6.334172925520785e-06,
      "loss": 0.6538,
      "step": 2679
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.10400773530343,
      "learning_rate": 6.3316576368532814e-06,
      "loss": 0.5831,
      "step": 2680
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9954560921615357,
      "learning_rate": 6.329141985405529e-06,
      "loss": 0.6026,
      "step": 2681
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9920694254942484,
      "learning_rate": 6.326625971862863e-06,
      "loss": 0.6456,
      "step": 2682
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1117445777285977,
      "learning_rate": 6.324109596910713e-06,
      "loss": 0.6861,
      "step": 2683
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0007172599680345,
      "learning_rate": 6.32159286123461e-06,
      "loss": 0.6027,
      "step": 2684
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0442335287996225,
      "learning_rate": 6.319075765520179e-06,
      "loss": 0.7014,
      "step": 2685
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.024363800862652,
      "learning_rate": 6.316558310453153e-06,
      "loss": 0.6181,
      "step": 2686
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.996700915993852,
      "learning_rate": 6.314040496719349e-06,
      "loss": 0.5465,
      "step": 2687
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.04127223109327,
      "learning_rate": 6.3115223250046934e-06,
      "loss": 0.5966,
      "step": 2688
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0385564091664412,
      "learning_rate": 6.309003795995205e-06,
      "loss": 0.5778,
      "step": 2689
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9943036095306566,
      "learning_rate": 6.306484910376998e-06,
      "loss": 0.5587,
      "step": 2690
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9930851736935974,
      "learning_rate": 6.303965668836288e-06,
      "loss": 0.5281,
      "step": 2691
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.014775318818345,
      "learning_rate": 6.301446072059386e-06,
      "loss": 0.5402,
      "step": 2692
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0194681029197552,
      "learning_rate": 6.2989261207327e-06,
      "loss": 0.6484,
      "step": 2693
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9388331733262057,
      "learning_rate": 6.296405815542732e-06,
      "loss": 0.635,
      "step": 2694
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9511933598242038,
      "learning_rate": 6.293885157176087e-06,
      "loss": 0.546,
      "step": 2695
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.076864411551599,
      "learning_rate": 6.2913641463194566e-06,
      "loss": 0.6562,
      "step": 2696
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0022117574532048,
      "learning_rate": 6.288842783659638e-06,
      "loss": 0.6845,
      "step": 2697
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.041833791627389,
      "learning_rate": 6.286321069883517e-06,
      "loss": 0.6595,
      "step": 2698
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0451460609433734,
      "learning_rate": 6.283799005678077e-06,
      "loss": 0.6186,
      "step": 2699
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0462101091994838,
      "learning_rate": 6.281276591730403e-06,
      "loss": 0.6887,
      "step": 2700
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0730079743254441,
      "learning_rate": 6.278753828727664e-06,
      "loss": 0.5979,
      "step": 2701
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9358364909451595,
      "learning_rate": 6.276230717357136e-06,
      "loss": 0.6139,
      "step": 2702
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.043386788523056,
      "learning_rate": 6.273707258306178e-06,
      "loss": 0.6397,
      "step": 2703
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.070217159876887,
      "learning_rate": 6.271183452262255e-06,
      "loss": 0.7397,
      "step": 2704
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0371873081340126,
      "learning_rate": 6.268659299912918e-06,
      "loss": 0.6027,
      "step": 2705
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0835037923073132,
      "learning_rate": 6.266134801945819e-06,
      "loss": 0.6581,
      "step": 2706
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0377717305959842,
      "learning_rate": 6.263609959048696e-06,
      "loss": 0.5935,
      "step": 2707
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0555480715060301,
      "learning_rate": 6.261084771909391e-06,
      "loss": 0.6672,
      "step": 2708
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0046978247829903,
      "learning_rate": 6.25855924121583e-06,
      "loss": 0.6083,
      "step": 2709
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1137995500897437,
      "learning_rate": 6.25603336765604e-06,
      "loss": 0.686,
      "step": 2710
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0755274642395372,
      "learning_rate": 6.2535071519181385e-06,
      "loss": 0.5973,
      "step": 2711
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9401479869826727,
      "learning_rate": 6.250980594690335e-06,
      "loss": 0.5906,
      "step": 2712
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.951699673235168,
      "learning_rate": 6.248453696660934e-06,
      "loss": 0.5524,
      "step": 2713
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0949283102618816,
      "learning_rate": 6.245926458518333e-06,
      "loss": 0.6439,
      "step": 2714
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8792250544124379,
      "learning_rate": 6.243398880951019e-06,
      "loss": 0.508,
      "step": 2715
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.028113940702658,
      "learning_rate": 6.240870964647577e-06,
      "loss": 0.6541,
      "step": 2716
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0099267913794,
      "learning_rate": 6.2383427102966786e-06,
      "loss": 0.6453,
      "step": 2717
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0204399579000307,
      "learning_rate": 6.2358141185870915e-06,
      "loss": 0.5835,
      "step": 2718
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9616924762999283,
      "learning_rate": 6.233285190207673e-06,
      "loss": 0.5441,
      "step": 2719
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0113952959848211,
      "learning_rate": 6.2307559258473716e-06,
      "loss": 0.639,
      "step": 2720
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.054723157240936,
      "learning_rate": 6.228226326195232e-06,
      "loss": 0.7181,
      "step": 2721
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9157772165866018,
      "learning_rate": 6.225696391940383e-06,
      "loss": 0.5936,
      "step": 2722
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8993306648988618,
      "learning_rate": 6.223166123772051e-06,
      "loss": 0.573,
      "step": 2723
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0494719444982958,
      "learning_rate": 6.220635522379551e-06,
      "loss": 0.6398,
      "step": 2724
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.721185944079204,
      "learning_rate": 6.2181045884522876e-06,
      "loss": 0.5122,
      "step": 2725
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.022674034357875,
      "learning_rate": 6.215573322679756e-06,
      "loss": 0.6005,
      "step": 2726
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9943601408064036,
      "learning_rate": 6.213041725751543e-06,
      "loss": 0.5862,
      "step": 2727
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9925358434272786,
      "learning_rate": 6.210509798357328e-06,
      "loss": 0.6031,
      "step": 2728
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9451944177379101,
      "learning_rate": 6.207977541186876e-06,
      "loss": 0.59,
      "step": 2729
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.108988356036505,
      "learning_rate": 6.205444954930043e-06,
      "loss": 0.5949,
      "step": 2730
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8295013596724732,
      "learning_rate": 6.2029120402767765e-06,
      "loss": 0.5223,
      "step": 2731
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0252469321977662,
      "learning_rate": 6.2003787979171105e-06,
      "loss": 0.5757,
      "step": 2732
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.003224875173133,
      "learning_rate": 6.197845228541174e-06,
      "loss": 0.5775,
      "step": 2733
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9542687388224328,
      "learning_rate": 6.195311332839175e-06,
      "loss": 0.4823,
      "step": 2734
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0167995302973238,
      "learning_rate": 6.192777111501422e-06,
      "loss": 0.5735,
      "step": 2735
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0687491447292161,
      "learning_rate": 6.190242565218305e-06,
      "loss": 0.5961,
      "step": 2736
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9669557991716592,
      "learning_rate": 6.187707694680302e-06,
      "loss": 0.6276,
      "step": 2737
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.009573939839138,
      "learning_rate": 6.185172500577986e-06,
      "loss": 0.6418,
      "step": 2738
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9263174770171911,
      "learning_rate": 6.182636983602009e-06,
      "loss": 0.5746,
      "step": 2739
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.070686947758525,
      "learning_rate": 6.180101144443121e-06,
      "loss": 0.5811,
      "step": 2740
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0476264076354889,
      "learning_rate": 6.17756498379215e-06,
      "loss": 0.6015,
      "step": 2741
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.091797889353649,
      "learning_rate": 6.175028502340018e-06,
      "loss": 0.6599,
      "step": 2742
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.10336487253729,
      "learning_rate": 6.172491700777732e-06,
      "loss": 0.597,
      "step": 2743
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.026731675902572,
      "learning_rate": 6.169954579796387e-06,
      "loss": 0.6537,
      "step": 2744
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9839566667990703,
      "learning_rate": 6.167417140087163e-06,
      "loss": 0.5771,
      "step": 2745
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0680208728310914,
      "learning_rate": 6.164879382341331e-06,
      "loss": 0.6432,
      "step": 2746
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.943084573772298,
      "learning_rate": 6.162341307250246e-06,
      "loss": 0.6403,
      "step": 2747
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0021196809870656,
      "learning_rate": 6.159802915505347e-06,
      "loss": 0.602,
      "step": 2748
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0148224882586994,
      "learning_rate": 6.157264207798165e-06,
      "loss": 0.6468,
      "step": 2749
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9945659509598346,
      "learning_rate": 6.154725184820311e-06,
      "loss": 0.6677,
      "step": 2750
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.02768686070851,
      "learning_rate": 6.15218584726349e-06,
      "loss": 0.6318,
      "step": 2751
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9387833832900156,
      "learning_rate": 6.149646195819481e-06,
      "loss": 0.5321,
      "step": 2752
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9546907945735594,
      "learning_rate": 6.147106231180159e-06,
      "loss": 0.5739,
      "step": 2753
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0263359002574455,
      "learning_rate": 6.144565954037479e-06,
      "loss": 0.6006,
      "step": 2754
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9286238737150959,
      "learning_rate": 6.142025365083484e-06,
      "loss": 0.5959,
      "step": 2755
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0075802546700396,
      "learning_rate": 6.139484465010298e-06,
      "loss": 0.6407,
      "step": 2756
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0086601027313675,
      "learning_rate": 6.136943254510135e-06,
      "loss": 0.616,
      "step": 2757
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0534146431243012,
      "learning_rate": 6.134401734275288e-06,
      "loss": 0.5996,
      "step": 2758
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9413599539829495,
      "learning_rate": 6.13185990499814e-06,
      "loss": 0.5713,
      "step": 2759
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9861118215920378,
      "learning_rate": 6.129317767371153e-06,
      "loss": 0.633,
      "step": 2760
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0876144804774115,
      "learning_rate": 6.126775322086876e-06,
      "loss": 0.6214,
      "step": 2761
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9469400815668929,
      "learning_rate": 6.124232569837943e-06,
      "loss": 0.5566,
      "step": 2762
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.095637070381988,
      "learning_rate": 6.121689511317068e-06,
      "loss": 0.6115,
      "step": 2763
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9268069941241458,
      "learning_rate": 6.11914614721705e-06,
      "loss": 0.5637,
      "step": 2764
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0278119700210662,
      "learning_rate": 6.116602478230772e-06,
      "loss": 0.5636,
      "step": 2765
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9288487496563996,
      "learning_rate": 6.1140585050512e-06,
      "loss": 0.591,
      "step": 2766
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9332189838564758,
      "learning_rate": 6.111514228371381e-06,
      "loss": 0.5692,
      "step": 2767
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8064784567144704,
      "learning_rate": 6.108969648884449e-06,
      "loss": 0.4666,
      "step": 2768
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9990096279018127,
      "learning_rate": 6.106424767283615e-06,
      "loss": 0.5489,
      "step": 2769
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6825400765201701,
      "learning_rate": 6.103879584262176e-06,
      "loss": 0.4957,
      "step": 2770
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.034556765734561,
      "learning_rate": 6.101334100513508e-06,
      "loss": 0.5957,
      "step": 2771
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9979857980217465,
      "learning_rate": 6.098788316731074e-06,
      "loss": 0.6099,
      "step": 2772
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9479816871057386,
      "learning_rate": 6.096242233608414e-06,
      "loss": 0.6554,
      "step": 2773
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0210469732623622,
      "learning_rate": 6.0936958518391505e-06,
      "loss": 0.621,
      "step": 2774
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6353273894022693,
      "learning_rate": 6.0911491721169906e-06,
      "loss": 0.5044,
      "step": 2775
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9369604918482168,
      "learning_rate": 6.088602195135717e-06,
      "loss": 0.5492,
      "step": 2776
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0151224502365879,
      "learning_rate": 6.086054921589198e-06,
      "loss": 0.6041,
      "step": 2777
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9946196746334682,
      "learning_rate": 6.083507352171382e-06,
      "loss": 0.5566,
      "step": 2778
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9408918036691608,
      "learning_rate": 6.080959487576297e-06,
      "loss": 0.5011,
      "step": 2779
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8987212710381357,
      "learning_rate": 6.078411328498049e-06,
      "loss": 0.5316,
      "step": 2780
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9843685472932292,
      "learning_rate": 6.075862875630829e-06,
      "loss": 0.5432,
      "step": 2781
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.10341551933072,
      "learning_rate": 6.073314129668907e-06,
      "loss": 0.6315,
      "step": 2782
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9665471873690706,
      "learning_rate": 6.070765091306628e-06,
      "loss": 0.5118,
      "step": 2783
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9694075445181521,
      "learning_rate": 6.068215761238423e-06,
      "loss": 0.6138,
      "step": 2784
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9658980548077123,
      "learning_rate": 6.065666140158799e-06,
      "loss": 0.6251,
      "step": 2785
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0337417258632708,
      "learning_rate": 6.063116228762343e-06,
      "loss": 0.6551,
      "step": 2786
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6889929695094947,
      "learning_rate": 6.060566027743721e-06,
      "loss": 0.528,
      "step": 2787
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9962554815461884,
      "learning_rate": 6.058015537797678e-06,
      "loss": 0.6426,
      "step": 2788
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9411791735795947,
      "learning_rate": 6.055464759619038e-06,
      "loss": 0.5736,
      "step": 2789
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9327620072832649,
      "learning_rate": 6.052913693902706e-06,
      "loss": 0.5934,
      "step": 2790
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6534439121209137,
      "learning_rate": 6.050362341343658e-06,
      "loss": 0.4981,
      "step": 2791
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.041894309473969,
      "learning_rate": 6.047810702636956e-06,
      "loss": 0.5498,
      "step": 2792
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.067873760685095,
      "learning_rate": 6.045258778477735e-06,
      "loss": 0.6422,
      "step": 2793
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9260814996905765,
      "learning_rate": 6.042706569561209e-06,
      "loss": 0.596,
      "step": 2794
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1061981823310687,
      "learning_rate": 6.040154076582672e-06,
      "loss": 0.6022,
      "step": 2795
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0867869512914343,
      "learning_rate": 6.0376013002374924e-06,
      "loss": 0.5846,
      "step": 2796
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1323472762612898,
      "learning_rate": 6.035048241221118e-06,
      "loss": 0.6493,
      "step": 2797
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9956746162157875,
      "learning_rate": 6.032494900229069e-06,
      "loss": 0.6626,
      "step": 2798
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0220848694246965,
      "learning_rate": 6.02994127795695e-06,
      "loss": 0.657,
      "step": 2799
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7519369907581069,
      "learning_rate": 6.027387375100435e-06,
      "loss": 0.4794,
      "step": 2800
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2510138519067262,
      "learning_rate": 6.024833192355282e-06,
      "loss": 0.6645,
      "step": 2801
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9978137367219454,
      "learning_rate": 6.022278730417315e-06,
      "loss": 0.6178,
      "step": 2802
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.879076805279253,
      "learning_rate": 6.019723989982444e-06,
      "loss": 0.6214,
      "step": 2803
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9526879217424523,
      "learning_rate": 6.0171689717466485e-06,
      "loss": 0.5267,
      "step": 2804
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.090810141366759,
      "learning_rate": 6.014613676405985e-06,
      "loss": 0.6342,
      "step": 2805
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9534369894209447,
      "learning_rate": 6.01205810465659e-06,
      "loss": 0.616,
      "step": 2806
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1192545139697414,
      "learning_rate": 6.009502257194669e-06,
      "loss": 0.7187,
      "step": 2807
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9557433902679736,
      "learning_rate": 6.0069461347165084e-06,
      "loss": 0.5943,
      "step": 2808
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0061525776555953,
      "learning_rate": 6.0043897379184615e-06,
      "loss": 0.662,
      "step": 2809
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7241871984498629,
      "learning_rate": 6.001833067496964e-06,
      "loss": 0.5247,
      "step": 2810
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9974530270710181,
      "learning_rate": 5.999276124148525e-06,
      "loss": 0.5649,
      "step": 2811
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9335744127886252,
      "learning_rate": 5.996718908569723e-06,
      "loss": 0.6428,
      "step": 2812
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6433130535567881,
      "learning_rate": 5.9941614214572155e-06,
      "loss": 0.4537,
      "step": 2813
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0485140470847456,
      "learning_rate": 5.9916036635077325e-06,
      "loss": 0.5831,
      "step": 2814
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6681916084973161,
      "learning_rate": 5.98904563541808e-06,
      "loss": 0.4931,
      "step": 2815
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.945187241558846,
      "learning_rate": 5.986487337885129e-06,
      "loss": 0.5558,
      "step": 2816
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0005775416640104,
      "learning_rate": 5.983928771605839e-06,
      "loss": 0.5851,
      "step": 2817
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9864974883270706,
      "learning_rate": 5.981369937277226e-06,
      "loss": 0.5818,
      "step": 2818
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9434374588602612,
      "learning_rate": 5.978810835596392e-06,
      "loss": 0.47,
      "step": 2819
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9362136348598245,
      "learning_rate": 5.976251467260505e-06,
      "loss": 0.588,
      "step": 2820
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9543007793584634,
      "learning_rate": 5.9736918329668094e-06,
      "loss": 0.6151,
      "step": 2821
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9769119770876596,
      "learning_rate": 5.971131933412617e-06,
      "loss": 0.5705,
      "step": 2822
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.065116302963071,
      "learning_rate": 5.9685717692953175e-06,
      "loss": 0.7212,
      "step": 2823
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9654754066893405,
      "learning_rate": 5.966011341312368e-06,
      "loss": 0.5285,
      "step": 2824
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0447922236417653,
      "learning_rate": 5.9634506501613006e-06,
      "loss": 0.6868,
      "step": 2825
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0907522488480732,
      "learning_rate": 5.960889696539719e-06,
      "loss": 0.5415,
      "step": 2826
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9861072523253314,
      "learning_rate": 5.958328481145297e-06,
      "loss": 0.6213,
      "step": 2827
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9826569307630312,
      "learning_rate": 5.955767004675781e-06,
      "loss": 0.5936,
      "step": 2828
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8880778126550937,
      "learning_rate": 5.953205267828985e-06,
      "loss": 0.5613,
      "step": 2829
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.061051830151999,
      "learning_rate": 5.950643271302802e-06,
      "loss": 0.666,
      "step": 2830
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9862833098359478,
      "learning_rate": 5.948081015795184e-06,
      "loss": 0.6226,
      "step": 2831
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0115595722546844,
      "learning_rate": 5.945518502004165e-06,
      "loss": 0.6191,
      "step": 2832
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9693306457139176,
      "learning_rate": 5.942955730627843e-06,
      "loss": 0.6121,
      "step": 2833
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6950079262184486,
      "learning_rate": 5.940392702364387e-06,
      "loss": 0.4948,
      "step": 2834
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0067215949947752,
      "learning_rate": 5.9378294179120355e-06,
      "loss": 0.6243,
      "step": 2835
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0260249904961083,
      "learning_rate": 5.935265877969101e-06,
      "loss": 0.6728,
      "step": 2836
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6741729777060154,
      "learning_rate": 5.932702083233959e-06,
      "loss": 0.4638,
      "step": 2837
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0443448028452096,
      "learning_rate": 5.9301380344050595e-06,
      "loss": 0.5679,
      "step": 2838
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0515865803300575,
      "learning_rate": 5.927573732180923e-06,
      "loss": 0.593,
      "step": 2839
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0009193903112188,
      "learning_rate": 5.925009177260128e-06,
      "loss": 0.5751,
      "step": 2840
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.040483844679302,
      "learning_rate": 5.922444370341339e-06,
      "loss": 0.6243,
      "step": 2841
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0243246247365925,
      "learning_rate": 5.919879312123276e-06,
      "loss": 0.679,
      "step": 2842
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0714410709058184,
      "learning_rate": 5.91731400330473e-06,
      "loss": 0.6655,
      "step": 2843
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6576500400723551,
      "learning_rate": 5.914748444584565e-06,
      "loss": 0.4861,
      "step": 2844
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9951407951673544,
      "learning_rate": 5.912182636661707e-06,
      "loss": 0.6328,
      "step": 2845
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.96199168652253,
      "learning_rate": 5.9096165802351555e-06,
      "loss": 0.673,
      "step": 2846
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0435814384747515,
      "learning_rate": 5.907050276003974e-06,
      "loss": 0.6391,
      "step": 2847
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9814886288719546,
      "learning_rate": 5.904483724667294e-06,
      "loss": 0.6067,
      "step": 2848
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9533904715207262,
      "learning_rate": 5.901916926924314e-06,
      "loss": 0.6376,
      "step": 2849
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.14037421516871,
      "learning_rate": 5.899349883474303e-06,
      "loss": 0.6104,
      "step": 2850
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1356659729894598,
      "learning_rate": 5.896782595016594e-06,
      "loss": 0.6797,
      "step": 2851
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9571935450125075,
      "learning_rate": 5.894215062250584e-06,
      "loss": 0.5546,
      "step": 2852
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0448126872165153,
      "learning_rate": 5.891647285875742e-06,
      "loss": 0.6283,
      "step": 2853
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0202340053770433,
      "learning_rate": 5.889079266591602e-06,
      "loss": 0.632,
      "step": 2854
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7245235718193215,
      "learning_rate": 5.886511005097762e-06,
      "loss": 0.5008,
      "step": 2855
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.037303779554695,
      "learning_rate": 5.883942502093887e-06,
      "loss": 0.7077,
      "step": 2856
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9626651622238032,
      "learning_rate": 5.881373758279709e-06,
      "loss": 0.5808,
      "step": 2857
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0799903077324495,
      "learning_rate": 5.878804774355024e-06,
      "loss": 0.5611,
      "step": 2858
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9119368640953579,
      "learning_rate": 5.876235551019692e-06,
      "loss": 0.5494,
      "step": 2859
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9325781799811966,
      "learning_rate": 5.873666088973644e-06,
      "loss": 0.6375,
      "step": 2860
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9580425266679156,
      "learning_rate": 5.871096388916873e-06,
      "loss": 0.5708,
      "step": 2861
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9949791635089573,
      "learning_rate": 5.86852645154943e-06,
      "loss": 0.6741,
      "step": 2862
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8429102684011858,
      "learning_rate": 5.865956277571441e-06,
      "loss": 0.5482,
      "step": 2863
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7153777408979878,
      "learning_rate": 5.863385867683093e-06,
      "loss": 0.4925,
      "step": 2864
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.021691894737005,
      "learning_rate": 5.860815222584635e-06,
      "loss": 0.5671,
      "step": 2865
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.04003435400023,
      "learning_rate": 5.858244342976383e-06,
      "loss": 0.5376,
      "step": 2866
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0153299438045589,
      "learning_rate": 5.855673229558711e-06,
      "loss": 0.6282,
      "step": 2867
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9760824607151967,
      "learning_rate": 5.853101883032069e-06,
      "loss": 0.5719,
      "step": 2868
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9229037560765895,
      "learning_rate": 5.850530304096956e-06,
      "loss": 0.5132,
      "step": 2869
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.134357325170705,
      "learning_rate": 5.847958493453941e-06,
      "loss": 0.6539,
      "step": 2870
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0145327777779447,
      "learning_rate": 5.845386451803661e-06,
      "loss": 0.6205,
      "step": 2871
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9325220494417922,
      "learning_rate": 5.8428141798468086e-06,
      "loss": 0.5998,
      "step": 2872
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9665798595348647,
      "learning_rate": 5.84024167828414e-06,
      "loss": 0.5656,
      "step": 2873
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9660355541230701,
      "learning_rate": 5.837668947816476e-06,
      "loss": 0.6228,
      "step": 2874
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9758271768920974,
      "learning_rate": 5.835095989144701e-06,
      "loss": 0.4941,
      "step": 2875
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9952800449274192,
      "learning_rate": 5.83252280296976e-06,
      "loss": 0.5789,
      "step": 2876
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0071803209791454,
      "learning_rate": 5.8299493899926584e-06,
      "loss": 0.6329,
      "step": 2877
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0330822679323406,
      "learning_rate": 5.827375750914463e-06,
      "loss": 0.5953,
      "step": 2878
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9920242991794651,
      "learning_rate": 5.82480188643631e-06,
      "loss": 0.5618,
      "step": 2879
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.908463964742955,
      "learning_rate": 5.822227797259385e-06,
      "loss": 0.5591,
      "step": 2880
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9734249425447155,
      "learning_rate": 5.819653484084944e-06,
      "loss": 0.5582,
      "step": 2881
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0482688264227953,
      "learning_rate": 5.817078947614299e-06,
      "loss": 0.6518,
      "step": 2882
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9982005040609158,
      "learning_rate": 5.814504188548828e-06,
      "loss": 0.5804,
      "step": 2883
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0406878893692733,
      "learning_rate": 5.811929207589963e-06,
      "loss": 0.6547,
      "step": 2884
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9929599008526521,
      "learning_rate": 5.8093540054392e-06,
      "loss": 0.6201,
      "step": 2885
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1267583672529835,
      "learning_rate": 5.806778582798096e-06,
      "loss": 0.6606,
      "step": 2886
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0224621836707042,
      "learning_rate": 5.804202940368267e-06,
      "loss": 0.624,
      "step": 2887
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9707155779879098,
      "learning_rate": 5.801627078851391e-06,
      "loss": 0.4935,
      "step": 2888
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6748279549493201,
      "learning_rate": 5.799050998949199e-06,
      "loss": 0.4731,
      "step": 2889
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.05949852412725,
      "learning_rate": 5.796474701363491e-06,
      "loss": 0.6366,
      "step": 2890
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9535495776999585,
      "learning_rate": 5.793898186796118e-06,
      "loss": 0.5655,
      "step": 2891
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9198737751283886,
      "learning_rate": 5.791321455948996e-06,
      "loss": 0.5518,
      "step": 2892
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0418289728507253,
      "learning_rate": 5.7887445095240954e-06,
      "loss": 0.6887,
      "step": 2893
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9784598122698184,
      "learning_rate": 5.78616734822345e-06,
      "loss": 0.5411,
      "step": 2894
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9686931736203592,
      "learning_rate": 5.783589972749149e-06,
      "loss": 0.6016,
      "step": 2895
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7065970217542885,
      "learning_rate": 5.781012383803337e-06,
      "loss": 0.4934,
      "step": 2896
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9767568755009245,
      "learning_rate": 5.7784345820882235e-06,
      "loss": 0.6501,
      "step": 2897
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9842211547307382,
      "learning_rate": 5.775856568306073e-06,
      "loss": 0.6504,
      "step": 2898
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9187687725756604,
      "learning_rate": 5.773278343159207e-06,
      "loss": 0.6305,
      "step": 2899
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9478112891604635,
      "learning_rate": 5.770699907350003e-06,
      "loss": 0.606,
      "step": 2900
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9970778068140749,
      "learning_rate": 5.768121261580901e-06,
      "loss": 0.6595,
      "step": 2901
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0027512925106365,
      "learning_rate": 5.765542406554394e-06,
      "loss": 0.611,
      "step": 2902
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6473420049760968,
      "learning_rate": 5.762963342973033e-06,
      "loss": 0.4431,
      "step": 2903
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9995884499091411,
      "learning_rate": 5.7603840715394264e-06,
      "loss": 0.616,
      "step": 2904
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9601077924899245,
      "learning_rate": 5.757804592956237e-06,
      "loss": 0.5939,
      "step": 2905
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9412940288685271,
      "learning_rate": 5.75522490792619e-06,
      "loss": 0.5359,
      "step": 2906
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9398322951543423,
      "learning_rate": 5.752645017152058e-06,
      "loss": 0.5587,
      "step": 2907
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0515066278753094,
      "learning_rate": 5.750064921336679e-06,
      "loss": 0.6334,
      "step": 2908
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9814663356076286,
      "learning_rate": 5.7474846211829376e-06,
      "loss": 0.5837,
      "step": 2909
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.975444376733813,
      "learning_rate": 5.7449041173937825e-06,
      "loss": 0.5639,
      "step": 2910
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9819154176096044,
      "learning_rate": 5.7423234106722105e-06,
      "loss": 0.58,
      "step": 2911
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9355672498184845,
      "learning_rate": 5.739742501721282e-06,
      "loss": 0.6102,
      "step": 2912
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.042837747463878,
      "learning_rate": 5.7371613912441015e-06,
      "loss": 0.6252,
      "step": 2913
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9489000653581348,
      "learning_rate": 5.73458007994384e-06,
      "loss": 0.5359,
      "step": 2914
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0667492205073674,
      "learning_rate": 5.731998568523716e-06,
      "loss": 0.5718,
      "step": 2915
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0658525638087435,
      "learning_rate": 5.729416857687002e-06,
      "loss": 0.572,
      "step": 2916
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.906314885453991,
      "learning_rate": 5.726834948137033e-06,
      "loss": 0.5979,
      "step": 2917
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6527965997067648,
      "learning_rate": 5.7242528405771865e-06,
      "loss": 0.5006,
      "step": 2918
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9745269175018161,
      "learning_rate": 5.721670535710902e-06,
      "loss": 0.59,
      "step": 2919
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9606615458988463,
      "learning_rate": 5.7190880342416725e-06,
      "loss": 0.5138,
      "step": 2920
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9215391451222117,
      "learning_rate": 5.7165053368730395e-06,
      "loss": 0.5162,
      "step": 2921
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9784007233180673,
      "learning_rate": 5.713922444308601e-06,
      "loss": 0.6312,
      "step": 2922
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0467927211741697,
      "learning_rate": 5.711339357252011e-06,
      "loss": 0.6162,
      "step": 2923
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0149131340285158,
      "learning_rate": 5.708756076406972e-06,
      "loss": 0.6831,
      "step": 2924
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9229247982096688,
      "learning_rate": 5.706172602477239e-06,
      "loss": 0.565,
      "step": 2925
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9466669804342748,
      "learning_rate": 5.703588936166625e-06,
      "loss": 0.5746,
      "step": 2926
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0500888447560603,
      "learning_rate": 5.701005078178989e-06,
      "loss": 0.6598,
      "step": 2927
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.971042840476184,
      "learning_rate": 5.6984210292182485e-06,
      "loss": 0.6464,
      "step": 2928
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6460691474961241,
      "learning_rate": 5.695836789988365e-06,
      "loss": 0.4842,
      "step": 2929
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9339738886626757,
      "learning_rate": 5.6932523611933595e-06,
      "loss": 0.5478,
      "step": 2930
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0590730115011948,
      "learning_rate": 5.690667743537303e-06,
      "loss": 0.6744,
      "step": 2931
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0031172910016795,
      "learning_rate": 5.688082937724314e-06,
      "loss": 0.5452,
      "step": 2932
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9859394480193693,
      "learning_rate": 5.685497944458564e-06,
      "loss": 0.6006,
      "step": 2933
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0720091468741184,
      "learning_rate": 5.68291276444428e-06,
      "loss": 0.686,
      "step": 2934
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0638350428115457,
      "learning_rate": 5.680327398385735e-06,
      "loss": 0.6147,
      "step": 2935
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0443623935295188,
      "learning_rate": 5.677741846987251e-06,
      "loss": 0.5785,
      "step": 2936
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9460562663119576,
      "learning_rate": 5.675156110953208e-06,
      "loss": 0.591,
      "step": 2937
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9840086478433832,
      "learning_rate": 5.672570190988027e-06,
      "loss": 0.5703,
      "step": 2938
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9655825815455247,
      "learning_rate": 5.6699840877961884e-06,
      "loss": 0.6394,
      "step": 2939
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9383340524609113,
      "learning_rate": 5.667397802082216e-06,
      "loss": 0.5584,
      "step": 2940
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6824580848495375,
      "learning_rate": 5.664811334550685e-06,
      "loss": 0.4696,
      "step": 2941
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9734839878046732,
      "learning_rate": 5.6622246859062205e-06,
      "loss": 0.6223,
      "step": 2942
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9976258656055123,
      "learning_rate": 5.6596378568534984e-06,
      "loss": 0.5596,
      "step": 2943
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1616181664783893,
      "learning_rate": 5.657050848097239e-06,
      "loss": 0.6638,
      "step": 2944
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.054097453906456,
      "learning_rate": 5.654463660342219e-06,
      "loss": 0.5254,
      "step": 2945
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9798027750585353,
      "learning_rate": 5.651876294293256e-06,
      "loss": 0.5791,
      "step": 2946
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.034999689480227,
      "learning_rate": 5.649288750655222e-06,
      "loss": 0.61,
      "step": 2947
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9716036140434269,
      "learning_rate": 5.646701030133036e-06,
      "loss": 0.6397,
      "step": 2948
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.023438271244348,
      "learning_rate": 5.64411313343166e-06,
      "loss": 0.6162,
      "step": 2949
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.011932914375525,
      "learning_rate": 5.641525061256113e-06,
      "loss": 0.5884,
      "step": 2950
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0566773371453246,
      "learning_rate": 5.638936814311455e-06,
      "loss": 0.5718,
      "step": 2951
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9660460607165937,
      "learning_rate": 5.636348393302798e-06,
      "loss": 0.6475,
      "step": 2952
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9415786387167717,
      "learning_rate": 5.633759798935296e-06,
      "loss": 0.628,
      "step": 2953
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0725500159297616,
      "learning_rate": 5.631171031914155e-06,
      "loss": 0.6822,
      "step": 2954
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.046320208223975,
      "learning_rate": 5.628582092944628e-06,
      "loss": 0.5688,
      "step": 2955
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.984616930248111,
      "learning_rate": 5.625992982732011e-06,
      "loss": 0.6191,
      "step": 2956
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.014814606010456,
      "learning_rate": 5.62340370198165e-06,
      "loss": 0.5978,
      "step": 2957
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0526415672892597,
      "learning_rate": 5.620814251398936e-06,
      "loss": 0.6085,
      "step": 2958
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9986443605901345,
      "learning_rate": 5.6182246316893086e-06,
      "loss": 0.622,
      "step": 2959
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9748661628529793,
      "learning_rate": 5.615634843558248e-06,
      "loss": 0.5738,
      "step": 2960
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0362174043701942,
      "learning_rate": 5.613044887711288e-06,
      "loss": 0.6715,
      "step": 2961
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0854737732401636,
      "learning_rate": 5.6104547648539995e-06,
      "loss": 0.6683,
      "step": 2962
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9787029480476527,
      "learning_rate": 5.6078644756920066e-06,
      "loss": 0.5822,
      "step": 2963
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0308883299220237,
      "learning_rate": 5.605274020930975e-06,
      "loss": 0.6706,
      "step": 2964
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0433236334143845,
      "learning_rate": 5.6026834012766155e-06,
      "loss": 0.5991,
      "step": 2965
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1390599260896506,
      "learning_rate": 5.600092617434686e-06,
      "loss": 0.6231,
      "step": 2966
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9848717212754493,
      "learning_rate": 5.597501670110984e-06,
      "loss": 0.5963,
      "step": 2967
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0317528875631188,
      "learning_rate": 5.594910560011356e-06,
      "loss": 0.6093,
      "step": 2968
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9590709450128417,
      "learning_rate": 5.592319287841694e-06,
      "loss": 0.6131,
      "step": 2969
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9969229727161726,
      "learning_rate": 5.589727854307932e-06,
      "loss": 0.5557,
      "step": 2970
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0186536378162427,
      "learning_rate": 5.587136260116044e-06,
      "loss": 0.591,
      "step": 2971
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9823264422474592,
      "learning_rate": 5.584544505972056e-06,
      "loss": 0.6239,
      "step": 2972
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9911094580054917,
      "learning_rate": 5.58195259258203e-06,
      "loss": 0.5878,
      "step": 2973
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9698722558679959,
      "learning_rate": 5.5793605206520776e-06,
      "loss": 0.6192,
      "step": 2974
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.027272684064404,
      "learning_rate": 5.576768290888348e-06,
      "loss": 0.6053,
      "step": 2975
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9149003480513932,
      "learning_rate": 5.574175903997038e-06,
      "loss": 0.5614,
      "step": 2976
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.011394312007025,
      "learning_rate": 5.571583360684385e-06,
      "loss": 0.604,
      "step": 2977
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9940550443182878,
      "learning_rate": 5.568990661656668e-06,
      "loss": 0.6088,
      "step": 2978
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9971869309873015,
      "learning_rate": 5.566397807620209e-06,
      "loss": 0.6147,
      "step": 2979
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1043937946642919,
      "learning_rate": 5.563804799281377e-06,
      "loss": 0.6082,
      "step": 2980
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9330190083213578,
      "learning_rate": 5.561211637346576e-06,
      "loss": 0.5153,
      "step": 2981
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.973954067031446,
      "learning_rate": 5.558618322522253e-06,
      "loss": 0.6328,
      "step": 2982
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0117180263642327,
      "learning_rate": 5.556024855514904e-06,
      "loss": 0.6376,
      "step": 2983
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0677488480021098,
      "learning_rate": 5.553431237031054e-06,
      "loss": 0.6833,
      "step": 2984
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6389923541318332,
      "learning_rate": 5.5508374677772806e-06,
      "loss": 0.4692,
      "step": 2985
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9729807122934176,
      "learning_rate": 5.548243548460197e-06,
      "loss": 0.5252,
      "step": 2986
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9365279704550078,
      "learning_rate": 5.545649479786459e-06,
      "loss": 0.5309,
      "step": 2987
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0561609247980046,
      "learning_rate": 5.543055262462762e-06,
      "loss": 0.6651,
      "step": 2988
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9665930783657178,
      "learning_rate": 5.540460897195842e-06,
      "loss": 0.5441,
      "step": 2989
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0446924921264358,
      "learning_rate": 5.537866384692474e-06,
      "loss": 0.6535,
      "step": 2990
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9863250907770889,
      "learning_rate": 5.535271725659478e-06,
      "loss": 0.5865,
      "step": 2991
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9839658211112265,
      "learning_rate": 5.53267692080371e-06,
      "loss": 0.5171,
      "step": 2992
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.007925281747917,
      "learning_rate": 5.530081970832063e-06,
      "loss": 0.6195,
      "step": 2993
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.609024433343191,
      "learning_rate": 5.527486876451478e-06,
      "loss": 0.4639,
      "step": 2994
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9271836706238477,
      "learning_rate": 5.524891638368926e-06,
      "loss": 0.517,
      "step": 2995
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.994170977424247,
      "learning_rate": 5.522296257291422e-06,
      "loss": 0.6015,
      "step": 2996
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1118335253691134,
      "learning_rate": 5.519700733926022e-06,
      "loss": 0.6087,
      "step": 2997
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6571332502038161,
      "learning_rate": 5.517105068979816e-06,
      "loss": 0.4622,
      "step": 2998
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9272507738478353,
      "learning_rate": 5.514509263159935e-06,
      "loss": 0.6516,
      "step": 2999
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0462342525215655,
      "learning_rate": 5.511913317173548e-06,
      "loss": 0.6291,
      "step": 3000
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.936869178129046,
      "learning_rate": 5.509317231727863e-06,
      "loss": 0.5081,
      "step": 3001
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.029494923829399,
      "learning_rate": 5.506721007530125e-06,
      "loss": 0.5566,
      "step": 3002
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9725678518237109,
      "learning_rate": 5.504124645287616e-06,
      "loss": 0.569,
      "step": 3003
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0306481944792303,
      "learning_rate": 5.501528145707657e-06,
      "loss": 0.5281,
      "step": 3004
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9443282740213054,
      "learning_rate": 5.498931509497607e-06,
      "loss": 0.6326,
      "step": 3005
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9038622756152291,
      "learning_rate": 5.4963347373648615e-06,
      "loss": 0.5238,
      "step": 3006
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9778305294949731,
      "learning_rate": 5.49373783001685e-06,
      "loss": 0.5942,
      "step": 3007
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9653584012944149,
      "learning_rate": 5.4911407881610455e-06,
      "loss": 0.613,
      "step": 3008
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0003753597047107,
      "learning_rate": 5.48854361250495e-06,
      "loss": 0.6187,
      "step": 3009
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.070530346537155,
      "learning_rate": 5.48594630375611e-06,
      "loss": 0.6225,
      "step": 3010
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9367117191076113,
      "learning_rate": 5.4833488626221e-06,
      "loss": 0.5321,
      "step": 3011
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7029591564399963,
      "learning_rate": 5.480751289810537e-06,
      "loss": 0.5008,
      "step": 3012
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0371518262102353,
      "learning_rate": 5.478153586029069e-06,
      "loss": 0.5597,
      "step": 3013
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.070595877657481,
      "learning_rate": 5.4755557519853854e-06,
      "loss": 0.6365,
      "step": 3014
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0666471393524088,
      "learning_rate": 5.472957788387206e-06,
      "loss": 0.6621,
      "step": 3015
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.013622463994857,
      "learning_rate": 5.470359695942289e-06,
      "loss": 0.6112,
      "step": 3016
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1751072458337048,
      "learning_rate": 5.467761475358423e-06,
      "loss": 0.6891,
      "step": 3017
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9358695068154921,
      "learning_rate": 5.465163127343438e-06,
      "loss": 0.5813,
      "step": 3018
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.047044540762912,
      "learning_rate": 5.462564652605197e-06,
      "loss": 0.6275,
      "step": 3019
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9099149676891721,
      "learning_rate": 5.459966051851589e-06,
      "loss": 0.5174,
      "step": 3020
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9671128862280134,
      "learning_rate": 5.457367325790555e-06,
      "loss": 0.6282,
      "step": 3021
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.113845901514489,
      "learning_rate": 5.454768475130051e-06,
      "loss": 0.646,
      "step": 3022
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0459357585850415,
      "learning_rate": 5.452169500578079e-06,
      "loss": 0.6456,
      "step": 3023
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.922231377513865,
      "learning_rate": 5.449570402842671e-06,
      "loss": 0.5915,
      "step": 3024
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9392697318301192,
      "learning_rate": 5.446971182631893e-06,
      "loss": 0.6032,
      "step": 3025
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1390953849131236,
      "learning_rate": 5.444371840653843e-06,
      "loss": 0.6191,
      "step": 3026
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9539435396481226,
      "learning_rate": 5.441772377616656e-06,
      "loss": 0.5335,
      "step": 3027
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0552984394827207,
      "learning_rate": 5.439172794228495e-06,
      "loss": 0.5907,
      "step": 3028
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9659820268226376,
      "learning_rate": 5.436573091197559e-06,
      "loss": 0.6172,
      "step": 3029
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8864818214482333,
      "learning_rate": 5.43397326923208e-06,
      "loss": 0.4983,
      "step": 3030
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0274360810404572,
      "learning_rate": 5.431373329040317e-06,
      "loss": 0.5689,
      "step": 3031
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9950136640603137,
      "learning_rate": 5.42877327133057e-06,
      "loss": 0.5067,
      "step": 3032
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9591756514245183,
      "learning_rate": 5.426173096811166e-06,
      "loss": 0.5689,
      "step": 3033
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0299236156333356,
      "learning_rate": 5.423572806190461e-06,
      "loss": 0.6553,
      "step": 3034
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0771270259381647,
      "learning_rate": 5.420972400176848e-06,
      "loss": 0.6876,
      "step": 3035
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9524806173014413,
      "learning_rate": 5.41837187947875e-06,
      "loss": 0.5057,
      "step": 3036
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.647660965685755,
      "learning_rate": 5.41577124480462e-06,
      "loss": 0.4907,
      "step": 3037
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0137313883631436,
      "learning_rate": 5.413170496862941e-06,
      "loss": 0.6225,
      "step": 3038
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.986362827004136,
      "learning_rate": 5.4105696363622284e-06,
      "loss": 0.6063,
      "step": 3039
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1103650922467134,
      "learning_rate": 5.4079686640110305e-06,
      "loss": 0.687,
      "step": 3040
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9551687305777193,
      "learning_rate": 5.405367580517924e-06,
      "loss": 0.6027,
      "step": 3041
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9952775875541707,
      "learning_rate": 5.402766386591511e-06,
      "loss": 0.5803,
      "step": 3042
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.966615474914266,
      "learning_rate": 5.400165082940436e-06,
      "loss": 0.6194,
      "step": 3043
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0046841789499001,
      "learning_rate": 5.397563670273359e-06,
      "loss": 0.6401,
      "step": 3044
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9366289362668158,
      "learning_rate": 5.394962149298981e-06,
      "loss": 0.5246,
      "step": 3045
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9783160899744681,
      "learning_rate": 5.392360520726027e-06,
      "loss": 0.6505,
      "step": 3046
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9770758549753037,
      "learning_rate": 5.38975878526325e-06,
      "loss": 0.6113,
      "step": 3047
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0213627205930274,
      "learning_rate": 5.387156943619437e-06,
      "loss": 0.5857,
      "step": 3048
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0750538410657855,
      "learning_rate": 5.384554996503401e-06,
      "loss": 0.4824,
      "step": 3049
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9825977248911135,
      "learning_rate": 5.381952944623982e-06,
      "loss": 0.6208,
      "step": 3050
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9226608540394012,
      "learning_rate": 5.3793507886900535e-06,
      "loss": 0.604,
      "step": 3051
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9753211072132864,
      "learning_rate": 5.3767485294105135e-06,
      "loss": 0.596,
      "step": 3052
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0082598425233422,
      "learning_rate": 5.374146167494286e-06,
      "loss": 0.5621,
      "step": 3053
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.976669264124965,
      "learning_rate": 5.371543703650332e-06,
      "loss": 0.5898,
      "step": 3054
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.972239347324036,
      "learning_rate": 5.368941138587631e-06,
      "loss": 0.5775,
      "step": 3055
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9672805681186081,
      "learning_rate": 5.366338473015195e-06,
      "loss": 0.6467,
      "step": 3056
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9966797134327225,
      "learning_rate": 5.3637357076420595e-06,
      "loss": 0.618,
      "step": 3057
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.02542895058069,
      "learning_rate": 5.36113284317729e-06,
      "loss": 0.5879,
      "step": 3058
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0151860729866955,
      "learning_rate": 5.358529880329982e-06,
      "loss": 0.5825,
      "step": 3059
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0601162942182552,
      "learning_rate": 5.355926819809248e-06,
      "loss": 0.5881,
      "step": 3060
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9645723235164121,
      "learning_rate": 5.3533236623242405e-06,
      "loss": 0.5126,
      "step": 3061
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0200621614675258,
      "learning_rate": 5.3507204085841256e-06,
      "loss": 0.5828,
      "step": 3062
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9255197244764944,
      "learning_rate": 5.348117059298105e-06,
      "loss": 0.6006,
      "step": 3063
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0290562933389957,
      "learning_rate": 5.345513615175401e-06,
      "loss": 0.6803,
      "step": 3064
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9543669687354582,
      "learning_rate": 5.342910076925264e-06,
      "loss": 0.636,
      "step": 3065
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1155655322458797,
      "learning_rate": 5.3403064452569685e-06,
      "loss": 0.6432,
      "step": 3066
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.046370806515194,
      "learning_rate": 5.337702720879819e-06,
      "loss": 0.6251,
      "step": 3067
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6187576687078945,
      "learning_rate": 5.335098904503139e-06,
      "loss": 0.4402,
      "step": 3068
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0319952196748736,
      "learning_rate": 5.332494996836279e-06,
      "loss": 0.5977,
      "step": 3069
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0407564670659808,
      "learning_rate": 5.3298909985886195e-06,
      "loss": 0.5804,
      "step": 3070
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0196830922707836,
      "learning_rate": 5.327286910469556e-06,
      "loss": 0.6094,
      "step": 3071
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0014782755851042,
      "learning_rate": 5.3246827331885165e-06,
      "loss": 0.603,
      "step": 3072
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1232367538615327,
      "learning_rate": 5.322078467454951e-06,
      "loss": 0.6418,
      "step": 3073
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0375479765497255,
      "learning_rate": 5.3194741139783324e-06,
      "loss": 0.6545,
      "step": 3074
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9727802927555881,
      "learning_rate": 5.316869673468159e-06,
      "loss": 0.5534,
      "step": 3075
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0580887244480295,
      "learning_rate": 5.314265146633954e-06,
      "loss": 0.5961,
      "step": 3076
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0156370684938056,
      "learning_rate": 5.311660534185258e-06,
      "loss": 0.5958,
      "step": 3077
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0269353595009565,
      "learning_rate": 5.3090558368316415e-06,
      "loss": 0.6334,
      "step": 3078
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.009717712809675,
      "learning_rate": 5.306451055282696e-06,
      "loss": 0.6815,
      "step": 3079
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0712421098110942,
      "learning_rate": 5.303846190248035e-06,
      "loss": 0.6059,
      "step": 3080
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9813017799082245,
      "learning_rate": 5.301241242437299e-06,
      "loss": 0.6531,
      "step": 3081
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0460458862001938,
      "learning_rate": 5.298636212560143e-06,
      "loss": 0.7166,
      "step": 3082
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1095162091104769,
      "learning_rate": 5.29603110132625e-06,
      "loss": 0.6719,
      "step": 3083
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0901207350495221,
      "learning_rate": 5.293425909445326e-06,
      "loss": 0.6551,
      "step": 3084
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0677799275461581,
      "learning_rate": 5.290820637627095e-06,
      "loss": 0.6109,
      "step": 3085
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1125464838436239,
      "learning_rate": 5.288215286581306e-06,
      "loss": 0.6873,
      "step": 3086
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.006909826725494,
      "learning_rate": 5.28560985701773e-06,
      "loss": 0.5868,
      "step": 3087
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.969084402435138,
      "learning_rate": 5.283004349646154e-06,
      "loss": 0.6709,
      "step": 3088
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8987830500424991,
      "learning_rate": 5.280398765176392e-06,
      "loss": 0.6198,
      "step": 3089
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9752011274024958,
      "learning_rate": 5.277793104318279e-06,
      "loss": 0.641,
      "step": 3090
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.957811544843052,
      "learning_rate": 5.275187367781664e-06,
      "loss": 0.5438,
      "step": 3091
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9993246830984682,
      "learning_rate": 5.272581556276428e-06,
      "loss": 0.6294,
      "step": 3092
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9265692398970193,
      "learning_rate": 5.26997567051246e-06,
      "loss": 0.5151,
      "step": 3093
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0231563731950963,
      "learning_rate": 5.267369711199678e-06,
      "loss": 0.6316,
      "step": 3094
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9871853345824139,
      "learning_rate": 5.2647636790480175e-06,
      "loss": 0.5517,
      "step": 3095
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0448852004407465,
      "learning_rate": 5.26215757476743e-06,
      "loss": 0.6879,
      "step": 3096
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1488055347980102,
      "learning_rate": 5.259551399067895e-06,
      "loss": 0.6267,
      "step": 3097
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0150334138255017,
      "learning_rate": 5.256945152659404e-06,
      "loss": 0.58,
      "step": 3098
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9554687164052483,
      "learning_rate": 5.254338836251971e-06,
      "loss": 0.5876,
      "step": 3099
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.912083311863043,
      "learning_rate": 5.251732450555626e-06,
      "loss": 0.6198,
      "step": 3100
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9516639827121232,
      "learning_rate": 5.249125996280426e-06,
      "loss": 0.5529,
      "step": 3101
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0612828208102298,
      "learning_rate": 5.246519474136433e-06,
      "loss": 0.6349,
      "step": 3102
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0379666184225487,
      "learning_rate": 5.243912884833744e-06,
      "loss": 0.6012,
      "step": 3103
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9509187204776852,
      "learning_rate": 5.241306229082459e-06,
      "loss": 0.558,
      "step": 3104
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0056121610890216,
      "learning_rate": 5.2386995075927074e-06,
      "loss": 0.5798,
      "step": 3105
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7191349098112629,
      "learning_rate": 5.236092721074629e-06,
      "loss": 0.4819,
      "step": 3106
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0226666027622446,
      "learning_rate": 5.233485870238385e-06,
      "loss": 0.571,
      "step": 3107
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1289147239160917,
      "learning_rate": 5.2308789557941546e-06,
      "loss": 0.5952,
      "step": 3108
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.010917397290751,
      "learning_rate": 5.228271978452133e-06,
      "loss": 0.6295,
      "step": 3109
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0888974285976984,
      "learning_rate": 5.225664938922531e-06,
      "loss": 0.5459,
      "step": 3110
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9572816224649667,
      "learning_rate": 5.223057837915579e-06,
      "loss": 0.5315,
      "step": 3111
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9918979931515884,
      "learning_rate": 5.220450676141525e-06,
      "loss": 0.5214,
      "step": 3112
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9856326420778148,
      "learning_rate": 5.217843454310628e-06,
      "loss": 0.6189,
      "step": 3113
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9542165125860546,
      "learning_rate": 5.215236173133172e-06,
      "loss": 0.5794,
      "step": 3114
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9316873099032466,
      "learning_rate": 5.2126288333194455e-06,
      "loss": 0.551,
      "step": 3115
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8982918487614081,
      "learning_rate": 5.210021435579766e-06,
      "loss": 0.5019,
      "step": 3116
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0252266053892645,
      "learning_rate": 5.207413980624459e-06,
      "loss": 0.6235,
      "step": 3117
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0880284171151882,
      "learning_rate": 5.204806469163865e-06,
      "loss": 0.6904,
      "step": 3118
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0381679697602924,
      "learning_rate": 5.202198901908343e-06,
      "loss": 0.6294,
      "step": 3119
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9200688863420144,
      "learning_rate": 5.199591279568268e-06,
      "loss": 0.5163,
      "step": 3120
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9293583855976372,
      "learning_rate": 5.196983602854024e-06,
      "loss": 0.5243,
      "step": 3121
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0254220433600487,
      "learning_rate": 5.1943758724760185e-06,
      "loss": 0.6222,
      "step": 3122
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.047226987845554,
      "learning_rate": 5.191768089144668e-06,
      "loss": 0.5761,
      "step": 3123
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9147856117445466,
      "learning_rate": 5.189160253570402e-06,
      "loss": 0.4939,
      "step": 3124
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7415240357698905,
      "learning_rate": 5.186552366463671e-06,
      "loss": 0.4881,
      "step": 3125
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6322239423134934,
      "learning_rate": 5.18394442853493e-06,
      "loss": 0.4703,
      "step": 3126
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9832375534920925,
      "learning_rate": 5.181336440494659e-06,
      "loss": 0.5202,
      "step": 3127
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0929385540764123,
      "learning_rate": 5.178728403053342e-06,
      "loss": 0.7039,
      "step": 3128
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0110602711965202,
      "learning_rate": 5.176120316921481e-06,
      "loss": 0.6393,
      "step": 3129
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0634345474252345,
      "learning_rate": 5.173512182809592e-06,
      "loss": 0.6479,
      "step": 3130
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0545830695103025,
      "learning_rate": 5.170904001428201e-06,
      "loss": 0.6091,
      "step": 3131
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9651603731949634,
      "learning_rate": 5.168295773487848e-06,
      "loss": 0.551,
      "step": 3132
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.910173616825386,
      "learning_rate": 5.165687499699088e-06,
      "loss": 0.5797,
      "step": 3133
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0465306525543627,
      "learning_rate": 5.163079180772486e-06,
      "loss": 0.6389,
      "step": 3134
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0140264239980943,
      "learning_rate": 5.160470817418619e-06,
      "loss": 0.5651,
      "step": 3135
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0865930450129375,
      "learning_rate": 5.15786241034808e-06,
      "loss": 0.7361,
      "step": 3136
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.073680740320688,
      "learning_rate": 5.155253960271466e-06,
      "loss": 0.6918,
      "step": 3137
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8618014706965529,
      "learning_rate": 5.152645467899397e-06,
      "loss": 0.5424,
      "step": 3138
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1115292451014795,
      "learning_rate": 5.150036933942492e-06,
      "loss": 0.7406,
      "step": 3139
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0586927228750298,
      "learning_rate": 5.147428359111391e-06,
      "loss": 0.6194,
      "step": 3140
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9786524039263109,
      "learning_rate": 5.144819744116742e-06,
      "loss": 0.5958,
      "step": 3141
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9043824250121534,
      "learning_rate": 5.142211089669202e-06,
      "loss": 0.5477,
      "step": 3142
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0340517383668435,
      "learning_rate": 5.1396023964794415e-06,
      "loss": 0.5947,
      "step": 3143
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9702461993920194,
      "learning_rate": 5.1369936652581395e-06,
      "loss": 0.5259,
      "step": 3144
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8553168746533659,
      "learning_rate": 5.134384896715987e-06,
      "loss": 0.4923,
      "step": 3145
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9816506245386065,
      "learning_rate": 5.131776091563685e-06,
      "loss": 0.5675,
      "step": 3146
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9669388424297567,
      "learning_rate": 5.129167250511943e-06,
      "loss": 0.5951,
      "step": 3147
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9871660852011059,
      "learning_rate": 5.12655837427148e-06,
      "loss": 0.6006,
      "step": 3148
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9990625613082773,
      "learning_rate": 5.123949463553029e-06,
      "loss": 0.5699,
      "step": 3149
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0329120251833712,
      "learning_rate": 5.121340519067327e-06,
      "loss": 0.6798,
      "step": 3150
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0425169803595384,
      "learning_rate": 5.118731541525123e-06,
      "loss": 0.5663,
      "step": 3151
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0871819823524007,
      "learning_rate": 5.116122531637174e-06,
      "loss": 0.6908,
      "step": 3152
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.078285683265701,
      "learning_rate": 5.113513490114246e-06,
      "loss": 0.676,
      "step": 3153
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0036849719504581,
      "learning_rate": 5.110904417667115e-06,
      "loss": 0.6757,
      "step": 3154
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0388476133534623,
      "learning_rate": 5.108295315006563e-06,
      "loss": 0.6373,
      "step": 3155
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0630591065915274,
      "learning_rate": 5.1056861828433815e-06,
      "loss": 0.5922,
      "step": 3156
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9346870041670945,
      "learning_rate": 5.1030770218883706e-06,
      "loss": 0.6295,
      "step": 3157
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0441141778922003,
      "learning_rate": 5.1004678328523395e-06,
      "loss": 0.6233,
      "step": 3158
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.034541901945305,
      "learning_rate": 5.097858616446099e-06,
      "loss": 0.6038,
      "step": 3159
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0204589788283998,
      "learning_rate": 5.0952493733804755e-06,
      "loss": 0.615,
      "step": 3160
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.931207029440427,
      "learning_rate": 5.092640104366297e-06,
      "loss": 0.5718,
      "step": 3161
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9649478944400361,
      "learning_rate": 5.0900308101143996e-06,
      "loss": 0.5562,
      "step": 3162
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.002578387231302,
      "learning_rate": 5.087421491335629e-06,
      "loss": 0.6287,
      "step": 3163
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0391535216531402,
      "learning_rate": 5.0848121487408316e-06,
      "loss": 0.6283,
      "step": 3164
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0140145805553982,
      "learning_rate": 5.08220278304087e-06,
      "loss": 0.584,
      "step": 3165
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.015182019752588,
      "learning_rate": 5.079593394946603e-06,
      "loss": 0.5474,
      "step": 3166
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.982440242242882,
      "learning_rate": 5.076983985168901e-06,
      "loss": 0.5816,
      "step": 3167
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1400493975062624,
      "learning_rate": 5.074374554418641e-06,
      "loss": 0.646,
      "step": 3168
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9698206550883761,
      "learning_rate": 5.071765103406702e-06,
      "loss": 0.5911,
      "step": 3169
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9663288582076922,
      "learning_rate": 5.06915563284397e-06,
      "loss": 0.6061,
      "step": 3170
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.073261526336657,
      "learning_rate": 5.066546143441336e-06,
      "loss": 0.6141,
      "step": 3171
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0030036543102403,
      "learning_rate": 5.0639366359097e-06,
      "loss": 0.6036,
      "step": 3172
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0723402760222287,
      "learning_rate": 5.061327110959961e-06,
      "loss": 0.5882,
      "step": 3173
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9133739858984177,
      "learning_rate": 5.058717569303027e-06,
      "loss": 0.5806,
      "step": 3174
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0029731177779446,
      "learning_rate": 5.056108011649807e-06,
      "loss": 0.6411,
      "step": 3175
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0338771199589651,
      "learning_rate": 5.053498438711221e-06,
      "loss": 0.6601,
      "step": 3176
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9925496123672746,
      "learning_rate": 5.050888851198183e-06,
      "loss": 0.5516,
      "step": 3177
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.022091983083876,
      "learning_rate": 5.0482792498216194e-06,
      "loss": 0.6339,
      "step": 3178
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9656249352266356,
      "learning_rate": 5.045669635292458e-06,
      "loss": 0.6412,
      "step": 3179
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0099292343527497,
      "learning_rate": 5.04306000832163e-06,
      "loss": 0.5973,
      "step": 3180
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9672537550692064,
      "learning_rate": 5.040450369620068e-06,
      "loss": 0.5604,
      "step": 3181
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.038917159440407,
      "learning_rate": 5.03784071989871e-06,
      "loss": 0.5997,
      "step": 3182
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9619137737291765,
      "learning_rate": 5.0352310598684965e-06,
      "loss": 0.5323,
      "step": 3183
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0424061215150913,
      "learning_rate": 5.032621390240371e-06,
      "loss": 0.6757,
      "step": 3184
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9815577959975248,
      "learning_rate": 5.030011711725281e-06,
      "loss": 0.5911,
      "step": 3185
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0146655243825957,
      "learning_rate": 5.027402025034171e-06,
      "loss": 0.6315,
      "step": 3186
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1577283696456933,
      "learning_rate": 5.024792330877997e-06,
      "loss": 0.6354,
      "step": 3187
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8999827968225897,
      "learning_rate": 5.022182629967707e-06,
      "loss": 0.5922,
      "step": 3188
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9843898738010718,
      "learning_rate": 5.019572923014258e-06,
      "loss": 0.6193,
      "step": 3189
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9105745404761089,
      "learning_rate": 5.016963210728604e-06,
      "loss": 0.5151,
      "step": 3190
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.937101883668275,
      "learning_rate": 5.014353493821705e-06,
      "loss": 0.4985,
      "step": 3191
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0613820986548346,
      "learning_rate": 5.011743773004518e-06,
      "loss": 0.573,
      "step": 3192
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9101921625096693,
      "learning_rate": 5.009134048988004e-06,
      "loss": 0.5697,
      "step": 3193
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9344871409151593,
      "learning_rate": 5.006524322483123e-06,
      "loss": 0.5767,
      "step": 3194
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9516814451468132,
      "learning_rate": 5.0039145942008364e-06,
      "loss": 0.5703,
      "step": 3195
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0388140777675396,
      "learning_rate": 5.001304864852106e-06,
      "loss": 0.665,
      "step": 3196
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9952440897527149,
      "learning_rate": 4.998695135147896e-06,
      "loss": 0.6188,
      "step": 3197
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9119715043850809,
      "learning_rate": 4.996085405799166e-06,
      "loss": 0.5529,
      "step": 3198
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.050781916803946,
      "learning_rate": 4.993475677516878e-06,
      "loss": 0.62,
      "step": 3199
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0843701732318667,
      "learning_rate": 4.990865951011999e-06,
      "loss": 0.5864,
      "step": 3200
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9863101523621242,
      "learning_rate": 4.9882562269954835e-06,
      "loss": 0.5736,
      "step": 3201
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9186119294044961,
      "learning_rate": 4.985646506178296e-06,
      "loss": 0.5541,
      "step": 3202
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9620687350222229,
      "learning_rate": 4.983036789271398e-06,
      "loss": 0.5768,
      "step": 3203
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2517755544620321,
      "learning_rate": 4.980427076985744e-06,
      "loss": 0.6364,
      "step": 3204
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9517456926756165,
      "learning_rate": 4.977817370032294e-06,
      "loss": 0.5365,
      "step": 3205
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1757747256515527,
      "learning_rate": 4.975207669122005e-06,
      "loss": 0.6424,
      "step": 3206
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9722261868761642,
      "learning_rate": 4.97259797496583e-06,
      "loss": 0.5859,
      "step": 3207
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9288341838015219,
      "learning_rate": 4.9699882882747205e-06,
      "loss": 0.6219,
      "step": 3208
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.993202144225267,
      "learning_rate": 4.967378609759629e-06,
      "loss": 0.6632,
      "step": 3209
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0292518335109053,
      "learning_rate": 4.964768940131505e-06,
      "loss": 0.6361,
      "step": 3210
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7057615342850861,
      "learning_rate": 4.962159280101292e-06,
      "loss": 0.4751,
      "step": 3211
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0525475881626118,
      "learning_rate": 4.959549630379934e-06,
      "loss": 0.5953,
      "step": 3212
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.018774125172252,
      "learning_rate": 4.9569399916783724e-06,
      "loss": 0.5984,
      "step": 3213
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6713126892689764,
      "learning_rate": 4.954330364707543e-06,
      "loss": 0.4722,
      "step": 3214
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0152952164386781,
      "learning_rate": 4.9517207501783805e-06,
      "loss": 0.6113,
      "step": 3215
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0100901898664556,
      "learning_rate": 4.949111148801819e-06,
      "loss": 0.6073,
      "step": 3216
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1189124080852684,
      "learning_rate": 4.9465015612887815e-06,
      "loss": 0.6122,
      "step": 3217
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0416949190818834,
      "learning_rate": 4.9438919883501934e-06,
      "loss": 0.6916,
      "step": 3218
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0052468492721593,
      "learning_rate": 4.941282430696976e-06,
      "loss": 0.6138,
      "step": 3219
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0211165296109515,
      "learning_rate": 4.938672889040041e-06,
      "loss": 0.5694,
      "step": 3220
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0033456810706858,
      "learning_rate": 4.936063364090302e-06,
      "loss": 0.5724,
      "step": 3221
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0531567301778528,
      "learning_rate": 4.933453856558666e-06,
      "loss": 0.6856,
      "step": 3222
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0257883738659475,
      "learning_rate": 4.930844367156032e-06,
      "loss": 0.6476,
      "step": 3223
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6845580322843383,
      "learning_rate": 4.9282348965933e-06,
      "loss": 0.4974,
      "step": 3224
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7260622160382391,
      "learning_rate": 4.925625445581361e-06,
      "loss": 0.4977,
      "step": 3225
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1068457428246035,
      "learning_rate": 4.923016014831099e-06,
      "loss": 0.5365,
      "step": 3226
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9134538705991346,
      "learning_rate": 4.9204066050533975e-06,
      "loss": 0.5367,
      "step": 3227
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9905446525890542,
      "learning_rate": 4.91779721695913e-06,
      "loss": 0.5526,
      "step": 3228
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0281560905594054,
      "learning_rate": 4.915187851259169e-06,
      "loss": 0.4344,
      "step": 3229
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9233087237956519,
      "learning_rate": 4.9125785086643726e-06,
      "loss": 0.5204,
      "step": 3230
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1068483085479515,
      "learning_rate": 4.9099691898856e-06,
      "loss": 0.659,
      "step": 3231
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6528755477179455,
      "learning_rate": 4.907359895633705e-06,
      "loss": 0.4504,
      "step": 3232
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0446952793453832,
      "learning_rate": 4.904750626619525e-06,
      "loss": 0.6392,
      "step": 3233
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.065630310872365,
      "learning_rate": 4.902141383553901e-06,
      "loss": 0.6076,
      "step": 3234
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9767672369534619,
      "learning_rate": 4.899532167147662e-06,
      "loss": 0.5521,
      "step": 3235
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.948386066730927,
      "learning_rate": 4.89692297811163e-06,
      "loss": 0.5934,
      "step": 3236
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0257924378445473,
      "learning_rate": 4.8943138171566184e-06,
      "loss": 0.5762,
      "step": 3237
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0828713787209612,
      "learning_rate": 4.89170468499344e-06,
      "loss": 0.5946,
      "step": 3238
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0160062577190154,
      "learning_rate": 4.889095582332887e-06,
      "loss": 0.6492,
      "step": 3239
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9624144541492015,
      "learning_rate": 4.886486509885755e-06,
      "loss": 0.5235,
      "step": 3240
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0337714842579497,
      "learning_rate": 4.883877468362828e-06,
      "loss": 0.6651,
      "step": 3241
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8944878521173915,
      "learning_rate": 4.88126845847488e-06,
      "loss": 0.5663,
      "step": 3242
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8898746825233013,
      "learning_rate": 4.878659480932674e-06,
      "loss": 0.5219,
      "step": 3243
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0036830134717925,
      "learning_rate": 4.876050536446973e-06,
      "loss": 0.5806,
      "step": 3244
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0040041799254726,
      "learning_rate": 4.8734416257285215e-06,
      "loss": 0.6191,
      "step": 3245
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.941530140974121,
      "learning_rate": 4.870832749488058e-06,
      "loss": 0.5844,
      "step": 3246
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0787821864249088,
      "learning_rate": 4.868223908436316e-06,
      "loss": 0.6222,
      "step": 3247
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0279489181115704,
      "learning_rate": 4.865615103284014e-06,
      "loss": 0.6378,
      "step": 3248
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9474272284976109,
      "learning_rate": 4.863006334741861e-06,
      "loss": 0.5683,
      "step": 3249
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0845749118054895,
      "learning_rate": 4.860397603520559e-06,
      "loss": 0.7109,
      "step": 3250
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0129500644018108,
      "learning_rate": 4.8577889103308e-06,
      "loss": 0.5639,
      "step": 3251
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8838735599091417,
      "learning_rate": 4.855180255883259e-06,
      "loss": 0.5763,
      "step": 3252
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9813482493430584,
      "learning_rate": 4.852571640888609e-06,
      "loss": 0.5639,
      "step": 3253
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0430583684115946,
      "learning_rate": 4.84996306605751e-06,
      "loss": 0.6719,
      "step": 3254
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0370844068598817,
      "learning_rate": 4.847354532100606e-06,
      "loss": 0.668,
      "step": 3255
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8569399735079533,
      "learning_rate": 4.844746039728535e-06,
      "loss": 0.5862,
      "step": 3256
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9243387475529347,
      "learning_rate": 4.842137589651923e-06,
      "loss": 0.5553,
      "step": 3257
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9799758505005088,
      "learning_rate": 4.8395291825813824e-06,
      "loss": 0.5284,
      "step": 3258
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0533121346361678,
      "learning_rate": 4.836920819227514e-06,
      "loss": 0.6536,
      "step": 3259
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9686528667314844,
      "learning_rate": 4.834312500300915e-06,
      "loss": 0.5889,
      "step": 3260
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9546470682833343,
      "learning_rate": 4.831704226512153e-06,
      "loss": 0.589,
      "step": 3261
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0321028834380768,
      "learning_rate": 4.829095998571801e-06,
      "loss": 0.6534,
      "step": 3262
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.035261629521912,
      "learning_rate": 4.826487817190411e-06,
      "loss": 0.6646,
      "step": 3263
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9813499141933322,
      "learning_rate": 4.8238796830785206e-06,
      "loss": 0.5839,
      "step": 3264
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0050005536683733,
      "learning_rate": 4.821271596946659e-06,
      "loss": 0.5929,
      "step": 3265
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9756024965647901,
      "learning_rate": 4.818663559505341e-06,
      "loss": 0.571,
      "step": 3266
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9410676858082435,
      "learning_rate": 4.816055571465071e-06,
      "loss": 0.5833,
      "step": 3267
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.092260096522639,
      "learning_rate": 4.813447633536331e-06,
      "loss": 0.5887,
      "step": 3268
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6489076504273431,
      "learning_rate": 4.810839746429598e-06,
      "loss": 0.4856,
      "step": 3269
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.986971256592421,
      "learning_rate": 4.808231910855335e-06,
      "loss": 0.5939,
      "step": 3270
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0511774516352927,
      "learning_rate": 4.805624127523982e-06,
      "loss": 0.6158,
      "step": 3271
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0591539859023469,
      "learning_rate": 4.8030163971459765e-06,
      "loss": 0.5564,
      "step": 3272
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9158695754390288,
      "learning_rate": 4.800408720431735e-06,
      "loss": 0.4901,
      "step": 3273
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9316861937665096,
      "learning_rate": 4.797801098091659e-06,
      "loss": 0.6633,
      "step": 3274
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9837307286668362,
      "learning_rate": 4.795193530836136e-06,
      "loss": 0.6788,
      "step": 3275
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8716925796675218,
      "learning_rate": 4.792586019375543e-06,
      "loss": 0.5431,
      "step": 3276
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.98272962390744,
      "learning_rate": 4.789978564420235e-06,
      "loss": 0.5959,
      "step": 3277
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.096442303039265,
      "learning_rate": 4.787371166680555e-06,
      "loss": 0.5049,
      "step": 3278
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9383788517359899,
      "learning_rate": 4.784763826866831e-06,
      "loss": 0.5299,
      "step": 3279
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9420626470428047,
      "learning_rate": 4.782156545689373e-06,
      "loss": 0.5454,
      "step": 3280
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9999786944956942,
      "learning_rate": 4.779549323858476e-06,
      "loss": 0.6702,
      "step": 3281
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9382697957196202,
      "learning_rate": 4.776942162084423e-06,
      "loss": 0.5119,
      "step": 3282
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0501426760544719,
      "learning_rate": 4.77433506107747e-06,
      "loss": 0.6212,
      "step": 3283
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0119859481314675,
      "learning_rate": 4.771728021547868e-06,
      "loss": 0.534,
      "step": 3284
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9955955700051046,
      "learning_rate": 4.769121044205847e-06,
      "loss": 0.5964,
      "step": 3285
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0270516391757718,
      "learning_rate": 4.766514129761616e-06,
      "loss": 0.5367,
      "step": 3286
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9794585512204484,
      "learning_rate": 4.763907278925372e-06,
      "loss": 0.6205,
      "step": 3287
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9887812462142733,
      "learning_rate": 4.761300492407293e-06,
      "loss": 0.6344,
      "step": 3288
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9956573898457196,
      "learning_rate": 4.758693770917543e-06,
      "loss": 0.6225,
      "step": 3289
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0156228494740815,
      "learning_rate": 4.756087115166257e-06,
      "loss": 0.5827,
      "step": 3290
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9688455199353587,
      "learning_rate": 4.753480525863566e-06,
      "loss": 0.5658,
      "step": 3291
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0028057696677601,
      "learning_rate": 4.750874003719577e-06,
      "loss": 0.6364,
      "step": 3292
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0241697916084316,
      "learning_rate": 4.7482675494443745e-06,
      "loss": 0.6023,
      "step": 3293
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0284079232661567,
      "learning_rate": 4.745661163748031e-06,
      "loss": 0.6009,
      "step": 3294
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.97762384613863,
      "learning_rate": 4.743054847340598e-06,
      "loss": 0.5339,
      "step": 3295
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8942975996140001,
      "learning_rate": 4.740448600932107e-06,
      "loss": 0.5167,
      "step": 3296
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.02946323026704,
      "learning_rate": 4.73784242523257e-06,
      "loss": 0.6356,
      "step": 3297
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9673391064227175,
      "learning_rate": 4.735236320951986e-06,
      "loss": 0.6504,
      "step": 3298
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0053082070466397,
      "learning_rate": 4.7326302888003224e-06,
      "loss": 0.6361,
      "step": 3299
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.01238243880198,
      "learning_rate": 4.730024329487541e-06,
      "loss": 0.6365,
      "step": 3300
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.079870481415195,
      "learning_rate": 4.7274184437235744e-06,
      "loss": 0.6607,
      "step": 3301
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0158539646895635,
      "learning_rate": 4.7248126322183364e-06,
      "loss": 0.5392,
      "step": 3302
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0954856586988109,
      "learning_rate": 4.722206895681723e-06,
      "loss": 0.5435,
      "step": 3303
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9715745977540837,
      "learning_rate": 4.71960123482361e-06,
      "loss": 0.6219,
      "step": 3304
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0108381823401609,
      "learning_rate": 4.716995650353847e-06,
      "loss": 0.6217,
      "step": 3305
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0106127951154473,
      "learning_rate": 4.714390142982272e-06,
      "loss": 0.6135,
      "step": 3306
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0648348077552454,
      "learning_rate": 4.711784713418694e-06,
      "loss": 0.7526,
      "step": 3307
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9697685467554417,
      "learning_rate": 4.709179362372906e-06,
      "loss": 0.5674,
      "step": 3308
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.030153011804173,
      "learning_rate": 4.706574090554676e-06,
      "loss": 0.648,
      "step": 3309
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9772541393119392,
      "learning_rate": 4.70396889867375e-06,
      "loss": 0.5966,
      "step": 3310
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9598646896327911,
      "learning_rate": 4.70136378743986e-06,
      "loss": 0.6555,
      "step": 3311
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0508878543051865,
      "learning_rate": 4.698758757562703e-06,
      "loss": 0.5364,
      "step": 3312
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8655864938193625,
      "learning_rate": 4.696153809751964e-06,
      "loss": 0.5215,
      "step": 3313
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7203948335746837,
      "learning_rate": 4.693548944717305e-06,
      "loss": 0.4929,
      "step": 3314
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0472455215044212,
      "learning_rate": 4.69094416316836e-06,
      "loss": 0.6037,
      "step": 3315
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.057798255155583,
      "learning_rate": 4.688339465814744e-06,
      "loss": 0.6236,
      "step": 3316
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0698489687274464,
      "learning_rate": 4.685734853366049e-06,
      "loss": 0.6243,
      "step": 3317
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0206028978032256,
      "learning_rate": 4.683130326531842e-06,
      "loss": 0.6048,
      "step": 3318
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0458362084130033,
      "learning_rate": 4.6805258860216675e-06,
      "loss": 0.6091,
      "step": 3319
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9851619905752268,
      "learning_rate": 4.677921532545052e-06,
      "loss": 0.6214,
      "step": 3320
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0055450197061575,
      "learning_rate": 4.675317266811485e-06,
      "loss": 0.6123,
      "step": 3321
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0690971487930612,
      "learning_rate": 4.672713089530445e-06,
      "loss": 0.5572,
      "step": 3322
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.06309775329184,
      "learning_rate": 4.670109001411383e-06,
      "loss": 0.6638,
      "step": 3323
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9820673733588248,
      "learning_rate": 4.6675050031637216e-06,
      "loss": 0.635,
      "step": 3324
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9899523766928819,
      "learning_rate": 4.664901095496863e-06,
      "loss": 0.608,
      "step": 3325
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0711515637688442,
      "learning_rate": 4.662297279120181e-06,
      "loss": 0.6321,
      "step": 3326
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0245246481248924,
      "learning_rate": 4.659693554743032e-06,
      "loss": 0.6343,
      "step": 3327
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0193057985995089,
      "learning_rate": 4.657089923074737e-06,
      "loss": 0.5832,
      "step": 3328
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.849949264677271,
      "learning_rate": 4.6544863848246e-06,
      "loss": 0.4311,
      "step": 3329
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0023897712017473,
      "learning_rate": 4.651882940701897e-06,
      "loss": 0.5576,
      "step": 3330
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0759814866015502,
      "learning_rate": 4.649279591415876e-06,
      "loss": 0.6501,
      "step": 3331
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0042620186316635,
      "learning_rate": 4.646676337675762e-06,
      "loss": 0.677,
      "step": 3332
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.728400960748559,
      "learning_rate": 4.644073180190753e-06,
      "loss": 0.4642,
      "step": 3333
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9771090554812161,
      "learning_rate": 4.64147011967002e-06,
      "loss": 0.532,
      "step": 3334
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9699961074665734,
      "learning_rate": 4.638867156822711e-06,
      "loss": 0.5055,
      "step": 3335
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0131552898961167,
      "learning_rate": 4.636264292357943e-06,
      "loss": 0.5773,
      "step": 3336
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.992264366617067,
      "learning_rate": 4.633661526984808e-06,
      "loss": 0.6353,
      "step": 3337
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9507375563520974,
      "learning_rate": 4.63105886141237e-06,
      "loss": 0.5232,
      "step": 3338
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9884793775634488,
      "learning_rate": 4.628456296349669e-06,
      "loss": 0.6421,
      "step": 3339
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9765297203754258,
      "learning_rate": 4.6258538325057145e-06,
      "loss": 0.6018,
      "step": 3340
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.917021338555848,
      "learning_rate": 4.623251470589488e-06,
      "loss": 0.6092,
      "step": 3341
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1175379450255176,
      "learning_rate": 4.62064921130995e-06,
      "loss": 0.66,
      "step": 3342
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0230437865126476,
      "learning_rate": 4.618047055376019e-06,
      "loss": 0.5707,
      "step": 3343
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8731650199015155,
      "learning_rate": 4.615445003496602e-06,
      "loss": 0.504,
      "step": 3344
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0984480208303067,
      "learning_rate": 4.612843056380564e-06,
      "loss": 0.5766,
      "step": 3345
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0019576403667154,
      "learning_rate": 4.610241214736751e-06,
      "loss": 0.4803,
      "step": 3346
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0358686086219877,
      "learning_rate": 4.607639479273976e-06,
      "loss": 0.5832,
      "step": 3347
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9844074066270181,
      "learning_rate": 4.605037850701019e-06,
      "loss": 0.6311,
      "step": 3348
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0050693618593929,
      "learning_rate": 4.602436329726643e-06,
      "loss": 0.6498,
      "step": 3349
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9800617155975043,
      "learning_rate": 4.599834917059565e-06,
      "loss": 0.6901,
      "step": 3350
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9202540112595412,
      "learning_rate": 4.597233613408488e-06,
      "loss": 0.5405,
      "step": 3351
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0028899741437136,
      "learning_rate": 4.5946324194820795e-06,
      "loss": 0.6265,
      "step": 3352
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9885265067391772,
      "learning_rate": 4.592031335988971e-06,
      "loss": 0.6625,
      "step": 3353
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9184750724152221,
      "learning_rate": 4.589430363637773e-06,
      "loss": 0.5673,
      "step": 3354
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8810295814819818,
      "learning_rate": 4.5868295031370625e-06,
      "loss": 0.5849,
      "step": 3355
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9551424365884923,
      "learning_rate": 4.584228755195382e-06,
      "loss": 0.5919,
      "step": 3356
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.939564277898375,
      "learning_rate": 4.581628120521251e-06,
      "loss": 0.5476,
      "step": 3357
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0297112285903718,
      "learning_rate": 4.579027599823153e-06,
      "loss": 0.5699,
      "step": 3358
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9664894153938252,
      "learning_rate": 4.5764271938095405e-06,
      "loss": 0.5522,
      "step": 3359
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0272219839356316,
      "learning_rate": 4.573826903188836e-06,
      "loss": 0.6462,
      "step": 3360
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6298293614313001,
      "learning_rate": 4.571226728669432e-06,
      "loss": 0.4497,
      "step": 3361
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9555385825041479,
      "learning_rate": 4.568626670959684e-06,
      "loss": 0.5651,
      "step": 3362
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0653822428767306,
      "learning_rate": 4.566026730767922e-06,
      "loss": 0.6795,
      "step": 3363
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0785622232610894,
      "learning_rate": 4.5634269088024434e-06,
      "loss": 0.6548,
      "step": 3364
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0189051865487468,
      "learning_rate": 4.560827205771506e-06,
      "loss": 0.5992,
      "step": 3365
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9748788212494538,
      "learning_rate": 4.558227622383345e-06,
      "loss": 0.5657,
      "step": 3366
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9933371883944114,
      "learning_rate": 4.555628159346158e-06,
      "loss": 0.6283,
      "step": 3367
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0467909845670023,
      "learning_rate": 4.55302881736811e-06,
      "loss": 0.6138,
      "step": 3368
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0160078135696642,
      "learning_rate": 4.550429597157331e-06,
      "loss": 0.5951,
      "step": 3369
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0118974586137033,
      "learning_rate": 4.547830499421921e-06,
      "loss": 0.5559,
      "step": 3370
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9811066188380705,
      "learning_rate": 4.545231524869952e-06,
      "loss": 0.5351,
      "step": 3371
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9789421129622694,
      "learning_rate": 4.542632674209447e-06,
      "loss": 0.5498,
      "step": 3372
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9327448785732281,
      "learning_rate": 4.54003394814841e-06,
      "loss": 0.4642,
      "step": 3373
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0037809812317164,
      "learning_rate": 4.537435347394807e-06,
      "loss": 0.5621,
      "step": 3374
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9440147702420087,
      "learning_rate": 4.534836872656563e-06,
      "loss": 0.5742,
      "step": 3375
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0015228739286448,
      "learning_rate": 4.532238524641578e-06,
      "loss": 0.5516,
      "step": 3376
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.94497275046947,
      "learning_rate": 4.529640304057714e-06,
      "loss": 0.5333,
      "step": 3377
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0944652085870414,
      "learning_rate": 4.527042211612796e-06,
      "loss": 0.6344,
      "step": 3378
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9708824750761053,
      "learning_rate": 4.524444248014615e-06,
      "loss": 0.6294,
      "step": 3379
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1014475061735598,
      "learning_rate": 4.521846413970932e-06,
      "loss": 0.6365,
      "step": 3380
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0909982689524242,
      "learning_rate": 4.519248710189465e-06,
      "loss": 0.6249,
      "step": 3381
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6756509597258938,
      "learning_rate": 4.516651137377902e-06,
      "loss": 0.523,
      "step": 3382
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0012806898756457,
      "learning_rate": 4.514053696243893e-06,
      "loss": 0.6284,
      "step": 3383
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.054176104004274,
      "learning_rate": 4.511456387495052e-06,
      "loss": 0.6339,
      "step": 3384
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9832878833525789,
      "learning_rate": 4.508859211838955e-06,
      "loss": 0.5306,
      "step": 3385
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0438038956253828,
      "learning_rate": 4.50626216998315e-06,
      "loss": 0.6075,
      "step": 3386
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9999624363758526,
      "learning_rate": 4.503665262635141e-06,
      "loss": 0.5194,
      "step": 3387
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.005021878357175,
      "learning_rate": 4.501068490502394e-06,
      "loss": 0.5998,
      "step": 3388
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.981212671047851,
      "learning_rate": 4.498471854292344e-06,
      "loss": 0.5643,
      "step": 3389
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9639825568043952,
      "learning_rate": 4.495875354712386e-06,
      "loss": 0.5536,
      "step": 3390
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0418486837601173,
      "learning_rate": 4.493278992469877e-06,
      "loss": 0.7085,
      "step": 3391
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0306037469831133,
      "learning_rate": 4.4906827682721375e-06,
      "loss": 0.5731,
      "step": 3392
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0268770565798349,
      "learning_rate": 4.488086682826454e-06,
      "loss": 0.6459,
      "step": 3393
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9843481887490956,
      "learning_rate": 4.485490736840066e-06,
      "loss": 0.6283,
      "step": 3394
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7154780086435534,
      "learning_rate": 4.482894931020185e-06,
      "loss": 0.4953,
      "step": 3395
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0165835042789737,
      "learning_rate": 4.48029926607398e-06,
      "loss": 0.633,
      "step": 3396
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0775379195150174,
      "learning_rate": 4.477703742708579e-06,
      "loss": 0.6045,
      "step": 3397
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0117171399178724,
      "learning_rate": 4.475108361631076e-06,
      "loss": 0.5863,
      "step": 3398
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0064772000644282,
      "learning_rate": 4.472513123548525e-06,
      "loss": 0.6036,
      "step": 3399
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0637247101489924,
      "learning_rate": 4.469918029167939e-06,
      "loss": 0.6411,
      "step": 3400
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0126899482249887,
      "learning_rate": 4.467323079196292e-06,
      "loss": 0.6374,
      "step": 3401
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9913440953164321,
      "learning_rate": 4.464728274340525e-06,
      "loss": 0.5539,
      "step": 3402
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9362319292359428,
      "learning_rate": 4.462133615307527e-06,
      "loss": 0.5683,
      "step": 3403
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0612359807843708,
      "learning_rate": 4.45953910280416e-06,
      "loss": 0.6249,
      "step": 3404
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9448788436125708,
      "learning_rate": 4.4569447375372396e-06,
      "loss": 0.5398,
      "step": 3405
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9947467746395797,
      "learning_rate": 4.454350520213543e-06,
      "loss": 0.5844,
      "step": 3406
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6441468058866067,
      "learning_rate": 4.451756451539804e-06,
      "loss": 0.4942,
      "step": 3407
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0237361743752325,
      "learning_rate": 4.44916253222272e-06,
      "loss": 0.5599,
      "step": 3408
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9830253723537834,
      "learning_rate": 4.446568762968947e-06,
      "loss": 0.6138,
      "step": 3409
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.011656096356316,
      "learning_rate": 4.443975144485099e-06,
      "loss": 0.6024,
      "step": 3410
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.007068221894919,
      "learning_rate": 4.441381677477748e-06,
      "loss": 0.6564,
      "step": 3411
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0135195242171868,
      "learning_rate": 4.438788362653426e-06,
      "loss": 0.6029,
      "step": 3412
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.028571937740489,
      "learning_rate": 4.436195200718625e-06,
      "loss": 0.5722,
      "step": 3413
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.979875824782129,
      "learning_rate": 4.43360219237979e-06,
      "loss": 0.5387,
      "step": 3414
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0245338249646139,
      "learning_rate": 4.431009338343335e-06,
      "loss": 0.5966,
      "step": 3415
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9860851904399448,
      "learning_rate": 4.428416639315616e-06,
      "loss": 0.6053,
      "step": 3416
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.992850656210538,
      "learning_rate": 4.425824096002962e-06,
      "loss": 0.575,
      "step": 3417
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.026673132900477,
      "learning_rate": 4.423231709111653e-06,
      "loss": 0.6488,
      "step": 3418
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9146066478984853,
      "learning_rate": 4.420639479347924e-06,
      "loss": 0.5271,
      "step": 3419
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0383789795319225,
      "learning_rate": 4.4180474074179705e-06,
      "loss": 0.6633,
      "step": 3420
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0144456681292144,
      "learning_rate": 4.415455494027946e-06,
      "loss": 0.5719,
      "step": 3421
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0103578156144246,
      "learning_rate": 4.412863739883958e-06,
      "loss": 0.6137,
      "step": 3422
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0063630950354066,
      "learning_rate": 4.410272145692069e-06,
      "loss": 0.6186,
      "step": 3423
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0293821623419879,
      "learning_rate": 4.407680712158308e-06,
      "loss": 0.7032,
      "step": 3424
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6757480844729872,
      "learning_rate": 4.405089439988645e-06,
      "loss": 0.4973,
      "step": 3425
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.061662291940576,
      "learning_rate": 4.402498329889018e-06,
      "loss": 0.5593,
      "step": 3426
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9970134233280021,
      "learning_rate": 4.399907382565316e-06,
      "loss": 0.6097,
      "step": 3427
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.892037453657559,
      "learning_rate": 4.397316598723385e-06,
      "loss": 0.5661,
      "step": 3428
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1205638085893797,
      "learning_rate": 4.394725979069026e-06,
      "loss": 0.5886,
      "step": 3429
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.021584968969128,
      "learning_rate": 4.392135524307993e-06,
      "loss": 0.5904,
      "step": 3430
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0056127530205334,
      "learning_rate": 4.389545235146003e-06,
      "loss": 0.5391,
      "step": 3431
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9808933888149748,
      "learning_rate": 4.386955112288714e-06,
      "loss": 0.6426,
      "step": 3432
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0552660627224566,
      "learning_rate": 4.384365156441752e-06,
      "loss": 0.5795,
      "step": 3433
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3802332678351499,
      "learning_rate": 4.381775368310694e-06,
      "loss": 0.543,
      "step": 3434
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0804355686806488,
      "learning_rate": 4.3791857486010655e-06,
      "loss": 0.6103,
      "step": 3435
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0482504271482156,
      "learning_rate": 4.3765962980183515e-06,
      "loss": 0.6564,
      "step": 3436
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0255054973674034,
      "learning_rate": 4.374007017267991e-06,
      "loss": 0.5455,
      "step": 3437
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9489932088476616,
      "learning_rate": 4.371417907055373e-06,
      "loss": 0.5522,
      "step": 3438
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0175646448103681,
      "learning_rate": 4.3688289680858456e-06,
      "loss": 0.6356,
      "step": 3439
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0588952872130237,
      "learning_rate": 4.366240201064705e-06,
      "loss": 0.6179,
      "step": 3440
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.973997577847627,
      "learning_rate": 4.363651606697204e-06,
      "loss": 0.5551,
      "step": 3441
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9578925516715439,
      "learning_rate": 4.361063185688546e-06,
      "loss": 0.5291,
      "step": 3442
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0363903348962193,
      "learning_rate": 4.358474938743889e-06,
      "loss": 0.6254,
      "step": 3443
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0615438370162738,
      "learning_rate": 4.355886866568342e-06,
      "loss": 0.6722,
      "step": 3444
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9227947161993313,
      "learning_rate": 4.353298969866966e-06,
      "loss": 0.5703,
      "step": 3445
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9539663780722605,
      "learning_rate": 4.350711249344778e-06,
      "loss": 0.5724,
      "step": 3446
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0945919909583495,
      "learning_rate": 4.348123705706745e-06,
      "loss": 0.6078,
      "step": 3447
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9609037824822799,
      "learning_rate": 4.345536339657783e-06,
      "loss": 0.5971,
      "step": 3448
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9621630552697684,
      "learning_rate": 4.342949151902761e-06,
      "loss": 0.5836,
      "step": 3449
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0440108468976308,
      "learning_rate": 4.340362143146504e-06,
      "loss": 0.5329,
      "step": 3450
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0398718137167298,
      "learning_rate": 4.337775314093781e-06,
      "loss": 0.6393,
      "step": 3451
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1203469021416326,
      "learning_rate": 4.335188665449316e-06,
      "loss": 0.6242,
      "step": 3452
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9763641048524768,
      "learning_rate": 4.3326021979177865e-06,
      "loss": 0.5806,
      "step": 3453
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0863615471001702,
      "learning_rate": 4.330015912203812e-06,
      "loss": 0.5869,
      "step": 3454
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.11805496575195,
      "learning_rate": 4.327429809011973e-06,
      "loss": 0.6434,
      "step": 3455
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9528676039345545,
      "learning_rate": 4.324843889046795e-06,
      "loss": 0.6018,
      "step": 3456
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0746723173661208,
      "learning_rate": 4.322258153012751e-06,
      "loss": 0.6162,
      "step": 3457
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0703732010328004,
      "learning_rate": 4.319672601614268e-06,
      "loss": 0.6293,
      "step": 3458
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0053156864252262,
      "learning_rate": 4.317087235555722e-06,
      "loss": 0.6019,
      "step": 3459
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9783536035908239,
      "learning_rate": 4.314502055541437e-06,
      "loss": 0.5834,
      "step": 3460
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0588505318675427,
      "learning_rate": 4.311917062275688e-06,
      "loss": 0.6204,
      "step": 3461
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.083726156048756,
      "learning_rate": 4.309332256462699e-06,
      "loss": 0.5946,
      "step": 3462
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0111729871369115,
      "learning_rate": 4.306747638806641e-06,
      "loss": 0.6374,
      "step": 3463
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.991857460491777,
      "learning_rate": 4.304163210011636e-06,
      "loss": 0.6378,
      "step": 3464
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9798702516168581,
      "learning_rate": 4.301578970781753e-06,
      "loss": 0.6079,
      "step": 3465
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9378730855677958,
      "learning_rate": 4.2989949218210125e-06,
      "loss": 0.5304,
      "step": 3466
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.029181924622109,
      "learning_rate": 4.2964110638333755e-06,
      "loss": 0.6053,
      "step": 3467
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1268163214369247,
      "learning_rate": 4.29382739752276e-06,
      "loss": 0.6988,
      "step": 3468
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0241383694819495,
      "learning_rate": 4.2912439235930305e-06,
      "loss": 0.5318,
      "step": 3469
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9936435664348292,
      "learning_rate": 4.2886606427479905e-06,
      "loss": 0.5999,
      "step": 3470
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0208262019693501,
      "learning_rate": 4.2860775556913995e-06,
      "loss": 0.6244,
      "step": 3471
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.984765494006994,
      "learning_rate": 4.283494663126962e-06,
      "loss": 0.576,
      "step": 3472
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.046474820262225,
      "learning_rate": 4.28091196575833e-06,
      "loss": 0.6252,
      "step": 3473
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.010594877905954,
      "learning_rate": 4.278329464289098e-06,
      "loss": 0.6172,
      "step": 3474
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9429245023346525,
      "learning_rate": 4.275747159422815e-06,
      "loss": 0.5658,
      "step": 3475
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9897712266118776,
      "learning_rate": 4.273165051862969e-06,
      "loss": 0.5238,
      "step": 3476
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.031600282400674,
      "learning_rate": 4.270583142312998e-06,
      "loss": 0.6536,
      "step": 3477
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0006061102746264,
      "learning_rate": 4.268001431476286e-06,
      "loss": 0.6083,
      "step": 3478
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9932404654108857,
      "learning_rate": 4.265419920056162e-06,
      "loss": 0.6448,
      "step": 3479
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0086057263584498,
      "learning_rate": 4.262838608755899e-06,
      "loss": 0.6089,
      "step": 3480
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0471477291438624,
      "learning_rate": 4.2602574982787216e-06,
      "loss": 0.6173,
      "step": 3481
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.017407652146029,
      "learning_rate": 4.257676589327791e-06,
      "loss": 0.5096,
      "step": 3482
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0065396804957187,
      "learning_rate": 4.255095882606219e-06,
      "loss": 0.5754,
      "step": 3483
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0247080178098593,
      "learning_rate": 4.252515378817062e-06,
      "loss": 0.6433,
      "step": 3484
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0642729450424664,
      "learning_rate": 4.2499350786633235e-06,
      "loss": 0.6071,
      "step": 3485
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.997166962929344,
      "learning_rate": 4.247354982847942e-06,
      "loss": 0.6168,
      "step": 3486
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9611128289541894,
      "learning_rate": 4.244775092073811e-06,
      "loss": 0.445,
      "step": 3487
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0471474507389487,
      "learning_rate": 4.242195407043763e-06,
      "loss": 0.6741,
      "step": 3488
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9978503201383152,
      "learning_rate": 4.239615928460574e-06,
      "loss": 0.5656,
      "step": 3489
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.66782530124989,
      "learning_rate": 4.2370366570269676e-06,
      "loss": 0.4941,
      "step": 3490
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.000474006680463,
      "learning_rate": 4.234457593445608e-06,
      "loss": 0.5469,
      "step": 3491
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0182215005965956,
      "learning_rate": 4.2318787384191e-06,
      "loss": 0.5881,
      "step": 3492
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9944937184679367,
      "learning_rate": 4.229300092649997e-06,
      "loss": 0.6067,
      "step": 3493
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9841926805046696,
      "learning_rate": 4.226721656840796e-06,
      "loss": 0.6258,
      "step": 3494
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6554381127531465,
      "learning_rate": 4.224143431693929e-06,
      "loss": 0.5057,
      "step": 3495
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9548726496469082,
      "learning_rate": 4.2215654179117765e-06,
      "loss": 0.6388,
      "step": 3496
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0564436382185023,
      "learning_rate": 4.2189876161966655e-06,
      "loss": 0.6052,
      "step": 3497
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9518821541021305,
      "learning_rate": 4.216410027250853e-06,
      "loss": 0.5519,
      "step": 3498
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0375068284240994,
      "learning_rate": 4.213832651776551e-06,
      "loss": 0.6657,
      "step": 3499
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9552643862231862,
      "learning_rate": 4.211255490475905e-06,
      "loss": 0.5794,
      "step": 3500
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1619851083376305,
      "learning_rate": 4.208678544051005e-06,
      "loss": 0.7086,
      "step": 3501
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9559149368495451,
      "learning_rate": 4.2061018132038825e-06,
      "loss": 0.6336,
      "step": 3502
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0075580160388367,
      "learning_rate": 4.203525298636512e-06,
      "loss": 0.6515,
      "step": 3503
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9957626187323152,
      "learning_rate": 4.2009490010508026e-06,
      "loss": 0.6039,
      "step": 3504
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.003979400032938,
      "learning_rate": 4.19837292114861e-06,
      "loss": 0.5286,
      "step": 3505
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9512930436167252,
      "learning_rate": 4.195797059631733e-06,
      "loss": 0.501,
      "step": 3506
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9785113594830841,
      "learning_rate": 4.1932214172019056e-06,
      "loss": 0.6031,
      "step": 3507
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0802192359246439,
      "learning_rate": 4.190645994560802e-06,
      "loss": 0.6857,
      "step": 3508
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0677456288820848,
      "learning_rate": 4.188070792410039e-06,
      "loss": 0.5758,
      "step": 3509
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0527727098660917,
      "learning_rate": 4.185495811451175e-06,
      "loss": 0.6547,
      "step": 3510
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0358121332402606,
      "learning_rate": 4.182921052385702e-06,
      "loss": 0.4871,
      "step": 3511
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9389598835819879,
      "learning_rate": 4.180346515915057e-06,
      "loss": 0.5542,
      "step": 3512
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9216510573199962,
      "learning_rate": 4.177772202740617e-06,
      "loss": 0.5612,
      "step": 3513
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.91114928686773,
      "learning_rate": 4.175198113563692e-06,
      "loss": 0.5907,
      "step": 3514
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0660798702695307,
      "learning_rate": 4.172624249085537e-06,
      "loss": 0.5298,
      "step": 3515
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0037855041451813,
      "learning_rate": 4.170050610007344e-06,
      "loss": 0.4824,
      "step": 3516
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9942996574008681,
      "learning_rate": 4.167477197030242e-06,
      "loss": 0.559,
      "step": 3517
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0305142592115522,
      "learning_rate": 4.164904010855299e-06,
      "loss": 0.6005,
      "step": 3518
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.039342522561636,
      "learning_rate": 4.162331052183526e-06,
      "loss": 0.589,
      "step": 3519
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.022367113488299,
      "learning_rate": 4.159758321715862e-06,
      "loss": 0.5999,
      "step": 3520
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1344523916684348,
      "learning_rate": 4.157185820153193e-06,
      "loss": 0.6175,
      "step": 3521
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9002174403567248,
      "learning_rate": 4.154613548196341e-06,
      "loss": 0.5595,
      "step": 3522
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9545828627968804,
      "learning_rate": 4.15204150654606e-06,
      "loss": 0.5816,
      "step": 3523
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.969562756038056,
      "learning_rate": 4.149469695903047e-06,
      "loss": 0.5626,
      "step": 3524
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.989309040393606,
      "learning_rate": 4.146898116967932e-06,
      "loss": 0.5724,
      "step": 3525
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0436471992420429,
      "learning_rate": 4.1443267704412895e-06,
      "loss": 0.5546,
      "step": 3526
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9799223718088593,
      "learning_rate": 4.141755657023618e-06,
      "loss": 0.5413,
      "step": 3527
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.955671423575773,
      "learning_rate": 4.139184777415365e-06,
      "loss": 0.4824,
      "step": 3528
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0084400676403826,
      "learning_rate": 4.1366141323169076e-06,
      "loss": 0.6352,
      "step": 3529
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.991495550938125,
      "learning_rate": 4.1340437224285594e-06,
      "loss": 0.5215,
      "step": 3530
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0321700138078431,
      "learning_rate": 4.131473548450571e-06,
      "loss": 0.6117,
      "step": 3531
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0529163700653292,
      "learning_rate": 4.12890361108313e-06,
      "loss": 0.6181,
      "step": 3532
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0788572242787289,
      "learning_rate": 4.126333911026357e-06,
      "loss": 0.638,
      "step": 3533
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0270532378003343,
      "learning_rate": 4.123764448980308e-06,
      "loss": 0.5967,
      "step": 3534
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1040814437559843,
      "learning_rate": 4.1211952256449796e-06,
      "loss": 0.6446,
      "step": 3535
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9512877802554228,
      "learning_rate": 4.118626241720293e-06,
      "loss": 0.5045,
      "step": 3536
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0681199635328404,
      "learning_rate": 4.116057497906114e-06,
      "loss": 0.594,
      "step": 3537
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9467210690602159,
      "learning_rate": 4.11348899490224e-06,
      "loss": 0.5965,
      "step": 3538
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0219927264139612,
      "learning_rate": 4.1109207334084e-06,
      "loss": 0.5924,
      "step": 3539
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9286100046742463,
      "learning_rate": 4.1083527141242594e-06,
      "loss": 0.4836,
      "step": 3540
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.930663264630551,
      "learning_rate": 4.105784937749419e-06,
      "loss": 0.5266,
      "step": 3541
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6851119629114353,
      "learning_rate": 4.103217404983409e-06,
      "loss": 0.476,
      "step": 3542
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0943651099493412,
      "learning_rate": 4.100650116525698e-06,
      "loss": 0.5977,
      "step": 3543
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0136223504813284,
      "learning_rate": 4.098083073075686e-06,
      "loss": 0.6293,
      "step": 3544
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0694835676136176,
      "learning_rate": 4.0955162753327085e-06,
      "loss": 0.5317,
      "step": 3545
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.034493704087418,
      "learning_rate": 4.092949723996028e-06,
      "loss": 0.6597,
      "step": 3546
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9557363981776348,
      "learning_rate": 4.0903834197648444e-06,
      "loss": 0.5837,
      "step": 3547
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0476567015524871,
      "learning_rate": 4.087817363338294e-06,
      "loss": 0.5776,
      "step": 3548
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0357392109925678,
      "learning_rate": 4.085251555415437e-06,
      "loss": 0.5624,
      "step": 3549
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1162700070216063,
      "learning_rate": 4.08268599669527e-06,
      "loss": 0.6595,
      "step": 3550
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0642952795890377,
      "learning_rate": 4.080120687876726e-06,
      "loss": 0.5921,
      "step": 3551
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0513458247298086,
      "learning_rate": 4.077555629658662e-06,
      "loss": 0.584,
      "step": 3552
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0129740644197829,
      "learning_rate": 4.074990822739871e-06,
      "loss": 0.5914,
      "step": 3553
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.920181825425386,
      "learning_rate": 4.072426267819081e-06,
      "loss": 0.5212,
      "step": 3554
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9690598513532722,
      "learning_rate": 4.069861965594941e-06,
      "loss": 0.589,
      "step": 3555
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9973525279064102,
      "learning_rate": 4.067297916766042e-06,
      "loss": 0.5975,
      "step": 3556
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0645226484452224,
      "learning_rate": 4.0647341220309024e-06,
      "loss": 0.618,
      "step": 3557
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9400911540186323,
      "learning_rate": 4.062170582087965e-06,
      "loss": 0.5783,
      "step": 3558
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0386451100228673,
      "learning_rate": 4.059607297635615e-06,
      "loss": 0.592,
      "step": 3559
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0033600419007418,
      "learning_rate": 4.057044269372159e-06,
      "loss": 0.638,
      "step": 3560
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0538372979874435,
      "learning_rate": 4.054481497995836e-06,
      "loss": 0.6638,
      "step": 3561
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9525545187657446,
      "learning_rate": 4.051918984204817e-06,
      "loss": 0.5553,
      "step": 3562
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9285094776304778,
      "learning_rate": 4.049356728697199e-06,
      "loss": 0.6766,
      "step": 3563
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6687189343103905,
      "learning_rate": 4.046794732171016e-06,
      "loss": 0.4598,
      "step": 3564
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.926237272006665,
      "learning_rate": 4.04423299532422e-06,
      "loss": 0.5816,
      "step": 3565
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9623821162953281,
      "learning_rate": 4.041671518854704e-06,
      "loss": 0.6149,
      "step": 3566
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.870138482166875,
      "learning_rate": 4.0391103034602826e-06,
      "loss": 0.4738,
      "step": 3567
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0343250936188184,
      "learning_rate": 4.036549349838701e-06,
      "loss": 0.6726,
      "step": 3568
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9308460059310422,
      "learning_rate": 4.033988658687634e-06,
      "loss": 0.5156,
      "step": 3569
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9493513943874937,
      "learning_rate": 4.031428230704686e-06,
      "loss": 0.5419,
      "step": 3570
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0127058052191331,
      "learning_rate": 4.028868066587384e-06,
      "loss": 0.5775,
      "step": 3571
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0794241081298137,
      "learning_rate": 4.026308167033191e-06,
      "loss": 0.6077,
      "step": 3572
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6378646439566517,
      "learning_rate": 4.023748532739496e-06,
      "loss": 0.4758,
      "step": 3573
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0945332890865889,
      "learning_rate": 4.0211891644036085e-06,
      "loss": 0.6466,
      "step": 3574
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.669016074431678,
      "learning_rate": 4.018630062722774e-06,
      "loss": 0.5188,
      "step": 3575
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.082421646202419,
      "learning_rate": 4.016071228394164e-06,
      "loss": 0.6084,
      "step": 3576
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6510619577249291,
      "learning_rate": 4.013512662114871e-06,
      "loss": 0.5087,
      "step": 3577
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8774765745410158,
      "learning_rate": 4.010954364581921e-06,
      "loss": 0.6094,
      "step": 3578
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0372123766379873,
      "learning_rate": 4.008396336492269e-06,
      "loss": 0.6835,
      "step": 3579
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9911761701962242,
      "learning_rate": 4.005838578542785e-06,
      "loss": 0.5903,
      "step": 3580
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.029811107875484,
      "learning_rate": 4.003281091430279e-06,
      "loss": 0.6103,
      "step": 3581
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8812597368343738,
      "learning_rate": 4.000723875851477e-06,
      "loss": 0.5839,
      "step": 3582
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0692610629856572,
      "learning_rate": 3.998166932503037e-06,
      "loss": 0.6543,
      "step": 3583
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0636536097536027,
      "learning_rate": 3.99561026208154e-06,
      "loss": 0.5822,
      "step": 3584
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0355956029911022,
      "learning_rate": 3.993053865283492e-06,
      "loss": 0.6446,
      "step": 3585
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9493343317904043,
      "learning_rate": 3.9904977428053325e-06,
      "loss": 0.4843,
      "step": 3586
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9952773328140285,
      "learning_rate": 3.9879418953434105e-06,
      "loss": 0.5672,
      "step": 3587
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0224048810022972,
      "learning_rate": 3.9853863235940145e-06,
      "loss": 0.582,
      "step": 3588
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.025766569928607,
      "learning_rate": 3.982831028253354e-06,
      "loss": 0.5604,
      "step": 3589
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8726111707183295,
      "learning_rate": 3.980276010017559e-06,
      "loss": 0.4965,
      "step": 3590
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9606217338927152,
      "learning_rate": 3.977721269582686e-06,
      "loss": 0.5225,
      "step": 3591
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9886663499745755,
      "learning_rate": 3.975166807644721e-06,
      "loss": 0.6256,
      "step": 3592
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0166462128536062,
      "learning_rate": 3.972612624899566e-06,
      "loss": 0.6158,
      "step": 3593
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6613396165630537,
      "learning_rate": 3.97005872204305e-06,
      "loss": 0.4713,
      "step": 3594
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9502643499529743,
      "learning_rate": 3.967505099770932e-06,
      "loss": 0.5892,
      "step": 3595
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.960340871389481,
      "learning_rate": 3.964951758778885e-06,
      "loss": 0.654,
      "step": 3596
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0456755589248152,
      "learning_rate": 3.962398699762508e-06,
      "loss": 0.6544,
      "step": 3597
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7261441018983972,
      "learning_rate": 3.95984592341733e-06,
      "loss": 0.4937,
      "step": 3598
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9485254547007809,
      "learning_rate": 3.957293430438792e-06,
      "loss": 0.4863,
      "step": 3599
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9374146634641488,
      "learning_rate": 3.954741221522266e-06,
      "loss": 0.5973,
      "step": 3600
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.045305859404248,
      "learning_rate": 3.952189297363047e-06,
      "loss": 0.6296,
      "step": 3601
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0073437740663223,
      "learning_rate": 3.949637658656343e-06,
      "loss": 0.63,
      "step": 3602
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.029205877083787,
      "learning_rate": 3.947086306097295e-06,
      "loss": 0.6812,
      "step": 3603
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9654550308141081,
      "learning_rate": 3.944535240380961e-06,
      "loss": 0.5523,
      "step": 3604
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.95038839953288,
      "learning_rate": 3.941984462202323e-06,
      "loss": 0.5832,
      "step": 3605
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0247128814822408,
      "learning_rate": 3.939433972256281e-06,
      "loss": 0.6253,
      "step": 3606
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.006875444451758,
      "learning_rate": 3.936883771237658e-06,
      "loss": 0.5686,
      "step": 3607
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0501704226400725,
      "learning_rate": 3.934333859841204e-06,
      "loss": 0.6426,
      "step": 3608
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0335160748811831,
      "learning_rate": 3.931784238761579e-06,
      "loss": 0.6587,
      "step": 3609
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0398723581307618,
      "learning_rate": 3.929234908693373e-06,
      "loss": 0.6476,
      "step": 3610
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9720836322917509,
      "learning_rate": 3.9266858703310965e-06,
      "loss": 0.5354,
      "step": 3611
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0235817592106464,
      "learning_rate": 3.924137124369172e-06,
      "loss": 0.581,
      "step": 3612
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9344374769381187,
      "learning_rate": 3.9215886715019525e-06,
      "loss": 0.5651,
      "step": 3613
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0419043147510474,
      "learning_rate": 3.919040512423706e-06,
      "loss": 0.6655,
      "step": 3614
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9554011474037226,
      "learning_rate": 3.91649264782862e-06,
      "loss": 0.5231,
      "step": 3615
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1408753992748843,
      "learning_rate": 3.913945078410802e-06,
      "loss": 0.6954,
      "step": 3616
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.019527657464595,
      "learning_rate": 3.911397804864285e-06,
      "loss": 0.6179,
      "step": 3617
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.00836818832811,
      "learning_rate": 3.908850827883012e-06,
      "loss": 0.5775,
      "step": 3618
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0138906764525937,
      "learning_rate": 3.90630414816085e-06,
      "loss": 0.6412,
      "step": 3619
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6430768421241214,
      "learning_rate": 3.9037577663915885e-06,
      "loss": 0.4665,
      "step": 3620
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0093415253321745,
      "learning_rate": 3.901211683268928e-06,
      "loss": 0.6549,
      "step": 3621
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9538775654820494,
      "learning_rate": 3.898665899486493e-06,
      "loss": 0.507,
      "step": 3622
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0054704400063228,
      "learning_rate": 3.896120415737825e-06,
      "loss": 0.629,
      "step": 3623
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0034105497009433,
      "learning_rate": 3.893575232716387e-06,
      "loss": 0.6086,
      "step": 3624
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0017484607329237,
      "learning_rate": 3.891030351115552e-06,
      "loss": 0.58,
      "step": 3625
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9608848017939492,
      "learning_rate": 3.888485771628618e-06,
      "loss": 0.604,
      "step": 3626
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.090375844173031,
      "learning_rate": 3.885941494948802e-06,
      "loss": 0.6254,
      "step": 3627
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8865414601880237,
      "learning_rate": 3.8833975217692285e-06,
      "loss": 0.5594,
      "step": 3628
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.028636722987518,
      "learning_rate": 3.88085385278295e-06,
      "loss": 0.6588,
      "step": 3629
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0516760461341588,
      "learning_rate": 3.878310488682934e-06,
      "loss": 0.6799,
      "step": 3630
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.054240845472082,
      "learning_rate": 3.875767430162058e-06,
      "loss": 0.6201,
      "step": 3631
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9821712063432436,
      "learning_rate": 3.873224677913124e-06,
      "loss": 0.6511,
      "step": 3632
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0828149086576655,
      "learning_rate": 3.870682232628848e-06,
      "loss": 0.6449,
      "step": 3633
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9587686176002178,
      "learning_rate": 3.8681400950018615e-06,
      "loss": 0.6138,
      "step": 3634
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9906677234458398,
      "learning_rate": 3.865598265724713e-06,
      "loss": 0.5973,
      "step": 3635
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9693627283449526,
      "learning_rate": 3.8630567454898676e-06,
      "loss": 0.5637,
      "step": 3636
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9635958691555139,
      "learning_rate": 3.860515534989704e-06,
      "loss": 0.5521,
      "step": 3637
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9776853309359553,
      "learning_rate": 3.857974634916517e-06,
      "loss": 0.6092,
      "step": 3638
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9919823100920274,
      "learning_rate": 3.8554340459625235e-06,
      "loss": 0.6157,
      "step": 3639
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9592799809249056,
      "learning_rate": 3.852893768819843e-06,
      "loss": 0.5655,
      "step": 3640
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0302782328014588,
      "learning_rate": 3.85035380418052e-06,
      "loss": 0.5606,
      "step": 3641
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9471612063342956,
      "learning_rate": 3.847814152736512e-06,
      "loss": 0.5105,
      "step": 3642
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0683907731608226,
      "learning_rate": 3.84527481517969e-06,
      "loss": 0.5703,
      "step": 3643
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9812978222579662,
      "learning_rate": 3.842735792201837e-06,
      "loss": 0.5607,
      "step": 3644
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9649701688631073,
      "learning_rate": 3.840197084494653e-06,
      "loss": 0.5839,
      "step": 3645
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1144162779059659,
      "learning_rate": 3.8376586927497565e-06,
      "loss": 0.6317,
      "step": 3646
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0057852413562807,
      "learning_rate": 3.835120617658669e-06,
      "loss": 0.5729,
      "step": 3647
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6248378964088227,
      "learning_rate": 3.832582859912838e-06,
      "loss": 0.4767,
      "step": 3648
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0457974670464516,
      "learning_rate": 3.8300454202036155e-06,
      "loss": 0.5634,
      "step": 3649
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0244525669570272,
      "learning_rate": 3.82750829922227e-06,
      "loss": 0.6233,
      "step": 3650
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0319010038810321,
      "learning_rate": 3.824971497659983e-06,
      "loss": 0.5837,
      "step": 3651
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.985724371482178,
      "learning_rate": 3.8224350162078526e-06,
      "loss": 0.5603,
      "step": 3652
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9773095017922779,
      "learning_rate": 3.81989885555688e-06,
      "loss": 0.587,
      "step": 3653
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0466489033617228,
      "learning_rate": 3.81736301639799e-06,
      "loss": 0.6294,
      "step": 3654
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0013083651569465,
      "learning_rate": 3.814827499422016e-06,
      "loss": 0.5524,
      "step": 3655
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0693882044740042,
      "learning_rate": 3.8122923053196984e-06,
      "loss": 0.6253,
      "step": 3656
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0463034654771313,
      "learning_rate": 3.809757434781697e-06,
      "loss": 0.6604,
      "step": 3657
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.027755000773748,
      "learning_rate": 3.8072228884985803e-06,
      "loss": 0.6065,
      "step": 3658
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0537201421606006,
      "learning_rate": 3.8046886671608264e-06,
      "loss": 0.5925,
      "step": 3659
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9882733363588228,
      "learning_rate": 3.802154771458828e-06,
      "loss": 0.6477,
      "step": 3660
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.035328119940369,
      "learning_rate": 3.7996212020828915e-06,
      "loss": 0.6528,
      "step": 3661
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0391902086360811,
      "learning_rate": 3.797087959723225e-06,
      "loss": 0.5394,
      "step": 3662
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0101724814285764,
      "learning_rate": 3.7945550450699585e-06,
      "loss": 0.6649,
      "step": 3663
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.636921762329395,
      "learning_rate": 3.7920224588131256e-06,
      "loss": 0.4822,
      "step": 3664
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.039649732775452,
      "learning_rate": 3.7894902016426738e-06,
      "loss": 0.6275,
      "step": 3665
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0176045175959179,
      "learning_rate": 3.786958274248458e-06,
      "loss": 0.5725,
      "step": 3666
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.954283123285502,
      "learning_rate": 3.7844266773202448e-06,
      "loss": 0.5167,
      "step": 3667
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.998656315280911,
      "learning_rate": 3.7818954115477158e-06,
      "loss": 0.5774,
      "step": 3668
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.857171099232043,
      "learning_rate": 3.7793644776204503e-06,
      "loss": 0.5464,
      "step": 3669
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9685128508313092,
      "learning_rate": 3.7768338762279493e-06,
      "loss": 0.5844,
      "step": 3670
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9451842011379081,
      "learning_rate": 3.7743036080596184e-06,
      "loss": 0.5733,
      "step": 3671
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0049774831781737,
      "learning_rate": 3.77177367380477e-06,
      "loss": 0.6287,
      "step": 3672
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9708478743800151,
      "learning_rate": 3.7692440741526293e-06,
      "loss": 0.6192,
      "step": 3673
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0671482179780776,
      "learning_rate": 3.76671480979233e-06,
      "loss": 0.5902,
      "step": 3674
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0640032070625849,
      "learning_rate": 3.7641858814129093e-06,
      "loss": 0.5895,
      "step": 3675
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9086950026591353,
      "learning_rate": 3.7616572897033223e-06,
      "loss": 0.5054,
      "step": 3676
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8865936394059376,
      "learning_rate": 3.7591290353524247e-06,
      "loss": 0.4583,
      "step": 3677
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0264330824186085,
      "learning_rate": 3.7566011190489815e-06,
      "loss": 0.5498,
      "step": 3678
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.043436189062516,
      "learning_rate": 3.7540735414816685e-06,
      "loss": 0.5719,
      "step": 3679
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.919636737589086,
      "learning_rate": 3.7515463033390676e-06,
      "loss": 0.5255,
      "step": 3680
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9250224772550386,
      "learning_rate": 3.7490194053096668e-06,
      "loss": 0.496,
      "step": 3681
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.939856344823606,
      "learning_rate": 3.7464928480818623e-06,
      "loss": 0.4919,
      "step": 3682
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0346223794439071,
      "learning_rate": 3.7439666323439603e-06,
      "loss": 0.6701,
      "step": 3683
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.933259065258618,
      "learning_rate": 3.741440758784172e-06,
      "loss": 0.6088,
      "step": 3684
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0570004191780462,
      "learning_rate": 3.738915228090611e-06,
      "loss": 0.6379,
      "step": 3685
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0318288603357135,
      "learning_rate": 3.736390040951304e-06,
      "loss": 0.6213,
      "step": 3686
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.091764499969712,
      "learning_rate": 3.733865198054184e-06,
      "loss": 0.6524,
      "step": 3687
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.072082840247095,
      "learning_rate": 3.7313407000870826e-06,
      "loss": 0.5865,
      "step": 3688
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9434084236197783,
      "learning_rate": 3.728816547737745e-06,
      "loss": 0.6035,
      "step": 3689
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0818286902320455,
      "learning_rate": 3.7262927416938234e-06,
      "loss": 0.6139,
      "step": 3690
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9334099724350631,
      "learning_rate": 3.7237692826428662e-06,
      "loss": 0.5408,
      "step": 3691
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0566960558738816,
      "learning_rate": 3.721246171272336e-06,
      "loss": 0.6409,
      "step": 3692
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0305981051713555,
      "learning_rate": 3.718723408269599e-06,
      "loss": 0.582,
      "step": 3693
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0092975564584266,
      "learning_rate": 3.7162009943219234e-06,
      "loss": 0.5413,
      "step": 3694
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9449371866673747,
      "learning_rate": 3.7136789301164854e-06,
      "loss": 0.5441,
      "step": 3695
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0709692124389771,
      "learning_rate": 3.7111572163403653e-06,
      "loss": 0.5534,
      "step": 3696
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9555718633938672,
      "learning_rate": 3.7086358536805455e-06,
      "loss": 0.566,
      "step": 3697
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9575918702428928,
      "learning_rate": 3.7061148428239147e-06,
      "loss": 0.6039,
      "step": 3698
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.908542315185394,
      "learning_rate": 3.7035941844572687e-06,
      "loss": 0.5145,
      "step": 3699
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.985297275370482,
      "learning_rate": 3.701073879267302e-06,
      "loss": 0.5886,
      "step": 3700
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9234656082570784,
      "learning_rate": 3.698553927940615e-06,
      "loss": 0.4895,
      "step": 3701
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0285237296142236,
      "learning_rate": 3.6960343311637132e-06,
      "loss": 0.6208,
      "step": 3702
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0562931810182334,
      "learning_rate": 3.6935150896230045e-06,
      "loss": 0.6351,
      "step": 3703
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.308257613000332,
      "learning_rate": 3.6909962040047964e-06,
      "loss": 0.6813,
      "step": 3704
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0124062074607452,
      "learning_rate": 3.6884776749953065e-06,
      "loss": 0.6446,
      "step": 3705
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9711714314337834,
      "learning_rate": 3.6859595032806518e-06,
      "loss": 0.6212,
      "step": 3706
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0920171599617163,
      "learning_rate": 3.683441689546849e-06,
      "loss": 0.6771,
      "step": 3707
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9369927068211686,
      "learning_rate": 3.6809242344798207e-06,
      "loss": 0.5389,
      "step": 3708
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9367938641030269,
      "learning_rate": 3.6784071387653926e-06,
      "loss": 0.6136,
      "step": 3709
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0040852131615712,
      "learning_rate": 3.675890403089289e-06,
      "loss": 0.533,
      "step": 3710
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.112536930927805,
      "learning_rate": 3.6733740281371377e-06,
      "loss": 0.6125,
      "step": 3711
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0274323160056607,
      "learning_rate": 3.670858014594473e-06,
      "loss": 0.6369,
      "step": 3712
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9903198715359927,
      "learning_rate": 3.66834236314672e-06,
      "loss": 0.5086,
      "step": 3713
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0374422086389528,
      "learning_rate": 3.665827074479215e-06,
      "loss": 0.6532,
      "step": 3714
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0019424313550227,
      "learning_rate": 3.6633121492771933e-06,
      "loss": 0.542,
      "step": 3715
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6828897592530904,
      "learning_rate": 3.6607975882257875e-06,
      "loss": 0.4592,
      "step": 3716
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0281136626208112,
      "learning_rate": 3.6582833920100336e-06,
      "loss": 0.5634,
      "step": 3717
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9760001597668401,
      "learning_rate": 3.655769561314869e-06,
      "loss": 0.5393,
      "step": 3718
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.027252856895966,
      "learning_rate": 3.653256096825129e-06,
      "loss": 0.5786,
      "step": 3719
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0461158386862597,
      "learning_rate": 3.6507429992255503e-06,
      "loss": 0.5831,
      "step": 3720
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9730807327154681,
      "learning_rate": 3.648230269200775e-06,
      "loss": 0.5938,
      "step": 3721
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.026528074930323,
      "learning_rate": 3.645717907435332e-06,
      "loss": 0.5867,
      "step": 3722
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9747641035795074,
      "learning_rate": 3.6432059146136633e-06,
      "loss": 0.5838,
      "step": 3723
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.032551067774948,
      "learning_rate": 3.6406942914201045e-06,
      "loss": 0.5871,
      "step": 3724
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0405188636025802,
      "learning_rate": 3.6381830385388907e-06,
      "loss": 0.5902,
      "step": 3725
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.035594332369752,
      "learning_rate": 3.635672156654154e-06,
      "loss": 0.6106,
      "step": 3726
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9461560612524657,
      "learning_rate": 3.6331616464499297e-06,
      "loss": 0.576,
      "step": 3727
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.069606405132498,
      "learning_rate": 3.6306515086101522e-06,
      "loss": 0.6559,
      "step": 3728
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9593178916229198,
      "learning_rate": 3.6281417438186484e-06,
      "loss": 0.4979,
      "step": 3729
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9365949825020783,
      "learning_rate": 3.6256323527591496e-06,
      "loss": 0.5148,
      "step": 3730
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9071196365865485,
      "learning_rate": 3.623123336115284e-06,
      "loss": 0.6425,
      "step": 3731
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8857162582489134,
      "learning_rate": 3.6206146945705735e-06,
      "loss": 0.5876,
      "step": 3732
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0537968052530504,
      "learning_rate": 3.6181064288084423e-06,
      "loss": 0.6381,
      "step": 3733
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0710083113371174,
      "learning_rate": 3.6155985395122157e-06,
      "loss": 0.5272,
      "step": 3734
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0030239065559137,
      "learning_rate": 3.613091027365104e-06,
      "loss": 0.6065,
      "step": 3735
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0140373771734972,
      "learning_rate": 3.610583893050229e-06,
      "loss": 0.6166,
      "step": 3736
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9945480461370594,
      "learning_rate": 3.6080771372506017e-06,
      "loss": 0.5035,
      "step": 3737
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.015427771330092,
      "learning_rate": 3.6055707606491297e-06,
      "loss": 0.5989,
      "step": 3738
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9661695759944301,
      "learning_rate": 3.6030647639286196e-06,
      "loss": 0.5115,
      "step": 3739
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9746666332410842,
      "learning_rate": 3.6005591477717766e-06,
      "loss": 0.5358,
      "step": 3740
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0397687798955546,
      "learning_rate": 3.598053912861196e-06,
      "loss": 0.5973,
      "step": 3741
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9967555690430571,
      "learning_rate": 3.5955490598793734e-06,
      "loss": 0.5378,
      "step": 3742
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.986029809508707,
      "learning_rate": 3.5930445895087017e-06,
      "loss": 0.5898,
      "step": 3743
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6719587740086863,
      "learning_rate": 3.5905405024314683e-06,
      "loss": 0.5099,
      "step": 3744
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0386979394093914,
      "learning_rate": 3.588036799329853e-06,
      "loss": 0.5404,
      "step": 3745
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.961626072896982,
      "learning_rate": 3.585533480885934e-06,
      "loss": 0.5343,
      "step": 3746
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0124418587997555,
      "learning_rate": 3.5830305477816863e-06,
      "loss": 0.5607,
      "step": 3747
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8621972663090065,
      "learning_rate": 3.580528000698975e-06,
      "loss": 0.505,
      "step": 3748
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6921688698729664,
      "learning_rate": 3.5780258403195635e-06,
      "loss": 0.4988,
      "step": 3749
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9674436575349495,
      "learning_rate": 3.5755240673251125e-06,
      "loss": 0.5687,
      "step": 3750
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0101681946376775,
      "learning_rate": 3.5730226823971693e-06,
      "loss": 0.5953,
      "step": 3751
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0591658946321683,
      "learning_rate": 3.5705216862171823e-06,
      "loss": 0.6232,
      "step": 3752
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0626841687338044,
      "learning_rate": 3.568021079466494e-06,
      "loss": 0.6207,
      "step": 3753
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.007311192565927,
      "learning_rate": 3.5655208628263345e-06,
      "loss": 0.5529,
      "step": 3754
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9236009604488256,
      "learning_rate": 3.563021036977834e-06,
      "loss": 0.4992,
      "step": 3755
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0674966716552712,
      "learning_rate": 3.560521602602014e-06,
      "loss": 0.5689,
      "step": 3756
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0254910971082718,
      "learning_rate": 3.5580225603797873e-06,
      "loss": 0.614,
      "step": 3757
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9855145166046688,
      "learning_rate": 3.5555239109919647e-06,
      "loss": 0.5802,
      "step": 3758
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0386564201646364,
      "learning_rate": 3.5530256551192467e-06,
      "loss": 0.5977,
      "step": 3759
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9745676066325097,
      "learning_rate": 3.5505277934422254e-06,
      "loss": 0.5789,
      "step": 3760
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.010156681798062,
      "learning_rate": 3.5480303266413884e-06,
      "loss": 0.5099,
      "step": 3761
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9424613633755482,
      "learning_rate": 3.545533255397112e-06,
      "loss": 0.4641,
      "step": 3762
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0760814249442308,
      "learning_rate": 3.5430365803896736e-06,
      "loss": 0.5708,
      "step": 3763
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0939593328510797,
      "learning_rate": 3.540540302299229e-06,
      "loss": 0.6189,
      "step": 3764
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0033054315733312,
      "learning_rate": 3.5380444218058374e-06,
      "loss": 0.6512,
      "step": 3765
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0390242538863326,
      "learning_rate": 3.5355489395894448e-06,
      "loss": 0.6106,
      "step": 3766
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9220934988649714,
      "learning_rate": 3.533053856329889e-06,
      "loss": 0.5308,
      "step": 3767
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.999644982687219,
      "learning_rate": 3.5305591727068984e-06,
      "loss": 0.6297,
      "step": 3768
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9681998955100389,
      "learning_rate": 3.5280648894000957e-06,
      "loss": 0.512,
      "step": 3769
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0945452887477485,
      "learning_rate": 3.5255710070889903e-06,
      "loss": 0.5709,
      "step": 3770
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6743879012061959,
      "learning_rate": 3.5230775264529837e-06,
      "loss": 0.5036,
      "step": 3771
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0304830272793255,
      "learning_rate": 3.520584448171375e-06,
      "loss": 0.5843,
      "step": 3772
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6462321439523159,
      "learning_rate": 3.518091772923339e-06,
      "loss": 0.4509,
      "step": 3773
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0385460809770495,
      "learning_rate": 3.515599501387954e-06,
      "loss": 0.6379,
      "step": 3774
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0030173222407714,
      "learning_rate": 3.5131076342441838e-06,
      "loss": 0.6234,
      "step": 3775
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6417501162560092,
      "learning_rate": 3.5106161721708797e-06,
      "loss": 0.4729,
      "step": 3776
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.668150555746875,
      "learning_rate": 3.508125115846785e-06,
      "loss": 0.5046,
      "step": 3777
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.974666462778021,
      "learning_rate": 3.5056344659505335e-06,
      "loss": 0.5618,
      "step": 3778
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.047299274571347,
      "learning_rate": 3.503144223160644e-06,
      "loss": 0.6341,
      "step": 3779
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.970053916181946,
      "learning_rate": 3.5006543881555304e-06,
      "loss": 0.5466,
      "step": 3780
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1330396493119341,
      "learning_rate": 3.4981649616134912e-06,
      "loss": 0.6785,
      "step": 3781
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.061182426738625,
      "learning_rate": 3.495675944212715e-06,
      "loss": 0.5731,
      "step": 3782
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.025664740388029,
      "learning_rate": 3.4931873366312785e-06,
      "loss": 0.6098,
      "step": 3783
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8937068979590761,
      "learning_rate": 3.490699139547146e-06,
      "loss": 0.5564,
      "step": 3784
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0820352482690225,
      "learning_rate": 3.4882113536381744e-06,
      "loss": 0.5599,
      "step": 3785
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0334731082963249,
      "learning_rate": 3.4857239795821003e-06,
      "loss": 0.6502,
      "step": 3786
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9884288461889096,
      "learning_rate": 3.483237018056556e-06,
      "loss": 0.5464,
      "step": 3787
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0365161320607403,
      "learning_rate": 3.480750469739059e-06,
      "loss": 0.6674,
      "step": 3788
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0941740081468672,
      "learning_rate": 3.478264335307011e-06,
      "loss": 0.6813,
      "step": 3789
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9652479966212781,
      "learning_rate": 3.475778615437706e-06,
      "loss": 0.5583,
      "step": 3790
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9919937186626112,
      "learning_rate": 3.4732933108083218e-06,
      "loss": 0.6403,
      "step": 3791
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0708996079670052,
      "learning_rate": 3.470808422095923e-06,
      "loss": 0.7034,
      "step": 3792
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9765262810141593,
      "learning_rate": 3.4683239499774606e-06,
      "loss": 0.5522,
      "step": 3793
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8977404724686663,
      "learning_rate": 3.465839895129779e-06,
      "loss": 0.4889,
      "step": 3794
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9280739806044486,
      "learning_rate": 3.463356258229596e-06,
      "loss": 0.5496,
      "step": 3795
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0657170922593124,
      "learning_rate": 3.4608730399535273e-06,
      "loss": 0.5261,
      "step": 3796
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9949565918874539,
      "learning_rate": 3.4583902409780693e-06,
      "loss": 0.6083,
      "step": 3797
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0440458759239095,
      "learning_rate": 3.4559078619796036e-06,
      "loss": 0.6129,
      "step": 3798
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7112464100255558,
      "learning_rate": 3.4534259036343996e-06,
      "loss": 0.4976,
      "step": 3799
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0007167626916826,
      "learning_rate": 3.450944366618613e-06,
      "loss": 0.4835,
      "step": 3800
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9983836366821283,
      "learning_rate": 3.4484632516082784e-06,
      "loss": 0.6277,
      "step": 3801
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0702889912854148,
      "learning_rate": 3.445982559279322e-06,
      "loss": 0.6222,
      "step": 3802
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9909125383805912,
      "learning_rate": 3.4435022903075536e-06,
      "loss": 0.5702,
      "step": 3803
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.092927579175884,
      "learning_rate": 3.441022445368668e-06,
      "loss": 0.6606,
      "step": 3804
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.997139841203399,
      "learning_rate": 3.4385430251382407e-06,
      "loss": 0.5478,
      "step": 3805
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.013557843900264,
      "learning_rate": 3.4360640302917353e-06,
      "loss": 0.5567,
      "step": 3806
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0851471345922516,
      "learning_rate": 3.433585461504499e-06,
      "loss": 0.5636,
      "step": 3807
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0669344730340906,
      "learning_rate": 3.4311073194517596e-06,
      "loss": 0.6132,
      "step": 3808
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.945417068996163,
      "learning_rate": 3.4286296048086343e-06,
      "loss": 0.5444,
      "step": 3809
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1078145189560855,
      "learning_rate": 3.42615231825012e-06,
      "loss": 0.6499,
      "step": 3810
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0910434657081942,
      "learning_rate": 3.423675460451097e-06,
      "loss": 0.5797,
      "step": 3811
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1091749285038728,
      "learning_rate": 3.4211990320863307e-06,
      "loss": 0.6325,
      "step": 3812
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9652520994772008,
      "learning_rate": 3.4187230338304684e-06,
      "loss": 0.5161,
      "step": 3813
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0093500327544223,
      "learning_rate": 3.416247466358039e-06,
      "loss": 0.6204,
      "step": 3814
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9716010010385902,
      "learning_rate": 3.413772330343455e-06,
      "loss": 0.5435,
      "step": 3815
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9334389388879336,
      "learning_rate": 3.4112976264610167e-06,
      "loss": 0.5108,
      "step": 3816
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0416967376825383,
      "learning_rate": 3.408823355384894e-06,
      "loss": 0.7029,
      "step": 3817
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9808993305224007,
      "learning_rate": 3.406349517789151e-06,
      "loss": 0.5983,
      "step": 3818
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9403708357453641,
      "learning_rate": 3.4038761143477296e-06,
      "loss": 0.6402,
      "step": 3819
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0255942724402032,
      "learning_rate": 3.4014031457344517e-06,
      "loss": 0.5946,
      "step": 3820
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0136123411115194,
      "learning_rate": 3.3989306126230226e-06,
      "loss": 0.6333,
      "step": 3821
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1014044416695128,
      "learning_rate": 3.3964585156870267e-06,
      "loss": 0.649,
      "step": 3822
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0704261066864926,
      "learning_rate": 3.393986855599936e-06,
      "loss": 0.6005,
      "step": 3823
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0293231100131388,
      "learning_rate": 3.391515633035093e-06,
      "loss": 0.6228,
      "step": 3824
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0497042683783568,
      "learning_rate": 3.389044848665731e-06,
      "loss": 0.5841,
      "step": 3825
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9889951022187614,
      "learning_rate": 3.3865745031649595e-06,
      "loss": 0.5989,
      "step": 3826
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.009225706833987,
      "learning_rate": 3.3841045972057663e-06,
      "loss": 0.587,
      "step": 3827
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6881615793553358,
      "learning_rate": 3.3816351314610235e-06,
      "loss": 0.4986,
      "step": 3828
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.981582931917428,
      "learning_rate": 3.379166106603482e-06,
      "loss": 0.6434,
      "step": 3829
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.999532755734759,
      "learning_rate": 3.3766975233057715e-06,
      "loss": 0.5965,
      "step": 3830
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9925510671524317,
      "learning_rate": 3.3742293822404005e-06,
      "loss": 0.6545,
      "step": 3831
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0584546470653944,
      "learning_rate": 3.371761684079763e-06,
      "loss": 0.611,
      "step": 3832
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9643934712837668,
      "learning_rate": 3.369294429496124e-06,
      "loss": 0.5979,
      "step": 3833
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1105021994254516,
      "learning_rate": 3.366827619161632e-06,
      "loss": 0.4818,
      "step": 3834
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0063002848521414,
      "learning_rate": 3.364361253748318e-06,
      "loss": 0.6076,
      "step": 3835
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9740458883440557,
      "learning_rate": 3.361895333928083e-06,
      "loss": 0.5727,
      "step": 3836
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.047983319551925,
      "learning_rate": 3.3594298603727126e-06,
      "loss": 0.6167,
      "step": 3837
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0238092677234603,
      "learning_rate": 3.356964833753875e-06,
      "loss": 0.5851,
      "step": 3838
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0280806689082074,
      "learning_rate": 3.3545002547431034e-06,
      "loss": 0.6203,
      "step": 3839
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0236209512934518,
      "learning_rate": 3.3520361240118216e-06,
      "loss": 0.6073,
      "step": 3840
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0531921026539857,
      "learning_rate": 3.3495724422313262e-06,
      "loss": 0.5258,
      "step": 3841
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0495024651792977,
      "learning_rate": 3.347109210072793e-06,
      "loss": 0.55,
      "step": 3842
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9861963312732116,
      "learning_rate": 3.3446464282072723e-06,
      "loss": 0.6009,
      "step": 3843
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0215939802065237,
      "learning_rate": 3.3421840973056935e-06,
      "loss": 0.6162,
      "step": 3844
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.096611425229634,
      "learning_rate": 3.3397222180388677e-06,
      "loss": 0.6058,
      "step": 3845
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9837178727566318,
      "learning_rate": 3.3372607910774726e-06,
      "loss": 0.5588,
      "step": 3846
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0430030139893576,
      "learning_rate": 3.3347998170920724e-06,
      "loss": 0.6685,
      "step": 3847
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0483787884757538,
      "learning_rate": 3.3323392967531043e-06,
      "loss": 0.6181,
      "step": 3848
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9593660690088923,
      "learning_rate": 3.32987923073088e-06,
      "loss": 0.5636,
      "step": 3849
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0171221211863994,
      "learning_rate": 3.327419619695591e-06,
      "loss": 0.6092,
      "step": 3850
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1164319239268619,
      "learning_rate": 3.3249604643173037e-06,
      "loss": 0.6894,
      "step": 3851
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0179417622328548,
      "learning_rate": 3.3225017652659577e-06,
      "loss": 0.4866,
      "step": 3852
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0404798651394507,
      "learning_rate": 3.3200435232113694e-06,
      "loss": 0.601,
      "step": 3853
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9838921438738345,
      "learning_rate": 3.3175857388232376e-06,
      "loss": 0.5197,
      "step": 3854
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.031423045261859,
      "learning_rate": 3.3151284127711227e-06,
      "loss": 0.5675,
      "step": 3855
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0434942394479056,
      "learning_rate": 3.312671545724474e-06,
      "loss": 0.5814,
      "step": 3856
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0556089090014797,
      "learning_rate": 3.3102151383526077e-06,
      "loss": 0.6185,
      "step": 3857
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0753541375345395,
      "learning_rate": 3.3077591913247166e-06,
      "loss": 0.5566,
      "step": 3858
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0163533202198782,
      "learning_rate": 3.305303705309868e-06,
      "loss": 0.5634,
      "step": 3859
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0907485922303994,
      "learning_rate": 3.3028486809770046e-06,
      "loss": 0.6315,
      "step": 3860
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.022000890092407,
      "learning_rate": 3.300394118994944e-06,
      "loss": 0.5632,
      "step": 3861
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.077016789156783,
      "learning_rate": 3.297940020032374e-06,
      "loss": 0.6239,
      "step": 3862
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7277498780297968,
      "learning_rate": 3.295486384757861e-06,
      "loss": 0.5215,
      "step": 3863
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9914723162909143,
      "learning_rate": 3.2930332138398422e-06,
      "loss": 0.5317,
      "step": 3864
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0380908649909524,
      "learning_rate": 3.2905805079466284e-06,
      "loss": 0.5879,
      "step": 3865
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0138116000033237,
      "learning_rate": 3.2881282677464034e-06,
      "loss": 0.6106,
      "step": 3866
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.961425936120296,
      "learning_rate": 3.2856764939072294e-06,
      "loss": 0.5415,
      "step": 3867
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6584740833075744,
      "learning_rate": 3.283225187097031e-06,
      "loss": 0.4866,
      "step": 3868
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9615580717935966,
      "learning_rate": 3.2807743479836155e-06,
      "loss": 0.5571,
      "step": 3869
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9550526246979025,
      "learning_rate": 3.27832397723466e-06,
      "loss": 0.5318,
      "step": 3870
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0677613108747284,
      "learning_rate": 3.27587407551771e-06,
      "loss": 0.6134,
      "step": 3871
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9922388490368044,
      "learning_rate": 3.273424643500187e-06,
      "loss": 0.5888,
      "step": 3872
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0540217881380682,
      "learning_rate": 3.2709756818493867e-06,
      "loss": 0.5908,
      "step": 3873
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1052795806249633,
      "learning_rate": 3.26852719123247e-06,
      "loss": 0.5983,
      "step": 3874
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9467924582273569,
      "learning_rate": 3.266079172316473e-06,
      "loss": 0.5271,
      "step": 3875
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9873677001461719,
      "learning_rate": 3.263631625768309e-06,
      "loss": 0.5883,
      "step": 3876
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0277961079433071,
      "learning_rate": 3.2611845522547503e-06,
      "loss": 0.5514,
      "step": 3877
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9640356475261352,
      "learning_rate": 3.2587379524424513e-06,
      "loss": 0.6418,
      "step": 3878
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9306280423357934,
      "learning_rate": 3.2562918269979334e-06,
      "loss": 0.5543,
      "step": 3879
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9874498620896189,
      "learning_rate": 3.253846176587586e-06,
      "loss": 0.6491,
      "step": 3880
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0175588846231591,
      "learning_rate": 3.251401001877673e-06,
      "loss": 0.6684,
      "step": 3881
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7127520043173042,
      "learning_rate": 3.2489563035343276e-06,
      "loss": 0.4573,
      "step": 3882
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.046964530426077,
      "learning_rate": 3.246512082223555e-06,
      "loss": 0.5647,
      "step": 3883
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0364448838734412,
      "learning_rate": 3.2440683386112238e-06,
      "loss": 0.5877,
      "step": 3884
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9810602879243683,
      "learning_rate": 3.24162507336308e-06,
      "loss": 0.5139,
      "step": 3885
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0750372116204037,
      "learning_rate": 3.2391822871447377e-06,
      "loss": 0.5118,
      "step": 3886
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0437136388529185,
      "learning_rate": 3.2367399806216765e-06,
      "loss": 0.6162,
      "step": 3887
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6642080921326198,
      "learning_rate": 3.234298154459249e-06,
      "loss": 0.4872,
      "step": 3888
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9747672796951058,
      "learning_rate": 3.231856809322677e-06,
      "loss": 0.5257,
      "step": 3889
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9604167273488331,
      "learning_rate": 3.229415945877048e-06,
      "loss": 0.4676,
      "step": 3890
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.040546834155076,
      "learning_rate": 3.226975564787322e-06,
      "loss": 0.5823,
      "step": 3891
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0658304899066287,
      "learning_rate": 3.224535666718327e-06,
      "loss": 0.5483,
      "step": 3892
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0114800986989505,
      "learning_rate": 3.2220962523347567e-06,
      "loss": 0.5761,
      "step": 3893
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0324232040563512,
      "learning_rate": 3.219657322301175e-06,
      "loss": 0.5825,
      "step": 3894
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1213028642852583,
      "learning_rate": 3.2172188772820154e-06,
      "loss": 0.6016,
      "step": 3895
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9161299632257038,
      "learning_rate": 3.214780917941575e-06,
      "loss": 0.4775,
      "step": 3896
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0684266603916466,
      "learning_rate": 3.212343444944022e-06,
      "loss": 0.572,
      "step": 3897
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.07930994972836,
      "learning_rate": 3.209906458953394e-06,
      "loss": 0.6814,
      "step": 3898
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9823765783505446,
      "learning_rate": 3.207469960633588e-06,
      "loss": 0.5735,
      "step": 3899
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9964728991615143,
      "learning_rate": 3.2050339506483774e-06,
      "loss": 0.5309,
      "step": 3900
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9701642956689647,
      "learning_rate": 3.2025984296613965e-06,
      "loss": 0.588,
      "step": 3901
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0205962542619393,
      "learning_rate": 3.200163398336151e-06,
      "loss": 0.5608,
      "step": 3902
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0095257759575251,
      "learning_rate": 3.1977288573360064e-06,
      "loss": 0.5441,
      "step": 3903
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9797308776978978,
      "learning_rate": 3.1952948073242006e-06,
      "loss": 0.5662,
      "step": 3904
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9433515548605147,
      "learning_rate": 3.19286124896384e-06,
      "loss": 0.6069,
      "step": 3905
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9823302178287,
      "learning_rate": 3.190428182917885e-06,
      "loss": 0.5448,
      "step": 3906
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0107005098866073,
      "learning_rate": 3.187995609849176e-06,
      "loss": 0.5432,
      "step": 3907
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9877543528653476,
      "learning_rate": 3.1855635304204113e-06,
      "loss": 0.5698,
      "step": 3908
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0127885371414798,
      "learning_rate": 3.1831319452941557e-06,
      "loss": 0.5861,
      "step": 3909
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1054590602451397,
      "learning_rate": 3.1807008551328407e-06,
      "loss": 0.6148,
      "step": 3910
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9365363089327245,
      "learning_rate": 3.1782702605987623e-06,
      "loss": 0.517,
      "step": 3911
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9800314041370826,
      "learning_rate": 3.175840162354081e-06,
      "loss": 0.6697,
      "step": 3912
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0280127140618784,
      "learning_rate": 3.1734105610608213e-06,
      "loss": 0.6088,
      "step": 3913
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0145710239859964,
      "learning_rate": 3.1709814573808766e-06,
      "loss": 0.5693,
      "step": 3914
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0268224268479194,
      "learning_rate": 3.168552851976e-06,
      "loss": 0.5772,
      "step": 3915
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0854266377996473,
      "learning_rate": 3.1661247455078097e-06,
      "loss": 0.6609,
      "step": 3916
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0064409912197665,
      "learning_rate": 3.163697138637791e-06,
      "loss": 0.6255,
      "step": 3917
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0181912941069655,
      "learning_rate": 3.161270032027289e-06,
      "loss": 0.653,
      "step": 3918
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0529235769020393,
      "learning_rate": 3.1588434263375146e-06,
      "loss": 0.6221,
      "step": 3919
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7093248634960904,
      "learning_rate": 3.156417322229543e-06,
      "loss": 0.4734,
      "step": 3920
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0317601315287412,
      "learning_rate": 3.153991720364313e-06,
      "loss": 0.5809,
      "step": 3921
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9941643082894005,
      "learning_rate": 3.151566621402622e-06,
      "loss": 0.6072,
      "step": 3922
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9527915073207505,
      "learning_rate": 3.1491420260051362e-06,
      "loss": 0.5579,
      "step": 3923
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9632165856033459,
      "learning_rate": 3.146717934832383e-06,
      "loss": 0.544,
      "step": 3924
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0069444861692525,
      "learning_rate": 3.1442943485447493e-06,
      "loss": 0.5907,
      "step": 3925
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0434127370633832,
      "learning_rate": 3.1418712678024866e-06,
      "loss": 0.6144,
      "step": 3926
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.048718618294521,
      "learning_rate": 3.1394486932657133e-06,
      "loss": 0.5173,
      "step": 3927
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6605594509395147,
      "learning_rate": 3.137026625594399e-06,
      "loss": 0.4878,
      "step": 3928
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1391659174370299,
      "learning_rate": 3.1346050654483867e-06,
      "loss": 0.5102,
      "step": 3929
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9176096080073862,
      "learning_rate": 3.132184013487375e-06,
      "loss": 0.5154,
      "step": 3930
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.003544181392987,
      "learning_rate": 3.129763470370924e-06,
      "loss": 0.4953,
      "step": 3931
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9733027541948882,
      "learning_rate": 3.1273434367584567e-06,
      "loss": 0.6208,
      "step": 3932
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0427152419050678,
      "learning_rate": 3.124923913309259e-06,
      "loss": 0.583,
      "step": 3933
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0123207416349886,
      "learning_rate": 3.1225049006824724e-06,
      "loss": 0.646,
      "step": 3934
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0700395644681315,
      "learning_rate": 3.1200863995371035e-06,
      "loss": 0.569,
      "step": 3935
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0121122100806814,
      "learning_rate": 3.1176684105320208e-06,
      "loss": 0.5601,
      "step": 3936
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0354226557393735,
      "learning_rate": 3.1152509343259494e-06,
      "loss": 0.5317,
      "step": 3937
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0493509529674856,
      "learning_rate": 3.112833971577478e-06,
      "loss": 0.6224,
      "step": 3938
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0685350193886183,
      "learning_rate": 3.1104175229450517e-06,
      "loss": 0.6386,
      "step": 3939
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9923083837423622,
      "learning_rate": 3.1080015890869796e-06,
      "loss": 0.5829,
      "step": 3940
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9283431805266033,
      "learning_rate": 3.1055861706614264e-06,
      "loss": 0.5328,
      "step": 3941
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0854677475824348,
      "learning_rate": 3.1031712683264204e-06,
      "loss": 0.5204,
      "step": 3942
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0388740428344243,
      "learning_rate": 3.1007568827398495e-06,
      "loss": 0.4937,
      "step": 3943
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9577553843347524,
      "learning_rate": 3.0983430145594547e-06,
      "loss": 0.5325,
      "step": 3944
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0226161745670226,
      "learning_rate": 3.0959296644428427e-06,
      "loss": 0.5775,
      "step": 3945
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0168065622439244,
      "learning_rate": 3.0935168330474763e-06,
      "loss": 0.673,
      "step": 3946
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9236844414712579,
      "learning_rate": 3.0911045210306767e-06,
      "loss": 0.5031,
      "step": 3947
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9867505138359844,
      "learning_rate": 3.088692729049624e-06,
      "loss": 0.5637,
      "step": 3948
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6791108286637646,
      "learning_rate": 3.0862814577613598e-06,
      "loss": 0.4872,
      "step": 3949
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6642796346266474,
      "learning_rate": 3.083870707822776e-06,
      "loss": 0.5134,
      "step": 3950
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9905004600933245,
      "learning_rate": 3.08146047989063e-06,
      "loss": 0.5607,
      "step": 3951
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.060459845436706,
      "learning_rate": 3.079050774621536e-06,
      "loss": 0.698,
      "step": 3952
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9989552984061584,
      "learning_rate": 3.0766415926719606e-06,
      "loss": 0.6178,
      "step": 3953
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9602586761919033,
      "learning_rate": 3.074232934698234e-06,
      "loss": 0.6212,
      "step": 3954
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.004330893952061,
      "learning_rate": 3.07182480135654e-06,
      "loss": 0.5922,
      "step": 3955
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9897670689631785,
      "learning_rate": 3.06941719330292e-06,
      "loss": 0.6009,
      "step": 3956
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1091726371428576,
      "learning_rate": 3.067010111193272e-06,
      "loss": 0.6615,
      "step": 3957
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0335384186321204,
      "learning_rate": 3.0646035556833563e-06,
      "loss": 0.5858,
      "step": 3958
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0714992736812141,
      "learning_rate": 3.0621975274287784e-06,
      "loss": 0.5414,
      "step": 3959
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0241482478513784,
      "learning_rate": 3.059792027085011e-06,
      "loss": 0.6138,
      "step": 3960
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9956902669802108,
      "learning_rate": 3.0573870553073776e-06,
      "loss": 0.5709,
      "step": 3961
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1193646097576968,
      "learning_rate": 3.0549826127510595e-06,
      "loss": 0.5772,
      "step": 3962
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8937665335720016,
      "learning_rate": 3.0525787000710915e-06,
      "loss": 0.4973,
      "step": 3963
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0077464705286991,
      "learning_rate": 3.0501753179223657e-06,
      "loss": 0.5847,
      "step": 3964
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0737823871175538,
      "learning_rate": 3.0477724669596326e-06,
      "loss": 0.6278,
      "step": 3965
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0488158934169045,
      "learning_rate": 3.045370147837492e-06,
      "loss": 0.6242,
      "step": 3966
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0433275138668243,
      "learning_rate": 3.042968361210403e-06,
      "loss": 0.5703,
      "step": 3967
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0373335544904982,
      "learning_rate": 3.040567107732679e-06,
      "loss": 0.6565,
      "step": 3968
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9717788618526988,
      "learning_rate": 3.0381663880584855e-06,
      "loss": 0.5418,
      "step": 3969
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9742336211818127,
      "learning_rate": 3.0357662028418455e-06,
      "loss": 0.5567,
      "step": 3970
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.995405615684413,
      "learning_rate": 3.0333665527366394e-06,
      "loss": 0.6194,
      "step": 3971
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0705520500341474,
      "learning_rate": 3.0309674383965915e-06,
      "loss": 0.6058,
      "step": 3972
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.008180411432752,
      "learning_rate": 3.0285688604752916e-06,
      "loss": 0.6149,
      "step": 3973
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7135981161783805,
      "learning_rate": 3.026170819626178e-06,
      "loss": 0.5309,
      "step": 3974
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0364132318562187,
      "learning_rate": 3.0237733165025408e-06,
      "loss": 0.5973,
      "step": 3975
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9770705726950235,
      "learning_rate": 3.021376351757527e-06,
      "loss": 0.5969,
      "step": 3976
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.980248473453085,
      "learning_rate": 3.0189799260441377e-06,
      "loss": 0.6109,
      "step": 3977
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0503109441728466,
      "learning_rate": 3.0165840400152218e-06,
      "loss": 0.6146,
      "step": 3978
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0329639379462412,
      "learning_rate": 3.014188694323486e-06,
      "loss": 0.6247,
      "step": 3979
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9636408239824905,
      "learning_rate": 3.0117938896214904e-06,
      "loss": 0.5129,
      "step": 3980
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1911600335662196,
      "learning_rate": 3.0093996265616447e-06,
      "loss": 0.6054,
      "step": 3981
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9808958923708414,
      "learning_rate": 3.007005905796212e-06,
      "loss": 0.5768,
      "step": 3982
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0026946143582824,
      "learning_rate": 3.0046127279773067e-06,
      "loss": 0.6459,
      "step": 3983
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.08459436468111,
      "learning_rate": 3.002220093756899e-06,
      "loss": 0.6993,
      "step": 3984
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9606463638486709,
      "learning_rate": 2.999828003786806e-06,
      "loss": 0.6063,
      "step": 3985
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0152998593346465,
      "learning_rate": 2.9974364587186988e-06,
      "loss": 0.5774,
      "step": 3986
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9389022774674889,
      "learning_rate": 2.995045459204104e-06,
      "loss": 0.5674,
      "step": 3987
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0183193081616835,
      "learning_rate": 2.9926550058943905e-06,
      "loss": 0.585,
      "step": 3988
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0289483491356517,
      "learning_rate": 2.9902650994407867e-06,
      "loss": 0.6559,
      "step": 3989
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1998811268913283,
      "learning_rate": 2.9878757404943694e-06,
      "loss": 0.6646,
      "step": 3990
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9602306860241301,
      "learning_rate": 2.985486929706064e-06,
      "loss": 0.6002,
      "step": 3991
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9152847388631519,
      "learning_rate": 2.9830986677266495e-06,
      "loss": 0.4856,
      "step": 3992
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0392895417871526,
      "learning_rate": 2.980710955206755e-06,
      "loss": 0.5952,
      "step": 3993
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9815355673280974,
      "learning_rate": 2.9783237927968567e-06,
      "loss": 0.5494,
      "step": 3994
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0115383196015817,
      "learning_rate": 2.9759371811472857e-06,
      "loss": 0.5873,
      "step": 3995
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0855040415203778,
      "learning_rate": 2.9735511209082213e-06,
      "loss": 0.5879,
      "step": 3996
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9648672459405413,
      "learning_rate": 2.9711656127296895e-06,
      "loss": 0.5235,
      "step": 3997
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9857473710790505,
      "learning_rate": 2.968780657261571e-06,
      "loss": 0.554,
      "step": 3998
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0269452393286644,
      "learning_rate": 2.96639625515359e-06,
      "loss": 0.6423,
      "step": 3999
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9932749260085405,
      "learning_rate": 2.9640124070553296e-06,
      "loss": 0.6052,
      "step": 4000
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9086579152455732,
      "learning_rate": 2.961629113616209e-06,
      "loss": 0.5286,
      "step": 4001
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0370745166057331,
      "learning_rate": 2.959246375485506e-06,
      "loss": 0.5563,
      "step": 4002
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0665460025653735,
      "learning_rate": 2.9568641933123456e-06,
      "loss": 0.6464,
      "step": 4003
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0087462408363064,
      "learning_rate": 2.954482567745697e-06,
      "loss": 0.6011,
      "step": 4004
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.062212148392431,
      "learning_rate": 2.9521014994343823e-06,
      "loss": 0.6257,
      "step": 4005
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0860046954246445,
      "learning_rate": 2.9497209890270704e-06,
      "loss": 0.5646,
      "step": 4006
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9864791856581164,
      "learning_rate": 2.947341037172277e-06,
      "loss": 0.5725,
      "step": 4007
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9408093108832384,
      "learning_rate": 2.944961644518366e-06,
      "loss": 0.5837,
      "step": 4008
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8584641385170075,
      "learning_rate": 2.942582811713553e-06,
      "loss": 0.6019,
      "step": 4009
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9590888743148718,
      "learning_rate": 2.940204539405892e-06,
      "loss": 0.5878,
      "step": 4010
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0804387029583662,
      "learning_rate": 2.937826828243294e-06,
      "loss": 0.6017,
      "step": 4011
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0725208438126408,
      "learning_rate": 2.9354496788735145e-06,
      "loss": 0.582,
      "step": 4012
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9758441344069746,
      "learning_rate": 2.9330730919441498e-06,
      "loss": 0.5477,
      "step": 4013
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9948540696366719,
      "learning_rate": 2.9306970681026503e-06,
      "loss": 0.5982,
      "step": 4014
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.017758082730898,
      "learning_rate": 2.9283216079963108e-06,
      "loss": 0.5707,
      "step": 4015
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9648955590752109,
      "learning_rate": 2.9259467122722705e-06,
      "loss": 0.5523,
      "step": 4016
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0696241005428277,
      "learning_rate": 2.9235723815775167e-06,
      "loss": 0.6234,
      "step": 4017
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.075589223409787,
      "learning_rate": 2.9211986165588856e-06,
      "loss": 0.6035,
      "step": 4018
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9699281154448633,
      "learning_rate": 2.9188254178630526e-06,
      "loss": 0.5683,
      "step": 4019
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0653538926361192,
      "learning_rate": 2.916452786136542e-06,
      "loss": 0.5727,
      "step": 4020
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1077431501808022,
      "learning_rate": 2.914080722025728e-06,
      "loss": 0.6357,
      "step": 4021
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0366285857823563,
      "learning_rate": 2.9117092261768247e-06,
      "loss": 0.57,
      "step": 4022
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1035121598330206,
      "learning_rate": 2.9093382992358897e-06,
      "loss": 0.5655,
      "step": 4023
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.021557033555813,
      "learning_rate": 2.906967941848834e-06,
      "loss": 0.5644,
      "step": 4024
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9723155030057634,
      "learning_rate": 2.9045981546614057e-06,
      "loss": 0.5638,
      "step": 4025
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9999347167003046,
      "learning_rate": 2.9022289383191972e-06,
      "loss": 0.5834,
      "step": 4026
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9184491080564697,
      "learning_rate": 2.899860293467652e-06,
      "loss": 0.4453,
      "step": 4027
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.098049304349839,
      "learning_rate": 2.897492220752057e-06,
      "loss": 0.6185,
      "step": 4028
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6522077055101206,
      "learning_rate": 2.8951247208175337e-06,
      "loss": 0.4705,
      "step": 4029
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9601248699824723,
      "learning_rate": 2.8927577943090574e-06,
      "loss": 0.5736,
      "step": 4030
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9295371322619084,
      "learning_rate": 2.890391441871446e-06,
      "loss": 0.5192,
      "step": 4031
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9424014440626866,
      "learning_rate": 2.8880256641493582e-06,
      "loss": 0.5074,
      "step": 4032
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9760377335761027,
      "learning_rate": 2.885660461787294e-06,
      "loss": 0.5554,
      "step": 4033
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0681492263783305,
      "learning_rate": 2.8832958354296048e-06,
      "loss": 0.5766,
      "step": 4034
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0247553021239837,
      "learning_rate": 2.880931785720474e-06,
      "loss": 0.523,
      "step": 4035
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0980706991646645,
      "learning_rate": 2.8785683133039404e-06,
      "loss": 0.6152,
      "step": 4036
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0251535049515124,
      "learning_rate": 2.876205418823875e-06,
      "loss": 0.5336,
      "step": 4037
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.066769795922078,
      "learning_rate": 2.8738431029239954e-06,
      "loss": 0.6633,
      "step": 4038
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9513954635247176,
      "learning_rate": 2.8714813662478633e-06,
      "loss": 0.5442,
      "step": 4039
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0569619391592409,
      "learning_rate": 2.869120209438879e-06,
      "loss": 0.5756,
      "step": 4040
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.998953772481972,
      "learning_rate": 2.8667596331402892e-06,
      "loss": 0.645,
      "step": 4041
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9452384310233856,
      "learning_rate": 2.864399637995176e-06,
      "loss": 0.4889,
      "step": 4042
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6030112875811883,
      "learning_rate": 2.8620402246464717e-06,
      "loss": 0.4678,
      "step": 4043
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0583223163384945,
      "learning_rate": 2.8596813937369437e-06,
      "loss": 0.5462,
      "step": 4044
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9372717342109275,
      "learning_rate": 2.8573231459092007e-06,
      "loss": 0.5508,
      "step": 4045
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6291393750141842,
      "learning_rate": 2.854965481805697e-06,
      "loss": 0.4609,
      "step": 4046
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0262129556727317,
      "learning_rate": 2.852608402068725e-06,
      "loss": 0.5961,
      "step": 4047
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9380202927331176,
      "learning_rate": 2.8502519073404155e-06,
      "loss": 0.4935,
      "step": 4048
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0678738327192216,
      "learning_rate": 2.847895998262744e-06,
      "loss": 0.6459,
      "step": 4049
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.01267225095389,
      "learning_rate": 2.8455406754775305e-06,
      "loss": 0.6172,
      "step": 4050
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0619557618270221,
      "learning_rate": 2.8431859396264203e-06,
      "loss": 0.6012,
      "step": 4051
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8820615161607135,
      "learning_rate": 2.8408317913509137e-06,
      "loss": 0.5129,
      "step": 4052
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9836737696913604,
      "learning_rate": 2.8384782312923466e-06,
      "loss": 0.5429,
      "step": 4053
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0486368746643207,
      "learning_rate": 2.836125260091892e-06,
      "loss": 0.6453,
      "step": 4054
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.973587137202366,
      "learning_rate": 2.8337728783905618e-06,
      "loss": 0.5924,
      "step": 4055
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0595626153235649,
      "learning_rate": 2.8314210868292145e-06,
      "loss": 0.6533,
      "step": 4056
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0163422289707038,
      "learning_rate": 2.82906988604854e-06,
      "loss": 0.5278,
      "step": 4057
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9642361096096631,
      "learning_rate": 2.8267192766890685e-06,
      "loss": 0.5671,
      "step": 4058
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9973913156262183,
      "learning_rate": 2.824369259391173e-06,
      "loss": 0.5524,
      "step": 4059
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6535756186752398,
      "learning_rate": 2.822019834795065e-06,
      "loss": 0.4992,
      "step": 4060
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9676996624162714,
      "learning_rate": 2.8196710035407905e-06,
      "loss": 0.5423,
      "step": 4061
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0111555335197016,
      "learning_rate": 2.8173227662682332e-06,
      "loss": 0.5783,
      "step": 4062
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0528855400098003,
      "learning_rate": 2.814975123617123e-06,
      "loss": 0.6183,
      "step": 4063
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9720294574597103,
      "learning_rate": 2.8126280762270177e-06,
      "loss": 0.5766,
      "step": 4064
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0117059077883155,
      "learning_rate": 2.810281624737321e-06,
      "loss": 0.5884,
      "step": 4065
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0029032986306197,
      "learning_rate": 2.8079357697872696e-06,
      "loss": 0.543,
      "step": 4066
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9829190310875199,
      "learning_rate": 2.8055905120159375e-06,
      "loss": 0.5926,
      "step": 4067
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0012520707616437,
      "learning_rate": 2.803245852062241e-06,
      "loss": 0.5412,
      "step": 4068
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.922010983816059,
      "learning_rate": 2.8009017905649283e-06,
      "loss": 0.5624,
      "step": 4069
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9398984862427566,
      "learning_rate": 2.7985583281625844e-06,
      "loss": 0.5919,
      "step": 4070
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.99623431936887,
      "learning_rate": 2.7962154654936352e-06,
      "loss": 0.6051,
      "step": 4071
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0172174272942505,
      "learning_rate": 2.7938732031963443e-06,
      "loss": 0.5767,
      "step": 4072
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1255140562845085,
      "learning_rate": 2.791531541908802e-06,
      "loss": 0.6609,
      "step": 4073
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0834849077560833,
      "learning_rate": 2.7891904822689443e-06,
      "loss": 0.6653,
      "step": 4074
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.993396970970261,
      "learning_rate": 2.7868500249145414e-06,
      "loss": 0.5427,
      "step": 4075
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0394902440121143,
      "learning_rate": 2.784510170483199e-06,
      "loss": 0.5476,
      "step": 4076
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0395937735266692,
      "learning_rate": 2.7821709196123535e-06,
      "loss": 0.6027,
      "step": 4077
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.023831495368322,
      "learning_rate": 2.779832272939285e-06,
      "loss": 0.6551,
      "step": 4078
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.996346818136337,
      "learning_rate": 2.7774942311011082e-06,
      "loss": 0.5874,
      "step": 4079
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1227017854867904,
      "learning_rate": 2.7751567947347624e-06,
      "loss": 0.6561,
      "step": 4080
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1112045278347196,
      "learning_rate": 2.772819964477035e-06,
      "loss": 0.6021,
      "step": 4081
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0112670827851002,
      "learning_rate": 2.7704837409645425e-06,
      "loss": 0.5412,
      "step": 4082
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0267388027631472,
      "learning_rate": 2.768148124833736e-06,
      "loss": 0.6338,
      "step": 4083
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0503775356111906,
      "learning_rate": 2.765813116720901e-06,
      "loss": 0.6238,
      "step": 4084
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9453234809026312,
      "learning_rate": 2.7634787172621593e-06,
      "loss": 0.544,
      "step": 4085
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9783486332969259,
      "learning_rate": 2.761144927093464e-06,
      "loss": 0.6603,
      "step": 4086
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9646492267888072,
      "learning_rate": 2.7588117468506064e-06,
      "loss": 0.5816,
      "step": 4087
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0643843033456615,
      "learning_rate": 2.756479177169208e-06,
      "loss": 0.5559,
      "step": 4088
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.107682969498182,
      "learning_rate": 2.7541472186847224e-06,
      "loss": 0.6535,
      "step": 4089
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9439996384254786,
      "learning_rate": 2.751815872032444e-06,
      "loss": 0.4856,
      "step": 4090
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0541343974608182,
      "learning_rate": 2.7494851378474936e-06,
      "loss": 0.6625,
      "step": 4091
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9710797886490599,
      "learning_rate": 2.7471550167648255e-06,
      "loss": 0.5354,
      "step": 4092
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9632435199968827,
      "learning_rate": 2.744825509419231e-06,
      "loss": 0.5757,
      "step": 4093
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9891147592514674,
      "learning_rate": 2.742496616445335e-06,
      "loss": 0.6176,
      "step": 4094
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0284624614492337,
      "learning_rate": 2.740168338477587e-06,
      "loss": 0.5448,
      "step": 4095
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0120196330006106,
      "learning_rate": 2.7378406761502747e-06,
      "loss": 0.55,
      "step": 4096
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0207633426175418,
      "learning_rate": 2.7355136300975214e-06,
      "loss": 0.596,
      "step": 4097
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0036759728682145,
      "learning_rate": 2.733187200953276e-06,
      "loss": 0.6742,
      "step": 4098
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9684029672783085,
      "learning_rate": 2.7308613893513205e-06,
      "loss": 0.6145,
      "step": 4099
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6860798733077779,
      "learning_rate": 2.7285361959252716e-06,
      "loss": 0.4903,
      "step": 4100
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0249399111039097,
      "learning_rate": 2.72621162130858e-06,
      "loss": 0.5744,
      "step": 4101
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9675987015377583,
      "learning_rate": 2.723887666134516e-06,
      "loss": 0.5363,
      "step": 4102
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9601710511198105,
      "learning_rate": 2.721564331036194e-06,
      "loss": 0.4931,
      "step": 4103
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0002350678264316,
      "learning_rate": 2.719241616646555e-06,
      "loss": 0.5632,
      "step": 4104
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0268711646908977,
      "learning_rate": 2.716919523598369e-06,
      "loss": 0.519,
      "step": 4105
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0328383060684727,
      "learning_rate": 2.7145980525242367e-06,
      "loss": 0.5732,
      "step": 4106
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8830074006332264,
      "learning_rate": 2.712277204056594e-06,
      "loss": 0.5377,
      "step": 4107
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9633476677697133,
      "learning_rate": 2.7099569788277023e-06,
      "loss": 0.581,
      "step": 4108
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.651712147164041,
      "learning_rate": 2.7076373774696527e-06,
      "loss": 0.4652,
      "step": 4109
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1104690933267418,
      "learning_rate": 2.705318400614374e-06,
      "loss": 0.6867,
      "step": 4110
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.062440968540112,
      "learning_rate": 2.703000048893613e-06,
      "loss": 0.5979,
      "step": 4111
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9005534877991328,
      "learning_rate": 2.7006823229389588e-06,
      "loss": 0.5076,
      "step": 4112
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0484482210944337,
      "learning_rate": 2.6983652233818205e-06,
      "loss": 0.6101,
      "step": 4113
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.038326363492528,
      "learning_rate": 2.6960487508534383e-06,
      "loss": 0.6463,
      "step": 4114
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8697546720999839,
      "learning_rate": 2.693732905984885e-06,
      "loss": 0.5255,
      "step": 4115
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0339550298516553,
      "learning_rate": 2.6914176894070644e-06,
      "loss": 0.5927,
      "step": 4116
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0518696530453056,
      "learning_rate": 2.6891031017506986e-06,
      "loss": 0.6093,
      "step": 4117
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9746747746976071,
      "learning_rate": 2.686789143646347e-06,
      "loss": 0.5765,
      "step": 4118
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9545353133327568,
      "learning_rate": 2.6844758157244e-06,
      "loss": 0.4967,
      "step": 4119
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0112389334627743,
      "learning_rate": 2.6821631186150676e-06,
      "loss": 0.5435,
      "step": 4120
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9078802657303984,
      "learning_rate": 2.6798510529483913e-06,
      "loss": 0.5022,
      "step": 4121
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.019573944843636,
      "learning_rate": 2.6775396193542436e-06,
      "loss": 0.6076,
      "step": 4122
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0762076879137368,
      "learning_rate": 2.675228818462327e-06,
      "loss": 0.5781,
      "step": 4123
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9626516893990852,
      "learning_rate": 2.6729186509021587e-06,
      "loss": 0.6379,
      "step": 4124
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0440073208478957,
      "learning_rate": 2.670609117303096e-06,
      "loss": 0.6741,
      "step": 4125
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9351192821926619,
      "learning_rate": 2.668300218294322e-06,
      "loss": 0.5034,
      "step": 4126
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9997945000517795,
      "learning_rate": 2.6659919545048424e-06,
      "loss": 0.6288,
      "step": 4127
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0364822621832968,
      "learning_rate": 2.663684326563489e-06,
      "loss": 0.5917,
      "step": 4128
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0171578350152857,
      "learning_rate": 2.6613773350989293e-06,
      "loss": 0.6218,
      "step": 4129
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0506184309431916,
      "learning_rate": 2.6590709807396464e-06,
      "loss": 0.571,
      "step": 4130
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9731584372164224,
      "learning_rate": 2.656765264113955e-06,
      "loss": 0.5707,
      "step": 4131
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0349383518243065,
      "learning_rate": 2.65446018585e-06,
      "loss": 0.5882,
      "step": 4132
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9412264459653292,
      "learning_rate": 2.652155746575743e-06,
      "loss": 0.5388,
      "step": 4133
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0045372851596146,
      "learning_rate": 2.6498519469189814e-06,
      "loss": 0.5531,
      "step": 4134
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1290671236547234,
      "learning_rate": 2.6475487875073318e-06,
      "loss": 0.6585,
      "step": 4135
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0472501482244814,
      "learning_rate": 2.6452462689682364e-06,
      "loss": 0.5867,
      "step": 4136
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9992871552936109,
      "learning_rate": 2.6429443919289676e-06,
      "loss": 0.5509,
      "step": 4137
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9673557712221101,
      "learning_rate": 2.640643157016618e-06,
      "loss": 0.5319,
      "step": 4138
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0872471036352866,
      "learning_rate": 2.63834256485811e-06,
      "loss": 0.6222,
      "step": 4139
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9953755076305282,
      "learning_rate": 2.636042616080185e-06,
      "loss": 0.5661,
      "step": 4140
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0337555029160617,
      "learning_rate": 2.6337433113094164e-06,
      "loss": 0.5521,
      "step": 4141
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9571091209345833,
      "learning_rate": 2.6314446511721957e-06,
      "loss": 0.5773,
      "step": 4142
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9922458353879587,
      "learning_rate": 2.6291466362947392e-06,
      "loss": 0.4751,
      "step": 4143
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.005154945708394,
      "learning_rate": 2.626849267303093e-06,
      "loss": 0.6096,
      "step": 4144
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9603122056870393,
      "learning_rate": 2.624552544823126e-06,
      "loss": 0.5546,
      "step": 4145
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.998811112173916,
      "learning_rate": 2.622256469480522e-06,
      "loss": 0.6307,
      "step": 4146
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9617728011031288,
      "learning_rate": 2.6199610419007983e-06,
      "loss": 0.591,
      "step": 4147
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0381645044997831,
      "learning_rate": 2.617666262709294e-06,
      "loss": 0.5982,
      "step": 4148
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9839611724703446,
      "learning_rate": 2.61537213253117e-06,
      "loss": 0.5437,
      "step": 4149
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.11930758711911,
      "learning_rate": 2.6130786519914074e-06,
      "loss": 0.6655,
      "step": 4150
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.025176827269826,
      "learning_rate": 2.6107858217148185e-06,
      "loss": 0.5655,
      "step": 4151
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9888260855867205,
      "learning_rate": 2.6084936423260303e-06,
      "loss": 0.5546,
      "step": 4152
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9496668109439113,
      "learning_rate": 2.606202114449495e-06,
      "loss": 0.5856,
      "step": 4153
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.958141652196475,
      "learning_rate": 2.60391123870949e-06,
      "loss": 0.5585,
      "step": 4154
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0491309200885355,
      "learning_rate": 2.601621015730111e-06,
      "loss": 0.6014,
      "step": 4155
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9512044178850638,
      "learning_rate": 2.5993314461352802e-06,
      "loss": 0.5381,
      "step": 4156
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0685864043490851,
      "learning_rate": 2.597042530548737e-06,
      "loss": 0.608,
      "step": 4157
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9045842016294855,
      "learning_rate": 2.5947542695940485e-06,
      "loss": 0.4794,
      "step": 4158
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0339508753645148,
      "learning_rate": 2.5924666638945973e-06,
      "loss": 0.6068,
      "step": 4159
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9335131117724038,
      "learning_rate": 2.5901797140735895e-06,
      "loss": 0.5294,
      "step": 4160
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0336950839868349,
      "learning_rate": 2.5878934207540564e-06,
      "loss": 0.552,
      "step": 4161
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9638734238060196,
      "learning_rate": 2.5856077845588433e-06,
      "loss": 0.6377,
      "step": 4162
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9913705357017132,
      "learning_rate": 2.5833228061106253e-06,
      "loss": 0.605,
      "step": 4163
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0559041525037611,
      "learning_rate": 2.5810384860318904e-06,
      "loss": 0.532,
      "step": 4164
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9787071825297399,
      "learning_rate": 2.57875482494495e-06,
      "loss": 0.575,
      "step": 4165
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.065966937096762,
      "learning_rate": 2.576471823471938e-06,
      "loss": 0.54,
      "step": 4166
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9863645580418978,
      "learning_rate": 2.5741894822348073e-06,
      "loss": 0.5754,
      "step": 4167
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9871672428386502,
      "learning_rate": 2.571907801855327e-06,
      "loss": 0.6072,
      "step": 4168
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0015140921790995,
      "learning_rate": 2.5696267829550926e-06,
      "loss": 0.4829,
      "step": 4169
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9119101577402052,
      "learning_rate": 2.5673464261555174e-06,
      "loss": 0.5422,
      "step": 4170
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0047353643708197,
      "learning_rate": 2.5650667320778333e-06,
      "loss": 0.5304,
      "step": 4171
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1171378668114342,
      "learning_rate": 2.562787701343088e-06,
      "loss": 0.6919,
      "step": 4172
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9707952096669856,
      "learning_rate": 2.5605093345721564e-06,
      "loss": 0.6045,
      "step": 4173
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0235726104671539,
      "learning_rate": 2.558231632385728e-06,
      "loss": 0.5485,
      "step": 4174
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9825484094276393,
      "learning_rate": 2.5559545954043086e-06,
      "loss": 0.5019,
      "step": 4175
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9468994100485664,
      "learning_rate": 2.5536782242482294e-06,
      "loss": 0.4679,
      "step": 4176
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0674821700239088,
      "learning_rate": 2.5514025195376336e-06,
      "loss": 0.5489,
      "step": 4177
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9846847122935792,
      "learning_rate": 2.5491274818924893e-06,
      "loss": 0.6565,
      "step": 4178
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2642556262049445,
      "learning_rate": 2.5468531119325756e-06,
      "loss": 0.5754,
      "step": 4179
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9995403192376702,
      "learning_rate": 2.544579410277498e-06,
      "loss": 0.5443,
      "step": 4180
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6405436646215436,
      "learning_rate": 2.542306377546673e-06,
      "loss": 0.5105,
      "step": 4181
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0270509837663513,
      "learning_rate": 2.540034014359335e-06,
      "loss": 0.5144,
      "step": 4182
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.950759222975867,
      "learning_rate": 2.537762321334543e-06,
      "loss": 0.5047,
      "step": 4183
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6680771031194185,
      "learning_rate": 2.535491299091165e-06,
      "loss": 0.4919,
      "step": 4184
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9932084463512395,
      "learning_rate": 2.533220948247893e-06,
      "loss": 0.5669,
      "step": 4185
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0558010412566432,
      "learning_rate": 2.530951269423233e-06,
      "loss": 0.5509,
      "step": 4186
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0070353108861123,
      "learning_rate": 2.528682263235504e-06,
      "loss": 0.6196,
      "step": 4187
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0515677106285464,
      "learning_rate": 2.526413930302851e-06,
      "loss": 0.6381,
      "step": 4188
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.109079647613207,
      "learning_rate": 2.5241462712432284e-06,
      "loss": 0.5929,
      "step": 4189
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.06401687367605,
      "learning_rate": 2.5218792866744068e-06,
      "loss": 0.5805,
      "step": 4190
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0427495065370334,
      "learning_rate": 2.5196129772139766e-06,
      "loss": 0.6248,
      "step": 4191
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9698619245859882,
      "learning_rate": 2.517347343479346e-06,
      "loss": 0.5889,
      "step": 4192
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0173573977091384,
      "learning_rate": 2.5150823860877345e-06,
      "loss": 0.5907,
      "step": 4193
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0205188937172416,
      "learning_rate": 2.5128181056561755e-06,
      "loss": 0.6145,
      "step": 4194
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0859000590577317,
      "learning_rate": 2.5105545028015266e-06,
      "loss": 0.6684,
      "step": 4195
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.067539270807866,
      "learning_rate": 2.508291578140453e-06,
      "loss": 0.5944,
      "step": 4196
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9812208485344739,
      "learning_rate": 2.5060293322894353e-06,
      "loss": 0.5039,
      "step": 4197
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0730453026024784,
      "learning_rate": 2.503767765864774e-06,
      "loss": 0.6313,
      "step": 4198
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.055682591029642,
      "learning_rate": 2.5015068794825853e-06,
      "loss": 0.6127,
      "step": 4199
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.017364367452376,
      "learning_rate": 2.4992466737587932e-06,
      "loss": 0.5112,
      "step": 4200
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9725027893805415,
      "learning_rate": 2.4969871493091398e-06,
      "loss": 0.5759,
      "step": 4201
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1145453529875105,
      "learning_rate": 2.494728306749184e-06,
      "loss": 0.5794,
      "step": 4202
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1103276577931425,
      "learning_rate": 2.4924701466942957e-06,
      "loss": 0.6807,
      "step": 4203
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9818666726430761,
      "learning_rate": 2.490212669759658e-06,
      "loss": 0.5569,
      "step": 4204
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.044396937141999,
      "learning_rate": 2.487955876560274e-06,
      "loss": 0.5911,
      "step": 4205
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0532239681032975,
      "learning_rate": 2.485699767710952e-06,
      "loss": 0.6668,
      "step": 4206
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0466868715744622,
      "learning_rate": 2.4834443438263224e-06,
      "loss": 0.5529,
      "step": 4207
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1177457339927725,
      "learning_rate": 2.4811896055208227e-06,
      "loss": 0.5821,
      "step": 4208
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9621071520259866,
      "learning_rate": 2.4789355534087038e-06,
      "loss": 0.603,
      "step": 4209
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.976281559879944,
      "learning_rate": 2.476682188104036e-06,
      "loss": 0.5211,
      "step": 4210
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9640236686646355,
      "learning_rate": 2.4744295102206954e-06,
      "loss": 0.5287,
      "step": 4211
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0616524712151376,
      "learning_rate": 2.4721775203723726e-06,
      "loss": 0.5912,
      "step": 4212
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9782968114215164,
      "learning_rate": 2.4699262191725726e-06,
      "loss": 0.5875,
      "step": 4213
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6527304073611394,
      "learning_rate": 2.467675607234615e-06,
      "loss": 0.4926,
      "step": 4214
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9931967102263395,
      "learning_rate": 2.465425685171625e-06,
      "loss": 0.5837,
      "step": 4215
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8878886028237913,
      "learning_rate": 2.463176453596543e-06,
      "loss": 0.618,
      "step": 4216
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.095364786205635,
      "learning_rate": 2.4609279131221243e-06,
      "loss": 0.6461,
      "step": 4217
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0093553868229896,
      "learning_rate": 2.4586800643609326e-06,
      "loss": 0.6479,
      "step": 4218
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9645670020073929,
      "learning_rate": 2.456432907925341e-06,
      "loss": 0.6111,
      "step": 4219
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.861160689218319,
      "learning_rate": 2.454186444427539e-06,
      "loss": 0.4385,
      "step": 4220
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9280475433651341,
      "learning_rate": 2.4519406744795275e-06,
      "loss": 0.6049,
      "step": 4221
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0764823811922422,
      "learning_rate": 2.4496955986931143e-06,
      "loss": 0.5823,
      "step": 4222
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6800656064268441,
      "learning_rate": 2.447451217679917e-06,
      "loss": 0.4759,
      "step": 4223
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6103010032403269,
      "learning_rate": 2.445207532051372e-06,
      "loss": 0.4488,
      "step": 4224
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0608226400989327,
      "learning_rate": 2.442964542418718e-06,
      "loss": 0.604,
      "step": 4225
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.968264105863904,
      "learning_rate": 2.4407222493930063e-06,
      "loss": 0.6438,
      "step": 4226
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.124588890962765,
      "learning_rate": 2.438480653585102e-06,
      "loss": 0.6845,
      "step": 4227
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.043963024970984,
      "learning_rate": 2.436239755605675e-06,
      "loss": 0.6434,
      "step": 4228
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0137092436854946,
      "learning_rate": 2.433999556065211e-06,
      "loss": 0.506,
      "step": 4229
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9688744037280497,
      "learning_rate": 2.4317600555739997e-06,
      "loss": 0.5737,
      "step": 4230
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9374011263255312,
      "learning_rate": 2.4295212547421415e-06,
      "loss": 0.5451,
      "step": 4231
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0755521488943034,
      "learning_rate": 2.427283154179551e-06,
      "loss": 0.5794,
      "step": 4232
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9039070093462079,
      "learning_rate": 2.4250457544959465e-06,
      "loss": 0.5334,
      "step": 4233
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0078130332867878,
      "learning_rate": 2.422809056300856e-06,
      "loss": 0.5164,
      "step": 4234
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9737893491190986,
      "learning_rate": 2.4205730602036194e-06,
      "loss": 0.4805,
      "step": 4235
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.045161043056691,
      "learning_rate": 2.4183377668133843e-06,
      "loss": 0.6696,
      "step": 4236
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0509801752114005,
      "learning_rate": 2.4161031767391067e-06,
      "loss": 0.5917,
      "step": 4237
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0751212508222223,
      "learning_rate": 2.4138692905895473e-06,
      "loss": 0.5391,
      "step": 4238
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1186338260123807,
      "learning_rate": 2.4116361089732815e-06,
      "loss": 0.6193,
      "step": 4239
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0955150998655903,
      "learning_rate": 2.4094036324986888e-06,
      "loss": 0.583,
      "step": 4240
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9768281747867427,
      "learning_rate": 2.4071718617739542e-06,
      "loss": 0.5097,
      "step": 4241
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0416028918638447,
      "learning_rate": 2.404940797407077e-06,
      "loss": 0.5973,
      "step": 4242
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.071215029253211,
      "learning_rate": 2.4027104400058615e-06,
      "loss": 0.6126,
      "step": 4243
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0016019832517598,
      "learning_rate": 2.4004807901779164e-06,
      "loss": 0.6109,
      "step": 4244
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0528640004682408,
      "learning_rate": 2.3982518485306587e-06,
      "loss": 0.5732,
      "step": 4245
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.10580606238819,
      "learning_rate": 2.3960236156713175e-06,
      "loss": 0.6594,
      "step": 4246
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0321446845105122,
      "learning_rate": 2.3937960922069213e-06,
      "loss": 0.5996,
      "step": 4247
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0518940432588118,
      "learning_rate": 2.391569278744309e-06,
      "loss": 0.5591,
      "step": 4248
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0105563382628076,
      "learning_rate": 2.3893431758901285e-06,
      "loss": 0.5654,
      "step": 4249
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0848417325617543,
      "learning_rate": 2.387117784250828e-06,
      "loss": 0.5923,
      "step": 4250
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0369607119868975,
      "learning_rate": 2.38489310443267e-06,
      "loss": 0.6218,
      "step": 4251
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0129207689590631,
      "learning_rate": 2.382669137041716e-06,
      "loss": 0.6521,
      "step": 4252
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0740799521333386,
      "learning_rate": 2.3804458826838337e-06,
      "loss": 0.572,
      "step": 4253
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0385039706048727,
      "learning_rate": 2.3782233419647043e-06,
      "loss": 0.6048,
      "step": 4254
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9763869625370863,
      "learning_rate": 2.376001515489806e-06,
      "loss": 0.5225,
      "step": 4255
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0787852225201433,
      "learning_rate": 2.3737804038644236e-06,
      "loss": 0.5765,
      "step": 4256
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6449513645642397,
      "learning_rate": 2.371560007693651e-06,
      "loss": 0.487,
      "step": 4257
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0375524730435397,
      "learning_rate": 2.3693403275823883e-06,
      "loss": 0.5515,
      "step": 4258
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.080629989742259,
      "learning_rate": 2.367121364135335e-06,
      "loss": 0.6488,
      "step": 4259
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0748228522831667,
      "learning_rate": 2.3649031179569954e-06,
      "loss": 0.6009,
      "step": 4260
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.009638178228135,
      "learning_rate": 2.3626855896516855e-06,
      "loss": 0.5926,
      "step": 4261
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0532746470902474,
      "learning_rate": 2.360468779823519e-06,
      "loss": 0.5781,
      "step": 4262
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0876090007156665,
      "learning_rate": 2.3582526890764135e-06,
      "loss": 0.6065,
      "step": 4263
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9688587447321712,
      "learning_rate": 2.356037318014096e-06,
      "loss": 0.5459,
      "step": 4264
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9871704478051079,
      "learning_rate": 2.3538226672400982e-06,
      "loss": 0.5321,
      "step": 4265
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.925710277647997,
      "learning_rate": 2.3516087373577447e-06,
      "loss": 0.5454,
      "step": 4266
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0066311533423127,
      "learning_rate": 2.3493955289701744e-06,
      "loss": 0.503,
      "step": 4267
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9836004265667032,
      "learning_rate": 2.347183042680328e-06,
      "loss": 0.4816,
      "step": 4268
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0465209394909063,
      "learning_rate": 2.3449712790909465e-06,
      "loss": 0.5548,
      "step": 4269
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0514379669767695,
      "learning_rate": 2.3427602388045723e-06,
      "loss": 0.6567,
      "step": 4270
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9945471342843365,
      "learning_rate": 2.3405499224235583e-06,
      "loss": 0.57,
      "step": 4271
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6141325194890646,
      "learning_rate": 2.3383403305500523e-06,
      "loss": 0.4448,
      "step": 4272
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0514987457407428,
      "learning_rate": 2.3361314637860104e-06,
      "loss": 0.6204,
      "step": 4273
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0134794630826691,
      "learning_rate": 2.3339233227331887e-06,
      "loss": 0.6331,
      "step": 4274
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9600739970389689,
      "learning_rate": 2.331715907993142e-06,
      "loss": 0.541,
      "step": 4275
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0062314876880716,
      "learning_rate": 2.329509220167236e-06,
      "loss": 0.5615,
      "step": 4276
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9285020977391022,
      "learning_rate": 2.327303259856629e-06,
      "loss": 0.4819,
      "step": 4277
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9850668225425437,
      "learning_rate": 2.3250980276622893e-06,
      "loss": 0.5898,
      "step": 4278
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0288642390567753,
      "learning_rate": 2.322893524184979e-06,
      "loss": 0.5794,
      "step": 4279
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9388705756172252,
      "learning_rate": 2.320689750025269e-06,
      "loss": 0.4809,
      "step": 4280
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0188909249248297,
      "learning_rate": 2.3184867057835274e-06,
      "loss": 0.5767,
      "step": 4281
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9773519638522371,
      "learning_rate": 2.316284392059921e-06,
      "loss": 0.5615,
      "step": 4282
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0141484454483456,
      "learning_rate": 2.3140828094544253e-06,
      "loss": 0.6172,
      "step": 4283
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1581460539414263,
      "learning_rate": 2.3118819585668106e-06,
      "loss": 0.5405,
      "step": 4284
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9408137050777134,
      "learning_rate": 2.3096818399966465e-06,
      "loss": 0.5941,
      "step": 4285
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.029944524956399,
      "learning_rate": 2.3074824543433084e-06,
      "loss": 0.5928,
      "step": 4286
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1642417762581787,
      "learning_rate": 2.305283802205973e-06,
      "loss": 0.5989,
      "step": 4287
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0751821084317132,
      "learning_rate": 2.3030858841836063e-06,
      "loss": 0.6036,
      "step": 4288
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9729737236314856,
      "learning_rate": 2.300888700874986e-06,
      "loss": 0.5738,
      "step": 4289
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9590697990400221,
      "learning_rate": 2.2986922528786865e-06,
      "loss": 0.5249,
      "step": 4290
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.987308828859753,
      "learning_rate": 2.2964965407930796e-06,
      "loss": 0.6709,
      "step": 4291
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0200935595026772,
      "learning_rate": 2.2943015652163355e-06,
      "loss": 0.5552,
      "step": 4292
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.071280413417095,
      "learning_rate": 2.29210732674643e-06,
      "loss": 0.6347,
      "step": 4293
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1153660272231845,
      "learning_rate": 2.289913825981132e-06,
      "loss": 0.5771,
      "step": 4294
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0132108436006937,
      "learning_rate": 2.2877210635180098e-06,
      "loss": 0.5826,
      "step": 4295
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1056380368813161,
      "learning_rate": 2.2855290399544346e-06,
      "loss": 0.631,
      "step": 4296
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.021821561320138,
      "learning_rate": 2.2833377558875754e-06,
      "loss": 0.6237,
      "step": 4297
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9785955829227968,
      "learning_rate": 2.281147211914397e-06,
      "loss": 0.5759,
      "step": 4298
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9965246853753821,
      "learning_rate": 2.278957408631662e-06,
      "loss": 0.5739,
      "step": 4299
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9673545572633372,
      "learning_rate": 2.276768346635936e-06,
      "loss": 0.5616,
      "step": 4300
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9844179247709128,
      "learning_rate": 2.2745800265235773e-06,
      "loss": 0.5781,
      "step": 4301
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0278060625703047,
      "learning_rate": 2.2723924488907477e-06,
      "loss": 0.5597,
      "step": 4302
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.095728777239182,
      "learning_rate": 2.2702056143334025e-06,
      "loss": 0.6157,
      "step": 4303
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0684900911957962,
      "learning_rate": 2.268019523447292e-06,
      "loss": 0.6255,
      "step": 4304
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.057841643317205,
      "learning_rate": 2.265834176827974e-06,
      "loss": 0.5698,
      "step": 4305
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0313746198145808,
      "learning_rate": 2.263649575070794e-06,
      "loss": 0.5369,
      "step": 4306
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9958966049203255,
      "learning_rate": 2.261465718770895e-06,
      "loss": 0.5767,
      "step": 4307
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0000605341500162,
      "learning_rate": 2.259282608523223e-06,
      "loss": 0.5821,
      "step": 4308
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9888116323403905,
      "learning_rate": 2.25710024492252e-06,
      "loss": 0.5996,
      "step": 4309
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0104775279837934,
      "learning_rate": 2.254918628563315e-06,
      "loss": 0.581,
      "step": 4310
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6139474355332751,
      "learning_rate": 2.2527377600399446e-06,
      "loss": 0.4727,
      "step": 4311
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0181499617222898,
      "learning_rate": 2.2505576399465384e-06,
      "loss": 0.6151,
      "step": 4312
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1593228908256783,
      "learning_rate": 2.2483782688770208e-06,
      "loss": 0.6212,
      "step": 4313
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9952250140501226,
      "learning_rate": 2.246199647425109e-06,
      "loss": 0.5851,
      "step": 4314
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9699933128754111,
      "learning_rate": 2.2440217761843244e-06,
      "loss": 0.5349,
      "step": 4315
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0204612824565595,
      "learning_rate": 2.241844655747977e-06,
      "loss": 0.561,
      "step": 4316
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0372634162655423,
      "learning_rate": 2.2396682867091723e-06,
      "loss": 0.6273,
      "step": 4317
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9423573528464719,
      "learning_rate": 2.237492669660815e-06,
      "loss": 0.5285,
      "step": 4318
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.957634302418033,
      "learning_rate": 2.235317805195606e-06,
      "loss": 0.4913,
      "step": 4319
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0166342127527939,
      "learning_rate": 2.2331436939060354e-06,
      "loss": 0.522,
      "step": 4320
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1101188505160835,
      "learning_rate": 2.2309703363843893e-06,
      "loss": 0.6177,
      "step": 4321
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9244857061004736,
      "learning_rate": 2.2287977332227543e-06,
      "loss": 0.6043,
      "step": 4322
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.045407396206926,
      "learning_rate": 2.2266258850130055e-06,
      "loss": 0.5881,
      "step": 4323
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1283888141294376,
      "learning_rate": 2.224454792346813e-06,
      "loss": 0.597,
      "step": 4324
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0062355093952606,
      "learning_rate": 2.2222844558156444e-06,
      "loss": 0.6127,
      "step": 4325
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0296407087292927,
      "learning_rate": 2.2201148760107568e-06,
      "loss": 0.59,
      "step": 4326
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9510830070004738,
      "learning_rate": 2.217946053523207e-06,
      "loss": 0.4927,
      "step": 4327
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1201027069792013,
      "learning_rate": 2.2157779889438393e-06,
      "loss": 0.6687,
      "step": 4328
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9754356050119517,
      "learning_rate": 2.213610682863293e-06,
      "loss": 0.548,
      "step": 4329
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.898160362247851,
      "learning_rate": 2.2114441358720046e-06,
      "loss": 0.5496,
      "step": 4330
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6505670553356139,
      "learning_rate": 2.209278348560204e-06,
      "loss": 0.4776,
      "step": 4331
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0686122207263347,
      "learning_rate": 2.2071133215179043e-06,
      "loss": 0.5962,
      "step": 4332
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.024448749725595,
      "learning_rate": 2.204949055334922e-06,
      "loss": 0.606,
      "step": 4333
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0705894874019695,
      "learning_rate": 2.202785550600865e-06,
      "loss": 0.617,
      "step": 4334
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1051852097843025,
      "learning_rate": 2.20062280790513e-06,
      "loss": 0.6864,
      "step": 4335
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8997067110572752,
      "learning_rate": 2.198460827836905e-06,
      "loss": 0.588,
      "step": 4336
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0083473424166143,
      "learning_rate": 2.1962996109851757e-06,
      "loss": 0.5863,
      "step": 4337
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9787921633341143,
      "learning_rate": 2.1941391579387204e-06,
      "loss": 0.561,
      "step": 4338
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1118588866905192,
      "learning_rate": 2.1919794692860992e-06,
      "loss": 0.6166,
      "step": 4339
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0413830326226816,
      "learning_rate": 2.189820545615674e-06,
      "loss": 0.5707,
      "step": 4340
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.998681071817628,
      "learning_rate": 2.187662387515597e-06,
      "loss": 0.5303,
      "step": 4341
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.940447028568086,
      "learning_rate": 2.1855049955738077e-06,
      "loss": 0.5048,
      "step": 4342
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9218426914007682,
      "learning_rate": 2.1833483703780383e-06,
      "loss": 0.5227,
      "step": 4343
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9716910169013515,
      "learning_rate": 2.1811925125158154e-06,
      "loss": 0.5136,
      "step": 4344
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9314152784886676,
      "learning_rate": 2.1790374225744537e-06,
      "loss": 0.5814,
      "step": 4345
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.98972658540485,
      "learning_rate": 2.1768831011410567e-06,
      "loss": 0.525,
      "step": 4346
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9605991279313632,
      "learning_rate": 2.174729548802524e-06,
      "loss": 0.5193,
      "step": 4347
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.078976858560916,
      "learning_rate": 2.1725767661455393e-06,
      "loss": 0.595,
      "step": 4348
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6176878710571864,
      "learning_rate": 2.170424753756584e-06,
      "loss": 0.4309,
      "step": 4349
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.979086988625556,
      "learning_rate": 2.1682735122219236e-06,
      "loss": 0.5947,
      "step": 4350
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9957832274404707,
      "learning_rate": 2.1661230421276134e-06,
      "loss": 0.565,
      "step": 4351
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9467792076769326,
      "learning_rate": 2.163973344059503e-06,
      "loss": 0.5047,
      "step": 4352
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1187075589984037,
      "learning_rate": 2.1618244186032328e-06,
      "loss": 0.6377,
      "step": 4353
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1229383198837464,
      "learning_rate": 2.159676266344222e-06,
      "loss": 0.63,
      "step": 4354
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0570499616688716,
      "learning_rate": 2.15752888786769e-06,
      "loss": 0.5871,
      "step": 4355
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0281149312130586,
      "learning_rate": 2.155382283758644e-06,
      "loss": 0.5501,
      "step": 4356
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7026934038065543,
      "learning_rate": 2.153236454601877e-06,
      "loss": 0.45,
      "step": 4357
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9983436858585417,
      "learning_rate": 2.1510914009819684e-06,
      "loss": 0.5696,
      "step": 4358
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0748328530273754,
      "learning_rate": 2.148947123483293e-06,
      "loss": 0.5201,
      "step": 4359
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9379861049735776,
      "learning_rate": 2.146803622690015e-06,
      "loss": 0.5862,
      "step": 4360
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9897210265157065,
      "learning_rate": 2.144660899186075e-06,
      "loss": 0.5239,
      "step": 4361
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.118871363739599,
      "learning_rate": 2.1425189535552143e-06,
      "loss": 0.5699,
      "step": 4362
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9390580780275393,
      "learning_rate": 2.1403777863809594e-06,
      "loss": 0.6308,
      "step": 4363
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9429519936250398,
      "learning_rate": 2.1382373982466213e-06,
      "loss": 0.5525,
      "step": 4364
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.103850596742432,
      "learning_rate": 2.1360977897352992e-06,
      "loss": 0.685,
      "step": 4365
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9594066969595867,
      "learning_rate": 2.1339589614298844e-06,
      "loss": 0.5418,
      "step": 4366
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9897303244456891,
      "learning_rate": 2.131820913913052e-06,
      "loss": 0.6349,
      "step": 4367
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0708054786150407,
      "learning_rate": 2.129683647767262e-06,
      "loss": 0.6332,
      "step": 4368
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0103269301851554,
      "learning_rate": 2.1275471635747695e-06,
      "loss": 0.5688,
      "step": 4369
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9747713866377347,
      "learning_rate": 2.125411461917607e-06,
      "loss": 0.5695,
      "step": 4370
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0066227660160711,
      "learning_rate": 2.1232765433776033e-06,
      "loss": 0.5459,
      "step": 4371
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0510292753178838,
      "learning_rate": 2.121142408536366e-06,
      "loss": 0.5927,
      "step": 4372
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9575320067407117,
      "learning_rate": 2.1190090579752908e-06,
      "loss": 0.608,
      "step": 4373
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9561336488417097,
      "learning_rate": 2.116876492275565e-06,
      "loss": 0.5464,
      "step": 4374
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1413714150507352,
      "learning_rate": 2.114744712018155e-06,
      "loss": 0.6146,
      "step": 4375
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0571317172190016,
      "learning_rate": 2.1126137177838197e-06,
      "loss": 0.6558,
      "step": 4376
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0514399351145252,
      "learning_rate": 2.110483510153097e-06,
      "loss": 0.5739,
      "step": 4377
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0652555507786652,
      "learning_rate": 2.108354089706318e-06,
      "loss": 0.5817,
      "step": 4378
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.074606568415449,
      "learning_rate": 2.1062254570235934e-06,
      "loss": 0.6285,
      "step": 4379
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9495002344301338,
      "learning_rate": 2.1040976126848208e-06,
      "loss": 0.5789,
      "step": 4380
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9365276027535807,
      "learning_rate": 2.1019705572696836e-06,
      "loss": 0.5452,
      "step": 4381
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1532710146626786,
      "learning_rate": 2.099844291357655e-06,
      "loss": 0.6162,
      "step": 4382
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9566003170692688,
      "learning_rate": 2.0977188155279815e-06,
      "loss": 0.612,
      "step": 4383
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.064451984457309,
      "learning_rate": 2.095594130359704e-06,
      "loss": 0.52,
      "step": 4384
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9721030269934459,
      "learning_rate": 2.093470236431648e-06,
      "loss": 0.5816,
      "step": 4385
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0263124023787649,
      "learning_rate": 2.0913471343224183e-06,
      "loss": 0.6105,
      "step": 4386
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0360248695281542,
      "learning_rate": 2.0892248246104047e-06,
      "loss": 0.5846,
      "step": 4387
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9599371113155775,
      "learning_rate": 2.0871033078737875e-06,
      "loss": 0.5488,
      "step": 4388
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0629735296058986,
      "learning_rate": 2.0849825846905223e-06,
      "loss": 0.5669,
      "step": 4389
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.037818797492482,
      "learning_rate": 2.0828626556383534e-06,
      "loss": 0.551,
      "step": 4390
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.001975935238851,
      "learning_rate": 2.0807435212948094e-06,
      "loss": 0.5208,
      "step": 4391
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0102894404564389,
      "learning_rate": 2.078625182237199e-06,
      "loss": 0.5894,
      "step": 4392
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9488442323108185,
      "learning_rate": 2.0765076390426187e-06,
      "loss": 0.5129,
      "step": 4393
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.013179202559768,
      "learning_rate": 2.074390892287944e-06,
      "loss": 0.5734,
      "step": 4394
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0036265240563709,
      "learning_rate": 2.0722749425498332e-06,
      "loss": 0.599,
      "step": 4395
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0307054782037015,
      "learning_rate": 2.0701597904047332e-06,
      "loss": 0.6025,
      "step": 4396
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0780990306793368,
      "learning_rate": 2.068045436428866e-06,
      "loss": 0.6645,
      "step": 4397
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.031579837261566,
      "learning_rate": 2.0659318811982433e-06,
      "loss": 0.5413,
      "step": 4398
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9942503653210693,
      "learning_rate": 2.0638191252886526e-06,
      "loss": 0.6074,
      "step": 4399
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.973968583670157,
      "learning_rate": 2.06170716927567e-06,
      "loss": 0.5242,
      "step": 4400
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.040580949940331,
      "learning_rate": 2.0595960137346494e-06,
      "loss": 0.5865,
      "step": 4401
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9637768632991652,
      "learning_rate": 2.0574856592407255e-06,
      "loss": 0.5554,
      "step": 4402
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9941152391196982,
      "learning_rate": 2.0553761063688204e-06,
      "loss": 0.5806,
      "step": 4403
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.048443298701311,
      "learning_rate": 2.0532673556936333e-06,
      "loss": 0.6085,
      "step": 4404
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1441486685205917,
      "learning_rate": 2.051159407789644e-06,
      "loss": 0.6071,
      "step": 4405
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0288277200021714,
      "learning_rate": 2.0490522632311173e-06,
      "loss": 0.5338,
      "step": 4406
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0605927369688255,
      "learning_rate": 2.0469459225920987e-06,
      "loss": 0.5226,
      "step": 4407
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.063853287339275,
      "learning_rate": 2.0448403864464123e-06,
      "loss": 0.6315,
      "step": 4408
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9568009469530431,
      "learning_rate": 2.0427356553676625e-06,
      "loss": 0.5455,
      "step": 4409
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9576294182197103,
      "learning_rate": 2.0406317299292394e-06,
      "loss": 0.5512,
      "step": 4410
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1457119517243424,
      "learning_rate": 2.0385286107043086e-06,
      "loss": 0.6531,
      "step": 4411
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9673911766082235,
      "learning_rate": 2.0364262982658155e-06,
      "loss": 0.5254,
      "step": 4412
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1156946921945767,
      "learning_rate": 2.0343247931864927e-06,
      "loss": 0.6367,
      "step": 4413
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.023979847371299,
      "learning_rate": 2.0322240960388436e-06,
      "loss": 0.6001,
      "step": 4414
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1182797265223214,
      "learning_rate": 2.030124207395159e-06,
      "loss": 0.612,
      "step": 4415
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0626283932530691,
      "learning_rate": 2.0280251278275043e-06,
      "loss": 0.6504,
      "step": 4416
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9977280018997456,
      "learning_rate": 2.0259268579077295e-06,
      "loss": 0.5169,
      "step": 4417
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9332323780212993,
      "learning_rate": 2.0238293982074593e-06,
      "loss": 0.5012,
      "step": 4418
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0325264759062336,
      "learning_rate": 2.0217327492980977e-06,
      "loss": 0.6486,
      "step": 4419
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.040979137140815,
      "learning_rate": 2.0196369117508336e-06,
      "loss": 0.5435,
      "step": 4420
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9525682949156852,
      "learning_rate": 2.017541886136627e-06,
      "loss": 0.5725,
      "step": 4421
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9340170760755132,
      "learning_rate": 2.0154476730262244e-06,
      "loss": 0.5903,
      "step": 4422
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0271993330307705,
      "learning_rate": 2.013354272990145e-06,
      "loss": 0.6254,
      "step": 4423
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.078332499629157,
      "learning_rate": 2.011261686598688e-06,
      "loss": 0.5352,
      "step": 4424
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0487546083841914,
      "learning_rate": 2.009169914421934e-06,
      "loss": 0.6782,
      "step": 4425
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9895600591381529,
      "learning_rate": 2.0070789570297377e-06,
      "loss": 0.5546,
      "step": 4426
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0038166561717847,
      "learning_rate": 2.004988814991732e-06,
      "loss": 0.673,
      "step": 4427
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0515045812428239,
      "learning_rate": 2.002899488877332e-06,
      "loss": 0.5804,
      "step": 4428
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.948833130784876,
      "learning_rate": 2.000810979255728e-06,
      "loss": 0.5491,
      "step": 4429
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0846637818769351,
      "learning_rate": 1.9987232866958863e-06,
      "loss": 0.6186,
      "step": 4430
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9651124960764791,
      "learning_rate": 1.9966364117665505e-06,
      "loss": 0.5391,
      "step": 4431
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0853738563186297,
      "learning_rate": 1.9945503550362453e-06,
      "loss": 0.6483,
      "step": 4432
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0792623974146165,
      "learning_rate": 1.99246511707327e-06,
      "loss": 0.6042,
      "step": 4433
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0440470598253824,
      "learning_rate": 1.990380698445698e-06,
      "loss": 0.5338,
      "step": 4434
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9688458860415644,
      "learning_rate": 1.988297099721384e-06,
      "loss": 0.5584,
      "step": 4435
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9717499282735592,
      "learning_rate": 1.9862143214679606e-06,
      "loss": 0.5282,
      "step": 4436
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0275641805829114,
      "learning_rate": 1.984132364252831e-06,
      "loss": 0.5551,
      "step": 4437
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6679739863897243,
      "learning_rate": 1.9820512286431764e-06,
      "loss": 0.4942,
      "step": 4438
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9798136990781231,
      "learning_rate": 1.979970915205959e-06,
      "loss": 0.5139,
      "step": 4439
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0515300364735751,
      "learning_rate": 1.9778914245079127e-06,
      "loss": 0.635,
      "step": 4440
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9938738727041243,
      "learning_rate": 1.975812757115545e-06,
      "loss": 0.6163,
      "step": 4441
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.062051535182483,
      "learning_rate": 1.9737349135951463e-06,
      "loss": 0.6618,
      "step": 4442
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0235032628986918,
      "learning_rate": 1.971657894512775e-06,
      "loss": 0.5621,
      "step": 4443
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9913046908992199,
      "learning_rate": 1.9695817004342715e-06,
      "loss": 0.5873,
      "step": 4444
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0113034241138232,
      "learning_rate": 1.9675063319252467e-06,
      "loss": 0.5999,
      "step": 4445
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9625358337048183,
      "learning_rate": 1.9654317895510867e-06,
      "loss": 0.5033,
      "step": 4446
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9084037734582943,
      "learning_rate": 1.9633580738769577e-06,
      "loss": 0.4943,
      "step": 4447
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1276189660817588,
      "learning_rate": 1.961285185467794e-06,
      "loss": 0.6612,
      "step": 4448
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9858220928266545,
      "learning_rate": 1.959213124888307e-06,
      "loss": 0.5048,
      "step": 4449
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9688420214644086,
      "learning_rate": 1.957141892702985e-06,
      "loss": 0.5402,
      "step": 4450
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.104374233471592,
      "learning_rate": 1.9550714894760913e-06,
      "loss": 0.6701,
      "step": 4451
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.981839711976729,
      "learning_rate": 1.9530019157716547e-06,
      "loss": 0.5938,
      "step": 4452
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6822615439349085,
      "learning_rate": 1.9509331721534873e-06,
      "loss": 0.4551,
      "step": 4453
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1094499461385463,
      "learning_rate": 1.9488652591851732e-06,
      "loss": 0.6013,
      "step": 4454
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9779556835374327,
      "learning_rate": 1.9467981774300676e-06,
      "loss": 0.4785,
      "step": 4455
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.995450621811704,
      "learning_rate": 1.944731927451299e-06,
      "loss": 0.5694,
      "step": 4456
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.051327162307236,
      "learning_rate": 1.942666509811772e-06,
      "loss": 0.5931,
      "step": 4457
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0767348300926474,
      "learning_rate": 1.940601925074165e-06,
      "loss": 0.6805,
      "step": 4458
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1608250767833006,
      "learning_rate": 1.9385381738009273e-06,
      "loss": 0.6607,
      "step": 4459
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9909462952714984,
      "learning_rate": 1.936475256554279e-06,
      "loss": 0.5638,
      "step": 4460
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0592733690936156,
      "learning_rate": 1.9344131738962183e-06,
      "loss": 0.498,
      "step": 4461
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0938205556088303,
      "learning_rate": 1.932351926388513e-06,
      "loss": 0.654,
      "step": 4462
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.064658127442864,
      "learning_rate": 1.9302915145927015e-06,
      "loss": 0.5682,
      "step": 4463
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.038664699191305,
      "learning_rate": 1.9282319390701e-06,
      "loss": 0.6282,
      "step": 4464
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0010602205718497,
      "learning_rate": 1.92617320038179e-06,
      "loss": 0.6063,
      "step": 4465
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0443817895368612,
      "learning_rate": 1.924115299088633e-06,
      "loss": 0.6471,
      "step": 4466
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0522851667469617,
      "learning_rate": 1.9220582357512555e-06,
      "loss": 0.6579,
      "step": 4467
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.039322542338591,
      "learning_rate": 1.9200020109300567e-06,
      "loss": 0.5714,
      "step": 4468
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0459744987336177,
      "learning_rate": 1.917946625185213e-06,
      "loss": 0.5427,
      "step": 4469
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9745878427971678,
      "learning_rate": 1.9158920790766657e-06,
      "loss": 0.5372,
      "step": 4470
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0193460149233513,
      "learning_rate": 1.913838373164128e-06,
      "loss": 0.5104,
      "step": 4471
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9241309298565304,
      "learning_rate": 1.911785508007088e-06,
      "loss": 0.5137,
      "step": 4472
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9824555908978065,
      "learning_rate": 1.9097334841648064e-06,
      "loss": 0.5437,
      "step": 4473
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0243003199284917,
      "learning_rate": 1.9076823021963036e-06,
      "loss": 0.5912,
      "step": 4474
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9221318019575095,
      "learning_rate": 1.9056319626603826e-06,
      "loss": 0.5095,
      "step": 4475
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.048009878169771,
      "learning_rate": 1.9035824661156128e-06,
      "loss": 0.6009,
      "step": 4476
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9068629041941682,
      "learning_rate": 1.9015338131203325e-06,
      "loss": 0.5156,
      "step": 4477
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0595245521769754,
      "learning_rate": 1.8994860042326501e-06,
      "loss": 0.582,
      "step": 4478
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9963073402615256,
      "learning_rate": 1.8974390400104458e-06,
      "loss": 0.5351,
      "step": 4479
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9916796836916534,
      "learning_rate": 1.895392921011373e-06,
      "loss": 0.5337,
      "step": 4480
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.080926975716654,
      "learning_rate": 1.8933476477928447e-06,
      "loss": 0.6249,
      "step": 4481
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.054033648421879,
      "learning_rate": 1.8913032209120519e-06,
      "loss": 0.6573,
      "step": 4482
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.98402513733256,
      "learning_rate": 1.8892596409259556e-06,
      "loss": 0.6416,
      "step": 4483
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.058914311388047,
      "learning_rate": 1.8872169083912806e-06,
      "loss": 0.5672,
      "step": 4484
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0838310564215012,
      "learning_rate": 1.8851750238645228e-06,
      "loss": 0.5914,
      "step": 4485
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.144504469791738,
      "learning_rate": 1.8831339879019511e-06,
      "loss": 0.6833,
      "step": 4486
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1008378449182945,
      "learning_rate": 1.881093801059596e-06,
      "loss": 0.6428,
      "step": 4487
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1440230328453846,
      "learning_rate": 1.879054463893264e-06,
      "loss": 0.6747,
      "step": 4488
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9691767322484067,
      "learning_rate": 1.8770159769585261e-06,
      "loss": 0.5642,
      "step": 4489
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9882803546430539,
      "learning_rate": 1.8749783408107192e-06,
      "loss": 0.5864,
      "step": 4490
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0537737422081317,
      "learning_rate": 1.8729415560049563e-06,
      "loss": 0.6091,
      "step": 4491
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0202667004053467,
      "learning_rate": 1.8709056230961115e-06,
      "loss": 0.6378,
      "step": 4492
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9731768304347831,
      "learning_rate": 1.8688705426388264e-06,
      "loss": 0.6426,
      "step": 4493
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9973015812365162,
      "learning_rate": 1.8668363151875163e-06,
      "loss": 0.6096,
      "step": 4494
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8999630077015925,
      "learning_rate": 1.864802941296362e-06,
      "loss": 0.5093,
      "step": 4495
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.015264257127109,
      "learning_rate": 1.8627704215193082e-06,
      "loss": 0.6019,
      "step": 4496
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.950875653822951,
      "learning_rate": 1.8607387564100681e-06,
      "loss": 0.5558,
      "step": 4497
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6648692148125289,
      "learning_rate": 1.858707946522127e-06,
      "loss": 0.4642,
      "step": 4498
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0278146275732263,
      "learning_rate": 1.8566779924087315e-06,
      "loss": 0.5462,
      "step": 4499
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0370917727633262,
      "learning_rate": 1.8546488946228952e-06,
      "loss": 0.5279,
      "step": 4500
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0406354720486597,
      "learning_rate": 1.8526206537174019e-06,
      "loss": 0.6215,
      "step": 4501
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0023033611571697,
      "learning_rate": 1.8505932702448042e-06,
      "loss": 0.6234,
      "step": 4502
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6682318885547555,
      "learning_rate": 1.8485667447574101e-06,
      "loss": 0.4722,
      "step": 4503
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1003850765855483,
      "learning_rate": 1.8465410778073046e-06,
      "loss": 0.5873,
      "step": 4504
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.973967877420844,
      "learning_rate": 1.8445162699463365e-06,
      "loss": 0.5606,
      "step": 4505
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0884594394899498,
      "learning_rate": 1.842492321726118e-06,
      "loss": 0.5656,
      "step": 4506
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7179925487800976,
      "learning_rate": 1.8404692336980272e-06,
      "loss": 0.5039,
      "step": 4507
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0411976164716523,
      "learning_rate": 1.838447006413211e-06,
      "loss": 0.6579,
      "step": 4508
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9937321839601436,
      "learning_rate": 1.8364256404225777e-06,
      "loss": 0.5954,
      "step": 4509
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1133740945256696,
      "learning_rate": 1.834405136276806e-06,
      "loss": 0.5877,
      "step": 4510
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0179292627722212,
      "learning_rate": 1.8323854945263354e-06,
      "loss": 0.5665,
      "step": 4511
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0433403483680341,
      "learning_rate": 1.83036671572137e-06,
      "loss": 0.627,
      "step": 4512
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0358019075219171,
      "learning_rate": 1.828348800411885e-06,
      "loss": 0.5829,
      "step": 4513
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9492425744452456,
      "learning_rate": 1.8263317491476112e-06,
      "loss": 0.5176,
      "step": 4514
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0682577920385135,
      "learning_rate": 1.824315562478054e-06,
      "loss": 0.5408,
      "step": 4515
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1026432289393973,
      "learning_rate": 1.8223002409524736e-06,
      "loss": 0.6425,
      "step": 4516
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.043240757393761,
      "learning_rate": 1.8202857851199034e-06,
      "loss": 0.469,
      "step": 4517
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0258577824967918,
      "learning_rate": 1.818272195529135e-06,
      "loss": 0.5508,
      "step": 4518
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.074717332755621,
      "learning_rate": 1.816259472728723e-06,
      "loss": 0.6264,
      "step": 4519
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1268831032505362,
      "learning_rate": 1.8142476172669932e-06,
      "loss": 0.5856,
      "step": 4520
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1311594111881769,
      "learning_rate": 1.812236629692028e-06,
      "loss": 0.5408,
      "step": 4521
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9533680480349744,
      "learning_rate": 1.8102265105516741e-06,
      "loss": 0.5281,
      "step": 4522
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9735073767640183,
      "learning_rate": 1.8082172603935455e-06,
      "loss": 0.536,
      "step": 4523
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0303294669484855,
      "learning_rate": 1.8062088797650208e-06,
      "loss": 0.6707,
      "step": 4524
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.913712449516118,
      "learning_rate": 1.804201369213231e-06,
      "loss": 0.5109,
      "step": 4525
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.09304499984354,
      "learning_rate": 1.8021947292850811e-06,
      "loss": 0.5639,
      "step": 4526
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0055565905617532,
      "learning_rate": 1.8001889605272366e-06,
      "loss": 0.5189,
      "step": 4527
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6665609002423036,
      "learning_rate": 1.7981840634861225e-06,
      "loss": 0.4354,
      "step": 4528
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9525491987749914,
      "learning_rate": 1.7961800387079265e-06,
      "loss": 0.5369,
      "step": 4529
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.076572602414472,
      "learning_rate": 1.794176886738604e-06,
      "loss": 0.5392,
      "step": 4530
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9667290953092357,
      "learning_rate": 1.7921746081238656e-06,
      "loss": 0.5126,
      "step": 4531
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.008991971535808,
      "learning_rate": 1.7901732034091867e-06,
      "loss": 0.6412,
      "step": 4532
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1056895841284429,
      "learning_rate": 1.788172673139807e-06,
      "loss": 0.6228,
      "step": 4533
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9456323638609678,
      "learning_rate": 1.7861730178607274e-06,
      "loss": 0.6157,
      "step": 4534
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0313676991501415,
      "learning_rate": 1.784174238116707e-06,
      "loss": 0.5862,
      "step": 4535
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6908546940926898,
      "learning_rate": 1.7821763344522675e-06,
      "loss": 0.4944,
      "step": 4536
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.00372118721112,
      "learning_rate": 1.7801793074116958e-06,
      "loss": 0.5437,
      "step": 4537
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9820989058181645,
      "learning_rate": 1.778183157539034e-06,
      "loss": 0.6172,
      "step": 4538
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9793883487040431,
      "learning_rate": 1.7761878853780918e-06,
      "loss": 0.5017,
      "step": 4539
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9950062727924022,
      "learning_rate": 1.774193491472434e-06,
      "loss": 0.6108,
      "step": 4540
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0557316922369198,
      "learning_rate": 1.772199976365388e-06,
      "loss": 0.5814,
      "step": 4541
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0398378311937757,
      "learning_rate": 1.7702073406000447e-06,
      "loss": 0.6045,
      "step": 4542
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9760984174009428,
      "learning_rate": 1.768215584719251e-06,
      "loss": 0.5338,
      "step": 4543
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9503920570903627,
      "learning_rate": 1.7662247092656159e-06,
      "loss": 0.564,
      "step": 4544
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0012512473734976,
      "learning_rate": 1.7642347147815091e-06,
      "loss": 0.5631,
      "step": 4545
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9919883870155742,
      "learning_rate": 1.7622456018090638e-06,
      "loss": 0.5994,
      "step": 4546
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0007699799377054,
      "learning_rate": 1.7602573708901627e-06,
      "loss": 0.5202,
      "step": 4547
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9630859865282322,
      "learning_rate": 1.7582700225664574e-06,
      "loss": 0.5019,
      "step": 4548
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9886830241969943,
      "learning_rate": 1.7562835573793585e-06,
      "loss": 0.6152,
      "step": 4549
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9918944738179931,
      "learning_rate": 1.7542979758700323e-06,
      "loss": 0.5511,
      "step": 4550
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0186850525783768,
      "learning_rate": 1.752313278579404e-06,
      "loss": 0.6056,
      "step": 4551
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0692293347575916,
      "learning_rate": 1.7503294660481635e-06,
      "loss": 0.5979,
      "step": 4552
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9605800372580544,
      "learning_rate": 1.7483465388167542e-06,
      "loss": 0.5034,
      "step": 4553
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0168235329214348,
      "learning_rate": 1.746364497425378e-06,
      "loss": 0.623,
      "step": 4554
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.975057260985416,
      "learning_rate": 1.744383342414e-06,
      "loss": 0.4495,
      "step": 4555
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0235361254140782,
      "learning_rate": 1.7424030743223424e-06,
      "loss": 0.6545,
      "step": 4556
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0008824798564044,
      "learning_rate": 1.7404236936898837e-06,
      "loss": 0.5508,
      "step": 4557
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0077385296832553,
      "learning_rate": 1.7384452010558605e-06,
      "loss": 0.5707,
      "step": 4558
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0186551325364035,
      "learning_rate": 1.736467596959271e-06,
      "loss": 0.5517,
      "step": 4559
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.03883226524103,
      "learning_rate": 1.7344908819388678e-06,
      "loss": 0.5843,
      "step": 4560
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.033411049254337,
      "learning_rate": 1.7325150565331612e-06,
      "loss": 0.5348,
      "step": 4561
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.987120115160067,
      "learning_rate": 1.7305401212804235e-06,
      "loss": 0.5503,
      "step": 4562
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0726670284041682,
      "learning_rate": 1.7285660767186778e-06,
      "loss": 0.6184,
      "step": 4563
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.910549684606902,
      "learning_rate": 1.7265929233857125e-06,
      "loss": 0.5881,
      "step": 4564
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0959819749300486,
      "learning_rate": 1.7246206618190676e-06,
      "loss": 0.6323,
      "step": 4565
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6986199185576496,
      "learning_rate": 1.7226492925560385e-06,
      "loss": 0.4331,
      "step": 4566
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0016474581240205,
      "learning_rate": 1.7206788161336825e-06,
      "loss": 0.6044,
      "step": 4567
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.053930163379281,
      "learning_rate": 1.7187092330888155e-06,
      "loss": 0.5755,
      "step": 4568
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8222728565654627,
      "learning_rate": 1.7167405439579994e-06,
      "loss": 0.4907,
      "step": 4569
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.015326467079108,
      "learning_rate": 1.714772749277563e-06,
      "loss": 0.649,
      "step": 4570
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6309767094193446,
      "learning_rate": 1.7128058495835886e-06,
      "loss": 0.4573,
      "step": 4571
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.075567636986535,
      "learning_rate": 1.7108398454119134e-06,
      "loss": 0.6686,
      "step": 4572
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0369795383662215,
      "learning_rate": 1.7088747372981285e-06,
      "loss": 0.5756,
      "step": 4573
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0215088796316194,
      "learning_rate": 1.7069105257775859e-06,
      "loss": 0.5845,
      "step": 4574
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0149531918264814,
      "learning_rate": 1.7049472113853937e-06,
      "loss": 0.6007,
      "step": 4575
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9497023349984758,
      "learning_rate": 1.7029847946564066e-06,
      "loss": 0.5005,
      "step": 4576
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0430082551972177,
      "learning_rate": 1.7010232761252438e-06,
      "loss": 0.5598,
      "step": 4577
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.074633601607668,
      "learning_rate": 1.6990626563262797e-06,
      "loss": 0.5296,
      "step": 4578
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0226885140224544,
      "learning_rate": 1.6971029357936385e-06,
      "loss": 0.6796,
      "step": 4579
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9326700438658629,
      "learning_rate": 1.6951441150612008e-06,
      "loss": 0.5227,
      "step": 4580
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0244232479068807,
      "learning_rate": 1.6931861946626066e-06,
      "loss": 0.5533,
      "step": 4581
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9938922024662841,
      "learning_rate": 1.6912291751312465e-06,
      "loss": 0.5854,
      "step": 4582
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0084104109247258,
      "learning_rate": 1.6892730570002635e-06,
      "loss": 0.5213,
      "step": 4583
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9822383826338872,
      "learning_rate": 1.6873178408025625e-06,
      "loss": 0.5351,
      "step": 4584
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0777081440874667,
      "learning_rate": 1.6853635270707947e-06,
      "loss": 0.6314,
      "step": 4585
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8765669779006965,
      "learning_rate": 1.6834101163373716e-06,
      "loss": 0.4682,
      "step": 4586
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0546422316207198,
      "learning_rate": 1.6814576091344559e-06,
      "loss": 0.6105,
      "step": 4587
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0460391994620626,
      "learning_rate": 1.6795060059939617e-06,
      "loss": 0.629,
      "step": 4588
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9760723416311925,
      "learning_rate": 1.6775553074475631e-06,
      "loss": 0.5809,
      "step": 4589
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.004342744954546,
      "learning_rate": 1.6756055140266826e-06,
      "loss": 0.6542,
      "step": 4590
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9876713791277192,
      "learning_rate": 1.6736566262624959e-06,
      "loss": 0.6452,
      "step": 4591
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.030443627761999,
      "learning_rate": 1.6717086446859354e-06,
      "loss": 0.5704,
      "step": 4592
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.12360749664529,
      "learning_rate": 1.6697615698276864e-06,
      "loss": 0.5641,
      "step": 4593
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9941591519871712,
      "learning_rate": 1.6678154022181842e-06,
      "loss": 0.5799,
      "step": 4594
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9740282086805269,
      "learning_rate": 1.6658701423876172e-06,
      "loss": 0.5553,
      "step": 4595
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0068896513255563,
      "learning_rate": 1.6639257908659294e-06,
      "loss": 0.6338,
      "step": 4596
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0111096460758524,
      "learning_rate": 1.661982348182819e-06,
      "loss": 0.6418,
      "step": 4597
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9917540142360353,
      "learning_rate": 1.660039814867726e-06,
      "loss": 0.5822,
      "step": 4598
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9388540241978435,
      "learning_rate": 1.6580981914498545e-06,
      "loss": 0.5764,
      "step": 4599
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0216502354150883,
      "learning_rate": 1.6561574784581574e-06,
      "loss": 0.5439,
      "step": 4600
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0258042131043732,
      "learning_rate": 1.6542176764213364e-06,
      "loss": 0.6145,
      "step": 4601
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9457805795717893,
      "learning_rate": 1.6522787858678463e-06,
      "loss": 0.6158,
      "step": 4602
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0211130300977866,
      "learning_rate": 1.6503408073258964e-06,
      "loss": 0.5648,
      "step": 4603
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1836207721646184,
      "learning_rate": 1.648403741323445e-06,
      "loss": 0.5979,
      "step": 4604
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0726368798535384,
      "learning_rate": 1.6464675883881998e-06,
      "loss": 0.514,
      "step": 4605
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.001707961364406,
      "learning_rate": 1.6445323490476255e-06,
      "loss": 0.6539,
      "step": 4606
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0060576450149683,
      "learning_rate": 1.6425980238289324e-06,
      "loss": 0.6151,
      "step": 4607
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0317817956918052,
      "learning_rate": 1.6406646132590865e-06,
      "loss": 0.5337,
      "step": 4608
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0965988725387565,
      "learning_rate": 1.6387321178648003e-06,
      "loss": 0.666,
      "step": 4609
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0306348325177166,
      "learning_rate": 1.6368005381725372e-06,
      "loss": 0.5865,
      "step": 4610
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0350629539880678,
      "learning_rate": 1.6348698747085168e-06,
      "loss": 0.6547,
      "step": 4611
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.052627907632733,
      "learning_rate": 1.6329401279987023e-06,
      "loss": 0.5901,
      "step": 4612
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2953140377677184,
      "learning_rate": 1.6310112985688093e-06,
      "loss": 0.5436,
      "step": 4613
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0289040927026327,
      "learning_rate": 1.629083386944305e-06,
      "loss": 0.5794,
      "step": 4614
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0610051271619096,
      "learning_rate": 1.6271563936504082e-06,
      "loss": 0.5643,
      "step": 4615
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0501535680735308,
      "learning_rate": 1.6252303192120821e-06,
      "loss": 0.5454,
      "step": 4616
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0754897924423756,
      "learning_rate": 1.6233051641540415e-06,
      "loss": 0.6368,
      "step": 4617
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8599721851114317,
      "learning_rate": 1.6213809290007554e-06,
      "loss": 0.5357,
      "step": 4618
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.007530512748564,
      "learning_rate": 1.6194576142764363e-06,
      "loss": 0.6103,
      "step": 4619
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9223225239681228,
      "learning_rate": 1.617535220505046e-06,
      "loss": 0.486,
      "step": 4620
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.461727661840068,
      "learning_rate": 1.6156137482103006e-06,
      "loss": 0.6026,
      "step": 4621
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9702424794330411,
      "learning_rate": 1.6136931979156628e-06,
      "loss": 0.5598,
      "step": 4622
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0954686620276628,
      "learning_rate": 1.6117735701443416e-06,
      "loss": 0.5132,
      "step": 4623
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0320996486463314,
      "learning_rate": 1.6098548654192958e-06,
      "loss": 0.5218,
      "step": 4624
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9881557658220156,
      "learning_rate": 1.6079370842632358e-06,
      "loss": 0.6314,
      "step": 4625
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9993048510708834,
      "learning_rate": 1.606020227198617e-06,
      "loss": 0.5527,
      "step": 4626
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0049627443895865,
      "learning_rate": 1.604104294747642e-06,
      "loss": 0.6379,
      "step": 4627
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6653413309808434,
      "learning_rate": 1.6021892874322675e-06,
      "loss": 0.4696,
      "step": 4628
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9910164953967522,
      "learning_rate": 1.6002752057741906e-06,
      "loss": 0.5776,
      "step": 4629
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0135103755728023,
      "learning_rate": 1.5983620502948632e-06,
      "loss": 0.536,
      "step": 4630
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0056511059751256,
      "learning_rate": 1.59644982151548e-06,
      "loss": 0.547,
      "step": 4631
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0747398896875249,
      "learning_rate": 1.5945385199569836e-06,
      "loss": 0.5497,
      "step": 4632
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9650985332350642,
      "learning_rate": 1.5926281461400673e-06,
      "loss": 0.5782,
      "step": 4633
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0019111542824164,
      "learning_rate": 1.5907187005851676e-06,
      "loss": 0.539,
      "step": 4634
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6518307694193087,
      "learning_rate": 1.588810183812473e-06,
      "loss": 0.4661,
      "step": 4635
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0117046829158827,
      "learning_rate": 1.5869025963419122e-06,
      "loss": 0.6053,
      "step": 4636
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9912204196638209,
      "learning_rate": 1.584995938693169e-06,
      "loss": 0.6106,
      "step": 4637
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.034026816954276,
      "learning_rate": 1.5830902113856667e-06,
      "loss": 0.5606,
      "step": 4638
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0038063957306251,
      "learning_rate": 1.581185414938577e-06,
      "loss": 0.6445,
      "step": 4639
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0960056877163022,
      "learning_rate": 1.579281549870822e-06,
      "loss": 0.6455,
      "step": 4640
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0493224612790566,
      "learning_rate": 1.5773786167010657e-06,
      "loss": 0.636,
      "step": 4641
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0104172329272194,
      "learning_rate": 1.5754766159477174e-06,
      "loss": 0.5466,
      "step": 4642
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9364076472764195,
      "learning_rate": 1.5735755481289371e-06,
      "loss": 0.5073,
      "step": 4643
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6921175689223544,
      "learning_rate": 1.5716754137626284e-06,
      "loss": 0.5014,
      "step": 4644
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0571832740687772,
      "learning_rate": 1.5697762133664396e-06,
      "loss": 0.5495,
      "step": 4645
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0705032073965264,
      "learning_rate": 1.5678779474577633e-06,
      "loss": 0.6357,
      "step": 4646
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0527774804630468,
      "learning_rate": 1.5659806165537428e-06,
      "loss": 0.5679,
      "step": 4647
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6439222756368792,
      "learning_rate": 1.5640842211712614e-06,
      "loss": 0.4629,
      "step": 4648
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0353919131271123,
      "learning_rate": 1.5621887618269481e-06,
      "loss": 0.6412,
      "step": 4649
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9652472075259956,
      "learning_rate": 1.5602942390371817e-06,
      "loss": 0.5282,
      "step": 4650
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0112878234701226,
      "learning_rate": 1.558400653318079e-06,
      "loss": 0.5504,
      "step": 4651
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1112490357883558,
      "learning_rate": 1.556508005185508e-06,
      "loss": 0.618,
      "step": 4652
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.031563008682259,
      "learning_rate": 1.5546162951550759e-06,
      "loss": 0.6235,
      "step": 4653
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.100984084546513,
      "learning_rate": 1.5527255237421384e-06,
      "loss": 0.5266,
      "step": 4654
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9949047889332807,
      "learning_rate": 1.5508356914617933e-06,
      "loss": 0.5041,
      "step": 4655
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6986522745554307,
      "learning_rate": 1.5489467988288809e-06,
      "loss": 0.4694,
      "step": 4656
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0619180370878778,
      "learning_rate": 1.5470588463579906e-06,
      "loss": 0.6401,
      "step": 4657
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9846845786639264,
      "learning_rate": 1.5451718345634503e-06,
      "loss": 0.5767,
      "step": 4658
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0556847420811397,
      "learning_rate": 1.5432857639593362e-06,
      "loss": 0.567,
      "step": 4659
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0182353760342577,
      "learning_rate": 1.5414006350594658e-06,
      "loss": 0.5491,
      "step": 4660
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1140348348767246,
      "learning_rate": 1.5395164483773966e-06,
      "loss": 0.6311,
      "step": 4661
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0450967558161928,
      "learning_rate": 1.537633204426438e-06,
      "loss": 0.5454,
      "step": 4662
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9866874873012441,
      "learning_rate": 1.535750903719635e-06,
      "loss": 0.5485,
      "step": 4663
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9911353208078594,
      "learning_rate": 1.5338695467697767e-06,
      "loss": 0.5321,
      "step": 4664
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1003559874522615,
      "learning_rate": 1.531989134089399e-06,
      "loss": 0.5864,
      "step": 4665
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9172789887628691,
      "learning_rate": 1.5301096661907782e-06,
      "loss": 0.4962,
      "step": 4666
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9851700993836898,
      "learning_rate": 1.5282311435859325e-06,
      "loss": 0.6045,
      "step": 4667
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.031288476652932,
      "learning_rate": 1.526353566786622e-06,
      "loss": 0.572,
      "step": 4668
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9060648684565664,
      "learning_rate": 1.5244769363043527e-06,
      "loss": 0.5392,
      "step": 4669
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9992479866474782,
      "learning_rate": 1.5226012526503698e-06,
      "loss": 0.5973,
      "step": 4670
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0179862322703919,
      "learning_rate": 1.5207265163356588e-06,
      "loss": 0.538,
      "step": 4671
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0951226527434563,
      "learning_rate": 1.5188527278709514e-06,
      "loss": 0.5914,
      "step": 4672
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6836788356717511,
      "learning_rate": 1.5169798877667207e-06,
      "loss": 0.4824,
      "step": 4673
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.11121861056405,
      "learning_rate": 1.5151079965331788e-06,
      "loss": 0.5995,
      "step": 4674
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.985113203331552,
      "learning_rate": 1.5132370546802794e-06,
      "loss": 0.6253,
      "step": 4675
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0118519168362552,
      "learning_rate": 1.5113670627177202e-06,
      "loss": 0.6672,
      "step": 4676
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.039865145090428,
      "learning_rate": 1.5094980211549382e-06,
      "loss": 0.6125,
      "step": 4677
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1086955962485365,
      "learning_rate": 1.5076299305011095e-06,
      "loss": 0.5682,
      "step": 4678
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9723495519802594,
      "learning_rate": 1.5057627912651574e-06,
      "loss": 0.5224,
      "step": 4679
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.091586004249404,
      "learning_rate": 1.5038966039557384e-06,
      "loss": 0.6349,
      "step": 4680
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0099828131721313,
      "learning_rate": 1.502031369081257e-06,
      "loss": 0.5339,
      "step": 4681
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0287233837077,
      "learning_rate": 1.5001670871498524e-06,
      "loss": 0.5469,
      "step": 4682
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9446446320678786,
      "learning_rate": 1.4983037586694055e-06,
      "loss": 0.5334,
      "step": 4683
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9429833480287023,
      "learning_rate": 1.4964413841475412e-06,
      "loss": 0.607,
      "step": 4684
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0510871424396153,
      "learning_rate": 1.4945799640916198e-06,
      "loss": 0.5784,
      "step": 4685
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9808270452977982,
      "learning_rate": 1.4927194990087418e-06,
      "loss": 0.536,
      "step": 4686
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0326406840165137,
      "learning_rate": 1.4908599894057512e-06,
      "loss": 0.6012,
      "step": 4687
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0092856723731785,
      "learning_rate": 1.489001435789233e-06,
      "loss": 0.5617,
      "step": 4688
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0651477063697676,
      "learning_rate": 1.4871438386655018e-06,
      "loss": 0.585,
      "step": 4689
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1943771231322091,
      "learning_rate": 1.4852871985406208e-06,
      "loss": 0.5912,
      "step": 4690
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0801611439199188,
      "learning_rate": 1.4834315159203927e-06,
      "loss": 0.6097,
      "step": 4691
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0415334332116688,
      "learning_rate": 1.481576791310354e-06,
      "loss": 0.5874,
      "step": 4692
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9298210419573935,
      "learning_rate": 1.4797230252157818e-06,
      "loss": 0.5302,
      "step": 4693
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1297748903233178,
      "learning_rate": 1.4778702181416949e-06,
      "loss": 0.5628,
      "step": 4694
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9763905336524379,
      "learning_rate": 1.4760183705928499e-06,
      "loss": 0.5381,
      "step": 4695
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0617805259257787,
      "learning_rate": 1.4741674830737396e-06,
      "loss": 0.5564,
      "step": 4696
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0433474367308533,
      "learning_rate": 1.472317556088596e-06,
      "loss": 0.6013,
      "step": 4697
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9424849864818384,
      "learning_rate": 1.4704685901413928e-06,
      "loss": 0.6186,
      "step": 4698
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0660765502687504,
      "learning_rate": 1.468620585735837e-06,
      "loss": 0.5698,
      "step": 4699
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9741464983824553,
      "learning_rate": 1.466773543375376e-06,
      "loss": 0.4499,
      "step": 4700
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0339087777584084,
      "learning_rate": 1.4649274635631972e-06,
      "loss": 0.6527,
      "step": 4701
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.034828524824342,
      "learning_rate": 1.4630823468022204e-06,
      "loss": 0.5456,
      "step": 4702
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9564945923499976,
      "learning_rate": 1.4612381935951093e-06,
      "loss": 0.4925,
      "step": 4703
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9541351045378871,
      "learning_rate": 1.4593950044442612e-06,
      "loss": 0.6145,
      "step": 4704
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1259528414984303,
      "learning_rate": 1.4575527798518097e-06,
      "loss": 0.5822,
      "step": 4705
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.129462630285597,
      "learning_rate": 1.4557115203196304e-06,
      "loss": 0.6138,
      "step": 4706
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.047974713743179,
      "learning_rate": 1.453871226349332e-06,
      "loss": 0.5968,
      "step": 4707
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6920814329824146,
      "learning_rate": 1.45203189844226e-06,
      "loss": 0.5023,
      "step": 4708
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.055623339586582,
      "learning_rate": 1.4501935370994985e-06,
      "loss": 0.5413,
      "step": 4709
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0422968758341848,
      "learning_rate": 1.4483561428218717e-06,
      "loss": 0.5834,
      "step": 4710
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0129936279358076,
      "learning_rate": 1.4465197161099305e-06,
      "loss": 0.6008,
      "step": 4711
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0713758779015725,
      "learning_rate": 1.4446842574639708e-06,
      "loss": 0.6416,
      "step": 4712
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.979327402017095,
      "learning_rate": 1.4428497673840235e-06,
      "loss": 0.5835,
      "step": 4713
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0577682939072912,
      "learning_rate": 1.441016246369853e-06,
      "loss": 0.6223,
      "step": 4714
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.044786129858887,
      "learning_rate": 1.4391836949209597e-06,
      "loss": 0.5502,
      "step": 4715
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0510924896776026,
      "learning_rate": 1.437352113536582e-06,
      "loss": 0.582,
      "step": 4716
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9759427130140866,
      "learning_rate": 1.435521502715697e-06,
      "loss": 0.5725,
      "step": 4717
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9927967192431314,
      "learning_rate": 1.4336918629570069e-06,
      "loss": 0.6091,
      "step": 4718
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0140102930957529,
      "learning_rate": 1.431863194758959e-06,
      "loss": 0.5954,
      "step": 4719
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.080223294085055,
      "learning_rate": 1.4300354986197345e-06,
      "loss": 0.5361,
      "step": 4720
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6146691283157173,
      "learning_rate": 1.4282087750372475e-06,
      "loss": 0.4262,
      "step": 4721
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0592016983361388,
      "learning_rate": 1.4263830245091454e-06,
      "loss": 0.5903,
      "step": 4722
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0076396928016627,
      "learning_rate": 1.4245582475328156e-06,
      "loss": 0.6539,
      "step": 4723
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0053110030153154,
      "learning_rate": 1.4227344446053759e-06,
      "loss": 0.6438,
      "step": 4724
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0818826403837445,
      "learning_rate": 1.420911616223683e-06,
      "loss": 0.6361,
      "step": 4725
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1270284096979695,
      "learning_rate": 1.4190897628843242e-06,
      "loss": 0.6209,
      "step": 4726
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9876169412889974,
      "learning_rate": 1.4172688850836202e-06,
      "loss": 0.6053,
      "step": 4727
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0587138338951596,
      "learning_rate": 1.4154489833176322e-06,
      "loss": 0.488,
      "step": 4728
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0617718876916362,
      "learning_rate": 1.41363005808215e-06,
      "loss": 0.6007,
      "step": 4729
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.642851754740661,
      "learning_rate": 1.4118121098726972e-06,
      "loss": 0.4287,
      "step": 4730
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9811546491786569,
      "learning_rate": 1.4099951391845345e-06,
      "loss": 0.5174,
      "step": 4731
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9804370804678393,
      "learning_rate": 1.408179146512657e-06,
      "loss": 0.4698,
      "step": 4732
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0558167756053627,
      "learning_rate": 1.4063641323517886e-06,
      "loss": 0.6615,
      "step": 4733
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0553494180830536,
      "learning_rate": 1.4045500971963882e-06,
      "loss": 0.6089,
      "step": 4734
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9199044935692631,
      "learning_rate": 1.4027370415406528e-06,
      "loss": 0.543,
      "step": 4735
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9600058626192374,
      "learning_rate": 1.4009249658785058e-06,
      "loss": 0.5626,
      "step": 4736
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0085699880186747,
      "learning_rate": 1.399113870703605e-06,
      "loss": 0.5655,
      "step": 4737
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0819619886971028,
      "learning_rate": 1.3973037565093455e-06,
      "loss": 0.6312,
      "step": 4738
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0033080411648958,
      "learning_rate": 1.395494623788855e-06,
      "loss": 0.5609,
      "step": 4739
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0100537800487672,
      "learning_rate": 1.3936864730349842e-06,
      "loss": 0.6115,
      "step": 4740
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0955957047143055,
      "learning_rate": 1.3918793047403268e-06,
      "loss": 0.6378,
      "step": 4741
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9682088035205518,
      "learning_rate": 1.3900731193972073e-06,
      "loss": 0.5358,
      "step": 4742
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1086444138545544,
      "learning_rate": 1.3882679174976777e-06,
      "loss": 0.5694,
      "step": 4743
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0744788218048613,
      "learning_rate": 1.386463699533524e-06,
      "loss": 0.62,
      "step": 4744
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9546603740743992,
      "learning_rate": 1.3846604659962676e-06,
      "loss": 0.5554,
      "step": 4745
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0208412539181988,
      "learning_rate": 1.3828582173771576e-06,
      "loss": 0.5437,
      "step": 4746
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.990630020751945,
      "learning_rate": 1.3810569541671754e-06,
      "loss": 0.522,
      "step": 4747
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.953695675484812,
      "learning_rate": 1.3792566768570364e-06,
      "loss": 0.5056,
      "step": 4748
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9512996204025755,
      "learning_rate": 1.3774573859371842e-06,
      "loss": 0.5249,
      "step": 4749
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0491823377594638,
      "learning_rate": 1.3756590818977972e-06,
      "loss": 0.5555,
      "step": 4750
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6900097313162797,
      "learning_rate": 1.37386176522878e-06,
      "loss": 0.4798,
      "step": 4751
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0441613387976707,
      "learning_rate": 1.372065436419775e-06,
      "loss": 0.5971,
      "step": 4752
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9901339001285573,
      "learning_rate": 1.3702700959601483e-06,
      "loss": 0.5175,
      "step": 4753
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0070733461256531,
      "learning_rate": 1.368475744339003e-06,
      "loss": 0.6231,
      "step": 4754
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0273213000445216,
      "learning_rate": 1.366682382045168e-06,
      "loss": 0.6205,
      "step": 4755
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9223905965180255,
      "learning_rate": 1.364890009567204e-06,
      "loss": 0.5709,
      "step": 4756
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9979029804802128,
      "learning_rate": 1.3630986273934054e-06,
      "loss": 0.5988,
      "step": 4757
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0688403940386748,
      "learning_rate": 1.3613082360117924e-06,
      "loss": 0.5789,
      "step": 4758
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0226584745375502,
      "learning_rate": 1.3595188359101152e-06,
      "loss": 0.5907,
      "step": 4759
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.998762771165808,
      "learning_rate": 1.357730427575858e-06,
      "loss": 0.5086,
      "step": 4760
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.101812334761135,
      "learning_rate": 1.3559430114962345e-06,
      "loss": 0.5884,
      "step": 4761
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1066168158737948,
      "learning_rate": 1.3541565881581815e-06,
      "loss": 0.5909,
      "step": 4762
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6466744174642701,
      "learning_rate": 1.3523711580483717e-06,
      "loss": 0.4377,
      "step": 4763
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0994280606255211,
      "learning_rate": 1.350586721653207e-06,
      "loss": 0.5579,
      "step": 4764
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0602953577208425,
      "learning_rate": 1.3488032794588168e-06,
      "loss": 0.6064,
      "step": 4765
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.152778111492181,
      "learning_rate": 1.347020831951057e-06,
      "loss": 0.6428,
      "step": 4766
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1205416009889089,
      "learning_rate": 1.3452393796155194e-06,
      "loss": 0.5771,
      "step": 4767
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9762382449955147,
      "learning_rate": 1.343458922937519e-06,
      "loss": 0.4884,
      "step": 4768
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.049547374661361,
      "learning_rate": 1.341679462402099e-06,
      "loss": 0.5797,
      "step": 4769
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1160523816536585,
      "learning_rate": 1.3399009984940376e-06,
      "loss": 0.6091,
      "step": 4770
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9077240129334275,
      "learning_rate": 1.338123531697834e-06,
      "loss": 0.5351,
      "step": 4771
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9700415414759279,
      "learning_rate": 1.3363470624977221e-06,
      "loss": 0.5269,
      "step": 4772
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0629540853488928,
      "learning_rate": 1.3345715913776575e-06,
      "loss": 0.529,
      "step": 4773
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1110386214513914,
      "learning_rate": 1.3327971188213317e-06,
      "loss": 0.52,
      "step": 4774
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0340044516619724,
      "learning_rate": 1.3310236453121562e-06,
      "loss": 0.5697,
      "step": 4775
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1617930378848875,
      "learning_rate": 1.3292511713332767e-06,
      "loss": 0.6597,
      "step": 4776
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0218908291623725,
      "learning_rate": 1.3274796973675629e-06,
      "loss": 0.5914,
      "step": 4777
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9801214601429244,
      "learning_rate": 1.3257092238976122e-06,
      "loss": 0.5198,
      "step": 4778
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0268192281898316,
      "learning_rate": 1.3239397514057523e-06,
      "loss": 0.5969,
      "step": 4779
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0939345103262594,
      "learning_rate": 1.3221712803740356e-06,
      "loss": 0.5738,
      "step": 4780
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9043580143327139,
      "learning_rate": 1.3204038112842404e-06,
      "loss": 0.5433,
      "step": 4781
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0705442305321387,
      "learning_rate": 1.3186373446178757e-06,
      "loss": 0.6263,
      "step": 4782
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9372205255549791,
      "learning_rate": 1.3168718808561793e-06,
      "loss": 0.5278,
      "step": 4783
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0727860235877,
      "learning_rate": 1.3151074204801046e-06,
      "loss": 0.5905,
      "step": 4784
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9957230085026105,
      "learning_rate": 1.313343963970344e-06,
      "loss": 0.4986,
      "step": 4785
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.021198510233983,
      "learning_rate": 1.3115815118073116e-06,
      "loss": 0.5299,
      "step": 4786
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0195905084845667,
      "learning_rate": 1.3098200644711478e-06,
      "loss": 0.6474,
      "step": 4787
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0291298947187988,
      "learning_rate": 1.3080596224417174e-06,
      "loss": 0.5757,
      "step": 4788
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0861047473881578,
      "learning_rate": 1.3063001861986162e-06,
      "loss": 0.5584,
      "step": 4789
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0322124145542402,
      "learning_rate": 1.3045417562211616e-06,
      "loss": 0.6033,
      "step": 4790
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9895760867156055,
      "learning_rate": 1.3027843329883972e-06,
      "loss": 0.5136,
      "step": 4791
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0368715567197082,
      "learning_rate": 1.3010279169790947e-06,
      "loss": 0.5749,
      "step": 4792
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9821782597806226,
      "learning_rate": 1.2992725086717518e-06,
      "loss": 0.5896,
      "step": 4793
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9734868166777456,
      "learning_rate": 1.2975181085445887e-06,
      "loss": 0.5706,
      "step": 4794
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0566996243198126,
      "learning_rate": 1.2957647170755504e-06,
      "loss": 0.6214,
      "step": 4795
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0052147943887189,
      "learning_rate": 1.2940123347423133e-06,
      "loss": 0.5403,
      "step": 4796
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0247135774526586,
      "learning_rate": 1.292260962022272e-06,
      "loss": 0.6127,
      "step": 4797
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9880507581468722,
      "learning_rate": 1.2905105993925477e-06,
      "loss": 0.5573,
      "step": 4798
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0055752594790681,
      "learning_rate": 1.2887612473299905e-06,
      "loss": 0.6452,
      "step": 4799
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9570762270499491,
      "learning_rate": 1.2870129063111685e-06,
      "loss": 0.5762,
      "step": 4800
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1612245717039151,
      "learning_rate": 1.2852655768123811e-06,
      "loss": 0.5618,
      "step": 4801
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0780271235733663,
      "learning_rate": 1.2835192593096485e-06,
      "loss": 0.6211,
      "step": 4802
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0272610365790378,
      "learning_rate": 1.2817739542787134e-06,
      "loss": 0.6028,
      "step": 4803
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9308020187175112,
      "learning_rate": 1.2800296621950463e-06,
      "loss": 0.5242,
      "step": 4804
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9996328125949665,
      "learning_rate": 1.2782863835338444e-06,
      "loss": 0.646,
      "step": 4805
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0582797614044683,
      "learning_rate": 1.2765441187700179e-06,
      "loss": 0.6205,
      "step": 4806
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9939757887695724,
      "learning_rate": 1.2748028683782115e-06,
      "loss": 0.5454,
      "step": 4807
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0809741304944729,
      "learning_rate": 1.2730626328327906e-06,
      "loss": 0.5349,
      "step": 4808
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.016857539652991,
      "learning_rate": 1.2713234126078423e-06,
      "loss": 0.6498,
      "step": 4809
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9763743942256143,
      "learning_rate": 1.2695852081771758e-06,
      "loss": 0.5509,
      "step": 4810
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.897441652388817,
      "learning_rate": 1.267848020014329e-06,
      "loss": 0.4787,
      "step": 4811
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0361103215170073,
      "learning_rate": 1.266111848592561e-06,
      "loss": 0.5771,
      "step": 4812
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9979941579886407,
      "learning_rate": 1.2643766943848484e-06,
      "loss": 0.6079,
      "step": 4813
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1683656780290095,
      "learning_rate": 1.2626425578638973e-06,
      "loss": 0.6393,
      "step": 4814
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.164056577321666,
      "learning_rate": 1.2609094395021354e-06,
      "loss": 0.5224,
      "step": 4815
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9992703714677759,
      "learning_rate": 1.259177339771711e-06,
      "loss": 0.5495,
      "step": 4816
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6615808699624711,
      "learning_rate": 1.257446259144494e-06,
      "loss": 0.4415,
      "step": 4817
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1307706345043902,
      "learning_rate": 1.2557161980920824e-06,
      "loss": 0.6388,
      "step": 4818
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0865392636521165,
      "learning_rate": 1.2539871570857893e-06,
      "loss": 0.614,
      "step": 4819
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0310294755539249,
      "learning_rate": 1.252259136596653e-06,
      "loss": 0.5445,
      "step": 4820
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.003256213122981,
      "learning_rate": 1.250532137095436e-06,
      "loss": 0.5397,
      "step": 4821
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0131353005072248,
      "learning_rate": 1.2488061590526185e-06,
      "loss": 0.5955,
      "step": 4822
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1267250074824153,
      "learning_rate": 1.2470812029384071e-06,
      "loss": 0.571,
      "step": 4823
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0569334713592216,
      "learning_rate": 1.2453572692227257e-06,
      "loss": 0.5973,
      "step": 4824
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.090326288810078,
      "learning_rate": 1.2436343583752197e-06,
      "loss": 0.5725,
      "step": 4825
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0397862221096719,
      "learning_rate": 1.2419124708652607e-06,
      "loss": 0.5649,
      "step": 4826
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.013949833800736,
      "learning_rate": 1.2401916071619374e-06,
      "loss": 0.5968,
      "step": 4827
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7003614548727249,
      "learning_rate": 1.2384717677340585e-06,
      "loss": 0.4832,
      "step": 4828
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1993203312056389,
      "learning_rate": 1.2367529530501571e-06,
      "loss": 0.6199,
      "step": 4829
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0570309506820976,
      "learning_rate": 1.2350351635784875e-06,
      "loss": 0.6219,
      "step": 4830
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9740990567289188,
      "learning_rate": 1.2333183997870207e-06,
      "loss": 0.6379,
      "step": 4831
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9126363079168041,
      "learning_rate": 1.2316026621434502e-06,
      "loss": 0.494,
      "step": 4832
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0477745684724922,
      "learning_rate": 1.2298879511151906e-06,
      "loss": 0.5694,
      "step": 4833
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0782862611213115,
      "learning_rate": 1.2281742671693798e-06,
      "loss": 0.6169,
      "step": 4834
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0301941379476045,
      "learning_rate": 1.2264616107728666e-06,
      "loss": 0.5389,
      "step": 4835
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9593152225838427,
      "learning_rate": 1.2247499823922287e-06,
      "loss": 0.5486,
      "step": 4836
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0404751963798367,
      "learning_rate": 1.2230393824937631e-06,
      "loss": 0.5849,
      "step": 4837
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.041562605162502,
      "learning_rate": 1.2213298115434812e-06,
      "loss": 0.5269,
      "step": 4838
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0538988111924779,
      "learning_rate": 1.219621270007118e-06,
      "loss": 0.6464,
      "step": 4839
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0371658609559797,
      "learning_rate": 1.2179137583501282e-06,
      "loss": 0.623,
      "step": 4840
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9447912804203711,
      "learning_rate": 1.2162072770376848e-06,
      "loss": 0.4805,
      "step": 4841
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.111514848970632,
      "learning_rate": 1.2145018265346786e-06,
      "loss": 0.5777,
      "step": 4842
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9769974053682706,
      "learning_rate": 1.2127974073057241e-06,
      "loss": 0.5377,
      "step": 4843
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9862941487838945,
      "learning_rate": 1.2110940198151489e-06,
      "loss": 0.603,
      "step": 4844
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0130055728344844,
      "learning_rate": 1.2093916645270066e-06,
      "loss": 0.5406,
      "step": 4845
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0765615085256808,
      "learning_rate": 1.2076903419050629e-06,
      "loss": 0.6312,
      "step": 4846
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0232266963208256,
      "learning_rate": 1.2059900524128048e-06,
      "loss": 0.5312,
      "step": 4847
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9982036253533321,
      "learning_rate": 1.2042907965134404e-06,
      "loss": 0.6082,
      "step": 4848
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0435942369917213,
      "learning_rate": 1.2025925746698918e-06,
      "loss": 0.6235,
      "step": 4849
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9442277551631802,
      "learning_rate": 1.200895387344801e-06,
      "loss": 0.493,
      "step": 4850
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0198683266561288,
      "learning_rate": 1.1991992350005294e-06,
      "loss": 0.6288,
      "step": 4851
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0030860410892632,
      "learning_rate": 1.1975041180991576e-06,
      "loss": 0.5933,
      "step": 4852
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9540299322996658,
      "learning_rate": 1.19581003710248e-06,
      "loss": 0.5504,
      "step": 4853
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9785886419739728,
      "learning_rate": 1.1941169924720103e-06,
      "loss": 0.518,
      "step": 4854
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0224131136273988,
      "learning_rate": 1.1924249846689835e-06,
      "loss": 0.5245,
      "step": 4855
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.942766505277974,
      "learning_rate": 1.1907340141543466e-06,
      "loss": 0.5372,
      "step": 4856
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.137324298188351,
      "learning_rate": 1.189044081388766e-06,
      "loss": 0.6422,
      "step": 4857
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6170500544403583,
      "learning_rate": 1.1873551868326272e-06,
      "loss": 0.4597,
      "step": 4858
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.970180802878914,
      "learning_rate": 1.185667330946033e-06,
      "loss": 0.5872,
      "step": 4859
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9432730718589339,
      "learning_rate": 1.1839805141888012e-06,
      "loss": 0.507,
      "step": 4860
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.998595924766067,
      "learning_rate": 1.1822947370204647e-06,
      "loss": 0.5216,
      "step": 4861
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0692182074604981,
      "learning_rate": 1.180609999900279e-06,
      "loss": 0.5224,
      "step": 4862
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1028190461750578,
      "learning_rate": 1.1789263032872112e-06,
      "loss": 0.596,
      "step": 4863
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.972959880949075,
      "learning_rate": 1.1772436476399456e-06,
      "loss": 0.5082,
      "step": 4864
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.035319628683588,
      "learning_rate": 1.1755620334168866e-06,
      "loss": 0.6499,
      "step": 4865
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0631279209034292,
      "learning_rate": 1.1738814610761496e-06,
      "loss": 0.6126,
      "step": 4866
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0435218505550952,
      "learning_rate": 1.1722019310755717e-06,
      "loss": 0.6054,
      "step": 4867
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9892401393099272,
      "learning_rate": 1.1705234438727015e-06,
      "loss": 0.5468,
      "step": 4868
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.087570974149394,
      "learning_rate": 1.1688459999248042e-06,
      "loss": 0.6364,
      "step": 4869
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0356677646484551,
      "learning_rate": 1.1671695996888637e-06,
      "loss": 0.551,
      "step": 4870
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9773303193903512,
      "learning_rate": 1.165494243621576e-06,
      "loss": 0.4945,
      "step": 4871
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0469000872910896,
      "learning_rate": 1.1638199321793563e-06,
      "loss": 0.6011,
      "step": 4872
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6385754600794116,
      "learning_rate": 1.1621466658183306e-06,
      "loss": 0.4718,
      "step": 4873
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9796142423707253,
      "learning_rate": 1.1604744449943455e-06,
      "loss": 0.6125,
      "step": 4874
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9449258643922108,
      "learning_rate": 1.1588032701629592e-06,
      "loss": 0.4749,
      "step": 4875
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9889890888196375,
      "learning_rate": 1.1571331417794435e-06,
      "loss": 0.6487,
      "step": 4876
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0570777451080675,
      "learning_rate": 1.1554640602987905e-06,
      "loss": 0.587,
      "step": 4877
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9952627243540655,
      "learning_rate": 1.153796026175703e-06,
      "loss": 0.5791,
      "step": 4878
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9907503735854721,
      "learning_rate": 1.1521290398645978e-06,
      "loss": 0.5526,
      "step": 4879
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0404907860076669,
      "learning_rate": 1.150463101819609e-06,
      "loss": 0.6021,
      "step": 4880
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9774002395645606,
      "learning_rate": 1.1487982124945861e-06,
      "loss": 0.4974,
      "step": 4881
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0297807045282714,
      "learning_rate": 1.147134372343089e-06,
      "loss": 0.5757,
      "step": 4882
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1197887877110793,
      "learning_rate": 1.1454715818183927e-06,
      "loss": 0.6369,
      "step": 4883
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9772213488602425,
      "learning_rate": 1.1438098413734888e-06,
      "loss": 0.4955,
      "step": 4884
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0546868775547618,
      "learning_rate": 1.142149151461081e-06,
      "loss": 0.5916,
      "step": 4885
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0184133691652397,
      "learning_rate": 1.1404895125335859e-06,
      "loss": 0.5497,
      "step": 4886
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0336824100110868,
      "learning_rate": 1.1388309250431363e-06,
      "loss": 0.614,
      "step": 4887
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9875244583500011,
      "learning_rate": 1.1371733894415748e-06,
      "loss": 0.5408,
      "step": 4888
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1588956728125823,
      "learning_rate": 1.1355169061804632e-06,
      "loss": 0.6184,
      "step": 4889
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.994247924798844,
      "learning_rate": 1.1338614757110706e-06,
      "loss": 0.6087,
      "step": 4890
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.047226355969621,
      "learning_rate": 1.1322070984843837e-06,
      "loss": 0.5653,
      "step": 4891
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0332391159342798,
      "learning_rate": 1.1305537749510993e-06,
      "loss": 0.6379,
      "step": 4892
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9083993973695406,
      "learning_rate": 1.128901505561627e-06,
      "loss": 0.5347,
      "step": 4893
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0579780289118221,
      "learning_rate": 1.1272502907660937e-06,
      "loss": 0.5138,
      "step": 4894
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9773930014324707,
      "learning_rate": 1.1256001310143327e-06,
      "loss": 0.5644,
      "step": 4895
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6356582770548203,
      "learning_rate": 1.1239510267558962e-06,
      "loss": 0.4536,
      "step": 4896
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9905671215844932,
      "learning_rate": 1.1223029784400436e-06,
      "loss": 0.5845,
      "step": 4897
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0013842924128515,
      "learning_rate": 1.120655986515748e-06,
      "loss": 0.5283,
      "step": 4898
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9990999749162804,
      "learning_rate": 1.1190100514316977e-06,
      "loss": 0.5798,
      "step": 4899
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9457899880933173,
      "learning_rate": 1.1173651736362889e-06,
      "loss": 0.5252,
      "step": 4900
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9810971111077363,
      "learning_rate": 1.1157213535776312e-06,
      "loss": 0.5067,
      "step": 4901
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9905835598361677,
      "learning_rate": 1.114078591703548e-06,
      "loss": 0.5389,
      "step": 4902
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9524603670153281,
      "learning_rate": 1.1124368884615748e-06,
      "loss": 0.5458,
      "step": 4903
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0509611355104258,
      "learning_rate": 1.1107962442989518e-06,
      "loss": 0.5741,
      "step": 4904
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0205756768538135,
      "learning_rate": 1.1091566596626384e-06,
      "loss": 0.5731,
      "step": 4905
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9603002272771002,
      "learning_rate": 1.1075181349993042e-06,
      "loss": 0.583,
      "step": 4906
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9621613892360988,
      "learning_rate": 1.1058806707553266e-06,
      "loss": 0.5466,
      "step": 4907
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9918935232987662,
      "learning_rate": 1.104244267376795e-06,
      "loss": 0.6225,
      "step": 4908
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0730732088409445,
      "learning_rate": 1.1026089253095134e-06,
      "loss": 0.6161,
      "step": 4909
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.037998289469748,
      "learning_rate": 1.1009746449989916e-06,
      "loss": 0.6627,
      "step": 4910
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0111287066663015,
      "learning_rate": 1.0993414268904552e-06,
      "loss": 0.5532,
      "step": 4911
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0122731097177788,
      "learning_rate": 1.0977092714288345e-06,
      "loss": 0.5719,
      "step": 4912
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9441848115210496,
      "learning_rate": 1.0960781790587776e-06,
      "loss": 0.6116,
      "step": 4913
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9630579437179908,
      "learning_rate": 1.0944481502246368e-06,
      "loss": 0.5264,
      "step": 4914
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.118526695402412,
      "learning_rate": 1.0928191853704757e-06,
      "loss": 0.5988,
      "step": 4915
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0739731711401403,
      "learning_rate": 1.0911912849400712e-06,
      "loss": 0.6152,
      "step": 4916
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9748309955896621,
      "learning_rate": 1.089564449376907e-06,
      "loss": 0.5271,
      "step": 4917
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0871552072922026,
      "learning_rate": 1.0879386791241797e-06,
      "loss": 0.6041,
      "step": 4918
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9902936428489488,
      "learning_rate": 1.086313974624793e-06,
      "loss": 0.5552,
      "step": 4919
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9883066875537733,
      "learning_rate": 1.0846903363213595e-06,
      "loss": 0.5691,
      "step": 4920
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9851862367287978,
      "learning_rate": 1.083067764656206e-06,
      "loss": 0.5583,
      "step": 4921
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6792032706664342,
      "learning_rate": 1.0814462600713642e-06,
      "loss": 0.4752,
      "step": 4922
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0477393794242218,
      "learning_rate": 1.0798258230085756e-06,
      "loss": 0.6159,
      "step": 4923
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.090374123848842,
      "learning_rate": 1.078206453909293e-06,
      "loss": 0.6091,
      "step": 4924
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.082930731788072,
      "learning_rate": 1.0765881532146793e-06,
      "loss": 0.5768,
      "step": 4925
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.011558509102789,
      "learning_rate": 1.0749709213656001e-06,
      "loss": 0.6002,
      "step": 4926
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0534283111384604,
      "learning_rate": 1.0733547588026355e-06,
      "loss": 0.6246,
      "step": 4927
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0503463487545455,
      "learning_rate": 1.071739665966075e-06,
      "loss": 0.5964,
      "step": 4928
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.952565066555294,
      "learning_rate": 1.0701256432959123e-06,
      "loss": 0.4964,
      "step": 4929
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0351319118407494,
      "learning_rate": 1.0685126912318511e-06,
      "loss": 0.5809,
      "step": 4930
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0375116794214496,
      "learning_rate": 1.0669008102133044e-06,
      "loss": 0.6148,
      "step": 4931
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0185609557380317,
      "learning_rate": 1.0652900006793953e-06,
      "loss": 0.5386,
      "step": 4932
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0583794530259187,
      "learning_rate": 1.0636802630689508e-06,
      "loss": 0.6591,
      "step": 4933
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9618537872001663,
      "learning_rate": 1.0620715978205066e-06,
      "loss": 0.4807,
      "step": 4934
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0793447640657676,
      "learning_rate": 1.0604640053723098e-06,
      "loss": 0.5949,
      "step": 4935
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9692603450797759,
      "learning_rate": 1.058857486162312e-06,
      "loss": 0.4843,
      "step": 4936
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9515855143046564,
      "learning_rate": 1.0572520406281716e-06,
      "loss": 0.4982,
      "step": 4937
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8428090222497984,
      "learning_rate": 1.0556476692072598e-06,
      "loss": 0.4593,
      "step": 4938
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1489433698340852,
      "learning_rate": 1.0540443723366478e-06,
      "loss": 0.6145,
      "step": 4939
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.040249326547989,
      "learning_rate": 1.0524421504531212e-06,
      "loss": 0.5862,
      "step": 4940
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1446455674362987,
      "learning_rate": 1.0508410039931683e-06,
      "loss": 0.5761,
      "step": 4941
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0084965514507758,
      "learning_rate": 1.0492409333929833e-06,
      "loss": 0.5588,
      "step": 4942
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0031684686432332,
      "learning_rate": 1.0476419390884723e-06,
      "loss": 0.5603,
      "step": 4943
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9861083791968547,
      "learning_rate": 1.046044021515245e-06,
      "loss": 0.5194,
      "step": 4944
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9840869949261994,
      "learning_rate": 1.0444471811086166e-06,
      "loss": 0.5077,
      "step": 4945
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0438772357655342,
      "learning_rate": 1.0428514183036109e-06,
      "loss": 0.5302,
      "step": 4946
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6761246331243677,
      "learning_rate": 1.0412567335349616e-06,
      "loss": 0.5009,
      "step": 4947
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0899582308742042,
      "learning_rate": 1.0396631272370982e-06,
      "loss": 0.5148,
      "step": 4948
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9175558586235654,
      "learning_rate": 1.0380705998441664e-06,
      "loss": 0.4998,
      "step": 4949
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0027869980696893,
      "learning_rate": 1.0364791517900164e-06,
      "loss": 0.5917,
      "step": 4950
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0236817363530604,
      "learning_rate": 1.0348887835082e-06,
      "loss": 0.5375,
      "step": 4951
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1092562323388573,
      "learning_rate": 1.0332994954319763e-06,
      "loss": 0.6763,
      "step": 4952
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.096851544727364,
      "learning_rate": 1.031711287994313e-06,
      "loss": 0.6057,
      "step": 4953
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0081914274933619,
      "learning_rate": 1.0301241616278845e-06,
      "loss": 0.5634,
      "step": 4954
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9401353287228967,
      "learning_rate": 1.0285381167650615e-06,
      "loss": 0.5424,
      "step": 4955
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.081096666706904,
      "learning_rate": 1.0269531538379295e-06,
      "loss": 0.6202,
      "step": 4956
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0396890556432319,
      "learning_rate": 1.0253692732782778e-06,
      "loss": 0.5984,
      "step": 4957
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.036942196192119,
      "learning_rate": 1.0237864755175969e-06,
      "loss": 0.6014,
      "step": 4958
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9657725184447055,
      "learning_rate": 1.022204760987084e-06,
      "loss": 0.531,
      "step": 4959
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6186170350549213,
      "learning_rate": 1.0206241301176439e-06,
      "loss": 0.5075,
      "step": 4960
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9992049887019192,
      "learning_rate": 1.0190445833398814e-06,
      "loss": 0.6213,
      "step": 4961
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0269592697199725,
      "learning_rate": 1.0174661210841119e-06,
      "loss": 0.6366,
      "step": 4962
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0058328402061356,
      "learning_rate": 1.0158887437803499e-06,
      "loss": 0.6275,
      "step": 4963
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0966748109436368,
      "learning_rate": 1.0143124518583158e-06,
      "loss": 0.6268,
      "step": 4964
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1438911659012767,
      "learning_rate": 1.012737245747437e-06,
      "loss": 0.631,
      "step": 4965
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1132784331554506,
      "learning_rate": 1.0111631258768416e-06,
      "loss": 0.6037,
      "step": 4966
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.995544545072958,
      "learning_rate": 1.0095900926753632e-06,
      "loss": 0.5652,
      "step": 4967
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0700174365312556,
      "learning_rate": 1.0080181465715394e-06,
      "loss": 0.5622,
      "step": 4968
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0168888905909792,
      "learning_rate": 1.0064472879936132e-06,
      "loss": 0.5884,
      "step": 4969
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9641316570333548,
      "learning_rate": 1.0048775173695285e-06,
      "loss": 0.4881,
      "step": 4970
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9930317705232964,
      "learning_rate": 1.0033088351269338e-06,
      "loss": 0.583,
      "step": 4971
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9227064522184814,
      "learning_rate": 1.0017412416931826e-06,
      "loss": 0.5044,
      "step": 4972
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9721127347044944,
      "learning_rate": 1.0001747374953297e-06,
      "loss": 0.5121,
      "step": 4973
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9301788476217522,
      "learning_rate": 9.986093229601328e-07,
      "loss": 0.463,
      "step": 4974
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0340793639657742,
      "learning_rate": 9.970449985140557e-07,
      "loss": 0.5298,
      "step": 4975
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0368418654616984,
      "learning_rate": 9.954817645832654e-07,
      "loss": 0.6128,
      "step": 4976
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1220250053136023,
      "learning_rate": 9.939196215936253e-07,
      "loss": 0.6609,
      "step": 4977
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8750649879507977,
      "learning_rate": 9.923585699707084e-07,
      "loss": 0.534,
      "step": 4978
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0870802803605275,
      "learning_rate": 9.907986101397898e-07,
      "loss": 0.6635,
      "step": 4979
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0473413486167582,
      "learning_rate": 9.892397425258437e-07,
      "loss": 0.6698,
      "step": 4980
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0635177373556555,
      "learning_rate": 9.876819675535477e-07,
      "loss": 0.6301,
      "step": 4981
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9979297387600963,
      "learning_rate": 9.861252856472857e-07,
      "loss": 0.5618,
      "step": 4982
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6278419211672942,
      "learning_rate": 9.845696972311385e-07,
      "loss": 0.4588,
      "step": 4983
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9704510060528942,
      "learning_rate": 9.830152027288907e-07,
      "loss": 0.5604,
      "step": 4984
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9323322040649142,
      "learning_rate": 9.81461802564032e-07,
      "loss": 0.4842,
      "step": 4985
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9540889196475997,
      "learning_rate": 9.799094971597483e-07,
      "loss": 0.5627,
      "step": 4986
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9578975960902553,
      "learning_rate": 9.783582869389336e-07,
      "loss": 0.5684,
      "step": 4987
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9808323851785565,
      "learning_rate": 9.768081723241785e-07,
      "loss": 0.5474,
      "step": 4988
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.030864255329231,
      "learning_rate": 9.752591537377758e-07,
      "loss": 0.5895,
      "step": 4989
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9439998640126077,
      "learning_rate": 9.737112316017221e-07,
      "loss": 0.5285,
      "step": 4990
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9901448231053444,
      "learning_rate": 9.72164406337716e-07,
      "loss": 0.5821,
      "step": 4991
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.011354174543654,
      "learning_rate": 9.706186783671535e-07,
      "loss": 0.6013,
      "step": 4992
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9880411345959761,
      "learning_rate": 9.69074048111132e-07,
      "loss": 0.5288,
      "step": 4993
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9702355136333534,
      "learning_rate": 9.675305159904546e-07,
      "loss": 0.5088,
      "step": 4994
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0492935030882344,
      "learning_rate": 9.659880824256202e-07,
      "loss": 0.6256,
      "step": 4995
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1357593399319108,
      "learning_rate": 9.644467478368286e-07,
      "loss": 0.6289,
      "step": 4996
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1038472585163972,
      "learning_rate": 9.629065126439842e-07,
      "loss": 0.6172,
      "step": 4997
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.054392538073765,
      "learning_rate": 9.61367377266691e-07,
      "loss": 0.6636,
      "step": 4998
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9564277456754303,
      "learning_rate": 9.59829342124247e-07,
      "loss": 0.5632,
      "step": 4999
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0717058676526376,
      "learning_rate": 9.582924076356587e-07,
      "loss": 0.6295,
      "step": 5000
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0453968510410017,
      "learning_rate": 9.5675657421963e-07,
      "loss": 0.5734,
      "step": 5001
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0753582593412454,
      "learning_rate": 9.552218422945636e-07,
      "loss": 0.6152,
      "step": 5002
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0153285747586902,
      "learning_rate": 9.536882122785602e-07,
      "loss": 0.5109,
      "step": 5003
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0645105974326663,
      "learning_rate": 9.52155684589427e-07,
      "loss": 0.5695,
      "step": 5004
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.083354189531401,
      "learning_rate": 9.506242596446641e-07,
      "loss": 0.5537,
      "step": 5005
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0327720156730036,
      "learning_rate": 9.490939378614739e-07,
      "loss": 0.6127,
      "step": 5006
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9695052713525548,
      "learning_rate": 9.4756471965676e-07,
      "loss": 0.5779,
      "step": 5007
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6269649875165874,
      "learning_rate": 9.46036605447121e-07,
      "loss": 0.4677,
      "step": 5008
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9839377671206231,
      "learning_rate": 9.445095956488604e-07,
      "loss": 0.6018,
      "step": 5009
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0219459590451694,
      "learning_rate": 9.42983690677975e-07,
      "loss": 0.6332,
      "step": 5010
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9418588904057287,
      "learning_rate": 9.414588909501654e-07,
      "loss": 0.5481,
      "step": 5011
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0355557497987793,
      "learning_rate": 9.399351968808285e-07,
      "loss": 0.6182,
      "step": 5012
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0228155541236919,
      "learning_rate": 9.384126088850592e-07,
      "loss": 0.6801,
      "step": 5013
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0183066234612663,
      "learning_rate": 9.368911273776543e-07,
      "loss": 0.539,
      "step": 5014
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.055635608981327,
      "learning_rate": 9.35370752773106e-07,
      "loss": 0.6091,
      "step": 5015
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.068706245982093,
      "learning_rate": 9.338514854856073e-07,
      "loss": 0.5586,
      "step": 5016
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.03503789870056,
      "learning_rate": 9.323333259290484e-07,
      "loss": 0.6088,
      "step": 5017
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0312428312048019,
      "learning_rate": 9.308162745170163e-07,
      "loss": 0.5213,
      "step": 5018
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0199959250533983,
      "learning_rate": 9.293003316627985e-07,
      "loss": 0.51,
      "step": 5019
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.07047894711648,
      "learning_rate": 9.277854977793827e-07,
      "loss": 0.5867,
      "step": 5020
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0438064132575007,
      "learning_rate": 9.262717732794457e-07,
      "loss": 0.5915,
      "step": 5021
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1378786486287622,
      "learning_rate": 9.247591585753707e-07,
      "loss": 0.6518,
      "step": 5022
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0239993100400713,
      "learning_rate": 9.232476540792367e-07,
      "loss": 0.578,
      "step": 5023
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0379823234146377,
      "learning_rate": 9.217372602028185e-07,
      "loss": 0.5991,
      "step": 5024
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0185060104501789,
      "learning_rate": 9.202279773575873e-07,
      "loss": 0.5003,
      "step": 5025
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.025473491498839,
      "learning_rate": 9.187198059547153e-07,
      "loss": 0.5897,
      "step": 5026
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0339441141660415,
      "learning_rate": 9.172127464050701e-07,
      "loss": 0.5715,
      "step": 5027
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0518444237270355,
      "learning_rate": 9.157067991192137e-07,
      "loss": 0.6409,
      "step": 5028
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.111267927420634,
      "learning_rate": 9.14201964507409e-07,
      "loss": 0.5596,
      "step": 5029
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1076686365906552,
      "learning_rate": 9.126982429796172e-07,
      "loss": 0.607,
      "step": 5030
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0227393228347113,
      "learning_rate": 9.111956349454904e-07,
      "loss": 0.5185,
      "step": 5031
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0110920105080827,
      "learning_rate": 9.0969414081438e-07,
      "loss": 0.5662,
      "step": 5032
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9992000953081548,
      "learning_rate": 9.081937609953367e-07,
      "loss": 0.5491,
      "step": 5033
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0841741332390904,
      "learning_rate": 9.066944958971046e-07,
      "loss": 0.613,
      "step": 5034
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.919752911436365,
      "learning_rate": 9.051963459281232e-07,
      "loss": 0.5204,
      "step": 5035
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0950299932441798,
      "learning_rate": 9.03699311496532e-07,
      "loss": 0.6478,
      "step": 5036
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9942210681481561,
      "learning_rate": 9.022033930101625e-07,
      "loss": 0.594,
      "step": 5037
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0162820108552946,
      "learning_rate": 9.007085908765467e-07,
      "loss": 0.5698,
      "step": 5038
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0505083274964164,
      "learning_rate": 8.992149055029081e-07,
      "loss": 0.5232,
      "step": 5039
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9402461886859322,
      "learning_rate": 8.977223372961663e-07,
      "loss": 0.4646,
      "step": 5040
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0404755761060818,
      "learning_rate": 8.962308866629416e-07,
      "loss": 0.6223,
      "step": 5041
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.017697041412758,
      "learning_rate": 8.947405540095444e-07,
      "loss": 0.567,
      "step": 5042
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0305988427828752,
      "learning_rate": 8.932513397419812e-07,
      "loss": 0.5982,
      "step": 5043
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0136835639863133,
      "learning_rate": 8.917632442659563e-07,
      "loss": 0.4931,
      "step": 5044
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9680439121932444,
      "learning_rate": 8.902762679868687e-07,
      "loss": 0.5431,
      "step": 5045
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7138367926788204,
      "learning_rate": 8.887904113098111e-07,
      "loss": 0.5283,
      "step": 5046
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0111051074043325,
      "learning_rate": 8.873056746395703e-07,
      "loss": 0.658,
      "step": 5047
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.136741552687729,
      "learning_rate": 8.858220583806309e-07,
      "loss": 0.7093,
      "step": 5048
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0078540802622933,
      "learning_rate": 8.843395629371738e-07,
      "loss": 0.6422,
      "step": 5049
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0496875565093102,
      "learning_rate": 8.828581887130655e-07,
      "loss": 0.5896,
      "step": 5050
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0155197565061074,
      "learning_rate": 8.813779361118763e-07,
      "loss": 0.5786,
      "step": 5051
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1143675949384952,
      "learning_rate": 8.79898805536869e-07,
      "loss": 0.5729,
      "step": 5052
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0613994944710665,
      "learning_rate": 8.784207973909986e-07,
      "loss": 0.5842,
      "step": 5053
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0137764686236503,
      "learning_rate": 8.769439120769135e-07,
      "loss": 0.6061,
      "step": 5054
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.027095287741836,
      "learning_rate": 8.754681499969608e-07,
      "loss": 0.5739,
      "step": 5055
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0379607098038166,
      "learning_rate": 8.739935115531772e-07,
      "loss": 0.6461,
      "step": 5056
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.886098059639951,
      "learning_rate": 8.725199971472942e-07,
      "loss": 0.4745,
      "step": 5057
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0076398706435228,
      "learning_rate": 8.7104760718074e-07,
      "loss": 0.4842,
      "step": 5058
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0517277608006768,
      "learning_rate": 8.69576342054631e-07,
      "loss": 0.5611,
      "step": 5059
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0821786247663827,
      "learning_rate": 8.681062021697839e-07,
      "loss": 0.6009,
      "step": 5060
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0478631117812514,
      "learning_rate": 8.666371879267038e-07,
      "loss": 0.5638,
      "step": 5061
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.075180113684228,
      "learning_rate": 8.65169299725589e-07,
      "loss": 0.6464,
      "step": 5062
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.085654015984366,
      "learning_rate": 8.637025379663355e-07,
      "loss": 0.6513,
      "step": 5063
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0095910583274752,
      "learning_rate": 8.622369030485283e-07,
      "loss": 0.5331,
      "step": 5064
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9785614591180458,
      "learning_rate": 8.607723953714453e-07,
      "loss": 0.615,
      "step": 5065
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.984089999876762,
      "learning_rate": 8.593090153340611e-07,
      "loss": 0.5328,
      "step": 5066
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9109170058534404,
      "learning_rate": 8.578467633350407e-07,
      "loss": 0.547,
      "step": 5067
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.067273200051999,
      "learning_rate": 8.56385639772741e-07,
      "loss": 0.4914,
      "step": 5068
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9935404959291039,
      "learning_rate": 8.549256450452109e-07,
      "loss": 0.5732,
      "step": 5069
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0340418370817044,
      "learning_rate": 8.534667795501955e-07,
      "loss": 0.5325,
      "step": 5070
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9849397104151747,
      "learning_rate": 8.520090436851314e-07,
      "loss": 0.5574,
      "step": 5071
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0453074640754032,
      "learning_rate": 8.505524378471408e-07,
      "loss": 0.6129,
      "step": 5072
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9978611782867323,
      "learning_rate": 8.490969624330469e-07,
      "loss": 0.5705,
      "step": 5073
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9486073749000541,
      "learning_rate": 8.476426178393621e-07,
      "loss": 0.4886,
      "step": 5074
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.039823271279133,
      "learning_rate": 8.461894044622882e-07,
      "loss": 0.5588,
      "step": 5075
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7071239991668153,
      "learning_rate": 8.447373226977201e-07,
      "loss": 0.4618,
      "step": 5076
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0809769701938214,
      "learning_rate": 8.432863729412466e-07,
      "loss": 0.5829,
      "step": 5077
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0280928437236523,
      "learning_rate": 8.418365555881458e-07,
      "loss": 0.5774,
      "step": 5078
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1143425231876194,
      "learning_rate": 8.403878710333868e-07,
      "loss": 0.5906,
      "step": 5079
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.02656261952526,
      "learning_rate": 8.389403196716328e-07,
      "loss": 0.5758,
      "step": 5080
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1115519067503328,
      "learning_rate": 8.37493901897235e-07,
      "loss": 0.59,
      "step": 5081
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.012005791942428,
      "learning_rate": 8.360486181042399e-07,
      "loss": 0.5591,
      "step": 5082
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1303983119964944,
      "learning_rate": 8.346044686863808e-07,
      "loss": 0.587,
      "step": 5083
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6726855383353931,
      "learning_rate": 8.331614540370836e-07,
      "loss": 0.4923,
      "step": 5084
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0479890656908446,
      "learning_rate": 8.317195745494666e-07,
      "loss": 0.6057,
      "step": 5085
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0277858972007798,
      "learning_rate": 8.302788306163373e-07,
      "loss": 0.6242,
      "step": 5086
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0169531995334915,
      "learning_rate": 8.288392226301917e-07,
      "loss": 0.6007,
      "step": 5087
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0837306325390166,
      "learning_rate": 8.27400750983221e-07,
      "loss": 0.5331,
      "step": 5088
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9746158882202549,
      "learning_rate": 8.259634160673052e-07,
      "loss": 0.556,
      "step": 5089
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0441735335881588,
      "learning_rate": 8.245272182740133e-07,
      "loss": 0.6728,
      "step": 5090
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0151290827942323,
      "learning_rate": 8.230921579946027e-07,
      "loss": 0.5634,
      "step": 5091
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0652104511619458,
      "learning_rate": 8.21658235620027e-07,
      "loss": 0.4907,
      "step": 5092
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1011730919467875,
      "learning_rate": 8.202254515409247e-07,
      "loss": 0.5704,
      "step": 5093
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9569429331107602,
      "learning_rate": 8.187938061476242e-07,
      "loss": 0.5973,
      "step": 5094
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0668874503317447,
      "learning_rate": 8.173632998301468e-07,
      "loss": 0.591,
      "step": 5095
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.027311198385021,
      "learning_rate": 8.159339329782023e-07,
      "loss": 0.556,
      "step": 5096
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0591581798481031,
      "learning_rate": 8.145057059811895e-07,
      "loss": 0.603,
      "step": 5097
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.050515987303814,
      "learning_rate": 8.130786192281947e-07,
      "loss": 0.5685,
      "step": 5098
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0928953415925775,
      "learning_rate": 8.116526731079982e-07,
      "loss": 0.6424,
      "step": 5099
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9548030843388295,
      "learning_rate": 8.102278680090664e-07,
      "loss": 0.5206,
      "step": 5100
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6348145776017352,
      "learning_rate": 8.088042043195538e-07,
      "loss": 0.461,
      "step": 5101
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0000483019142208,
      "learning_rate": 8.073816824273078e-07,
      "loss": 0.551,
      "step": 5102
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0372890415216258,
      "learning_rate": 8.059603027198609e-07,
      "loss": 0.5141,
      "step": 5103
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9271157683425791,
      "learning_rate": 8.045400655844382e-07,
      "loss": 0.4809,
      "step": 5104
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.008115362066479,
      "learning_rate": 8.031209714079496e-07,
      "loss": 0.5326,
      "step": 5105
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9205227161731733,
      "learning_rate": 8.017030205769955e-07,
      "loss": 0.5881,
      "step": 5106
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.977175772526508,
      "learning_rate": 8.002862134778661e-07,
      "loss": 0.5581,
      "step": 5107
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9667871281550026,
      "learning_rate": 7.988705504965372e-07,
      "loss": 0.6322,
      "step": 5108
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0106334631779426,
      "learning_rate": 7.974560320186759e-07,
      "loss": 0.5755,
      "step": 5109
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0057387657596193,
      "learning_rate": 7.960426584296338e-07,
      "loss": 0.541,
      "step": 5110
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9860136698805336,
      "learning_rate": 7.946304301144564e-07,
      "loss": 0.5262,
      "step": 5111
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0051271369597181,
      "learning_rate": 7.932193474578709e-07,
      "loss": 0.6064,
      "step": 5112
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0697768423670966,
      "learning_rate": 7.918094108442947e-07,
      "loss": 0.6358,
      "step": 5113
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.039532736858145,
      "learning_rate": 7.904006206578358e-07,
      "loss": 0.5825,
      "step": 5114
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9846617040753357,
      "learning_rate": 7.889929772822857e-07,
      "loss": 0.5077,
      "step": 5115
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0837520996474137,
      "learning_rate": 7.875864811011247e-07,
      "loss": 0.6116,
      "step": 5116
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0520145630122697,
      "learning_rate": 7.861811324975221e-07,
      "loss": 0.6123,
      "step": 5117
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1221305956199863,
      "learning_rate": 7.847769318543346e-07,
      "loss": 0.6041,
      "step": 5118
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0052952155167758,
      "learning_rate": 7.833738795541046e-07,
      "loss": 0.5577,
      "step": 5119
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.028942333888086,
      "learning_rate": 7.819719759790606e-07,
      "loss": 0.6149,
      "step": 5120
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0188849906539101,
      "learning_rate": 7.805712215111216e-07,
      "loss": 0.5782,
      "step": 5121
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0703155929777164,
      "learning_rate": 7.791716165318913e-07,
      "loss": 0.5724,
      "step": 5122
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0205935358746852,
      "learning_rate": 7.777731614226596e-07,
      "loss": 0.5549,
      "step": 5123
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0734011355527322,
      "learning_rate": 7.763758565644064e-07,
      "loss": 0.5708,
      "step": 5124
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0720785869308467,
      "learning_rate": 7.749797023377931e-07,
      "loss": 0.5936,
      "step": 5125
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.990457673213393,
      "learning_rate": 7.735846991231738e-07,
      "loss": 0.5729,
      "step": 5126
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0477546517006748,
      "learning_rate": 7.721908473005829e-07,
      "loss": 0.5654,
      "step": 5127
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.051014318195096,
      "learning_rate": 7.707981472497467e-07,
      "loss": 0.5449,
      "step": 5128
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.115453626012281,
      "learning_rate": 7.694065993500732e-07,
      "loss": 0.6238,
      "step": 5129
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9984220253090474,
      "learning_rate": 7.680162039806588e-07,
      "loss": 0.5176,
      "step": 5130
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.016093153709853,
      "learning_rate": 7.666269615202865e-07,
      "loss": 0.5862,
      "step": 5131
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0499406862219314,
      "learning_rate": 7.652388723474224e-07,
      "loss": 0.6143,
      "step": 5132
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9165837686305148,
      "learning_rate": 7.638519368402225e-07,
      "loss": 0.451,
      "step": 5133
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.078373030892962,
      "learning_rate": 7.62466155376525e-07,
      "loss": 0.5965,
      "step": 5134
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0323207238559315,
      "learning_rate": 7.61081528333854e-07,
      "loss": 0.6433,
      "step": 5135
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9599229560634489,
      "learning_rate": 7.596980560894224e-07,
      "loss": 0.6164,
      "step": 5136
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9541945120732931,
      "learning_rate": 7.583157390201246e-07,
      "loss": 0.5529,
      "step": 5137
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9416686763491351,
      "learning_rate": 7.56934577502541e-07,
      "loss": 0.4379,
      "step": 5138
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9883795472859372,
      "learning_rate": 7.555545719129398e-07,
      "loss": 0.4941,
      "step": 5139
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9695803875549589,
      "learning_rate": 7.541757226272744e-07,
      "loss": 0.5427,
      "step": 5140
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9660900829162621,
      "learning_rate": 7.527980300211762e-07,
      "loss": 0.5015,
      "step": 5141
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9979269322415791,
      "learning_rate": 7.514214944699694e-07,
      "loss": 0.5207,
      "step": 5142
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.012012508514017,
      "learning_rate": 7.500461163486616e-07,
      "loss": 0.588,
      "step": 5143
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0523008410780177,
      "learning_rate": 7.486718960319428e-07,
      "loss": 0.6478,
      "step": 5144
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.066941122629187,
      "learning_rate": 7.472988338941861e-07,
      "loss": 0.6042,
      "step": 5145
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6423651733070286,
      "learning_rate": 7.459269303094552e-07,
      "loss": 0.4425,
      "step": 5146
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0770133591551112,
      "learning_rate": 7.445561856514916e-07,
      "loss": 0.5971,
      "step": 5147
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0174484184860932,
      "learning_rate": 7.431866002937254e-07,
      "loss": 0.5339,
      "step": 5148
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0791158691756308,
      "learning_rate": 7.41818174609269e-07,
      "loss": 0.6028,
      "step": 5149
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0049929975606402,
      "learning_rate": 7.404509089709194e-07,
      "loss": 0.5719,
      "step": 5150
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9952842135292382,
      "learning_rate": 7.390848037511578e-07,
      "loss": 0.5508,
      "step": 5151
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0396021794573012,
      "learning_rate": 7.377198593221474e-07,
      "loss": 0.6277,
      "step": 5152
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0124824095086598,
      "learning_rate": 7.363560760557392e-07,
      "loss": 0.6066,
      "step": 5153
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.14007864059285,
      "learning_rate": 7.349934543234621e-07,
      "loss": 0.6046,
      "step": 5154
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0547704247584988,
      "learning_rate": 7.336319944965353e-07,
      "loss": 0.6492,
      "step": 5155
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0498673742591713,
      "learning_rate": 7.32271696945856e-07,
      "loss": 0.6493,
      "step": 5156
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9812665219738321,
      "learning_rate": 7.30912562042006e-07,
      "loss": 0.613,
      "step": 5157
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0778590053183552,
      "learning_rate": 7.295545901552536e-07,
      "loss": 0.6437,
      "step": 5158
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1156138773829636,
      "learning_rate": 7.281977816555463e-07,
      "loss": 0.5987,
      "step": 5159
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0170515620023057,
      "learning_rate": 7.268421369125145e-07,
      "loss": 0.5749,
      "step": 5160
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0234315438274797,
      "learning_rate": 7.254876562954755e-07,
      "loss": 0.5794,
      "step": 5161
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.01689867505469,
      "learning_rate": 7.241343401734285e-07,
      "loss": 0.5766,
      "step": 5162
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6565716326712019,
      "learning_rate": 7.2278218891505e-07,
      "loss": 0.4686,
      "step": 5163
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9792088246171785,
      "learning_rate": 7.214312028887055e-07,
      "loss": 0.5413,
      "step": 5164
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6593046810552637,
      "learning_rate": 7.20081382462442e-07,
      "loss": 0.4763,
      "step": 5165
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.928707873666654,
      "learning_rate": 7.187327280039863e-07,
      "loss": 0.5257,
      "step": 5166
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9576603355790668,
      "learning_rate": 7.173852398807485e-07,
      "loss": 0.5072,
      "step": 5167
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0605474922685603,
      "learning_rate": 7.160389184598221e-07,
      "loss": 0.5857,
      "step": 5168
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0224333894207813,
      "learning_rate": 7.146937641079849e-07,
      "loss": 0.5833,
      "step": 5169
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6676944675232807,
      "learning_rate": 7.133497771916886e-07,
      "loss": 0.472,
      "step": 5170
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.122445741067534,
      "learning_rate": 7.120069580770755e-07,
      "loss": 0.6159,
      "step": 5171
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0375104525408834,
      "learning_rate": 7.10665307129968e-07,
      "loss": 0.6538,
      "step": 5172
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9761671422683879,
      "learning_rate": 7.093248247158663e-07,
      "loss": 0.5181,
      "step": 5173
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9456267845536658,
      "learning_rate": 7.079855111999545e-07,
      "loss": 0.5611,
      "step": 5174
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0457549808870232,
      "learning_rate": 7.066473669471008e-07,
      "loss": 0.5433,
      "step": 5175
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.095570837836015,
      "learning_rate": 7.053103923218501e-07,
      "loss": 0.6666,
      "step": 5176
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0542381801141385,
      "learning_rate": 7.039745876884335e-07,
      "loss": 0.547,
      "step": 5177
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1079500406046754,
      "learning_rate": 7.026399534107603e-07,
      "loss": 0.5841,
      "step": 5178
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9951796445141703,
      "learning_rate": 7.013064898524196e-07,
      "loss": 0.5908,
      "step": 5179
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9310716384104486,
      "learning_rate": 6.99974197376686e-07,
      "loss": 0.5281,
      "step": 5180
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0804487572968202,
      "learning_rate": 6.986430763465124e-07,
      "loss": 0.6431,
      "step": 5181
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.050529711038866,
      "learning_rate": 6.973131271245315e-07,
      "loss": 0.5737,
      "step": 5182
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0721895807143778,
      "learning_rate": 6.95984350073059e-07,
      "loss": 0.5865,
      "step": 5183
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.966865845983161,
      "learning_rate": 6.94656745554092e-07,
      "loss": 0.5353,
      "step": 5184
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9578115026971961,
      "learning_rate": 6.933303139293035e-07,
      "loss": 0.5209,
      "step": 5185
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.044091653804256,
      "learning_rate": 6.92005055560051e-07,
      "loss": 0.61,
      "step": 5186
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9787382138261209,
      "learning_rate": 6.906809708073736e-07,
      "loss": 0.597,
      "step": 5187
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9779440823391227,
      "learning_rate": 6.893580600319865e-07,
      "loss": 0.581,
      "step": 5188
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0423914030143775,
      "learning_rate": 6.880363235942861e-07,
      "loss": 0.5778,
      "step": 5189
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.132869353505299,
      "learning_rate": 6.867157618543513e-07,
      "loss": 0.6691,
      "step": 5190
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0642442805174035,
      "learning_rate": 6.853963751719417e-07,
      "loss": 0.6326,
      "step": 5191
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0168163899897797,
      "learning_rate": 6.840781639064897e-07,
      "loss": 0.5739,
      "step": 5192
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9972374342589777,
      "learning_rate": 6.827611284171154e-07,
      "loss": 0.6182,
      "step": 5193
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9677317311000039,
      "learning_rate": 6.814452690626161e-07,
      "loss": 0.5708,
      "step": 5194
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0617202236943117,
      "learning_rate": 6.801305862014667e-07,
      "loss": 0.6479,
      "step": 5195
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.107700966398236,
      "learning_rate": 6.788170801918231e-07,
      "loss": 0.5896,
      "step": 5196
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.678818516525114,
      "learning_rate": 6.775047513915218e-07,
      "loss": 0.4881,
      "step": 5197
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9851041418772838,
      "learning_rate": 6.76193600158076e-07,
      "loss": 0.5657,
      "step": 5198
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0029237829470858,
      "learning_rate": 6.748836268486797e-07,
      "loss": 0.5668,
      "step": 5199
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0000137716349071,
      "learning_rate": 6.735748318202062e-07,
      "loss": 0.5519,
      "step": 5200
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9298971378669835,
      "learning_rate": 6.722672154292065e-07,
      "loss": 0.491,
      "step": 5201
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6469655861912761,
      "learning_rate": 6.709607780319133e-07,
      "loss": 0.4532,
      "step": 5202
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0692189987071181,
      "learning_rate": 6.696555199842347e-07,
      "loss": 0.6318,
      "step": 5203
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9564436171121385,
      "learning_rate": 6.683514416417574e-07,
      "loss": 0.5563,
      "step": 5204
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9672300721502858,
      "learning_rate": 6.670485433597507e-07,
      "loss": 0.5317,
      "step": 5205
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9490091749015241,
      "learning_rate": 6.657468254931615e-07,
      "loss": 0.5369,
      "step": 5206
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0270188757465508,
      "learning_rate": 6.644462883966085e-07,
      "loss": 0.5934,
      "step": 5207
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.001116697757867,
      "learning_rate": 6.631469324243978e-07,
      "loss": 0.5315,
      "step": 5208
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1399380939896522,
      "learning_rate": 6.618487579305089e-07,
      "loss": 0.6084,
      "step": 5209
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9697330846666572,
      "learning_rate": 6.605517652686005e-07,
      "loss": 0.4909,
      "step": 5210
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.022122055597291,
      "learning_rate": 6.59255954792008e-07,
      "loss": 0.5588,
      "step": 5211
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.96744048994743,
      "learning_rate": 6.579613268537466e-07,
      "loss": 0.5232,
      "step": 5212
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.983173415945006,
      "learning_rate": 6.566678818065108e-07,
      "loss": 0.6211,
      "step": 5213
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9599810272101134,
      "learning_rate": 6.553756200026668e-07,
      "loss": 0.6128,
      "step": 5214
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0293366246887152,
      "learning_rate": 6.540845417942637e-07,
      "loss": 0.5605,
      "step": 5215
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9203235635111477,
      "learning_rate": 6.527946475330288e-07,
      "loss": 0.5024,
      "step": 5216
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0714088413915348,
      "learning_rate": 6.51505937570363e-07,
      "loss": 0.5708,
      "step": 5217
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0160707888060683,
      "learning_rate": 6.502184122573457e-07,
      "loss": 0.5779,
      "step": 5218
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.149816131632796,
      "learning_rate": 6.489320719447367e-07,
      "loss": 0.6189,
      "step": 5219
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6408134234686275,
      "learning_rate": 6.476469169829691e-07,
      "loss": 0.4619,
      "step": 5220
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6106888116828263,
      "learning_rate": 6.463629477221533e-07,
      "loss": 0.4358,
      "step": 5221
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9728220809545649,
      "learning_rate": 6.450801645120808e-07,
      "loss": 0.5557,
      "step": 5222
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0566893530617847,
      "learning_rate": 6.43798567702214e-07,
      "loss": 0.6574,
      "step": 5223
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9658884551846721,
      "learning_rate": 6.425181576416978e-07,
      "loss": 0.5301,
      "step": 5224
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0676052046611186,
      "learning_rate": 6.412389346793507e-07,
      "loss": 0.5757,
      "step": 5225
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9394752600821392,
      "learning_rate": 6.399608991636663e-07,
      "loss": 0.5734,
      "step": 5226
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.032603789988249,
      "learning_rate": 6.386840514428183e-07,
      "loss": 0.564,
      "step": 5227
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9877613769050175,
      "learning_rate": 6.374083918646557e-07,
      "loss": 0.5242,
      "step": 5228
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9692263331623233,
      "learning_rate": 6.361339207767031e-07,
      "loss": 0.5867,
      "step": 5229
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9198409120777696,
      "learning_rate": 6.348606385261602e-07,
      "loss": 0.5133,
      "step": 5230
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0632677504898758,
      "learning_rate": 6.335885454599061e-07,
      "loss": 0.6763,
      "step": 5231
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0128961745916334,
      "learning_rate": 6.323176419244925e-07,
      "loss": 0.6009,
      "step": 5232
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0621316668681846,
      "learning_rate": 6.310479282661485e-07,
      "loss": 0.5704,
      "step": 5233
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0288003693743362,
      "learning_rate": 6.297794048307798e-07,
      "loss": 0.6068,
      "step": 5234
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0464694220683792,
      "learning_rate": 6.285120719639693e-07,
      "loss": 0.6224,
      "step": 5235
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.991584675037723,
      "learning_rate": 6.27245930010969e-07,
      "loss": 0.6026,
      "step": 5236
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9413290003372221,
      "learning_rate": 6.259809793167127e-07,
      "loss": 0.5643,
      "step": 5237
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.952635429305962,
      "learning_rate": 6.247172202258095e-07,
      "loss": 0.5474,
      "step": 5238
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.036603394967367,
      "learning_rate": 6.234546530825408e-07,
      "loss": 0.6389,
      "step": 5239
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0002023264889008,
      "learning_rate": 6.221932782308637e-07,
      "loss": 0.4919,
      "step": 5240
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.995700157944254,
      "learning_rate": 6.209330960144139e-07,
      "loss": 0.6141,
      "step": 5241
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0086452883489023,
      "learning_rate": 6.196741067764977e-07,
      "loss": 0.5505,
      "step": 5242
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9757833782736011,
      "learning_rate": 6.184163108600988e-07,
      "loss": 0.5076,
      "step": 5243
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9632799634860469,
      "learning_rate": 6.171597086078767e-07,
      "loss": 0.5165,
      "step": 5244
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0346272079190675,
      "learning_rate": 6.159043003621628e-07,
      "loss": 0.5843,
      "step": 5245
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0415911331082455,
      "learning_rate": 6.146500864649668e-07,
      "loss": 0.5898,
      "step": 5246
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.007810722457339,
      "learning_rate": 6.133970672579693e-07,
      "loss": 0.6512,
      "step": 5247
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.058573892657414,
      "learning_rate": 6.121452430825287e-07,
      "loss": 0.5704,
      "step": 5248
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1039467023089087,
      "learning_rate": 6.108946142796757e-07,
      "loss": 0.5817,
      "step": 5249
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0738270460719461,
      "learning_rate": 6.096451811901155e-07,
      "loss": 0.5812,
      "step": 5250
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0503873475282457,
      "learning_rate": 6.083969441542298e-07,
      "loss": 0.5889,
      "step": 5251
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9341395579951143,
      "learning_rate": 6.071499035120703e-07,
      "loss": 0.4751,
      "step": 5252
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0049265052526277,
      "learning_rate": 6.059040596033682e-07,
      "loss": 0.6007,
      "step": 5253
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0368483016671652,
      "learning_rate": 6.046594127675242e-07,
      "loss": 0.5871,
      "step": 5254
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0767091744444386,
      "learning_rate": 6.034159633436132e-07,
      "loss": 0.5543,
      "step": 5255
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0497149171843612,
      "learning_rate": 6.021737116703868e-07,
      "loss": 0.5938,
      "step": 5256
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0049565644471161,
      "learning_rate": 6.009326580862696e-07,
      "loss": 0.5503,
      "step": 5257
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.060304302174428,
      "learning_rate": 5.996928029293559e-07,
      "loss": 0.5424,
      "step": 5258
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9904335420074606,
      "learning_rate": 5.984541465374172e-07,
      "loss": 0.5788,
      "step": 5259
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0404857441271587,
      "learning_rate": 5.972166892478998e-07,
      "loss": 0.5688,
      "step": 5260
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0255195350761621,
      "learning_rate": 5.959804313979195e-07,
      "loss": 0.4614,
      "step": 5261
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.056828412829864,
      "learning_rate": 5.947453733242659e-07,
      "loss": 0.5373,
      "step": 5262
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0243987665164414,
      "learning_rate": 5.935115153634058e-07,
      "loss": 0.5476,
      "step": 5263
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.01334532216102,
      "learning_rate": 5.922788578514737e-07,
      "loss": 0.5651,
      "step": 5264
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9994849118720266,
      "learning_rate": 5.910474011242801e-07,
      "loss": 0.5809,
      "step": 5265
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0000600061611418,
      "learning_rate": 5.898171455173074e-07,
      "loss": 0.5356,
      "step": 5266
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0362272088701074,
      "learning_rate": 5.885880913657127e-07,
      "loss": 0.5122,
      "step": 5267
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0763910483118724,
      "learning_rate": 5.873602390043231e-07,
      "loss": 0.5989,
      "step": 5268
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9344310290225571,
      "learning_rate": 5.861335887676389e-07,
      "loss": 0.5737,
      "step": 5269
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0824231357459353,
      "learning_rate": 5.84908140989835e-07,
      "loss": 0.5606,
      "step": 5270
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0275564466581721,
      "learning_rate": 5.836838960047558e-07,
      "loss": 0.6004,
      "step": 5271
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9437249392021032,
      "learning_rate": 5.824608541459192e-07,
      "loss": 0.5983,
      "step": 5272
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.037458432864212,
      "learning_rate": 5.812390157465169e-07,
      "loss": 0.6007,
      "step": 5273
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.976492695191397,
      "learning_rate": 5.8001838113941e-07,
      "loss": 0.5427,
      "step": 5274
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9688886953161026,
      "learning_rate": 5.78798950657134e-07,
      "loss": 0.5625,
      "step": 5275
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9946625119359499,
      "learning_rate": 5.775807246318954e-07,
      "loss": 0.5059,
      "step": 5276
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9868702518376952,
      "learning_rate": 5.76363703395571e-07,
      "loss": 0.5697,
      "step": 5277
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.067052010447291,
      "learning_rate": 5.751478872797128e-07,
      "loss": 0.5878,
      "step": 5278
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9366615099851899,
      "learning_rate": 5.739332766155419e-07,
      "loss": 0.5348,
      "step": 5279
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0780493273729084,
      "learning_rate": 5.727198717339511e-07,
      "loss": 0.6063,
      "step": 5280
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0689519579248485,
      "learning_rate": 5.715076729655056e-07,
      "loss": 0.622,
      "step": 5281
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0147821950795468,
      "learning_rate": 5.702966806404431e-07,
      "loss": 0.592,
      "step": 5282
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.048749534251478,
      "learning_rate": 5.690868950886702e-07,
      "loss": 0.553,
      "step": 5283
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6582095868611852,
      "learning_rate": 5.67878316639765e-07,
      "loss": 0.469,
      "step": 5284
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0903511460549942,
      "learning_rate": 5.666709456229797e-07,
      "loss": 0.6111,
      "step": 5285
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.069378836190153,
      "learning_rate": 5.654647823672337e-07,
      "loss": 0.6025,
      "step": 5286
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0498498453145473,
      "learning_rate": 5.642598272011196e-07,
      "loss": 0.6452,
      "step": 5287
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0204681124217996,
      "learning_rate": 5.630560804528995e-07,
      "loss": 0.5912,
      "step": 5288
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0768615642887107,
      "learning_rate": 5.6185354245051e-07,
      "loss": 0.6014,
      "step": 5289
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0105549987276876,
      "learning_rate": 5.606522135215531e-07,
      "loss": 0.5796,
      "step": 5290
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.060503094680933,
      "learning_rate": 5.594520939933041e-07,
      "loss": 0.6425,
      "step": 5291
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0585423246379893,
      "learning_rate": 5.582531841927097e-07,
      "loss": 0.569,
      "step": 5292
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9898097891778114,
      "learning_rate": 5.570554844463854e-07,
      "loss": 0.5129,
      "step": 5293
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.96660176974015,
      "learning_rate": 5.558589950806164e-07,
      "loss": 0.4866,
      "step": 5294
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.110907239102545,
      "learning_rate": 5.546637164213625e-07,
      "loss": 0.6565,
      "step": 5295
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0715278006922526,
      "learning_rate": 5.53469648794247e-07,
      "loss": 0.5875,
      "step": 5296
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9606097335561491,
      "learning_rate": 5.522767925245698e-07,
      "loss": 0.541,
      "step": 5297
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9436580348269376,
      "learning_rate": 5.510851479372959e-07,
      "loss": 0.4874,
      "step": 5298
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.960843959065725,
      "learning_rate": 5.498947153570622e-07,
      "loss": 0.4786,
      "step": 5299
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.134654081402097,
      "learning_rate": 5.487054951081772e-07,
      "loss": 0.5898,
      "step": 5300
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9818707185808371,
      "learning_rate": 5.475174875146156e-07,
      "loss": 0.5372,
      "step": 5301
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0922576665655455,
      "learning_rate": 5.463306929000228e-07,
      "loss": 0.606,
      "step": 5302
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0305282400865756,
      "learning_rate": 5.451451115877154e-07,
      "loss": 0.5586,
      "step": 5303
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0829463897755023,
      "learning_rate": 5.439607439006795e-07,
      "loss": 0.5664,
      "step": 5304
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0396069258770884,
      "learning_rate": 5.427775901615684e-07,
      "loss": 0.6382,
      "step": 5305
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0192800705378473,
      "learning_rate": 5.415956506927051e-07,
      "loss": 0.5376,
      "step": 5306
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1565156550638231,
      "learning_rate": 5.404149258160835e-07,
      "loss": 0.6994,
      "step": 5307
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0339829958337519,
      "learning_rate": 5.392354158533658e-07,
      "loss": 0.581,
      "step": 5308
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0403587546908875,
      "learning_rate": 5.380571211258811e-07,
      "loss": 0.5856,
      "step": 5309
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6954657542963856,
      "learning_rate": 5.36880041954631e-07,
      "loss": 0.5104,
      "step": 5310
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.121336678174201,
      "learning_rate": 5.357041786602851e-07,
      "loss": 0.6519,
      "step": 5311
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0613368977206599,
      "learning_rate": 5.345295315631805e-07,
      "loss": 0.5453,
      "step": 5312
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0162393875640843,
      "learning_rate": 5.33356100983321e-07,
      "loss": 0.533,
      "step": 5313
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0612525436499354,
      "learning_rate": 5.32183887240385e-07,
      "loss": 0.6138,
      "step": 5314
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0005071071691518,
      "learning_rate": 5.310128906537137e-07,
      "loss": 0.5782,
      "step": 5315
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9657993983141224,
      "learning_rate": 5.298431115423186e-07,
      "loss": 0.531,
      "step": 5316
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9996200117211376,
      "learning_rate": 5.286745502248819e-07,
      "loss": 0.6109,
      "step": 5317
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1776338300650258,
      "learning_rate": 5.27507207019749e-07,
      "loss": 0.5746,
      "step": 5318
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9429207567510018,
      "learning_rate": 5.263410822449388e-07,
      "loss": 0.5042,
      "step": 5319
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0915199418442072,
      "learning_rate": 5.251761762181351e-07,
      "loss": 0.5903,
      "step": 5320
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.038977501783133,
      "learning_rate": 5.240124892566895e-07,
      "loss": 0.5511,
      "step": 5321
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0594678368787669,
      "learning_rate": 5.228500216776239e-07,
      "loss": 0.5265,
      "step": 5322
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0666379055470785,
      "learning_rate": 5.216887737976256e-07,
      "loss": 0.6083,
      "step": 5323
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1512303845500023,
      "learning_rate": 5.205287459330499e-07,
      "loss": 0.6671,
      "step": 5324
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9340583537148618,
      "learning_rate": 5.193699383999213e-07,
      "loss": 0.577,
      "step": 5325
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2154686856510797,
      "learning_rate": 5.182123515139315e-07,
      "loss": 0.5092,
      "step": 5326
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0532878006381232,
      "learning_rate": 5.170559855904389e-07,
      "loss": 0.5884,
      "step": 5327
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0177303561681834,
      "learning_rate": 5.159008409444671e-07,
      "loss": 0.6046,
      "step": 5328
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0061290575507669,
      "learning_rate": 5.147469178907127e-07,
      "loss": 0.594,
      "step": 5329
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0150493950139718,
      "learning_rate": 5.135942167435342e-07,
      "loss": 0.5643,
      "step": 5330
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9799531989094968,
      "learning_rate": 5.124427378169588e-07,
      "loss": 0.566,
      "step": 5331
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0320827301044417,
      "learning_rate": 5.112924814246817e-07,
      "loss": 0.5371,
      "step": 5332
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0033996768121038,
      "learning_rate": 5.10143447880066e-07,
      "loss": 0.5938,
      "step": 5333
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0985550857635513,
      "learning_rate": 5.089956374961386e-07,
      "loss": 0.6197,
      "step": 5334
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9874049904785749,
      "learning_rate": 5.078490505855938e-07,
      "loss": 0.5124,
      "step": 5335
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0251815929947037,
      "learning_rate": 5.06703687460795e-07,
      "loss": 0.6097,
      "step": 5336
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0316808884098747,
      "learning_rate": 5.055595484337705e-07,
      "loss": 0.597,
      "step": 5337
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9861587380281057,
      "learning_rate": 5.044166338162143e-07,
      "loss": 0.5183,
      "step": 5338
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0567983884085228,
      "learning_rate": 5.03274943919489e-07,
      "loss": 0.6364,
      "step": 5339
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0491668352240564,
      "learning_rate": 5.021344790546212e-07,
      "loss": 0.5986,
      "step": 5340
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0756755698209064,
      "learning_rate": 5.009952395323065e-07,
      "loss": 0.6098,
      "step": 5341
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0201352465743787,
      "learning_rate": 4.998572256629047e-07,
      "loss": 0.6325,
      "step": 5342
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9624850896089041,
      "learning_rate": 4.987204377564409e-07,
      "loss": 0.5613,
      "step": 5343
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.065915688816882,
      "learning_rate": 4.975848761226088e-07,
      "loss": 0.6015,
      "step": 5344
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0281805141824745,
      "learning_rate": 4.964505410707655e-07,
      "loss": 0.4828,
      "step": 5345
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0205613111336664,
      "learning_rate": 4.95317432909937e-07,
      "loss": 0.5506,
      "step": 5346
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9234754546685896,
      "learning_rate": 4.941855519488109e-07,
      "loss": 0.488,
      "step": 5347
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.010656699995779,
      "learning_rate": 4.930548984957451e-07,
      "loss": 0.5806,
      "step": 5348
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0775370134740074,
      "learning_rate": 4.919254728587591e-07,
      "loss": 0.6418,
      "step": 5349
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9905056816255239,
      "learning_rate": 4.907972753455398e-07,
      "loss": 0.4806,
      "step": 5350
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.046818992747354,
      "learning_rate": 4.896703062634401e-07,
      "loss": 0.5669,
      "step": 5351
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0742812252935492,
      "learning_rate": 4.885445659194771e-07,
      "loss": 0.7122,
      "step": 5352
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0439717535472564,
      "learning_rate": 4.874200546203328e-07,
      "loss": 0.608,
      "step": 5353
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9994051914718778,
      "learning_rate": 4.86296772672355e-07,
      "loss": 0.6534,
      "step": 5354
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9724560046110385,
      "learning_rate": 4.851747203815588e-07,
      "loss": 0.5549,
      "step": 5355
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0702135479088555,
      "learning_rate": 4.840538980536203e-07,
      "loss": 0.5732,
      "step": 5356
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0320588451695234,
      "learning_rate": 4.829343059938818e-07,
      "loss": 0.5805,
      "step": 5357
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0605178500973564,
      "learning_rate": 4.818159445073528e-07,
      "loss": 0.5014,
      "step": 5358
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0981733393435582,
      "learning_rate": 4.806988138987051e-07,
      "loss": 0.5717,
      "step": 5359
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9879712150863855,
      "learning_rate": 4.795829144722752e-07,
      "loss": 0.5166,
      "step": 5360
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9558676793186869,
      "learning_rate": 4.784682465320661e-07,
      "loss": 0.5766,
      "step": 5361
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6665559118702703,
      "learning_rate": 4.773548103817421e-07,
      "loss": 0.4912,
      "step": 5362
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0936849109209077,
      "learning_rate": 4.762426063246367e-07,
      "loss": 0.6284,
      "step": 5363
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0379521015925315,
      "learning_rate": 4.75131634663743e-07,
      "loss": 0.6005,
      "step": 5364
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0525528268967095,
      "learning_rate": 4.740218957017201e-07,
      "loss": 0.6249,
      "step": 5365
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9808550876906315,
      "learning_rate": 4.729133897408933e-07,
      "loss": 0.5716,
      "step": 5366
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.075871209040523,
      "learning_rate": 4.7180611708324776e-07,
      "loss": 0.5349,
      "step": 5367
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1066723711838085,
      "learning_rate": 4.7070007803043714e-07,
      "loss": 0.6134,
      "step": 5368
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1210136016862404,
      "learning_rate": 4.6959527288377493e-07,
      "loss": 0.6187,
      "step": 5369
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0763661144842214,
      "learning_rate": 4.684917019442431e-07,
      "loss": 0.6473,
      "step": 5370
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1173034323796098,
      "learning_rate": 4.673893655124834e-07,
      "loss": 0.5817,
      "step": 5371
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0263195965203538,
      "learning_rate": 4.6628826388880165e-07,
      "loss": 0.4982,
      "step": 5372
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0620555346660505,
      "learning_rate": 4.651883973731708e-07,
      "loss": 0.6626,
      "step": 5373
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9040278249830518,
      "learning_rate": 4.640897662652227e-07,
      "loss": 0.4685,
      "step": 5374
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0041914536372203,
      "learning_rate": 4.629923708642542e-07,
      "loss": 0.5703,
      "step": 5375
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9535087821063442,
      "learning_rate": 4.618962114692277e-07,
      "loss": 0.5183,
      "step": 5376
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9820220585612087,
      "learning_rate": 4.608012883787688e-07,
      "loss": 0.514,
      "step": 5377
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0431003673985013,
      "learning_rate": 4.5970760189116057e-07,
      "loss": 0.5968,
      "step": 5378
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0299143341371995,
      "learning_rate": 4.5861515230435525e-07,
      "loss": 0.5257,
      "step": 5379
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0764525170430264,
      "learning_rate": 4.5752393991596754e-07,
      "loss": 0.6474,
      "step": 5380
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0602059020295858,
      "learning_rate": 4.5643396502327297e-07,
      "loss": 0.5415,
      "step": 5381
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9318694833006669,
      "learning_rate": 4.55345227923209e-07,
      "loss": 0.5686,
      "step": 5382
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0741222165021485,
      "learning_rate": 4.5425772891238065e-07,
      "loss": 0.6259,
      "step": 5383
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0363539626683014,
      "learning_rate": 4.5317146828704973e-07,
      "loss": 0.5537,
      "step": 5384
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0953554734378081,
      "learning_rate": 4.520864463431457e-07,
      "loss": 0.5768,
      "step": 5385
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6166165739699614,
      "learning_rate": 4.510026633762571e-07,
      "loss": 0.4494,
      "step": 5386
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0425424044417757,
      "learning_rate": 4.4992011968163775e-07,
      "loss": 0.6175,
      "step": 5387
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0230523111976728,
      "learning_rate": 4.488388155542012e-07,
      "loss": 0.6302,
      "step": 5388
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.04279325447922,
      "learning_rate": 4.47758751288524e-07,
      "loss": 0.5369,
      "step": 5389
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.02269064713844,
      "learning_rate": 4.466799271788469e-07,
      "loss": 0.5892,
      "step": 5390
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0196858339935873,
      "learning_rate": 4.4560234351906983e-07,
      "loss": 0.4966,
      "step": 5391
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9865520573336268,
      "learning_rate": 4.445260006027585e-07,
      "loss": 0.5919,
      "step": 5392
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9842238208294244,
      "learning_rate": 4.4345089872313674e-07,
      "loss": 0.5686,
      "step": 5393
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0293195850256684,
      "learning_rate": 4.4237703817309073e-07,
      "loss": 0.6088,
      "step": 5394
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9206359490846122,
      "learning_rate": 4.4130441924517263e-07,
      "loss": 0.45,
      "step": 5395
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.667148183892131,
      "learning_rate": 4.4023304223159203e-07,
      "loss": 0.5098,
      "step": 5396
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.067534368564209,
      "learning_rate": 4.3916290742421986e-07,
      "loss": 0.5546,
      "step": 5397
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0754095454079409,
      "learning_rate": 4.3809401511459237e-07,
      "loss": 0.5953,
      "step": 5398
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0153823284468526,
      "learning_rate": 4.3702636559390667e-07,
      "loss": 0.5817,
      "step": 5399
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0315366538737971,
      "learning_rate": 4.3595995915301614e-07,
      "loss": 0.5703,
      "step": 5400
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9291506792248045,
      "learning_rate": 4.348947960824412e-07,
      "loss": 0.5027,
      "step": 5401
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0274025123180561,
      "learning_rate": 4.3383087667236254e-07,
      "loss": 0.5889,
      "step": 5402
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9817287516962871,
      "learning_rate": 4.3276820121262053e-07,
      "loss": 0.5119,
      "step": 5403
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.02159256337688,
      "learning_rate": 4.3170676999271576e-07,
      "loss": 0.5819,
      "step": 5404
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9773692672498304,
      "learning_rate": 4.306465833018131e-07,
      "loss": 0.5016,
      "step": 5405
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0412070262846036,
      "learning_rate": 4.295876414287375e-07,
      "loss": 0.5804,
      "step": 5406
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0186674071232007,
      "learning_rate": 4.2852994466197064e-07,
      "loss": 0.5259,
      "step": 5407
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.033605603962162,
      "learning_rate": 4.274734932896607e-07,
      "loss": 0.5167,
      "step": 5408
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9921239516227279,
      "learning_rate": 4.264182875996142e-07,
      "loss": 0.5732,
      "step": 5409
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9394645921800621,
      "learning_rate": 4.253643278792974e-07,
      "loss": 0.5296,
      "step": 5410
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0792412043157142,
      "learning_rate": 4.243116144158377e-07,
      "loss": 0.5524,
      "step": 5411
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0621776967854895,
      "learning_rate": 4.232601474960246e-07,
      "loss": 0.5913,
      "step": 5412
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.107477053972103,
      "learning_rate": 4.2220992740630515e-07,
      "loss": 0.5944,
      "step": 5413
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.072704024203222,
      "learning_rate": 4.211609544327894e-07,
      "loss": 0.6419,
      "step": 5414
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0013870364173174,
      "learning_rate": 4.201132288612464e-07,
      "loss": 0.5553,
      "step": 5415
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0372138811562466,
      "learning_rate": 4.1906675097710457e-07,
      "loss": 0.5864,
      "step": 5416
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0014938965224214,
      "learning_rate": 4.180215210654548e-07,
      "loss": 0.6091,
      "step": 5417
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0344869093116396,
      "learning_rate": 4.1697753941104533e-07,
      "loss": 0.5855,
      "step": 5418
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9045712564435315,
      "learning_rate": 4.159348062982849e-07,
      "loss": 0.4658,
      "step": 5419
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9516960288834059,
      "learning_rate": 4.1489332201124443e-07,
      "loss": 0.5366,
      "step": 5420
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1475407599759215,
      "learning_rate": 4.138530868336532e-07,
      "loss": 0.5363,
      "step": 5421
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9804195425721148,
      "learning_rate": 4.1281410104889784e-07,
      "loss": 0.4947,
      "step": 5422
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0544655948697874,
      "learning_rate": 4.1177636494002747e-07,
      "loss": 0.6083,
      "step": 5423
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0599906459289223,
      "learning_rate": 4.107398787897515e-07,
      "loss": 0.5801,
      "step": 5424
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9417158938737237,
      "learning_rate": 4.097046428804363e-07,
      "loss": 0.5596,
      "step": 5425
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0926068634715345,
      "learning_rate": 4.0867065749410737e-07,
      "loss": 0.6613,
      "step": 5426
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9719320657022369,
      "learning_rate": 4.0763792291245274e-07,
      "loss": 0.5411,
      "step": 5427
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9530201623130976,
      "learning_rate": 4.066064394168184e-07,
      "loss": 0.5356,
      "step": 5428
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9727297474249349,
      "learning_rate": 4.055762072882058e-07,
      "loss": 0.4628,
      "step": 5429
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9847177137561993,
      "learning_rate": 4.045472268072803e-07,
      "loss": 0.6488,
      "step": 5430
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0834700565963222,
      "learning_rate": 4.0351949825436556e-07,
      "loss": 0.5347,
      "step": 5431
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0765626708944795,
      "learning_rate": 4.024930219094425e-07,
      "loss": 0.5857,
      "step": 5432
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0393289289976886,
      "learning_rate": 4.014677980521503e-07,
      "loss": 0.5744,
      "step": 5433
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9837184951009063,
      "learning_rate": 4.0044382696178994e-07,
      "loss": 0.5114,
      "step": 5434
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.016605576594332,
      "learning_rate": 3.994211089173189e-07,
      "loss": 0.5254,
      "step": 5435
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9228662738271468,
      "learning_rate": 3.9839964419735254e-07,
      "loss": 0.548,
      "step": 5436
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0090413072087565,
      "learning_rate": 3.9737943308016726e-07,
      "loss": 0.6058,
      "step": 5437
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0497552798598961,
      "learning_rate": 3.963604758436962e-07,
      "loss": 0.5549,
      "step": 5438
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.946658448727259,
      "learning_rate": 3.953427727655323e-07,
      "loss": 0.5162,
      "step": 5439
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.076480248934819,
      "learning_rate": 3.9432632412292504e-07,
      "loss": 0.6071,
      "step": 5440
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.04834389727141,
      "learning_rate": 3.933111301927828e-07,
      "loss": 0.49,
      "step": 5441
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0848161143569053,
      "learning_rate": 3.922971912516721e-07,
      "loss": 0.5789,
      "step": 5442
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0891721996182941,
      "learning_rate": 3.912845075758209e-07,
      "loss": 0.5624,
      "step": 5443
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.039746527048185,
      "learning_rate": 3.9027307944110804e-07,
      "loss": 0.6073,
      "step": 5444
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9698055791727057,
      "learning_rate": 3.892629071230758e-07,
      "loss": 0.5732,
      "step": 5445
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1164340624317148,
      "learning_rate": 3.8825399089692464e-07,
      "loss": 0.5743,
      "step": 5446
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.045446006316855,
      "learning_rate": 3.8724633103750964e-07,
      "loss": 0.6404,
      "step": 5447
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0923161128356194,
      "learning_rate": 3.862399278193446e-07,
      "loss": 0.672,
      "step": 5448
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0319511468126787,
      "learning_rate": 3.852347815166024e-07,
      "loss": 0.5949,
      "step": 5449
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.014788457900176,
      "learning_rate": 3.8423089240311404e-07,
      "loss": 0.502,
      "step": 5450
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9723411469674623,
      "learning_rate": 3.8322826075236295e-07,
      "loss": 0.5632,
      "step": 5451
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6112810697978414,
      "learning_rate": 3.822268868374962e-07,
      "loss": 0.4914,
      "step": 5452
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0534832844434676,
      "learning_rate": 3.8122677093131613e-07,
      "loss": 0.5993,
      "step": 5453
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0440273469845522,
      "learning_rate": 3.802279133062803e-07,
      "loss": 0.6233,
      "step": 5454
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9708111601216068,
      "learning_rate": 3.7923031423450495e-07,
      "loss": 0.5327,
      "step": 5455
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6143990052691758,
      "learning_rate": 3.782339739877655e-07,
      "loss": 0.4312,
      "step": 5456
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0049799616533042,
      "learning_rate": 3.772388928374909e-07,
      "loss": 0.5211,
      "step": 5457
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0328432953355067,
      "learning_rate": 3.7624507105476826e-07,
      "loss": 0.5667,
      "step": 5458
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0325620225997374,
      "learning_rate": 3.752525089103437e-07,
      "loss": 0.5312,
      "step": 5459
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9562727790235105,
      "learning_rate": 3.7426120667461706e-07,
      "loss": 0.5234,
      "step": 5460
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9864133114132222,
      "learning_rate": 3.7327116461764737e-07,
      "loss": 0.6322,
      "step": 5461
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6583603252530985,
      "learning_rate": 3.7228238300914934e-07,
      "loss": 0.4555,
      "step": 5462
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.960195546219715,
      "learning_rate": 3.7129486211849307e-07,
      "loss": 0.5506,
      "step": 5463
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.189643247130239,
      "learning_rate": 3.7030860221470787e-07,
      "loss": 0.6136,
      "step": 5464
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.01859904537151,
      "learning_rate": 3.6932360356647713e-07,
      "loss": 0.5096,
      "step": 5465
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.015535350943715,
      "learning_rate": 3.683398664421428e-07,
      "loss": 0.565,
      "step": 5466
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9517095635129854,
      "learning_rate": 3.673573911097006e-07,
      "loss": 0.5365,
      "step": 5467
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0241250877283186,
      "learning_rate": 3.663761778368052e-07,
      "loss": 0.5486,
      "step": 5468
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7023895722821127,
      "learning_rate": 3.653962268907657e-07,
      "loss": 0.4853,
      "step": 5469
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.044042763307488,
      "learning_rate": 3.6441753853854677e-07,
      "loss": 0.4482,
      "step": 5470
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0104596524046299,
      "learning_rate": 3.6344011304677074e-07,
      "loss": 0.5838,
      "step": 5471
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0166255412514318,
      "learning_rate": 3.624639506817174e-07,
      "loss": 0.5795,
      "step": 5472
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0335294877439671,
      "learning_rate": 3.614890517093161e-07,
      "loss": 0.6123,
      "step": 5473
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0554255229470013,
      "learning_rate": 3.605154163951591e-07,
      "loss": 0.5251,
      "step": 5474
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9840462754492645,
      "learning_rate": 3.595430450044912e-07,
      "loss": 0.5683,
      "step": 5475
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.029108872851345,
      "learning_rate": 3.585719378022129e-07,
      "loss": 0.5348,
      "step": 5476
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0053202782628397,
      "learning_rate": 3.576020950528797e-07,
      "loss": 0.5901,
      "step": 5477
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0730291520335882,
      "learning_rate": 3.566335170207053e-07,
      "loss": 0.5362,
      "step": 5478
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0823990179378389,
      "learning_rate": 3.5566620396955573e-07,
      "loss": 0.5647,
      "step": 5479
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0347269930065042,
      "learning_rate": 3.5470015616295315e-07,
      "loss": 0.492,
      "step": 5480
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6824849473063792,
      "learning_rate": 3.537353738640775e-07,
      "loss": 0.4767,
      "step": 5481
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6831128263896733,
      "learning_rate": 3.527718573357597e-07,
      "loss": 0.4596,
      "step": 5482
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9580094604741383,
      "learning_rate": 3.518096068404908e-07,
      "loss": 0.5817,
      "step": 5483
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1230928024331281,
      "learning_rate": 3.508486226404123e-07,
      "loss": 0.6442,
      "step": 5484
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0968922492014628,
      "learning_rate": 3.4988890499732477e-07,
      "loss": 0.6178,
      "step": 5485
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0698282548137314,
      "learning_rate": 3.489304541726801e-07,
      "loss": 0.5978,
      "step": 5486
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0803133582998532,
      "learning_rate": 3.479732704275862e-07,
      "loss": 0.5861,
      "step": 5487
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0588609594924376,
      "learning_rate": 3.470173540228089e-07,
      "loss": 0.547,
      "step": 5488
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0497306275020217,
      "learning_rate": 3.460627052187632e-07,
      "loss": 0.5671,
      "step": 5489
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6605524077733674,
      "learning_rate": 3.451093242755238e-07,
      "loss": 0.4743,
      "step": 5490
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1049236856108615,
      "learning_rate": 3.4415721145281743e-07,
      "loss": 0.5355,
      "step": 5491
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0411107866516922,
      "learning_rate": 3.4320636701002497e-07,
      "loss": 0.5872,
      "step": 5492
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0495261128448135,
      "learning_rate": 3.422567912061836e-07,
      "loss": 0.5362,
      "step": 5493
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.042951842294696,
      "learning_rate": 3.4130848429998533e-07,
      "loss": 0.6364,
      "step": 5494
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9684863664012521,
      "learning_rate": 3.403614465497718e-07,
      "loss": 0.5036,
      "step": 5495
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0407275384047345,
      "learning_rate": 3.3941567821354383e-07,
      "loss": 0.5326,
      "step": 5496
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9500087554589208,
      "learning_rate": 3.3847117954895536e-07,
      "loss": 0.5202,
      "step": 5497
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1121829882372587,
      "learning_rate": 3.375279508133139e-07,
      "loss": 0.6297,
      "step": 5498
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0294244345093888,
      "learning_rate": 3.3658599226357944e-07,
      "loss": 0.6812,
      "step": 5499
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6472027109529903,
      "learning_rate": 3.356453041563695e-07,
      "loss": 0.4464,
      "step": 5500
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0173436133692797,
      "learning_rate": 3.347058867479519e-07,
      "loss": 0.5746,
      "step": 5501
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0371058906915347,
      "learning_rate": 3.337677402942502e-07,
      "loss": 0.506,
      "step": 5502
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0806197988929243,
      "learning_rate": 3.3283086505084163e-07,
      "loss": 0.6032,
      "step": 5503
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.116689260823897,
      "learning_rate": 3.3189526127295644e-07,
      "loss": 0.6575,
      "step": 5504
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0214300518691564,
      "learning_rate": 3.3096092921547976e-07,
      "loss": 0.5944,
      "step": 5505
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1038338677836759,
      "learning_rate": 3.300278691329478e-07,
      "loss": 0.6476,
      "step": 5506
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0483367780975363,
      "learning_rate": 3.290960812795535e-07,
      "loss": 0.5709,
      "step": 5507
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0956658373423223,
      "learning_rate": 3.2816556590914096e-07,
      "loss": 0.5764,
      "step": 5508
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0122203818208813,
      "learning_rate": 3.2723632327520795e-07,
      "loss": 0.5847,
      "step": 5509
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1698777849723354,
      "learning_rate": 3.2630835363090584e-07,
      "loss": 0.6682,
      "step": 5510
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0689562929407017,
      "learning_rate": 3.2538165722903913e-07,
      "loss": 0.5433,
      "step": 5511
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9444796568777285,
      "learning_rate": 3.244562343220664e-07,
      "loss": 0.573,
      "step": 5512
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0536833407411912,
      "learning_rate": 3.235320851620971e-07,
      "loss": 0.5729,
      "step": 5513
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0027094273926629,
      "learning_rate": 3.2260921000089483e-07,
      "loss": 0.5171,
      "step": 5514
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0894450690659234,
      "learning_rate": 3.216876090898774e-07,
      "loss": 0.5276,
      "step": 5515
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0695732898158838,
      "learning_rate": 3.207672826801139e-07,
      "loss": 0.6677,
      "step": 5516
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.091673399881913,
      "learning_rate": 3.198482310223261e-07,
      "loss": 0.5787,
      "step": 5517
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0125337566118928,
      "learning_rate": 3.1893045436688863e-07,
      "loss": 0.5833,
      "step": 5518
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0889974030524212,
      "learning_rate": 3.1801395296383097e-07,
      "loss": 0.6043,
      "step": 5519
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1099818604557212,
      "learning_rate": 3.1709872706283173e-07,
      "loss": 0.6092,
      "step": 5520
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9557354878248199,
      "learning_rate": 3.161847769132237e-07,
      "loss": 0.4865,
      "step": 5521
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0861089240106898,
      "learning_rate": 3.152721027639932e-07,
      "loss": 0.5698,
      "step": 5522
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6722680754856171,
      "learning_rate": 3.1436070486377755e-07,
      "loss": 0.4502,
      "step": 5523
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9789450373599586,
      "learning_rate": 3.134505834608653e-07,
      "loss": 0.5599,
      "step": 5524
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0434220240551915,
      "learning_rate": 3.125417388031998e-07,
      "loss": 0.5947,
      "step": 5525
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0219079322471198,
      "learning_rate": 3.116341711383758e-07,
      "loss": 0.5705,
      "step": 5526
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.102326432719913,
      "learning_rate": 3.1072788071363936e-07,
      "loss": 0.5829,
      "step": 5527
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0237232203906357,
      "learning_rate": 3.0982286777588745e-07,
      "loss": 0.6001,
      "step": 5528
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0660494984694744,
      "learning_rate": 3.089191325716728e-07,
      "loss": 0.5844,
      "step": 5529
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1341907065540477,
      "learning_rate": 3.0801667534719694e-07,
      "loss": 0.5658,
      "step": 5530
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0596006387971575,
      "learning_rate": 3.0711549634831364e-07,
      "loss": 0.6221,
      "step": 5531
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.934146198553734,
      "learning_rate": 3.062155958205293e-07,
      "loss": 0.5726,
      "step": 5532
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6698126600777463,
      "learning_rate": 3.0531697400900117e-07,
      "loss": 0.4833,
      "step": 5533
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6466520817538444,
      "learning_rate": 3.044196311585401e-07,
      "loss": 0.4359,
      "step": 5534
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0625125926955123,
      "learning_rate": 3.0352356751360555e-07,
      "loss": 0.6247,
      "step": 5535
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9714020826596609,
      "learning_rate": 3.0262878331830946e-07,
      "loss": 0.543,
      "step": 5536
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0035036528019892,
      "learning_rate": 3.017352788164174e-07,
      "loss": 0.5282,
      "step": 5537
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0665749750397815,
      "learning_rate": 3.0084305425134406e-07,
      "loss": 0.6551,
      "step": 5538
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9332740941892883,
      "learning_rate": 2.999521098661551e-07,
      "loss": 0.4979,
      "step": 5539
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0170626745398943,
      "learning_rate": 2.990624459035685e-07,
      "loss": 0.5667,
      "step": 5540
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0543273011101513,
      "learning_rate": 2.981740626059548e-07,
      "loss": 0.5513,
      "step": 5541
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0200112625730604,
      "learning_rate": 2.972869602153333e-07,
      "loss": 0.581,
      "step": 5542
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0661442339344485,
      "learning_rate": 2.964011389733734e-07,
      "loss": 0.5549,
      "step": 5543
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9831828250171059,
      "learning_rate": 2.955165991214004e-07,
      "loss": 0.4838,
      "step": 5544
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0084705064476842,
      "learning_rate": 2.946333409003849e-07,
      "loss": 0.6019,
      "step": 5545
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0543011798622692,
      "learning_rate": 2.93751364550951e-07,
      "loss": 0.5785,
      "step": 5546
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0011303562065086,
      "learning_rate": 2.928706703133738e-07,
      "loss": 0.531,
      "step": 5547
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1228052869296419,
      "learning_rate": 2.919912584275797e-07,
      "loss": 0.5695,
      "step": 5548
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6572791098733273,
      "learning_rate": 2.911131291331437e-07,
      "loss": 0.4781,
      "step": 5549
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0779232607196123,
      "learning_rate": 2.902362826692923e-07,
      "loss": 0.6499,
      "step": 5550
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9756223479285945,
      "learning_rate": 2.8936071927490317e-07,
      "loss": 0.5049,
      "step": 5551
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9612104154297084,
      "learning_rate": 2.884864391885034e-07,
      "loss": 0.5759,
      "step": 5552
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0854430277396139,
      "learning_rate": 2.876134426482713e-07,
      "loss": 0.6429,
      "step": 5553
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0270442528963726,
      "learning_rate": 2.8674172989203497e-07,
      "loss": 0.5622,
      "step": 5554
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0382850024689978,
      "learning_rate": 2.858713011572728e-07,
      "loss": 0.6394,
      "step": 5555
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9819674903429195,
      "learning_rate": 2.850021566811145e-07,
      "loss": 0.561,
      "step": 5556
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6638302144387936,
      "learning_rate": 2.8413429670033855e-07,
      "loss": 0.4642,
      "step": 5557
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.061859273924965,
      "learning_rate": 2.832677214513735e-07,
      "loss": 0.6811,
      "step": 5558
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9164145949177618,
      "learning_rate": 2.8240243117029885e-07,
      "loss": 0.4756,
      "step": 5559
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9408193412622898,
      "learning_rate": 2.815384260928433e-07,
      "loss": 0.5358,
      "step": 5560
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.083725922040101,
      "learning_rate": 2.8067570645438467e-07,
      "loss": 0.575,
      "step": 5561
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0861298558444912,
      "learning_rate": 2.7981427248995276e-07,
      "loss": 0.5339,
      "step": 5562
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0973214650411918,
      "learning_rate": 2.7895412443422644e-07,
      "loss": 0.5715,
      "step": 5563
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0328810127482706,
      "learning_rate": 2.780952625215333e-07,
      "loss": 0.5915,
      "step": 5564
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0181382063910152,
      "learning_rate": 2.772376869858495e-07,
      "loss": 0.5834,
      "step": 5565
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.063112432318683,
      "learning_rate": 2.7638139806080435e-07,
      "loss": 0.5821,
      "step": 5566
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1131037332401148,
      "learning_rate": 2.755263959796739e-07,
      "loss": 0.665,
      "step": 5567
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1665945481067514,
      "learning_rate": 2.746726809753836e-07,
      "loss": 0.7384,
      "step": 5568
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9609751344829549,
      "learning_rate": 2.73820253280509e-07,
      "loss": 0.5228,
      "step": 5569
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9613750889826922,
      "learning_rate": 2.72969113127276e-07,
      "loss": 0.4558,
      "step": 5570
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.151759932071961,
      "learning_rate": 2.7211926074755755e-07,
      "loss": 0.7,
      "step": 5571
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9782703907592869,
      "learning_rate": 2.712706963728773e-07,
      "loss": 0.5761,
      "step": 5572
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9836208568725887,
      "learning_rate": 2.7042342023440803e-07,
      "loss": 0.5053,
      "step": 5573
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0118058314909428,
      "learning_rate": 2.6957743256297077e-07,
      "loss": 0.5513,
      "step": 5574
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9444850233991927,
      "learning_rate": 2.687327335890344e-07,
      "loss": 0.5098,
      "step": 5575
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6828821285754091,
      "learning_rate": 2.6788932354272104e-07,
      "loss": 0.4866,
      "step": 5576
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0841197724809626,
      "learning_rate": 2.6704720265379626e-07,
      "loss": 0.5151,
      "step": 5577
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0110367903493274,
      "learning_rate": 2.662063711516788e-07,
      "loss": 0.5921,
      "step": 5578
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0393891423510995,
      "learning_rate": 2.653668292654338e-07,
      "loss": 0.5062,
      "step": 5579
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9965735627587826,
      "learning_rate": 2.645285772237743e-07,
      "loss": 0.5224,
      "step": 5580
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9653412034617587,
      "learning_rate": 2.636916152550656e-07,
      "loss": 0.514,
      "step": 5581
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0418712480906847,
      "learning_rate": 2.6285594358731736e-07,
      "loss": 0.5581,
      "step": 5582
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0458123010412255,
      "learning_rate": 2.6202156244819035e-07,
      "loss": 0.5265,
      "step": 5583
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0082290737007817,
      "learning_rate": 2.611884720649921e-07,
      "loss": 0.6331,
      "step": 5584
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0674215917161387,
      "learning_rate": 2.6035667266468166e-07,
      "loss": 0.6253,
      "step": 5585
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0692431636148754,
      "learning_rate": 2.595261644738617e-07,
      "loss": 0.6049,
      "step": 5586
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9183243702528002,
      "learning_rate": 2.586969477187867e-07,
      "loss": 0.3919,
      "step": 5587
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9746585461481831,
      "learning_rate": 2.578690226253583e-07,
      "loss": 0.539,
      "step": 5588
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0121643076304234,
      "learning_rate": 2.570423894191254e-07,
      "loss": 0.482,
      "step": 5589
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9161540587789992,
      "learning_rate": 2.5621704832528573e-07,
      "loss": 0.595,
      "step": 5590
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0148002662535323,
      "learning_rate": 2.553929995686849e-07,
      "loss": 0.5615,
      "step": 5591
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0056509928856485,
      "learning_rate": 2.545702433738184e-07,
      "loss": 0.5691,
      "step": 5592
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0602073594496693,
      "learning_rate": 2.537487799648247e-07,
      "loss": 0.5574,
      "step": 5593
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0719631001202896,
      "learning_rate": 2.529286095654948e-07,
      "loss": 0.5199,
      "step": 5594
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0166341660136442,
      "learning_rate": 2.5210973239926553e-07,
      "loss": 0.5486,
      "step": 5595
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6271309046555144,
      "learning_rate": 2.5129214868922227e-07,
      "loss": 0.4733,
      "step": 5596
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9649522661143487,
      "learning_rate": 2.504758586580952e-07,
      "loss": 0.5531,
      "step": 5597
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0061987427835428,
      "learning_rate": 2.4966086252826706e-07,
      "loss": 0.578,
      "step": 5598
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.029425221736268,
      "learning_rate": 2.4884716052176237e-07,
      "loss": 0.673,
      "step": 5599
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0507613788297858,
      "learning_rate": 2.480347528602589e-07,
      "loss": 0.5702,
      "step": 5600
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9368570068431507,
      "learning_rate": 2.4722363976507677e-07,
      "loss": 0.5648,
      "step": 5601
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9922137994193313,
      "learning_rate": 2.4641382145718597e-07,
      "loss": 0.5856,
      "step": 5602
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9799842390263321,
      "learning_rate": 2.456052981572038e-07,
      "loss": 0.6119,
      "step": 5603
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2005713443172104,
      "learning_rate": 2.447980700853941e-07,
      "loss": 0.6263,
      "step": 5604
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9525423913853822,
      "learning_rate": 2.439921374616683e-07,
      "loss": 0.5675,
      "step": 5605
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0353348777385614,
      "learning_rate": 2.4318750050558336e-07,
      "loss": 0.6561,
      "step": 5606
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0041666195653278,
      "learning_rate": 2.423841594363469e-07,
      "loss": 0.5906,
      "step": 5607
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9715725656550747,
      "learning_rate": 2.415821144728092e-07,
      "loss": 0.5057,
      "step": 5608
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1820797523350357,
      "learning_rate": 2.4078136583347014e-07,
      "loss": 0.5647,
      "step": 5609
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9796705537021708,
      "learning_rate": 2.399819137364756e-07,
      "loss": 0.4694,
      "step": 5610
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9347358315090242,
      "learning_rate": 2.3918375839961927e-07,
      "loss": 0.5299,
      "step": 5611
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9991540671700192,
      "learning_rate": 2.3838690004033826e-07,
      "loss": 0.5376,
      "step": 5612
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1310369867254366,
      "learning_rate": 2.3759133887572117e-07,
      "loss": 0.6034,
      "step": 5613
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9215603401557225,
      "learning_rate": 2.3679707512250072e-07,
      "loss": 0.5203,
      "step": 5614
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0514627069091567,
      "learning_rate": 2.3600410899705383e-07,
      "loss": 0.5541,
      "step": 5615
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0350499966833382,
      "learning_rate": 2.3521244071540883e-07,
      "loss": 0.5925,
      "step": 5616
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.031460914229301,
      "learning_rate": 2.3442207049323707e-07,
      "loss": 0.5897,
      "step": 5617
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0078291330860885,
      "learning_rate": 2.3363299854585742e-07,
      "loss": 0.4935,
      "step": 5618
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0064122850718928,
      "learning_rate": 2.3284522508823403e-07,
      "loss": 0.5396,
      "step": 5619
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.057485856689331,
      "learning_rate": 2.320587503349797e-07,
      "loss": 0.544,
      "step": 5620
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9600073434806452,
      "learning_rate": 2.3127357450035025e-07,
      "loss": 0.495,
      "step": 5621
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0419411379111172,
      "learning_rate": 2.304896977982496e-07,
      "loss": 0.5436,
      "step": 5622
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6266454025245999,
      "learning_rate": 2.2970712044222742e-07,
      "loss": 0.4467,
      "step": 5623
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0552041361041875,
      "learning_rate": 2.2892584264548046e-07,
      "loss": 0.6166,
      "step": 5624
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0578903791416672,
      "learning_rate": 2.2814586462084897e-07,
      "loss": 0.5426,
      "step": 5625
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6218668535050852,
      "learning_rate": 2.2736718658082025e-07,
      "loss": 0.4685,
      "step": 5626
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0308873685227118,
      "learning_rate": 2.2658980873752902e-07,
      "loss": 0.5072,
      "step": 5627
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9872595760687692,
      "learning_rate": 2.2581373130275307e-07,
      "loss": 0.6359,
      "step": 5628
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9461594560136964,
      "learning_rate": 2.250389544879189e-07,
      "loss": 0.5133,
      "step": 5629
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9512881904970983,
      "learning_rate": 2.2426547850409598e-07,
      "loss": 0.5308,
      "step": 5630
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0140743877186862,
      "learning_rate": 2.234933035619996e-07,
      "loss": 0.5793,
      "step": 5631
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9687308362978071,
      "learning_rate": 2.2272242987199379e-07,
      "loss": 0.5817,
      "step": 5632
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0397100908649268,
      "learning_rate": 2.219528576440838e-07,
      "loss": 0.5632,
      "step": 5633
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9770546176333865,
      "learning_rate": 2.2118458708792256e-07,
      "loss": 0.552,
      "step": 5634
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9738079319773267,
      "learning_rate": 2.2041761841280873e-07,
      "loss": 0.5078,
      "step": 5635
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7228201611518009,
      "learning_rate": 2.196519518276874e-07,
      "loss": 0.4828,
      "step": 5636
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0118924593312792,
      "learning_rate": 2.1888758754114336e-07,
      "loss": 0.5149,
      "step": 5637
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9808364047946521,
      "learning_rate": 2.1812452576141285e-07,
      "loss": 0.5455,
      "step": 5638
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1022571526277614,
      "learning_rate": 2.1736276669637568e-07,
      "loss": 0.6014,
      "step": 5639
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.965735149413908,
      "learning_rate": 2.1660231055355473e-07,
      "loss": 0.5153,
      "step": 5640
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9606420238012073,
      "learning_rate": 2.1584315754011876e-07,
      "loss": 0.5643,
      "step": 5641
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0174527289956823,
      "learning_rate": 2.1508530786288284e-07,
      "loss": 0.5055,
      "step": 5642
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1231591946640616,
      "learning_rate": 2.143287617283074e-07,
      "loss": 0.6075,
      "step": 5643
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0155962549875126,
      "learning_rate": 2.135735193424937e-07,
      "loss": 0.4912,
      "step": 5644
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9496194154178266,
      "learning_rate": 2.1281958091119215e-07,
      "loss": 0.6203,
      "step": 5645
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0177730312752264,
      "learning_rate": 2.1206694663979676e-07,
      "loss": 0.566,
      "step": 5646
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9769314558605976,
      "learning_rate": 2.113156167333452e-07,
      "loss": 0.6019,
      "step": 5647
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.043883333046884,
      "learning_rate": 2.105655913965199e-07,
      "loss": 0.5344,
      "step": 5648
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0401883514881405,
      "learning_rate": 2.0981687083364955e-07,
      "loss": 0.5224,
      "step": 5649
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0067410360268318,
      "learning_rate": 2.0906945524870493e-07,
      "loss": 0.5839,
      "step": 5650
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9395106253423855,
      "learning_rate": 2.0832334484530424e-07,
      "loss": 0.5796,
      "step": 5651
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1258581587172127,
      "learning_rate": 2.0757853982670773e-07,
      "loss": 0.6504,
      "step": 5652
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.947602699409233,
      "learning_rate": 2.0683504039582024e-07,
      "loss": 0.5901,
      "step": 5653
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6669612913731927,
      "learning_rate": 2.0609284675519315e-07,
      "loss": 0.4684,
      "step": 5654
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0117106221943026,
      "learning_rate": 2.053519591070191e-07,
      "loss": 0.4657,
      "step": 5655
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0419734772538751,
      "learning_rate": 2.0461237765313612e-07,
      "loss": 0.5384,
      "step": 5656
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0364370514219166,
      "learning_rate": 2.0387410259502692e-07,
      "loss": 0.644,
      "step": 5657
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6109087541913896,
      "learning_rate": 2.0313713413381952e-07,
      "loss": 0.4458,
      "step": 5658
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0136842510244017,
      "learning_rate": 2.0240147247028164e-07,
      "loss": 0.5403,
      "step": 5659
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.047935503805689,
      "learning_rate": 2.016671178048296e-07,
      "loss": 0.5809,
      "step": 5660
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0568607901464904,
      "learning_rate": 2.0093407033752178e-07,
      "loss": 0.5509,
      "step": 5661
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9536937772409713,
      "learning_rate": 2.0020233026806003e-07,
      "loss": 0.4873,
      "step": 5662
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.016068484907409,
      "learning_rate": 1.9947189779579047e-07,
      "loss": 0.5177,
      "step": 5663
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.01372866943647,
      "learning_rate": 1.9874277311970335e-07,
      "loss": 0.5705,
      "step": 5664
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0964453122596354,
      "learning_rate": 1.9801495643843316e-07,
      "loss": 0.6649,
      "step": 5665
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0507715524085735,
      "learning_rate": 1.972884479502557e-07,
      "loss": 0.494,
      "step": 5666
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9795147367244392,
      "learning_rate": 1.965632478530921e-07,
      "loss": 0.5049,
      "step": 5667
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.918702237024255,
      "learning_rate": 1.9583935634450768e-07,
      "loss": 0.5642,
      "step": 5668
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.108550353338122,
      "learning_rate": 1.9511677362171078e-07,
      "loss": 0.5915,
      "step": 5669
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0336367266425566,
      "learning_rate": 1.943954998815506e-07,
      "loss": 0.6107,
      "step": 5670
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0163066736216737,
      "learning_rate": 1.9367553532052497e-07,
      "loss": 0.5603,
      "step": 5671
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9958782952940909,
      "learning_rate": 1.9295688013477031e-07,
      "loss": 0.5335,
      "step": 5672
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0419059487236775,
      "learning_rate": 1.9223953452006784e-07,
      "loss": 0.6191,
      "step": 5673
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9867827408125353,
      "learning_rate": 1.915234986718434e-07,
      "loss": 0.583,
      "step": 5674
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1035805609924016,
      "learning_rate": 1.908087727851632e-07,
      "loss": 0.5909,
      "step": 5675
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0640638087239234,
      "learning_rate": 1.9009535705474091e-07,
      "loss": 0.6196,
      "step": 5676
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9856650935100841,
      "learning_rate": 1.8938325167492888e-07,
      "loss": 0.6122,
      "step": 5677
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0406641386203577,
      "learning_rate": 1.8867245683972358e-07,
      "loss": 0.5901,
      "step": 5678
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0553211433519176,
      "learning_rate": 1.8796297274276677e-07,
      "loss": 0.6245,
      "step": 5679
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9562490584410693,
      "learning_rate": 1.8725479957734105e-07,
      "loss": 0.5086,
      "step": 5680
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.023846390618494,
      "learning_rate": 1.8654793753637156e-07,
      "loss": 0.5852,
      "step": 5681
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0790488532999576,
      "learning_rate": 1.8584238681242705e-07,
      "loss": 0.5812,
      "step": 5682
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0347801637233838,
      "learning_rate": 1.8513814759772042e-07,
      "loss": 0.5234,
      "step": 5683
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9608763991198356,
      "learning_rate": 1.844352200841043e-07,
      "loss": 0.5021,
      "step": 5684
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0555304915445853,
      "learning_rate": 1.8373360446307608e-07,
      "loss": 0.5191,
      "step": 5685
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9638197428882407,
      "learning_rate": 1.8303330092577453e-07,
      "loss": 0.5335,
      "step": 5686
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.072581173056278,
      "learning_rate": 1.823343096629837e-07,
      "loss": 0.6369,
      "step": 5687
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0848625696674354,
      "learning_rate": 1.8163663086512574e-07,
      "loss": 0.4998,
      "step": 5688
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9879044467968758,
      "learning_rate": 1.80940264722268e-07,
      "loss": 0.6271,
      "step": 5689
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.003648369682334,
      "learning_rate": 1.8024521142412155e-07,
      "loss": 0.6057,
      "step": 5690
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7008734723409707,
      "learning_rate": 1.7955147116003603e-07,
      "loss": 0.4891,
      "step": 5691
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.942205448049128,
      "learning_rate": 1.7885904411900633e-07,
      "loss": 0.5698,
      "step": 5692
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9677739316429045,
      "learning_rate": 1.7816793048966884e-07,
      "loss": 0.6258,
      "step": 5693
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.330364472853426,
      "learning_rate": 1.7747813046030184e-07,
      "loss": 0.5856,
      "step": 5694
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9871574962536415,
      "learning_rate": 1.7678964421882494e-07,
      "loss": 0.5688,
      "step": 5695
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0537375851310344,
      "learning_rate": 1.7610247195280207e-07,
      "loss": 0.5908,
      "step": 5696
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9527516972962567,
      "learning_rate": 1.7541661384943676e-07,
      "loss": 0.5401,
      "step": 5697
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9673482218804622,
      "learning_rate": 1.747320700955768e-07,
      "loss": 0.6761,
      "step": 5698
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0495548375639372,
      "learning_rate": 1.7404884087771023e-07,
      "loss": 0.5342,
      "step": 5699
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0395578120297386,
      "learning_rate": 1.7336692638196706e-07,
      "loss": 0.5758,
      "step": 5700
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0892405403178511,
      "learning_rate": 1.726863267941209e-07,
      "loss": 0.5799,
      "step": 5701
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9698192351580793,
      "learning_rate": 1.72007042299584e-07,
      "loss": 0.5248,
      "step": 5702
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0042716680228556,
      "learning_rate": 1.7132907308341383e-07,
      "loss": 0.626,
      "step": 5703
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0439854914496418,
      "learning_rate": 1.7065241933030664e-07,
      "loss": 0.6124,
      "step": 5704
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2328991660419708,
      "learning_rate": 1.6997708122460322e-07,
      "loss": 0.591,
      "step": 5705
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9988965526524847,
      "learning_rate": 1.6930305895028255e-07,
      "loss": 0.5668,
      "step": 5706
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9627770542481197,
      "learning_rate": 1.6863035269096774e-07,
      "loss": 0.5059,
      "step": 5707
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9788831261499522,
      "learning_rate": 1.6795896262992162e-07,
      "loss": 0.4812,
      "step": 5708
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9499693562799871,
      "learning_rate": 1.672888889500518e-07,
      "loss": 0.5326,
      "step": 5709
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0064958003630278,
      "learning_rate": 1.6662013183390112e-07,
      "loss": 0.5909,
      "step": 5710
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2256419873903477,
      "learning_rate": 1.6595269146366e-07,
      "loss": 0.5797,
      "step": 5711
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0126687153308378,
      "learning_rate": 1.6528656802115794e-07,
      "loss": 0.5904,
      "step": 5712
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.063817697119845,
      "learning_rate": 1.6462176168786371e-07,
      "loss": 0.5932,
      "step": 5713
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9987999802790458,
      "learning_rate": 1.6395827264488962e-07,
      "loss": 0.4862,
      "step": 5714
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9052428848949725,
      "learning_rate": 1.6329610107298833e-07,
      "loss": 0.4977,
      "step": 5715
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0204585195475446,
      "learning_rate": 1.6263524715255442e-07,
      "loss": 0.5578,
      "step": 5716
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0455663224062182,
      "learning_rate": 1.6197571106362054e-07,
      "loss": 0.5564,
      "step": 5717
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.036440472074687,
      "learning_rate": 1.6131749298586464e-07,
      "loss": 0.5709,
      "step": 5718
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0665461923133959,
      "learning_rate": 1.6066059309860273e-07,
      "loss": 0.639,
      "step": 5719
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9453739533683498,
      "learning_rate": 1.6000501158079217e-07,
      "loss": 0.5597,
      "step": 5720
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9294138299084963,
      "learning_rate": 1.5935074861103183e-07,
      "loss": 0.5368,
      "step": 5721
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0026262200191232,
      "learning_rate": 1.5869780436756078e-07,
      "loss": 0.6077,
      "step": 5722
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0199138356466153,
      "learning_rate": 1.5804617902825948e-07,
      "loss": 0.5314,
      "step": 5723
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6799222943976461,
      "learning_rate": 1.573958727706476e-07,
      "loss": 0.4525,
      "step": 5724
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9878787957496313,
      "learning_rate": 1.567468857718879e-07,
      "loss": 0.5243,
      "step": 5725
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0691929765081982,
      "learning_rate": 1.560992182087806e-07,
      "loss": 0.6273,
      "step": 5726
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9819726395135023,
      "learning_rate": 1.554528702577701e-07,
      "loss": 0.5903,
      "step": 5727
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9773675062929092,
      "learning_rate": 1.5480784209493837e-07,
      "loss": 0.5846,
      "step": 5728
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9414705551181769,
      "learning_rate": 1.5416413389600816e-07,
      "loss": 0.572,
      "step": 5729
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1266514391302263,
      "learning_rate": 1.5352174583634526e-07,
      "loss": 0.6172,
      "step": 5730
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0425727466367103,
      "learning_rate": 1.5288067809095196e-07,
      "loss": 0.6121,
      "step": 5731
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0771211729645591,
      "learning_rate": 1.5224093083447354e-07,
      "loss": 0.6217,
      "step": 5732
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9689854485630093,
      "learning_rate": 1.5160250424119505e-07,
      "loss": 0.4691,
      "step": 5733
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0183055889981414,
      "learning_rate": 1.509653984850412e-07,
      "loss": 0.5269,
      "step": 5734
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9650791639252029,
      "learning_rate": 1.5032961373957766e-07,
      "loss": 0.5759,
      "step": 5735
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5965172794468568,
      "learning_rate": 1.496951501780086e-07,
      "loss": 0.4555,
      "step": 5736
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.142808534888976,
      "learning_rate": 1.490620079731808e-07,
      "loss": 0.5938,
      "step": 5737
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0493650741030391,
      "learning_rate": 1.4843018729757853e-07,
      "loss": 0.5589,
      "step": 5738
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0182513073668857,
      "learning_rate": 1.4779968832332737e-07,
      "loss": 0.615,
      "step": 5739
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9343969332887979,
      "learning_rate": 1.4717051122219272e-07,
      "loss": 0.506,
      "step": 5740
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.016287246042637,
      "learning_rate": 1.4654265616557973e-07,
      "loss": 0.5642,
      "step": 5741
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.972119054787757,
      "learning_rate": 1.459161233245343e-07,
      "loss": 0.52,
      "step": 5742
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9775662336005666,
      "learning_rate": 1.4529091286973994e-07,
      "loss": 0.5977,
      "step": 5743
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0513593450720808,
      "learning_rate": 1.4466702497152208e-07,
      "loss": 0.5526,
      "step": 5744
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.061408736468417,
      "learning_rate": 1.4404445979984473e-07,
      "loss": 0.6584,
      "step": 5745
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9780178226874634,
      "learning_rate": 1.434232175243111e-07,
      "loss": 0.5526,
      "step": 5746
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9869974116198398,
      "learning_rate": 1.4280329831416585e-07,
      "loss": 0.5885,
      "step": 5747
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0554280892191013,
      "learning_rate": 1.4218470233829107e-07,
      "loss": 0.5779,
      "step": 5748
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0415155528439948,
      "learning_rate": 1.415674297652103e-07,
      "loss": 0.5735,
      "step": 5749
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0331095676887971,
      "learning_rate": 1.4095148076308518e-07,
      "loss": 0.6498,
      "step": 5750
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0421736968461854,
      "learning_rate": 1.4033685549971643e-07,
      "loss": 0.5419,
      "step": 5751
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0816728992584639,
      "learning_rate": 1.397235541425468e-07,
      "loss": 0.6044,
      "step": 5752
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.122301880380358,
      "learning_rate": 1.3911157685865483e-07,
      "loss": 0.6248,
      "step": 5753
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1438208355283332,
      "learning_rate": 1.3850092381476e-07,
      "loss": 0.5874,
      "step": 5754
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9871263683905064,
      "learning_rate": 1.3789159517722138e-07,
      "loss": 0.589,
      "step": 5755
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0060943984793174,
      "learning_rate": 1.3728359111203792e-07,
      "loss": 0.5474,
      "step": 5756
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0107584831464644,
      "learning_rate": 1.3667691178484598e-07,
      "loss": 0.5677,
      "step": 5757
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0161817245342006,
      "learning_rate": 1.360715573609206e-07,
      "loss": 0.5572,
      "step": 5758
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0308128861082348,
      "learning_rate": 1.354675280051787e-07,
      "loss": 0.5347,
      "step": 5759
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9786172304288007,
      "learning_rate": 1.348648238821737e-07,
      "loss": 0.5249,
      "step": 5760
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6227954474031245,
      "learning_rate": 1.3426344515609813e-07,
      "loss": 0.4707,
      "step": 5761
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6588872747239863,
      "learning_rate": 1.3366339199078538e-07,
      "loss": 0.4359,
      "step": 5762
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0073973195451362,
      "learning_rate": 1.3306466454970634e-07,
      "loss": 0.5871,
      "step": 5763
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0397322987053335,
      "learning_rate": 1.3246726299597058e-07,
      "loss": 0.6385,
      "step": 5764
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.063443920543598,
      "learning_rate": 1.318711874923262e-07,
      "loss": 0.6177,
      "step": 5765
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0609358052534776,
      "learning_rate": 1.312764382011611e-07,
      "loss": 0.6091,
      "step": 5766
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0591393382130974,
      "learning_rate": 1.3068301528450178e-07,
      "loss": 0.6093,
      "step": 5767
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9592990295281489,
      "learning_rate": 1.3009091890401226e-07,
      "loss": 0.4768,
      "step": 5768
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1286418862057355,
      "learning_rate": 1.2950014922099685e-07,
      "loss": 0.67,
      "step": 5769
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0664278554648636,
      "learning_rate": 1.2891070639639625e-07,
      "loss": 0.5342,
      "step": 5770
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9826537464362525,
      "learning_rate": 1.28322590590792e-07,
      "loss": 0.5697,
      "step": 5771
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9403362853099366,
      "learning_rate": 1.2773580196440262e-07,
      "loss": 0.571,
      "step": 5772
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9683736661096697,
      "learning_rate": 1.271503406770852e-07,
      "loss": 0.5607,
      "step": 5773
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0866259033120709,
      "learning_rate": 1.265662068883361e-07,
      "loss": 0.5242,
      "step": 5774
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.652245077830759,
      "learning_rate": 1.2598340075728967e-07,
      "loss": 0.4459,
      "step": 5775
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9748266944869681,
      "learning_rate": 1.254019224427172e-07,
      "loss": 0.5137,
      "step": 5776
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0717814362971791,
      "learning_rate": 1.2482177210303036e-07,
      "loss": 0.6168,
      "step": 5777
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0889820686915248,
      "learning_rate": 1.242429498962783e-07,
      "loss": 0.5683,
      "step": 5778
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0859607130585227,
      "learning_rate": 1.23665455980147e-07,
      "loss": 0.5559,
      "step": 5779
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0624214267699803,
      "learning_rate": 1.2308929051196296e-07,
      "loss": 0.5611,
      "step": 5780
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0238316214607182,
      "learning_rate": 1.2251445364868886e-07,
      "loss": 0.5959,
      "step": 5781
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.676344168231846,
      "learning_rate": 1.2194094554692614e-07,
      "loss": 0.4512,
      "step": 5782
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0026834136636622,
      "learning_rate": 1.213687663629143e-07,
      "loss": 0.5587,
      "step": 5783
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.982843265233904,
      "learning_rate": 1.2079791625253078e-07,
      "loss": 0.5644,
      "step": 5784
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9754653861413627,
      "learning_rate": 1.2022839537129127e-07,
      "loss": 0.5387,
      "step": 5785
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0522169351760533,
      "learning_rate": 1.196602038743483e-07,
      "loss": 0.6152,
      "step": 5786
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9751132834068625,
      "learning_rate": 1.1909334191649302e-07,
      "loss": 0.5185,
      "step": 5787
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0013185249832652,
      "learning_rate": 1.1852780965215471e-07,
      "loss": 0.5584,
      "step": 5788
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0304719621313692,
      "learning_rate": 1.1796360723539956e-07,
      "loss": 0.5886,
      "step": 5789
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9788791357776215,
      "learning_rate": 1.174007348199313e-07,
      "loss": 0.4906,
      "step": 5790
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0450787272003623,
      "learning_rate": 1.1683919255909337e-07,
      "loss": 0.5662,
      "step": 5791
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0408351561393512,
      "learning_rate": 1.1627898060586395e-07,
      "loss": 0.5815,
      "step": 5792
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0884220928793564,
      "learning_rate": 1.1572009911286097e-07,
      "loss": 0.57,
      "step": 5793
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0713184634646309,
      "learning_rate": 1.151625482323393e-07,
      "loss": 0.6207,
      "step": 5794
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1160283150969414,
      "learning_rate": 1.1460632811619021e-07,
      "loss": 0.6513,
      "step": 5795
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9962178038413154,
      "learning_rate": 1.1405143891594417e-07,
      "loss": 0.6872,
      "step": 5796
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.042475809805722,
      "learning_rate": 1.1349788078276857e-07,
      "loss": 0.5896,
      "step": 5797
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9681189102028814,
      "learning_rate": 1.129456538674667e-07,
      "loss": 0.5233,
      "step": 5798
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0004304131952244,
      "learning_rate": 1.1239475832048152e-07,
      "loss": 0.596,
      "step": 5799
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.082046235907496,
      "learning_rate": 1.1184519429189245e-07,
      "loss": 0.6274,
      "step": 5800
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9925946848209398,
      "learning_rate": 1.1129696193141414e-07,
      "loss": 0.5737,
      "step": 5801
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1232155854424806,
      "learning_rate": 1.107500613884016e-07,
      "loss": 0.5956,
      "step": 5802
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9863741399379653,
      "learning_rate": 1.1020449281184565e-07,
      "loss": 0.5304,
      "step": 5803
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.010276648973911,
      "learning_rate": 1.0966025635037403e-07,
      "loss": 0.5608,
      "step": 5804
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9572718717986605,
      "learning_rate": 1.0911735215225095e-07,
      "loss": 0.5191,
      "step": 5805
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9871357958617829,
      "learning_rate": 1.0857578036537919e-07,
      "loss": 0.5595,
      "step": 5806
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9691552688410355,
      "learning_rate": 1.0803554113729797e-07,
      "loss": 0.5508,
      "step": 5807
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9244410317940109,
      "learning_rate": 1.0749663461518344e-07,
      "loss": 0.4673,
      "step": 5808
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0252864380676119,
      "learning_rate": 1.0695906094584763e-07,
      "loss": 0.4839,
      "step": 5809
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0923606935668153,
      "learning_rate": 1.0642282027574168e-07,
      "loss": 0.6282,
      "step": 5810
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9959640840085532,
      "learning_rate": 1.0588791275095156e-07,
      "loss": 0.5691,
      "step": 5811
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0102238333022042,
      "learning_rate": 1.0535433851720067e-07,
      "loss": 0.6104,
      "step": 5812
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0374208702707612,
      "learning_rate": 1.0482209771985052e-07,
      "loss": 0.6268,
      "step": 5813
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.094712556063163,
      "learning_rate": 1.0429119050389624e-07,
      "loss": 0.5892,
      "step": 5814
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0082232269653209,
      "learning_rate": 1.0376161701397325e-07,
      "loss": 0.5411,
      "step": 5815
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9365700852822209,
      "learning_rate": 1.0323337739435169e-07,
      "loss": 0.5505,
      "step": 5816
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0761601481580647,
      "learning_rate": 1.0270647178893756e-07,
      "loss": 0.5205,
      "step": 5817
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9556843770183094,
      "learning_rate": 1.0218090034127603e-07,
      "loss": 0.5018,
      "step": 5818
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7275192326864739,
      "learning_rate": 1.0165666319454592e-07,
      "loss": 0.4696,
      "step": 5819
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.000482895881363,
      "learning_rate": 1.011337604915641e-07,
      "loss": 0.4971,
      "step": 5820
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9297404839360337,
      "learning_rate": 1.0061219237478382e-07,
      "loss": 0.4627,
      "step": 5821
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0507514866501615,
      "learning_rate": 1.0009195898629532e-07,
      "loss": 0.6638,
      "step": 5822
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9545611305913939,
      "learning_rate": 9.957306046782411e-08,
      "loss": 0.5016,
      "step": 5823
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0432008010995888,
      "learning_rate": 9.905549696073213e-08,
      "loss": 0.505,
      "step": 5824
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.189490967055027,
      "learning_rate": 9.85392686060177e-08,
      "loss": 0.6521,
      "step": 5825
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9706825595759822,
      "learning_rate": 9.802437554431665e-08,
      "loss": 0.5421,
      "step": 5826
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9949154007473566,
      "learning_rate": 9.7510817915899e-08,
      "loss": 0.6245,
      "step": 5827
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0633350709549745,
      "learning_rate": 9.699859586067229e-08,
      "loss": 0.5608,
      "step": 5828
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.165489803115361,
      "learning_rate": 9.648770951818098e-08,
      "loss": 0.6725,
      "step": 5829
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9568595411791763,
      "learning_rate": 9.597815902760321e-08,
      "loss": 0.5796,
      "step": 5830
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.158010647017492,
      "learning_rate": 9.54699445277546e-08,
      "loss": 0.5982,
      "step": 5831
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0634975790965278,
      "learning_rate": 9.496306615708773e-08,
      "loss": 0.5952,
      "step": 5832
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9860782899282723,
      "learning_rate": 9.44575240536899e-08,
      "loss": 0.5089,
      "step": 5833
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.989654658190415,
      "learning_rate": 9.395331835528431e-08,
      "loss": 0.5988,
      "step": 5834
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9790488626467019,
      "learning_rate": 9.345044919923108e-08,
      "loss": 0.5587,
      "step": 5835
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1034460187704282,
      "learning_rate": 9.294891672252448e-08,
      "loss": 0.5795,
      "step": 5836
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0095153520749391,
      "learning_rate": 9.244872106179748e-08,
      "loss": 0.5327,
      "step": 5837
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0255386318725004,
      "learning_rate": 9.194986235331604e-08,
      "loss": 0.6032,
      "step": 5838
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0842431932188958,
      "learning_rate": 9.145234073298314e-08,
      "loss": 0.5516,
      "step": 5839
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9539224188618249,
      "learning_rate": 9.095615633633814e-08,
      "loss": 0.469,
      "step": 5840
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1255803693516824,
      "learning_rate": 9.046130929855401e-08,
      "loss": 0.6669,
      "step": 5841
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1008710248296885,
      "learning_rate": 8.99677997544418e-08,
      "loss": 0.6232,
      "step": 5842
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0536205724554721,
      "learning_rate": 8.947562783844677e-08,
      "loss": 0.5702,
      "step": 5843
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0785764160775642,
      "learning_rate": 8.898479368464996e-08,
      "loss": 0.6063,
      "step": 5844
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0971332266116183,
      "learning_rate": 8.849529742676887e-08,
      "loss": 0.5751,
      "step": 5845
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0409547002377724,
      "learning_rate": 8.800713919815407e-08,
      "loss": 0.6088,
      "step": 5846
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9776980058593518,
      "learning_rate": 8.752031913179527e-08,
      "loss": 0.5463,
      "step": 5847
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.947627119924481,
      "learning_rate": 8.70348373603147e-08,
      "loss": 0.5821,
      "step": 5848
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0558757987215825,
      "learning_rate": 8.655069401597105e-08,
      "loss": 0.5534,
      "step": 5849
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.97947724745948,
      "learning_rate": 8.606788923065824e-08,
      "loss": 0.5325,
      "step": 5850
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.01175414455452,
      "learning_rate": 8.558642313590603e-08,
      "loss": 0.5486,
      "step": 5851
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0468003526242884,
      "learning_rate": 8.510629586287844e-08,
      "loss": 0.5141,
      "step": 5852
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9846697701846534,
      "learning_rate": 8.462750754237581e-08,
      "loss": 0.5417,
      "step": 5853
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9896046682575587,
      "learning_rate": 8.415005830483324e-08,
      "loss": 0.5053,
      "step": 5854
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0539101287136707,
      "learning_rate": 8.367394828032116e-08,
      "loss": 0.619,
      "step": 5855
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9329023943333974,
      "learning_rate": 8.319917759854413e-08,
      "loss": 0.5034,
      "step": 5856
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0244137264455406,
      "learning_rate": 8.272574638884423e-08,
      "loss": 0.6315,
      "step": 5857
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0128514210422985,
      "learning_rate": 8.225365478019664e-08,
      "loss": 0.617,
      "step": 5858
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.911444459882442,
      "learning_rate": 8.178290290121127e-08,
      "loss": 0.5044,
      "step": 5859
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0068624399303743,
      "learning_rate": 8.131349088013495e-08,
      "loss": 0.6689,
      "step": 5860
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2237416737767142,
      "learning_rate": 8.08454188448482e-08,
      "loss": 0.6577,
      "step": 5861
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.104144846841128,
      "learning_rate": 8.037868692286677e-08,
      "loss": 0.5315,
      "step": 5862
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1301577793979476,
      "learning_rate": 7.991329524134062e-08,
      "loss": 0.5844,
      "step": 5863
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0017748082345865,
      "learning_rate": 7.944924392705666e-08,
      "loss": 0.5658,
      "step": 5864
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0181163207448154,
      "learning_rate": 7.898653310643378e-08,
      "loss": 0.5521,
      "step": 5865
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1151265860275503,
      "learning_rate": 7.852516290552781e-08,
      "loss": 0.5873,
      "step": 5866
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9228355974580896,
      "learning_rate": 7.80651334500293e-08,
      "loss": 0.5088,
      "step": 5867
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9449438838911601,
      "learning_rate": 7.760644486526137e-08,
      "loss": 0.4749,
      "step": 5868
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1008864766550193,
      "learning_rate": 7.714909727618569e-08,
      "loss": 0.6327,
      "step": 5869
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9455451802672485,
      "learning_rate": 7.669309080739429e-08,
      "loss": 0.581,
      "step": 5870
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9312526334078194,
      "learning_rate": 7.623842558311668e-08,
      "loss": 0.5786,
      "step": 5871
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0410155610198526,
      "learning_rate": 7.578510172721598e-08,
      "loss": 0.5632,
      "step": 5872
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.979888148188077,
      "learning_rate": 7.533311936319121e-08,
      "loss": 0.5616,
      "step": 5873
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0364545794796396,
      "learning_rate": 7.48824786141733e-08,
      "loss": 0.5908,
      "step": 5874
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0378036120632044,
      "learning_rate": 7.443317960293018e-08,
      "loss": 0.5411,
      "step": 5875
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0180290679657185,
      "learning_rate": 7.39852224518628e-08,
      "loss": 0.5724,
      "step": 5876
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0102988192271567,
      "learning_rate": 7.353860728300743e-08,
      "loss": 0.5622,
      "step": 5877
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0540422263947613,
      "learning_rate": 7.3093334218034e-08,
      "loss": 0.5991,
      "step": 5878
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0745185538079773,
      "learning_rate": 7.264940337824767e-08,
      "loss": 0.6116,
      "step": 5879
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9289926911556593,
      "learning_rate": 7.22068148845867e-08,
      "loss": 0.5246,
      "step": 5880
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0698376288189828,
      "learning_rate": 7.176556885762465e-08,
      "loss": 0.6335,
      "step": 5881
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0860983073356667,
      "learning_rate": 7.132566541756925e-08,
      "loss": 0.6587,
      "step": 5882
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0037205561518752,
      "learning_rate": 7.088710468426241e-08,
      "loss": 0.6316,
      "step": 5883
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9654361445621602,
      "learning_rate": 7.044988677718023e-08,
      "loss": 0.4967,
      "step": 5884
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0913921664766342,
      "learning_rate": 7.001401181543243e-08,
      "loss": 0.6184,
      "step": 5885
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.04391949061577,
      "learning_rate": 6.957947991776403e-08,
      "loss": 0.592,
      "step": 5886
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0419761614941379,
      "learning_rate": 6.914629120255312e-08,
      "loss": 0.5841,
      "step": 5887
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.083068346517522,
      "learning_rate": 6.87144457878114e-08,
      "loss": 0.6112,
      "step": 5888
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0691754992437996,
      "learning_rate": 6.828394379118752e-08,
      "loss": 0.6316,
      "step": 5889
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.988836853077234,
      "learning_rate": 6.785478532995993e-08,
      "loss": 0.5444,
      "step": 5890
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0109803459973528,
      "learning_rate": 6.742697052104507e-08,
      "loss": 0.5392,
      "step": 5891
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.069822207359979,
      "learning_rate": 6.70004994809903e-08,
      "loss": 0.6435,
      "step": 5892
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0340221010661947,
      "learning_rate": 6.657537232597766e-08,
      "loss": 0.5679,
      "step": 5893
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0627810358568945,
      "learning_rate": 6.615158917182507e-08,
      "loss": 0.5926,
      "step": 5894
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.994793382533743,
      "learning_rate": 6.572915013398184e-08,
      "loss": 0.6034,
      "step": 5895
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.995469060372368,
      "learning_rate": 6.530805532753204e-08,
      "loss": 0.5311,
      "step": 5896
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0215230361837382,
      "learning_rate": 6.488830486719333e-08,
      "loss": 0.5905,
      "step": 5897
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.033712323412795,
      "learning_rate": 6.446989886731758e-08,
      "loss": 0.587,
      "step": 5898
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0223376808357088,
      "learning_rate": 6.405283744189027e-08,
      "loss": 0.5936,
      "step": 5899
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.00281986016423,
      "learning_rate": 6.363712070452999e-08,
      "loss": 0.5675,
      "step": 5900
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9823714504948695,
      "learning_rate": 6.322274876848944e-08,
      "loss": 0.5739,
      "step": 5901
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9993089425030285,
      "learning_rate": 6.280972174665611e-08,
      "loss": 0.5891,
      "step": 5902
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1230968393242087,
      "learning_rate": 6.239803975154835e-08,
      "loss": 0.5563,
      "step": 5903
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9949818208983318,
      "learning_rate": 6.19877028953203e-08,
      "loss": 0.534,
      "step": 5904
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0319019341731246,
      "learning_rate": 6.157871128975923e-08,
      "loss": 0.5755,
      "step": 5905
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0068243931849128,
      "learning_rate": 6.1171065046286e-08,
      "loss": 0.6079,
      "step": 5906
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.018423274934718,
      "learning_rate": 6.076476427595345e-08,
      "loss": 0.5996,
      "step": 5907
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0262417943226345,
      "learning_rate": 6.035980908945027e-08,
      "loss": 0.6006,
      "step": 5908
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0570430283449896,
      "learning_rate": 5.995619959709764e-08,
      "loss": 0.5896,
      "step": 5909
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0761374532356365,
      "learning_rate": 5.9553935908848724e-08,
      "loss": 0.5642,
      "step": 5910
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.122054330656811,
      "learning_rate": 5.915301813429197e-08,
      "loss": 0.5387,
      "step": 5911
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.042170259835966,
      "learning_rate": 5.875344638264835e-08,
      "loss": 0.6282,
      "step": 5912
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0077494408115886,
      "learning_rate": 5.835522076277189e-08,
      "loss": 0.5424,
      "step": 5913
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0230562817392699,
      "learning_rate": 5.795834138315137e-08,
      "loss": 0.603,
      "step": 5914
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0696472628994662,
      "learning_rate": 5.756280835190586e-08,
      "loss": 0.6047,
      "step": 5915
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0999515788597323,
      "learning_rate": 5.7168621776790836e-08,
      "loss": 0.5857,
      "step": 5916
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6678274730734242,
      "learning_rate": 5.6775781765193714e-08,
      "loss": 0.4339,
      "step": 5917
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9288694672933757,
      "learning_rate": 5.638428842413335e-08,
      "loss": 0.5441,
      "step": 5918
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9546129438615574,
      "learning_rate": 5.599414186026497e-08,
      "loss": 0.6183,
      "step": 5919
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1239097448911732,
      "learning_rate": 5.5605342179874676e-08,
      "loss": 0.6101,
      "step": 5920
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0612734427377728,
      "learning_rate": 5.5217889488882734e-08,
      "loss": 0.6476,
      "step": 5921
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0986691469737098,
      "learning_rate": 5.4831783892840275e-08,
      "loss": 0.535,
      "step": 5922
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0486178644805468,
      "learning_rate": 5.444702549693481e-08,
      "loss": 0.5622,
      "step": 5923
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0515931427585945,
      "learning_rate": 5.406361440598529e-08,
      "loss": 0.5598,
      "step": 5924
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0528764721211823,
      "learning_rate": 5.368155072444148e-08,
      "loss": 0.525,
      "step": 5925
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2108866243703074,
      "learning_rate": 5.330083455638957e-08,
      "loss": 0.5574,
      "step": 5926
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9822537781329186,
      "learning_rate": 5.2921466005547706e-08,
      "loss": 0.5721,
      "step": 5927
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1114821790273113,
      "learning_rate": 5.2543445175264883e-08,
      "loss": 0.5788,
      "step": 5928
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0349070822170474,
      "learning_rate": 5.216677216852539e-08,
      "loss": 0.5853,
      "step": 5929
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0309374259114716,
      "learning_rate": 5.179144708794437e-08,
      "loss": 0.5539,
      "step": 5930
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0121527434692474,
      "learning_rate": 5.141747003577224e-08,
      "loss": 0.5274,
      "step": 5931
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.036241680769594,
      "learning_rate": 5.104484111388919e-08,
      "loss": 0.6459,
      "step": 5932
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0473748186048766,
      "learning_rate": 5.067356042381011e-08,
      "loss": 0.5392,
      "step": 5933
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0136360800078754,
      "learning_rate": 5.0303628066681874e-08,
      "loss": 0.522,
      "step": 5934
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.012518778975016,
      "learning_rate": 4.9935044143284984e-08,
      "loss": 0.5662,
      "step": 5935
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9789977408359223,
      "learning_rate": 4.956780875403189e-08,
      "loss": 0.5089,
      "step": 5936
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0064219777351655,
      "learning_rate": 4.920192199896645e-08,
      "loss": 0.5407,
      "step": 5937
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0533164127089127,
      "learning_rate": 4.883738397776727e-08,
      "loss": 0.6073,
      "step": 5938
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0516117085533667,
      "learning_rate": 4.84741947897438e-08,
      "loss": 0.5379,
      "step": 5939
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9722539234449139,
      "learning_rate": 4.8112354533839664e-08,
      "loss": 0.559,
      "step": 5940
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0359129173947965,
      "learning_rate": 4.775186330862991e-08,
      "loss": 0.6012,
      "step": 5941
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1427769649015262,
      "learning_rate": 4.7392721212322076e-08,
      "loss": 0.6105,
      "step": 5942
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9837993358718313,
      "learning_rate": 4.703492834275625e-08,
      "loss": 0.5505,
      "step": 5943
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0015468148964117,
      "learning_rate": 4.6678484797405e-08,
      "loss": 0.5739,
      "step": 5944
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0974162753954928,
      "learning_rate": 4.6323390673373434e-08,
      "loss": 0.7025,
      "step": 5945
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9968159389320547,
      "learning_rate": 4.5969646067400285e-08,
      "loss": 0.5641,
      "step": 5946
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.035479829165455,
      "learning_rate": 4.561725107585346e-08,
      "loss": 0.5722,
      "step": 5947
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0281188604673126,
      "learning_rate": 4.5266205794735617e-08,
      "loss": 0.6042,
      "step": 5948
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.013322386425791,
      "learning_rate": 4.491651031968136e-08,
      "loss": 0.6053,
      "step": 5949
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6667967103538348,
      "learning_rate": 4.456816474595782e-08,
      "loss": 0.4382,
      "step": 5950
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0123668350722073,
      "learning_rate": 4.4221169168462975e-08,
      "loss": 0.4831,
      "step": 5951
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.102265518270378,
      "learning_rate": 4.387552368172898e-08,
      "loss": 0.5949,
      "step": 5952
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1509554567397842,
      "learning_rate": 4.35312283799183e-08,
      "loss": 0.6086,
      "step": 5953
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.015887155827983,
      "learning_rate": 4.318828335682701e-08,
      "loss": 0.5947,
      "step": 5954
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9909227706323633,
      "learning_rate": 4.28466887058826e-08,
      "loss": 0.5841,
      "step": 5955
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9510704396811385,
      "learning_rate": 4.250644452014507e-08,
      "loss": 0.5954,
      "step": 5956
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9436507208142618,
      "learning_rate": 4.216755089230584e-08,
      "loss": 0.5294,
      "step": 5957
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9636949924797883,
      "learning_rate": 4.183000791468994e-08,
      "loss": 0.6215,
      "step": 5958
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9996279796476775,
      "learning_rate": 4.149381567925215e-08,
      "loss": 0.5767,
      "step": 5959
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0244527099554446,
      "learning_rate": 4.1158974277580866e-08,
      "loss": 0.591,
      "step": 5960
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0087529619898716,
      "learning_rate": 4.0825483800895905e-08,
      "loss": 0.5452,
      "step": 5961
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0517600076489957,
      "learning_rate": 4.04933443400507e-08,
      "loss": 0.5124,
      "step": 5962
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0195473040672522,
      "learning_rate": 4.0162555985526766e-08,
      "loss": 0.532,
      "step": 5963
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9887309734278957,
      "learning_rate": 3.9833118827442565e-08,
      "loss": 0.55,
      "step": 5964
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0391874702430555,
      "learning_rate": 3.950503295554409e-08,
      "loss": 0.5959,
      "step": 5965
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6342514427984829,
      "learning_rate": 3.917829845921095e-08,
      "loss": 0.4595,
      "step": 5966
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.984273316744748,
      "learning_rate": 3.885291542745584e-08,
      "loss": 0.4875,
      "step": 5967
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.939514839659328,
      "learning_rate": 3.8528883948921183e-08,
      "loss": 0.5456,
      "step": 5968
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.015766670228296,
      "learning_rate": 3.8206204111882475e-08,
      "loss": 0.5938,
      "step": 5969
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6580458998161236,
      "learning_rate": 3.788487600424606e-08,
      "loss": 0.4838,
      "step": 5970
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.149977093564522,
      "learning_rate": 3.7564899713550815e-08,
      "loss": 0.6022,
      "step": 5971
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9457986939808961,
      "learning_rate": 3.724627532696812e-08,
      "loss": 0.4883,
      "step": 5972
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0280661459406126,
      "learning_rate": 3.6929002931297975e-08,
      "loss": 0.5579,
      "step": 5973
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0247971418262811,
      "learning_rate": 3.661308261297625e-08,
      "loss": 0.5172,
      "step": 5974
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0717537955333103,
      "learning_rate": 3.629851445806687e-08,
      "loss": 0.6657,
      "step": 5975
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1129269698927888,
      "learning_rate": 3.59852985522674e-08,
      "loss": 0.5521,
      "step": 5976
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.049013127165195,
      "learning_rate": 3.5673434980906806e-08,
      "loss": 0.6346,
      "step": 5977
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1147235334605654,
      "learning_rate": 3.536292382894435e-08,
      "loss": 0.6262,
      "step": 5978
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.649763230737416,
      "learning_rate": 3.505376518097292e-08,
      "loss": 0.4652,
      "step": 5979
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.028686180399358,
      "learning_rate": 3.474595912121514e-08,
      "loss": 0.5233,
      "step": 5980
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6687192185715938,
      "learning_rate": 3.443950573352672e-08,
      "loss": 0.4527,
      "step": 5981
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9934755332403051,
      "learning_rate": 3.41344051013931e-08,
      "loss": 0.568,
      "step": 5982
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.047055680393411,
      "learning_rate": 3.3830657307932224e-08,
      "loss": 0.5356,
      "step": 5983
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0645050359885233,
      "learning_rate": 3.352826243589346e-08,
      "loss": 0.6017,
      "step": 5984
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9829859783603944,
      "learning_rate": 3.322722056765759e-08,
      "loss": 0.5861,
      "step": 5985
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.933033284978129,
      "learning_rate": 3.292753178523733e-08,
      "loss": 0.4298,
      "step": 5986
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.990088279543204,
      "learning_rate": 3.262919617027516e-08,
      "loss": 0.486,
      "step": 5987
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1023272300908387,
      "learning_rate": 3.233221380404605e-08,
      "loss": 0.6508,
      "step": 5988
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0570378646831333,
      "learning_rate": 3.2036584767456965e-08,
      "loss": 0.6037,
      "step": 5989
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1745892198345733,
      "learning_rate": 3.1742309141044594e-08,
      "loss": 0.6475,
      "step": 5990
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.010077456937398,
      "learning_rate": 3.1449387004978125e-08,
      "loss": 0.5963,
      "step": 5991
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0686158008622504,
      "learning_rate": 3.115781843905763e-08,
      "loss": 0.6162,
      "step": 5992
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0950986324380942,
      "learning_rate": 3.086760352271401e-08,
      "loss": 0.6065,
      "step": 5993
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9944661068018521,
      "learning_rate": 3.057874233501068e-08,
      "loss": 0.5392,
      "step": 5994
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0005636005788723,
      "learning_rate": 3.0291234954640256e-08,
      "loss": 0.5621,
      "step": 5995
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0380618235767174,
      "learning_rate": 3.0005081459928976e-08,
      "loss": 0.5696,
      "step": 5996
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9716246179147302,
      "learning_rate": 2.9720281928831694e-08,
      "loss": 0.4748,
      "step": 5997
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1186111012642908,
      "learning_rate": 2.9436836438936356e-08,
      "loss": 0.5986,
      "step": 5998
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0699754116151239,
      "learning_rate": 2.9154745067460632e-08,
      "loss": 0.5617,
      "step": 5999
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0359113211246755,
      "learning_rate": 2.8874007891255273e-08,
      "loss": 0.645,
      "step": 6000
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9880499982328612,
      "learning_rate": 2.859462498679966e-08,
      "loss": 0.6027,
      "step": 6001
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0288891848585462,
      "learning_rate": 2.831659643020568e-08,
      "loss": 0.6045,
      "step": 6002
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1018752004272745,
      "learning_rate": 2.8039922297216638e-08,
      "loss": 0.5773,
      "step": 6003
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0683884670752244,
      "learning_rate": 2.776460266320502e-08,
      "loss": 0.6113,
      "step": 6004
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0517028291846477,
      "learning_rate": 2.7490637603176385e-08,
      "loss": 0.5133,
      "step": 6005
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0005565705303059,
      "learning_rate": 2.7218027191766027e-08,
      "loss": 0.5172,
      "step": 6006
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0184607254375562,
      "learning_rate": 2.6946771503240653e-08,
      "loss": 0.5846,
      "step": 6007
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0109404072443888,
      "learning_rate": 2.6676870611497817e-08,
      "loss": 0.5254,
      "step": 6008
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0353208929479085,
      "learning_rate": 2.6408324590065926e-08,
      "loss": 0.656,
      "step": 6009
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0101639011165042,
      "learning_rate": 2.6141133512103677e-08,
      "loss": 0.5457,
      "step": 6010
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0448329217035721,
      "learning_rate": 2.5875297450402848e-08,
      "loss": 0.6158,
      "step": 6011
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9986518312080574,
      "learning_rate": 2.5610816477382728e-08,
      "loss": 0.6002,
      "step": 6012
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0393833225269538,
      "learning_rate": 2.5347690665096236e-08,
      "loss": 0.555,
      "step": 6013
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0739016027471615,
      "learning_rate": 2.5085920085226035e-08,
      "loss": 0.529,
      "step": 6014
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1066436119120648,
      "learning_rate": 2.482550480908563e-08,
      "loss": 0.6016,
      "step": 6015
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1197941456947438,
      "learning_rate": 2.456644490761939e-08,
      "loss": 0.5893,
      "step": 6016
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9869868324157809,
      "learning_rate": 2.4308740451401413e-08,
      "loss": 0.5592,
      "step": 6017
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1033168990058149,
      "learning_rate": 2.4052391510638873e-08,
      "loss": 0.6427,
      "step": 6018
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.022031404359787,
      "learning_rate": 2.379739815516757e-08,
      "loss": 0.5448,
      "step": 6019
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9654395401859253,
      "learning_rate": 2.354376045445472e-08,
      "loss": 0.5427,
      "step": 6020
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9696483833118121,
      "learning_rate": 2.3291478477598383e-08,
      "loss": 0.5388,
      "step": 6021
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.042697620614537,
      "learning_rate": 2.304055229332691e-08,
      "loss": 0.5265,
      "step": 6022
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0403060616026418,
      "learning_rate": 2.279098197000007e-08,
      "loss": 0.5569,
      "step": 6023
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0929912311388212,
      "learning_rate": 2.254276757560736e-08,
      "loss": 0.5695,
      "step": 6024
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9993848019720345,
      "learning_rate": 2.2295909177769138e-08,
      "loss": 0.6064,
      "step": 6025
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0506677649019072,
      "learning_rate": 2.205040684373605e-08,
      "loss": 0.5429,
      "step": 6026
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0384675906242966,
      "learning_rate": 2.1806260640390155e-08,
      "loss": 0.6062,
      "step": 6027
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9247310579188803,
      "learning_rate": 2.15634706342438e-08,
      "loss": 0.4775,
      "step": 6028
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0665385886185803,
      "learning_rate": 2.1322036891439634e-08,
      "loss": 0.6205,
      "step": 6029
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1784458536650553,
      "learning_rate": 2.1081959477750604e-08,
      "loss": 0.6951,
      "step": 6030
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0927106332328886,
      "learning_rate": 2.0843238458580494e-08,
      "loss": 0.6427,
      "step": 6031
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.03369561541982,
      "learning_rate": 2.060587389896285e-08,
      "loss": 0.5974,
      "step": 6032
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0992525353131506,
      "learning_rate": 2.0369865863563708e-08,
      "loss": 0.6062,
      "step": 6033
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0417870840008874,
      "learning_rate": 2.0135214416677205e-08,
      "loss": 0.5374,
      "step": 6034
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0321039099791618,
      "learning_rate": 1.990191962222887e-08,
      "loss": 0.6301,
      "step": 6035
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0386259060362732,
      "learning_rate": 1.9669981543775085e-08,
      "loss": 0.5964,
      "step": 6036
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0568892260790435,
      "learning_rate": 1.943940024450197e-08,
      "loss": 0.6277,
      "step": 6037
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.060713907273282,
      "learning_rate": 1.9210175787226503e-08,
      "loss": 0.5961,
      "step": 6038
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9682470881197897,
      "learning_rate": 1.898230823439484e-08,
      "loss": 0.4966,
      "step": 6039
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.060418836250532,
      "learning_rate": 1.8755797648085105e-08,
      "loss": 0.5889,
      "step": 6040
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0359832530880622,
      "learning_rate": 1.8530644090005156e-08,
      "loss": 0.6253,
      "step": 6041
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0308354518405056,
      "learning_rate": 1.8306847621492042e-08,
      "loss": 0.5519,
      "step": 6042
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9826030975887552,
      "learning_rate": 1.808440830351532e-08,
      "loss": 0.5388,
      "step": 6043
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0209681616769581,
      "learning_rate": 1.7863326196673193e-08,
      "loss": 0.5491,
      "step": 6044
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0161220416178605,
      "learning_rate": 1.764360136119414e-08,
      "loss": 0.5689,
      "step": 6045
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0171699496113011,
      "learning_rate": 1.742523385693806e-08,
      "loss": 0.5552,
      "step": 6046
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0068480346224993,
      "learning_rate": 1.7208223743392927e-08,
      "loss": 0.4944,
      "step": 6047
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9376974781085439,
      "learning_rate": 1.6992571079679775e-08,
      "loss": 0.4934,
      "step": 6048
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.122715469435472,
      "learning_rate": 1.6778275924547726e-08,
      "loss": 0.5731,
      "step": 6049
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1345736771873207,
      "learning_rate": 1.656533833637619e-08,
      "loss": 0.6813,
      "step": 6050
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9876240670428966,
      "learning_rate": 1.6353758373175986e-08,
      "loss": 0.6256,
      "step": 6051
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0243227261975625,
      "learning_rate": 1.6143536092586564e-08,
      "loss": 0.6075,
      "step": 6052
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0491101871978217,
      "learning_rate": 1.593467155187933e-08,
      "loss": 0.6481,
      "step": 6053
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0048984113471824,
      "learning_rate": 1.5727164807953777e-08,
      "loss": 0.4735,
      "step": 6054
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9904350508901384,
      "learning_rate": 1.5521015917340787e-08,
      "loss": 0.5252,
      "step": 6055
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0773577469501434,
      "learning_rate": 1.5316224936200997e-08,
      "loss": 0.5856,
      "step": 6056
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.076435305942875,
      "learning_rate": 1.5112791920325332e-08,
      "loss": 0.5555,
      "step": 6057
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.055061134867831,
      "learning_rate": 1.491071692513446e-08,
      "loss": 0.5866,
      "step": 6058
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0378648709294953,
      "learning_rate": 1.4710000005678792e-08,
      "loss": 0.5823,
      "step": 6059
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0489727628892882,
      "learning_rate": 1.4510641216639587e-08,
      "loss": 0.4813,
      "step": 6060
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0058389814126618,
      "learning_rate": 1.4312640612327289e-08,
      "loss": 0.6395,
      "step": 6061
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9796919194829005,
      "learning_rate": 1.4115998246683193e-08,
      "loss": 0.5348,
      "step": 6062
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.07488176315803,
      "learning_rate": 1.3920714173278338e-08,
      "loss": 0.6353,
      "step": 6063
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9738456468264551,
      "learning_rate": 1.372678844531239e-08,
      "loss": 0.4619,
      "step": 6064
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9940628420085107,
      "learning_rate": 1.3534221115616975e-08,
      "loss": 0.5384,
      "step": 6065
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0138613082662014,
      "learning_rate": 1.3343012236652353e-08,
      "loss": 0.5462,
      "step": 6066
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1017571957182446,
      "learning_rate": 1.3153161860509078e-08,
      "loss": 0.6238,
      "step": 6067
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0066358316676416,
      "learning_rate": 1.2964670038908e-08,
      "loss": 0.4811,
      "step": 6068
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9908461624636431,
      "learning_rate": 1.2777536823199155e-08,
      "loss": 0.4938,
      "step": 6069
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9938861098008926,
      "learning_rate": 1.2591762264362872e-08,
      "loss": 0.5671,
      "step": 6070
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.117231795371279,
      "learning_rate": 1.2407346413009224e-08,
      "loss": 0.5714,
      "step": 6071
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0596293916296122,
      "learning_rate": 1.2224289319378025e-08,
      "loss": 0.5421,
      "step": 6072
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.027496352421042,
      "learning_rate": 1.204259103333938e-08,
      "loss": 0.6068,
      "step": 6073
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9735783994586902,
      "learning_rate": 1.1862251604393138e-08,
      "loss": 0.5381,
      "step": 6074
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9652057351408223,
      "learning_rate": 1.1683271081668334e-08,
      "loss": 0.5038,
      "step": 6075
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0287664508355712,
      "learning_rate": 1.1505649513923744e-08,
      "loss": 0.5579,
      "step": 6076
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1254638553424168,
      "learning_rate": 1.1329386949548993e-08,
      "loss": 0.5673,
      "step": 6077
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0621023278864372,
      "learning_rate": 1.115448343656289e-08,
      "loss": 0.5937,
      "step": 6078
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9439501980536436,
      "learning_rate": 1.098093902261399e-08,
      "loss": 0.5669,
      "step": 6079
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.022412626013796,
      "learning_rate": 1.0808753754980029e-08,
      "loss": 0.5074,
      "step": 6080
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9851438894304576,
      "learning_rate": 1.0637927680570149e-08,
      "loss": 0.5328,
      "step": 6081
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0571939724898407,
      "learning_rate": 1.0468460845921014e-08,
      "loss": 0.5309,
      "step": 6082
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0068854287604425,
      "learning_rate": 1.0300353297200139e-08,
      "loss": 0.5763,
      "step": 6083
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0573620959013172,
      "learning_rate": 1.0133605080204779e-08,
      "loss": 0.6127,
      "step": 6084
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1122790903781885,
      "learning_rate": 9.968216240361927e-09,
      "loss": 0.6164,
      "step": 6085
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0805030872753052,
      "learning_rate": 9.804186822728324e-09,
      "loss": 0.5554,
      "step": 6086
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.007465496823581,
      "learning_rate": 9.641516871989331e-09,
      "loss": 0.5334,
      "step": 6087
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0673019748820423,
      "learning_rate": 9.480206432461725e-09,
      "loss": 0.5737,
      "step": 6088
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9985507562910875,
      "learning_rate": 9.32025554809035e-09,
      "loss": 0.5345,
      "step": 6089
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9971575235860275,
      "learning_rate": 9.161664262449799e-09,
      "loss": 0.6047,
      "step": 6090
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9911103389218824,
      "learning_rate": 9.004432618745507e-09,
      "loss": 0.5889,
      "step": 6091
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0357703634874746,
      "learning_rate": 8.848560659810989e-09,
      "loss": 0.5911,
      "step": 6092
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9602679315792186,
      "learning_rate": 8.694048428110614e-09,
      "loss": 0.56,
      "step": 6093
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0373607960057563,
      "learning_rate": 8.540895965737928e-09,
      "loss": 0.5352,
      "step": 6094
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0091122134975479,
      "learning_rate": 8.38910331441567e-09,
      "loss": 0.6252,
      "step": 6095
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0091368786766277,
      "learning_rate": 8.23867051549576e-09,
      "loss": 0.5669,
      "step": 6096
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0580920145093027,
      "learning_rate": 8.089597609960976e-09,
      "loss": 0.6397,
      "step": 6097
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.935980110913327,
      "learning_rate": 7.94188463842216e-09,
      "loss": 0.505,
      "step": 6098
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.023334838052072,
      "learning_rate": 7.795531641121013e-09,
      "loss": 0.6305,
      "step": 6099
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9685433501730999,
      "learning_rate": 7.65053865792842e-09,
      "loss": 0.5018,
      "step": 6100
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.992892912556299,
      "learning_rate": 7.506905728343893e-09,
      "loss": 0.4982,
      "step": 6101
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9998551841717478,
      "learning_rate": 7.364632891496692e-09,
      "loss": 0.5191,
      "step": 6102
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.990183717577766,
      "learning_rate": 7.223720186146366e-09,
      "loss": 0.5313,
      "step": 6103
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0330959611771864,
      "learning_rate": 7.08416765068165e-09,
      "loss": 0.5294,
      "step": 6104
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.05190502969452,
      "learning_rate": 6.945975323119913e-09,
      "loss": 0.5541,
      "step": 6105
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.071577803082671,
      "learning_rate": 6.809143241109373e-09,
      "loss": 0.5759,
      "step": 6106
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.969015238135913,
      "learning_rate": 6.673671441925766e-09,
      "loss": 0.6539,
      "step": 6107
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0736419859618975,
      "learning_rate": 6.539559962476238e-09,
      "loss": 0.603,
      "step": 6108
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9281036570834018,
      "learning_rate": 6.4068088392960084e-09,
      "loss": 0.4484,
      "step": 6109
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9691735021225716,
      "learning_rate": 6.275418108550591e-09,
      "loss": 0.5434,
      "step": 6110
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.932596668774295,
      "learning_rate": 6.1453878060335755e-09,
      "loss": 0.4885,
      "step": 6111
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0500518374015095,
      "learning_rate": 6.0167179671694055e-09,
      "loss": 0.5763,
      "step": 6112
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6035114941550267,
      "learning_rate": 5.8894086270111505e-09,
      "loss": 0.4542,
      "step": 6113
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9972747102694812,
      "learning_rate": 5.7634598202416235e-09,
      "loss": 0.5713,
      "step": 6114
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.646022688923708,
      "learning_rate": 5.638871581172822e-09,
      "loss": 0.4433,
      "step": 6115
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0365478188837125,
      "learning_rate": 5.515643943745375e-09,
      "loss": 0.6057,
      "step": 6116
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1070703971096747,
      "learning_rate": 5.393776941530759e-09,
      "loss": 0.6072,
      "step": 6117
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0745398601819056,
      "learning_rate": 5.273270607727976e-09,
      "loss": 0.5967,
      "step": 6118
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.008255045280376,
      "learning_rate": 5.1541249751668745e-09,
      "loss": 0.5379,
      "step": 6119
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.933313812656356,
      "learning_rate": 5.0363400763059346e-09,
      "loss": 0.5529,
      "step": 6120
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7395389956068767,
      "learning_rate": 4.919915943232822e-09,
      "loss": 0.4884,
      "step": 6121
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.007876759717825,
      "learning_rate": 4.80485260766439e-09,
      "loss": 0.6308,
      "step": 6122
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.082508457015373,
      "learning_rate": 4.691150100948338e-09,
      "loss": 0.5723,
      "step": 6123
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6600835017230215,
      "learning_rate": 4.578808454058781e-09,
      "loss": 0.4713,
      "step": 6124
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0778406904243396,
      "learning_rate": 4.4678276976017895e-09,
      "loss": 0.5522,
      "step": 6125
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1528647064682462,
      "learning_rate": 4.358207861810959e-09,
      "loss": 0.5798,
      "step": 6126
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9963952404545122,
      "learning_rate": 4.249948976550178e-09,
      "loss": 0.5307,
      "step": 6127
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0747539120347367,
      "learning_rate": 4.143051071311965e-09,
      "loss": 0.5658,
      "step": 6128
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0301774275911888,
      "learning_rate": 4.037514175218027e-09,
      "loss": 0.596,
      "step": 6129
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0025789412957742,
      "learning_rate": 3.933338317019808e-09,
      "loss": 0.6152,
      "step": 6130
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1235622709269641,
      "learning_rate": 3.83052352509794e-09,
      "loss": 0.6478,
      "step": 6131
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0006984545537025,
      "learning_rate": 3.729069827461685e-09,
      "loss": 0.5367,
      "step": 6132
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9970821499202145,
      "learning_rate": 3.628977251749488e-09,
      "loss": 0.6065,
      "step": 6133
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.633620624780548,
      "learning_rate": 3.530245825229539e-09,
      "loss": 0.4632,
      "step": 6134
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.107699522009888,
      "learning_rate": 3.4328755747992103e-09,
      "loss": 0.6112,
      "step": 6135
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9713928310904248,
      "learning_rate": 3.336866526985061e-09,
      "loss": 0.558,
      "step": 6136
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.964809304299755,
      "learning_rate": 3.2422187079417242e-09,
      "loss": 0.5845,
      "step": 6137
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.953593170793836,
      "learning_rate": 3.148932143455241e-09,
      "loss": 0.5758,
      "step": 6138
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0668095124418397,
      "learning_rate": 3.0570068589380606e-09,
      "loss": 0.5583,
      "step": 6139
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.023734018974233,
      "learning_rate": 2.9664428794340393e-09,
      "loss": 0.6433,
      "step": 6140
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0148768386092069,
      "learning_rate": 2.877240229614553e-09,
      "loss": 0.4971,
      "step": 6141
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9460995846149767,
      "learning_rate": 2.7893989337818283e-09,
      "loss": 0.5436,
      "step": 6142
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1002763225031482,
      "learning_rate": 2.7029190158656125e-09,
      "loss": 0.597,
      "step": 6143
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9610353383184768,
      "learning_rate": 2.6178004994253936e-09,
      "loss": 0.5469,
      "step": 6144
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9914114508362192,
      "learning_rate": 2.5340434076503994e-09,
      "loss": 0.6009,
      "step": 6145
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6487762398526201,
      "learning_rate": 2.4516477633579338e-09,
      "loss": 0.5013,
      "step": 6146
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0156101100489658,
      "learning_rate": 2.370613588994486e-09,
      "loss": 0.484,
      "step": 6147
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1627599762560006,
      "learning_rate": 2.2909409066362854e-09,
      "loss": 0.6794,
      "step": 6148
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0733042254008707,
      "learning_rate": 2.2126297379887473e-09,
      "loss": 0.6338,
      "step": 6149
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.018659992916284,
      "learning_rate": 2.1356801043853624e-09,
      "loss": 0.5561,
      "step": 6150
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0653913053243522,
      "learning_rate": 2.060092026789917e-09,
      "loss": 0.5583,
      "step": 6151
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0322219392742877,
      "learning_rate": 1.9858655257942726e-09,
      "loss": 0.5764,
      "step": 6152
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0187572648277592,
      "learning_rate": 1.9130006216200314e-09,
      "loss": 0.6284,
      "step": 6153
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9741050372475817,
      "learning_rate": 1.841497334117426e-09,
      "loss": 0.5356,
      "step": 6154
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0528706205744032,
      "learning_rate": 1.771355682765874e-09,
      "loss": 0.6183,
      "step": 6155
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9529098490161597,
      "learning_rate": 1.7025756866739795e-09,
      "loss": 0.5154,
      "step": 6156
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0260655015735392,
      "learning_rate": 1.6351573645795316e-09,
      "loss": 0.6401,
      "step": 6157
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9785927370283991,
      "learning_rate": 1.5691007348489495e-09,
      "loss": 0.5262,
      "step": 6158
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0023156458056082,
      "learning_rate": 1.5044058154778385e-09,
      "loss": 0.5684,
      "step": 6159
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9956653057538485,
      "learning_rate": 1.441072624090989e-09,
      "loss": 0.5037,
      "step": 6160
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0443396664572782,
      "learning_rate": 1.3791011779423769e-09,
      "loss": 0.6064,
      "step": 6161
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.002802211768055,
      "learning_rate": 1.3184914939140537e-09,
      "loss": 0.5944,
      "step": 6162
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0187578677247746,
      "learning_rate": 1.2592435885178112e-09,
      "loss": 0.4841,
      "step": 6163
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0323323324330262,
      "learning_rate": 1.2013574778951819e-09,
      "loss": 0.5843,
      "step": 6164
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9583023689224676,
      "learning_rate": 1.1448331778152189e-09,
      "loss": 0.4871,
      "step": 6165
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.014101463885054,
      "learning_rate": 1.0896707036772703e-09,
      "loss": 0.5925,
      "step": 6166
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0861057934313525,
      "learning_rate": 1.0358700705082048e-09,
      "loss": 0.6003,
      "step": 6167
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0468390881227134,
      "learning_rate": 9.834312929657419e-10,
      "loss": 0.5825,
      "step": 6168
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0668872982331805,
      "learning_rate": 9.323543853351212e-10,
      "loss": 0.6826,
      "step": 6169
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9208274407055418,
      "learning_rate": 8.826393615318784e-10,
      "loss": 0.5703,
      "step": 6170
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.975338600482535,
      "learning_rate": 8.342862350985137e-10,
      "loss": 0.4622,
      "step": 6171
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.047378735295744,
      "learning_rate": 7.872950192083783e-10,
      "loss": 0.5905,
      "step": 6172
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9496929556563557,
      "learning_rate": 7.416657266634542e-10,
      "loss": 0.5127,
      "step": 6173
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9693853647972032,
      "learning_rate": 6.973983698943532e-10,
      "loss": 0.5019,
      "step": 6174
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0201178810878129,
      "learning_rate": 6.544929609597628e-10,
      "loss": 0.6044,
      "step": 6175
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0368535263982326,
      "learning_rate": 6.129495115497764e-10,
      "loss": 0.5626,
      "step": 6176
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0282627798580182,
      "learning_rate": 5.727680329808971e-10,
      "loss": 0.6514,
      "step": 6177
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0087698862022365,
      "learning_rate": 5.339485362004793e-10,
      "loss": 0.4787,
      "step": 6178
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.054843370275368,
      "learning_rate": 4.96491031782842e-10,
      "loss": 0.6147,
      "step": 6179
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1403561273488698,
      "learning_rate": 4.603955299337104e-10,
      "loss": 0.5943,
      "step": 6180
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6299107626400982,
      "learning_rate": 4.2566204048577473e-10,
      "loss": 0.441,
      "step": 6181
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0421485468693035,
      "learning_rate": 3.9229057290146587e-10,
      "loss": 0.5308,
      "step": 6182
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6571222166553881,
      "learning_rate": 3.6028113627240015e-10,
      "loss": 0.4514,
      "step": 6183
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0382762399662364,
      "learning_rate": 3.2963373931882425e-10,
      "loss": 0.5939,
      "step": 6184
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0353351838259006,
      "learning_rate": 3.003483903890603e-10,
      "loss": 0.5504,
      "step": 6185
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0037410242411964,
      "learning_rate": 2.724250974628362e-10,
      "loss": 0.5159,
      "step": 6186
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9437987787586366,
      "learning_rate": 2.458638681457348e-10,
      "loss": 0.5558,
      "step": 6187
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8987079984731946,
      "learning_rate": 2.2066470967418985e-10,
      "loss": 0.4366,
      "step": 6188
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1069131194271307,
      "learning_rate": 1.9682762891382045e-10,
      "loss": 0.6337,
      "step": 6189
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0059220665422945,
      "learning_rate": 1.74352632357766e-10,
      "loss": 0.5416,
      "step": 6190
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0178639414680455,
      "learning_rate": 1.5323972612890647e-10,
      "loss": 0.5728,
      "step": 6191
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0053577679578694,
      "learning_rate": 1.3348891597930734e-10,
      "loss": 0.5193,
      "step": 6192
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.157801710997135,
      "learning_rate": 1.1510020728910943e-10,
      "loss": 0.6408,
      "step": 6193
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0568511084265244,
      "learning_rate": 9.807360506874919e-11,
      "loss": 0.5634,
      "step": 6194
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0386398308947,
      "learning_rate": 8.240911395562823e-11,
      "loss": 0.5886,
      "step": 6195
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0195762522641245,
      "learning_rate": 6.810673821855407e-11,
      "loss": 0.6168,
      "step": 6196
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0220328269684127,
      "learning_rate": 5.516648175274419e-11,
      "loss": 0.5266,
      "step": 6197
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.059553168300954,
      "learning_rate": 4.358834808371182e-11,
      "loss": 0.5748,
      "step": 6198
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9465146315770474,
      "learning_rate": 3.3372340366155663e-11,
      "loss": 0.5277,
      "step": 6199
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0693354860113116,
      "learning_rate": 2.4518461382849745e-11,
      "loss": 0.6031,
      "step": 6200
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0011966740331144,
      "learning_rate": 1.702671354575358e-11,
      "loss": 0.6298,
      "step": 6201
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1349782264437707,
      "learning_rate": 1.0897098895457092e-11,
      "loss": 0.6135,
      "step": 6202
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0384279976212387,
      "learning_rate": 6.129619102845929e-12,
      "loss": 0.6528,
      "step": 6203
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0027548377380584,
      "learning_rate": 2.7242754663259207e-12,
      "loss": 0.5874,
      "step": 6204
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.139677711687267,
      "learning_rate": 6.810689129332915e-13,
      "loss": 0.5835,
      "step": 6205
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9912142881766588,
      "learning_rate": 0.0,
      "loss": 0.4097,
      "step": 6206
    },
    {
      "epoch": 1.0,
      "step": 6206,
      "total_flos": 9688178644156416.0,
      "train_loss": 0.6073194185630222,
      "train_runtime": 63272.6884,
      "train_samples_per_second": 12.555,
      "train_steps_per_second": 0.098
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 6206,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 2000,
  "total_flos": 9688178644156416.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}