{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.999936,
  "eval_steps": 500,
  "global_step": 7812,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 3.244857915065077,
      "learning_rate": 4.2553191489361707e-08,
      "loss": 0.3326,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.3796528671741006,
      "learning_rate": 8.510638297872341e-08,
      "loss": 0.3263,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.2157470725132815,
      "learning_rate": 1.276595744680851e-07,
      "loss": 0.3324,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.5518362485706065,
      "learning_rate": 1.7021276595744683e-07,
      "loss": 0.3558,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.641466657105259,
      "learning_rate": 2.1276595744680852e-07,
      "loss": 0.3157,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.2068875711126434,
      "learning_rate": 2.553191489361702e-07,
      "loss": 0.3638,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.268050471410201,
      "learning_rate": 2.9787234042553196e-07,
      "loss": 0.2877,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.5696445843591746,
      "learning_rate": 3.4042553191489365e-07,
      "loss": 0.3143,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.705456924384768,
      "learning_rate": 3.8297872340425535e-07,
      "loss": 0.9404,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.3427188612417122,
      "learning_rate": 4.2553191489361704e-07,
      "loss": 0.3347,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.539425821636402,
      "learning_rate": 4.6808510638297873e-07,
      "loss": 0.7912,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.807842663707556,
      "learning_rate": 5.106382978723404e-07,
      "loss": 0.7724,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.457675271118148,
      "learning_rate": 5.531914893617021e-07,
      "loss": 0.2783,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.0850231790595872,
      "learning_rate": 5.957446808510639e-07,
      "loss": 0.2995,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.916238684648272,
      "learning_rate": 6.382978723404255e-07,
      "loss": 0.3488,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.8819745176332163,
      "learning_rate": 6.808510638297873e-07,
      "loss": 0.369,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.338342159013587,
      "learning_rate": 7.234042553191489e-07,
      "loss": 0.8123,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.3126230238540995,
      "learning_rate": 7.659574468085107e-07,
      "loss": 0.348,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.8621818518844853,
      "learning_rate": 8.085106382978725e-07,
      "loss": 0.312,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.4770111002367607,
      "learning_rate": 8.510638297872341e-07,
      "loss": 0.3122,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 10.914669959064028,
      "learning_rate": 8.936170212765959e-07,
      "loss": 0.8715,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.409123002543557,
      "learning_rate": 9.361702127659575e-07,
      "loss": 0.3458,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.260333152994881,
      "learning_rate": 9.787234042553193e-07,
      "loss": 0.2945,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.3148630545389706,
      "learning_rate": 1.0212765957446809e-06,
      "loss": 0.2475,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.7470038884957892,
      "learning_rate": 1.0638297872340427e-06,
      "loss": 0.3398,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.350008162849608,
      "learning_rate": 1.1063829787234042e-06,
      "loss": 0.2568,
      "step": 26
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.5342902873429485,
      "learning_rate": 1.148936170212766e-06,
      "loss": 0.2741,
      "step": 27
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.3556163838485276,
      "learning_rate": 1.1914893617021278e-06,
      "loss": 0.2474,
      "step": 28
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.47788573430894,
      "learning_rate": 1.2340425531914894e-06,
      "loss": 0.2963,
      "step": 29
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.593658071899859,
      "learning_rate": 1.276595744680851e-06,
      "loss": 0.3584,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.2568678609992654,
      "learning_rate": 1.3191489361702128e-06,
      "loss": 0.2489,
      "step": 31
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.168832233908094,
      "learning_rate": 1.3617021276595746e-06,
      "loss": 0.2958,
      "step": 32
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.4930909670011894,
      "learning_rate": 1.4042553191489364e-06,
      "loss": 0.2644,
      "step": 33
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.030498167748424,
      "learning_rate": 1.4468085106382978e-06,
      "loss": 0.2271,
      "step": 34
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9038773918233505,
      "learning_rate": 1.4893617021276596e-06,
      "loss": 0.206,
      "step": 35
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.774959761851697,
      "learning_rate": 1.5319148936170214e-06,
      "loss": 0.7723,
      "step": 36
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.4052126502033415,
      "learning_rate": 1.5744680851063832e-06,
      "loss": 0.2742,
      "step": 37
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.3511488273339105,
      "learning_rate": 1.617021276595745e-06,
      "loss": 0.2953,
      "step": 38
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.2121469329122054,
      "learning_rate": 1.6595744680851064e-06,
      "loss": 0.2749,
      "step": 39
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4151205432579226,
      "learning_rate": 1.7021276595744682e-06,
      "loss": 0.2758,
      "step": 40
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.341694519446735,
      "learning_rate": 1.74468085106383e-06,
      "loss": 0.7656,
      "step": 41
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1094044927377964,
      "learning_rate": 1.7872340425531918e-06,
      "loss": 0.2448,
      "step": 42
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1473800203928004,
      "learning_rate": 1.8297872340425531e-06,
      "loss": 0.2236,
      "step": 43
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.174991390195286,
      "learning_rate": 1.872340425531915e-06,
      "loss": 0.2683,
      "step": 44
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.057572931958973,
      "learning_rate": 1.9148936170212767e-06,
      "loss": 0.2381,
      "step": 45
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.585805371858611,
      "learning_rate": 1.9574468085106385e-06,
      "loss": 0.9346,
      "step": 46
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.279273848738161,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.2524,
      "step": 47
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.155229156002533,
      "learning_rate": 2.0425531914893617e-06,
      "loss": 0.2451,
      "step": 48
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.17026363604632,
      "learning_rate": 2.0851063829787235e-06,
      "loss": 0.8356,
      "step": 49
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.167592070326722,
      "learning_rate": 2.1276595744680853e-06,
      "loss": 0.2869,
      "step": 50
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.095045215821379,
      "learning_rate": 2.170212765957447e-06,
      "loss": 0.7306,
      "step": 51
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0171381209340655,
      "learning_rate": 2.2127659574468085e-06,
      "loss": 0.2395,
      "step": 52
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3088596402147754,
      "learning_rate": 2.2553191489361703e-06,
      "loss": 0.2639,
      "step": 53
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.062940949220895,
      "learning_rate": 2.297872340425532e-06,
      "loss": 0.2248,
      "step": 54
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0891778627436257,
      "learning_rate": 2.340425531914894e-06,
      "loss": 0.2584,
      "step": 55
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.230953019207813,
      "learning_rate": 2.3829787234042557e-06,
      "loss": 0.206,
      "step": 56
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1187017087348754,
      "learning_rate": 2.425531914893617e-06,
      "loss": 0.2327,
      "step": 57
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.914968806461365,
      "learning_rate": 2.468085106382979e-06,
      "loss": 0.1882,
      "step": 58
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.992032363481915,
      "learning_rate": 2.5106382978723402e-06,
      "loss": 0.7963,
      "step": 59
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8635908207122678,
      "learning_rate": 2.553191489361702e-06,
      "loss": 0.192,
      "step": 60
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2773863918825703,
      "learning_rate": 2.595744680851064e-06,
      "loss": 0.2383,
      "step": 61
    },
    {
      "epoch": 0.01,
      "grad_norm": 21.386170035753963,
      "learning_rate": 2.6382978723404256e-06,
      "loss": 0.7039,
      "step": 62
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.298839638646067,
      "learning_rate": 2.6808510638297874e-06,
      "loss": 0.2528,
      "step": 63
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.9295642381237235,
      "learning_rate": 2.7234042553191492e-06,
      "loss": 0.6609,
      "step": 64
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.928227213888163,
      "learning_rate": 2.765957446808511e-06,
      "loss": 0.5931,
      "step": 65
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2757529718318237,
      "learning_rate": 2.808510638297873e-06,
      "loss": 0.2073,
      "step": 66
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.184391165578907,
      "learning_rate": 2.8510638297872346e-06,
      "loss": 0.8008,
      "step": 67
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1087274485387955,
      "learning_rate": 2.8936170212765956e-06,
      "loss": 0.2384,
      "step": 68
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.086816793722631,
      "learning_rate": 2.9361702127659574e-06,
      "loss": 0.1937,
      "step": 69
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.172398665409211,
      "learning_rate": 2.978723404255319e-06,
      "loss": 0.2041,
      "step": 70
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.249633459641971,
      "learning_rate": 3.021276595744681e-06,
      "loss": 0.6165,
      "step": 71
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.218716318621196,
      "learning_rate": 3.0638297872340428e-06,
      "loss": 0.219,
      "step": 72
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9232705852623733,
      "learning_rate": 3.1063829787234046e-06,
      "loss": 0.2323,
      "step": 73
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2020696005196325,
      "learning_rate": 3.1489361702127664e-06,
      "loss": 0.2621,
      "step": 74
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1252684117447944,
      "learning_rate": 3.191489361702128e-06,
      "loss": 0.2699,
      "step": 75
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.761823776727612,
      "learning_rate": 3.23404255319149e-06,
      "loss": 0.75,
      "step": 76
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.6816595882263785,
      "learning_rate": 3.276595744680851e-06,
      "loss": 0.7955,
      "step": 77
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2341609324238987,
      "learning_rate": 3.3191489361702127e-06,
      "loss": 0.2046,
      "step": 78
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.428648423996099,
      "learning_rate": 3.3617021276595745e-06,
      "loss": 0.7799,
      "step": 79
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.016166635584139,
      "learning_rate": 3.4042553191489363e-06,
      "loss": 0.634,
      "step": 80
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.673711156227421,
      "learning_rate": 3.446808510638298e-06,
      "loss": 0.6681,
      "step": 81
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.889480165348969,
      "learning_rate": 3.48936170212766e-06,
      "loss": 0.8764,
      "step": 82
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.042589177413689,
      "learning_rate": 3.5319148936170217e-06,
      "loss": 0.2189,
      "step": 83
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.941125614707088,
      "learning_rate": 3.5744680851063835e-06,
      "loss": 0.218,
      "step": 84
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.80791353000491,
      "learning_rate": 3.6170212765957453e-06,
      "loss": 0.6777,
      "step": 85
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.046022080239068,
      "learning_rate": 3.6595744680851063e-06,
      "loss": 0.2727,
      "step": 86
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9115861349343413,
      "learning_rate": 3.702127659574468e-06,
      "loss": 0.181,
      "step": 87
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9551967271301054,
      "learning_rate": 3.74468085106383e-06,
      "loss": 0.2447,
      "step": 88
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.166942644740949,
      "learning_rate": 3.7872340425531917e-06,
      "loss": 0.2096,
      "step": 89
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.19078935834791,
      "learning_rate": 3.8297872340425535e-06,
      "loss": 0.2189,
      "step": 90
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1554849413255757,
      "learning_rate": 3.872340425531915e-06,
      "loss": 0.2253,
      "step": 91
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7562888706367863,
      "learning_rate": 3.914893617021277e-06,
      "loss": 0.2232,
      "step": 92
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.296518751608334,
      "learning_rate": 3.957446808510639e-06,
      "loss": 0.2649,
      "step": 93
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.458212081729814,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.6265,
      "step": 94
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9933007914493597,
      "learning_rate": 4.042553191489362e-06,
      "loss": 0.2423,
      "step": 95
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.254621140055434,
      "learning_rate": 4.085106382978723e-06,
      "loss": 0.2456,
      "step": 96
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.158007331811539,
      "learning_rate": 4.127659574468085e-06,
      "loss": 0.1837,
      "step": 97
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0749989377852387,
      "learning_rate": 4.170212765957447e-06,
      "loss": 0.2422,
      "step": 98
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9481347063245378,
      "learning_rate": 4.212765957446809e-06,
      "loss": 0.1932,
      "step": 99
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.109451481450835,
      "learning_rate": 4.255319148936171e-06,
      "loss": 0.235,
      "step": 100
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2723924084966094,
      "learning_rate": 4.297872340425532e-06,
      "loss": 0.2785,
      "step": 101
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8626036174901135,
      "learning_rate": 4.340425531914894e-06,
      "loss": 0.2169,
      "step": 102
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.941958171361437,
      "learning_rate": 4.382978723404256e-06,
      "loss": 0.22,
      "step": 103
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.082513756491552,
      "learning_rate": 4.425531914893617e-06,
      "loss": 0.2458,
      "step": 104
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1742099877828642,
      "learning_rate": 4.468085106382979e-06,
      "loss": 0.2589,
      "step": 105
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.903007088646307,
      "learning_rate": 4.5106382978723406e-06,
      "loss": 0.5906,
      "step": 106
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1465335597740043,
      "learning_rate": 4.553191489361702e-06,
      "loss": 0.2875,
      "step": 107
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.126479674098374,
      "learning_rate": 4.595744680851064e-06,
      "loss": 0.2426,
      "step": 108
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9156346809651508,
      "learning_rate": 4.638297872340426e-06,
      "loss": 0.2291,
      "step": 109
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0647599891295627,
      "learning_rate": 4.680851063829788e-06,
      "loss": 0.235,
      "step": 110
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.275487636394458,
      "learning_rate": 4.7234042553191496e-06,
      "loss": 0.2358,
      "step": 111
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0041453101661775,
      "learning_rate": 4.765957446808511e-06,
      "loss": 0.2719,
      "step": 112
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.58820662044493,
      "learning_rate": 4.808510638297872e-06,
      "loss": 0.4871,
      "step": 113
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.083192917840147,
      "learning_rate": 4.851063829787234e-06,
      "loss": 0.2227,
      "step": 114
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.850303884232678,
      "learning_rate": 4.893617021276596e-06,
      "loss": 0.1991,
      "step": 115
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9779108736055786,
      "learning_rate": 4.936170212765958e-06,
      "loss": 0.229,
      "step": 116
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0044443500315854,
      "learning_rate": 4.9787234042553195e-06,
      "loss": 0.2124,
      "step": 117
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9346753303675834,
      "learning_rate": 5.0212765957446805e-06,
      "loss": 0.2469,
      "step": 118
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.118894962960385,
      "learning_rate": 5.063829787234042e-06,
      "loss": 0.2795,
      "step": 119
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1549600121369252,
      "learning_rate": 5.106382978723404e-06,
      "loss": 0.1999,
      "step": 120
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.826564325612661,
      "learning_rate": 5.148936170212766e-06,
      "loss": 0.6024,
      "step": 121
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.129644409177819,
      "learning_rate": 5.191489361702128e-06,
      "loss": 0.2805,
      "step": 122
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8916956862827945,
      "learning_rate": 5.2340425531914895e-06,
      "loss": 0.2113,
      "step": 123
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4220023057054636,
      "learning_rate": 5.276595744680851e-06,
      "loss": 0.262,
      "step": 124
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9516651234985947,
      "learning_rate": 5.319148936170213e-06,
      "loss": 0.252,
      "step": 125
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0336706374512956,
      "learning_rate": 5.361702127659575e-06,
      "loss": 0.2301,
      "step": 126
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7490989558568575,
      "learning_rate": 5.404255319148937e-06,
      "loss": 0.247,
      "step": 127
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8050687137485335,
      "learning_rate": 5.4468085106382985e-06,
      "loss": 0.1754,
      "step": 128
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.025542047359308,
      "learning_rate": 5.48936170212766e-06,
      "loss": 0.2149,
      "step": 129
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0206138335485813,
      "learning_rate": 5.531914893617022e-06,
      "loss": 0.2257,
      "step": 130
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.213914342200139,
      "learning_rate": 5.574468085106384e-06,
      "loss": 0.2676,
      "step": 131
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.823962483739315,
      "learning_rate": 5.617021276595746e-06,
      "loss": 0.7091,
      "step": 132
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.316356556889312,
      "learning_rate": 5.6595744680851075e-06,
      "loss": 0.5414,
      "step": 133
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.419035123669159,
      "learning_rate": 5.702127659574469e-06,
      "loss": 0.2458,
      "step": 134
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.268724840982987,
      "learning_rate": 5.744680851063831e-06,
      "loss": 0.25,
      "step": 135
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8091603808306242,
      "learning_rate": 5.787234042553191e-06,
      "loss": 0.2044,
      "step": 136
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.745916269467642,
      "learning_rate": 5.829787234042553e-06,
      "loss": 0.8075,
      "step": 137
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.104741917225633,
      "learning_rate": 5.872340425531915e-06,
      "loss": 0.6392,
      "step": 138
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0865093063259104,
      "learning_rate": 5.9148936170212766e-06,
      "loss": 0.2408,
      "step": 139
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.294678787471527,
      "learning_rate": 5.957446808510638e-06,
      "loss": 0.2399,
      "step": 140
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.10771239097608,
      "learning_rate": 6e-06,
      "loss": 0.6005,
      "step": 141
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0071993700208117,
      "learning_rate": 6.042553191489362e-06,
      "loss": 0.2275,
      "step": 142
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9565291106549876,
      "learning_rate": 6.085106382978724e-06,
      "loss": 0.2618,
      "step": 143
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9635409534417727,
      "learning_rate": 6.1276595744680855e-06,
      "loss": 0.2189,
      "step": 144
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.075273948670289,
      "learning_rate": 6.170212765957447e-06,
      "loss": 0.2429,
      "step": 145
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.32045309061444,
      "learning_rate": 6.212765957446809e-06,
      "loss": 0.6722,
      "step": 146
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.768828172329951,
      "learning_rate": 6.255319148936171e-06,
      "loss": 0.1514,
      "step": 147
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.224439887296485,
      "learning_rate": 6.297872340425533e-06,
      "loss": 0.7554,
      "step": 148
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.411390653506701,
      "learning_rate": 6.3404255319148945e-06,
      "loss": 0.6534,
      "step": 149
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.189631726977135,
      "learning_rate": 6.382978723404256e-06,
      "loss": 0.2565,
      "step": 150
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.294955028387845,
      "learning_rate": 6.425531914893618e-06,
      "loss": 0.56,
      "step": 151
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8980013838481027,
      "learning_rate": 6.46808510638298e-06,
      "loss": 0.2267,
      "step": 152
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.8466579792289,
      "learning_rate": 6.510638297872342e-06,
      "loss": 0.749,
      "step": 153
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.11525326024659,
      "learning_rate": 6.553191489361702e-06,
      "loss": 0.2326,
      "step": 154
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.15883153561582,
      "learning_rate": 6.595744680851064e-06,
      "loss": 0.7285,
      "step": 155
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.683074834999534,
      "learning_rate": 6.6382978723404254e-06,
      "loss": 0.5802,
      "step": 156
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0074245520317726,
      "learning_rate": 6.680851063829787e-06,
      "loss": 0.2292,
      "step": 157
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8945620199571194,
      "learning_rate": 6.723404255319149e-06,
      "loss": 0.198,
      "step": 158
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9420799735988334,
      "learning_rate": 6.765957446808511e-06,
      "loss": 0.2091,
      "step": 159
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.27420080305347,
      "learning_rate": 6.808510638297873e-06,
      "loss": 0.5562,
      "step": 160
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8784064647817886,
      "learning_rate": 6.8510638297872344e-06,
      "loss": 0.1778,
      "step": 161
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.856401343231877,
      "learning_rate": 6.893617021276596e-06,
      "loss": 0.2162,
      "step": 162
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.190077699718289,
      "learning_rate": 6.936170212765958e-06,
      "loss": 0.5247,
      "step": 163
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.43757433224057,
      "learning_rate": 6.97872340425532e-06,
      "loss": 0.7797,
      "step": 164
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.059504489207313,
      "learning_rate": 7.021276595744682e-06,
      "loss": 0.2438,
      "step": 165
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1041662956004634,
      "learning_rate": 7.0638297872340434e-06,
      "loss": 0.2175,
      "step": 166
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.003805013385646,
      "learning_rate": 7.106382978723405e-06,
      "loss": 0.2705,
      "step": 167
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.164713438513456,
      "learning_rate": 7.148936170212767e-06,
      "loss": 0.2828,
      "step": 168
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1535508558456122,
      "learning_rate": 7.191489361702129e-06,
      "loss": 0.2565,
      "step": 169
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2881809448540174,
      "learning_rate": 7.234042553191491e-06,
      "loss": 0.2849,
      "step": 170
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.820868253861557,
      "learning_rate": 7.2765957446808524e-06,
      "loss": 0.2685,
      "step": 171
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.954897296473696,
      "learning_rate": 7.3191489361702125e-06,
      "loss": 0.2305,
      "step": 172
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.7235345054891615,
      "learning_rate": 7.361702127659574e-06,
      "loss": 0.493,
      "step": 173
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9902388304779266,
      "learning_rate": 7.404255319148936e-06,
      "loss": 0.2584,
      "step": 174
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8303572374518624,
      "learning_rate": 7.446808510638298e-06,
      "loss": 0.2108,
      "step": 175
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9731273378747147,
      "learning_rate": 7.48936170212766e-06,
      "loss": 0.2125,
      "step": 176
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8631219570073085,
      "learning_rate": 7.5319148936170215e-06,
      "loss": 0.2221,
      "step": 177
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0314111750500894,
      "learning_rate": 7.574468085106383e-06,
      "loss": 0.216,
      "step": 178
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.891986355940946,
      "learning_rate": 7.617021276595745e-06,
      "loss": 0.2688,
      "step": 179
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7444067481784695,
      "learning_rate": 7.659574468085107e-06,
      "loss": 0.1798,
      "step": 180
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9703099487988733,
      "learning_rate": 7.702127659574469e-06,
      "loss": 0.2335,
      "step": 181
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8178894105568457,
      "learning_rate": 7.74468085106383e-06,
      "loss": 0.2477,
      "step": 182
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.292675005457148,
      "learning_rate": 7.787234042553192e-06,
      "loss": 0.2682,
      "step": 183
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8799450599693084,
      "learning_rate": 7.829787234042554e-06,
      "loss": 0.2515,
      "step": 184
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1837406714524965,
      "learning_rate": 7.872340425531916e-06,
      "loss": 0.2381,
      "step": 185
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.04877126821194,
      "learning_rate": 7.914893617021278e-06,
      "loss": 0.6117,
      "step": 186
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8696611057653603,
      "learning_rate": 7.95744680851064e-06,
      "loss": 0.1801,
      "step": 187
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9769999837556003,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.2302,
      "step": 188
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8543476685286233,
      "learning_rate": 8.042553191489363e-06,
      "loss": 0.181,
      "step": 189
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.230393664706476,
      "learning_rate": 8.085106382978723e-06,
      "loss": 0.6918,
      "step": 190
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0030468334589413,
      "learning_rate": 8.127659574468085e-06,
      "loss": 0.231,
      "step": 191
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.4961978411047765,
      "learning_rate": 8.170212765957447e-06,
      "loss": 0.6256,
      "step": 192
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.005835699184817,
      "learning_rate": 8.212765957446809e-06,
      "loss": 0.2475,
      "step": 193
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.796332319899966,
      "learning_rate": 8.25531914893617e-06,
      "loss": 0.2017,
      "step": 194
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.817583413553041,
      "learning_rate": 8.297872340425532e-06,
      "loss": 0.5699,
      "step": 195
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7452356821365222,
      "learning_rate": 8.340425531914894e-06,
      "loss": 0.1883,
      "step": 196
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8686676585075395,
      "learning_rate": 8.382978723404256e-06,
      "loss": 0.2404,
      "step": 197
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9024695201790194,
      "learning_rate": 8.425531914893618e-06,
      "loss": 0.2429,
      "step": 198
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9582625230810389,
      "learning_rate": 8.46808510638298e-06,
      "loss": 0.3154,
      "step": 199
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7679910980373847,
      "learning_rate": 8.510638297872341e-06,
      "loss": 0.2311,
      "step": 200
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.980049270810546,
      "learning_rate": 8.553191489361703e-06,
      "loss": 0.2309,
      "step": 201
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7362779049443697,
      "learning_rate": 8.595744680851065e-06,
      "loss": 0.2407,
      "step": 202
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0719835982506116,
      "learning_rate": 8.638297872340427e-06,
      "loss": 0.2346,
      "step": 203
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9949109037000248,
      "learning_rate": 8.680851063829788e-06,
      "loss": 0.2273,
      "step": 204
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7395363756444848,
      "learning_rate": 8.72340425531915e-06,
      "loss": 0.2143,
      "step": 205
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1111903315427147,
      "learning_rate": 8.765957446808512e-06,
      "loss": 0.2659,
      "step": 206
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8917781732335772,
      "learning_rate": 8.808510638297874e-06,
      "loss": 0.2132,
      "step": 207
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9748454264538753,
      "learning_rate": 8.851063829787234e-06,
      "loss": 0.2992,
      "step": 208
    },
    {
      "epoch": 0.03,
      "grad_norm": 194.71987735484336,
      "learning_rate": 8.893617021276596e-06,
      "loss": 0.899,
      "step": 209
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.047660242393871,
      "learning_rate": 8.936170212765958e-06,
      "loss": 0.2748,
      "step": 210
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.012010278036523,
      "learning_rate": 8.97872340425532e-06,
      "loss": 0.2342,
      "step": 211
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8608220617354208,
      "learning_rate": 9.021276595744681e-06,
      "loss": 0.2886,
      "step": 212
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8422758328798328,
      "learning_rate": 9.063829787234043e-06,
      "loss": 0.2506,
      "step": 213
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.588496407350151,
      "learning_rate": 9.106382978723405e-06,
      "loss": 0.4876,
      "step": 214
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9357656296746164,
      "learning_rate": 9.148936170212767e-06,
      "loss": 0.2598,
      "step": 215
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8753114028396736,
      "learning_rate": 9.191489361702128e-06,
      "loss": 0.2034,
      "step": 216
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9371071858170108,
      "learning_rate": 9.23404255319149e-06,
      "loss": 0.2766,
      "step": 217
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9001856909425852,
      "learning_rate": 9.276595744680852e-06,
      "loss": 0.2873,
      "step": 218
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8770162478692762,
      "learning_rate": 9.319148936170214e-06,
      "loss": 0.2341,
      "step": 219
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9263872952687788,
      "learning_rate": 9.361702127659576e-06,
      "loss": 0.2209,
      "step": 220
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.956416404093468,
      "learning_rate": 9.404255319148937e-06,
      "loss": 0.2694,
      "step": 221
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.968284211977155,
      "learning_rate": 9.446808510638299e-06,
      "loss": 0.4052,
      "step": 222
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.308317702913886,
      "learning_rate": 9.489361702127661e-06,
      "loss": 0.6113,
      "step": 223
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.924042726187051,
      "learning_rate": 9.531914893617023e-06,
      "loss": 0.2464,
      "step": 224
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1032697732300845,
      "learning_rate": 9.574468085106385e-06,
      "loss": 0.2631,
      "step": 225
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.165971099406852,
      "learning_rate": 9.617021276595745e-06,
      "loss": 0.5856,
      "step": 226
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1997421969535647,
      "learning_rate": 9.659574468085106e-06,
      "loss": 0.2478,
      "step": 227
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8401758983835823,
      "learning_rate": 9.702127659574468e-06,
      "loss": 0.2464,
      "step": 228
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7597768574753776,
      "learning_rate": 9.74468085106383e-06,
      "loss": 0.1968,
      "step": 229
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.659482407843473,
      "learning_rate": 9.787234042553192e-06,
      "loss": 0.2119,
      "step": 230
    },
    {
      "epoch": 0.03,
      "grad_norm": 13.477297357816997,
      "learning_rate": 9.829787234042554e-06,
      "loss": 0.7337,
      "step": 231
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.022402942617915,
      "learning_rate": 9.872340425531915e-06,
      "loss": 0.4864,
      "step": 232
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.011049972513032,
      "learning_rate": 9.914893617021277e-06,
      "loss": 0.721,
      "step": 233
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0292199102291475,
      "learning_rate": 9.957446808510639e-06,
      "loss": 0.2213,
      "step": 234
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.066598589038051,
      "learning_rate": 1e-05,
      "loss": 0.6586,
      "step": 235
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.025934201461913,
      "learning_rate": 9.999999570221018e-06,
      "loss": 0.2502,
      "step": 236
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9206168812789688,
      "learning_rate": 9.999998280884144e-06,
      "loss": 0.2546,
      "step": 237
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9709818066845146,
      "learning_rate": 9.999996131989602e-06,
      "loss": 0.2852,
      "step": 238
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6562152490845539,
      "learning_rate": 9.99999312353776e-06,
      "loss": 0.2317,
      "step": 239
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.808616904963193,
      "learning_rate": 9.999989255529133e-06,
      "loss": 0.2728,
      "step": 240
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.336734852226964,
      "learning_rate": 9.99998452796439e-06,
      "loss": 0.2476,
      "step": 241
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.738291971856586,
      "learning_rate": 9.99997894084434e-06,
      "loss": 0.2454,
      "step": 242
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.790128637748417,
      "learning_rate": 9.999972494169947e-06,
      "loss": 0.6086,
      "step": 243
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7308651790427243,
      "learning_rate": 9.999965187942317e-06,
      "loss": 0.1853,
      "step": 244
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.199818694569289,
      "learning_rate": 9.999957022162707e-06,
      "loss": 0.7039,
      "step": 245
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.805045865139413,
      "learning_rate": 9.99994799683252e-06,
      "loss": 0.2403,
      "step": 246
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9040398704910304,
      "learning_rate": 9.999938111953306e-06,
      "loss": 0.272,
      "step": 247
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.707134193314682,
      "learning_rate": 9.99992736752677e-06,
      "loss": 0.5544,
      "step": 248
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.85496132951861,
      "learning_rate": 9.999915763554754e-06,
      "loss": 0.2379,
      "step": 249
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7106851470309155,
      "learning_rate": 9.999903300039253e-06,
      "loss": 0.222,
      "step": 250
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7927942391846008,
      "learning_rate": 9.999889976982413e-06,
      "loss": 0.2565,
      "step": 251
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.35270498595909,
      "learning_rate": 9.99987579438652e-06,
      "loss": 0.465,
      "step": 252
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0421264989373906,
      "learning_rate": 9.999860752254016e-06,
      "loss": 0.2693,
      "step": 253
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0063233375561564,
      "learning_rate": 9.999844850587486e-06,
      "loss": 0.2323,
      "step": 254
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7517620790621067,
      "learning_rate": 9.99982808938966e-06,
      "loss": 0.1961,
      "step": 255
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.8631560015278845,
      "learning_rate": 9.999810468663424e-06,
      "loss": 0.6534,
      "step": 256
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.659747555517259,
      "learning_rate": 9.999791988411807e-06,
      "loss": 0.2167,
      "step": 257
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9200477802119444,
      "learning_rate": 9.999772648637984e-06,
      "loss": 0.2268,
      "step": 258
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0401747345588443,
      "learning_rate": 9.999752449345279e-06,
      "loss": 0.2096,
      "step": 259
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8945229111805526,
      "learning_rate": 9.999731390537168e-06,
      "loss": 0.281,
      "step": 260
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9457059902514418,
      "learning_rate": 9.999709472217268e-06,
      "loss": 0.2435,
      "step": 261
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6952377438783914,
      "learning_rate": 9.999686694389348e-06,
      "loss": 0.2306,
      "step": 262
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.24691913063009,
      "learning_rate": 9.999663057057324e-06,
      "loss": 0.4915,
      "step": 263
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8987471953894361,
      "learning_rate": 9.999638560225259e-06,
      "loss": 0.2321,
      "step": 264
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.413342892065498,
      "learning_rate": 9.999613203897365e-06,
      "loss": 0.6769,
      "step": 265
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.288546702612012,
      "learning_rate": 9.999586988078e-06,
      "loss": 0.2445,
      "step": 266
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7870833671337765,
      "learning_rate": 9.999559912771673e-06,
      "loss": 0.254,
      "step": 267
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6844753213037438,
      "learning_rate": 9.999531977983038e-06,
      "loss": 0.2024,
      "step": 268
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9173313591429482,
      "learning_rate": 9.999503183716894e-06,
      "loss": 0.2603,
      "step": 269
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.726543747182928,
      "learning_rate": 9.999473529978194e-06,
      "loss": 0.2407,
      "step": 270
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6807299636694464,
      "learning_rate": 9.999443016772037e-06,
      "loss": 0.2157,
      "step": 271
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.284256937205807,
      "learning_rate": 9.999411644103665e-06,
      "loss": 0.745,
      "step": 272
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.477556590967655,
      "learning_rate": 9.999379411978474e-06,
      "loss": 0.4085,
      "step": 273
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.618871519150576,
      "learning_rate": 9.999346320402003e-06,
      "loss": 0.1785,
      "step": 274
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.6243819270100595,
      "learning_rate": 9.999312369379944e-06,
      "loss": 0.5249,
      "step": 275
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8126700578326513,
      "learning_rate": 9.99927755891813e-06,
      "loss": 0.2243,
      "step": 276
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7789433293670927,
      "learning_rate": 9.99924188902255e-06,
      "loss": 0.2558,
      "step": 277
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9433483664557305,
      "learning_rate": 9.99920535969933e-06,
      "loss": 0.2519,
      "step": 278
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.814009823120058,
      "learning_rate": 9.999167970954756e-06,
      "loss": 0.5737,
      "step": 279
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7522706489891633,
      "learning_rate": 9.999129722795248e-06,
      "loss": 0.2434,
      "step": 280
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.437379162716988,
      "learning_rate": 9.999090615227389e-06,
      "loss": 0.6007,
      "step": 281
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.81216081425422,
      "learning_rate": 9.999050648257898e-06,
      "loss": 0.2532,
      "step": 282
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7104308549325318,
      "learning_rate": 9.999009821893648e-06,
      "loss": 0.2282,
      "step": 283
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7542161074730827,
      "learning_rate": 9.998968136141655e-06,
      "loss": 0.2561,
      "step": 284
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6453475939855429,
      "learning_rate": 9.998925591009086e-06,
      "loss": 0.2265,
      "step": 285
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6567352057052007,
      "learning_rate": 9.998882186503256e-06,
      "loss": 0.2249,
      "step": 286
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6255071459411878,
      "learning_rate": 9.998837922631625e-06,
      "loss": 0.2149,
      "step": 287
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6825610020121933,
      "learning_rate": 9.998792799401804e-06,
      "loss": 0.2298,
      "step": 288
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.337183886493983,
      "learning_rate": 9.998746816821551e-06,
      "loss": 0.7256,
      "step": 289
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.931722691029489,
      "learning_rate": 9.99869997489877e-06,
      "loss": 0.8301,
      "step": 290
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7287107805975577,
      "learning_rate": 9.99865227364151e-06,
      "loss": 0.2247,
      "step": 291
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9319242003334012,
      "learning_rate": 9.998603713057977e-06,
      "loss": 0.2857,
      "step": 292
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.01197114069046,
      "learning_rate": 9.998554293156518e-06,
      "loss": 0.2991,
      "step": 293
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.65955360504946,
      "learning_rate": 9.998504013945627e-06,
      "loss": 0.6374,
      "step": 294
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7556688425491125,
      "learning_rate": 9.998452875433948e-06,
      "loss": 0.2467,
      "step": 295
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.764544823505664,
      "learning_rate": 9.998400877630272e-06,
      "loss": 0.231,
      "step": 296
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8266365262597084,
      "learning_rate": 9.99834802054354e-06,
      "loss": 0.2375,
      "step": 297
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6754132697303046,
      "learning_rate": 9.998294304182837e-06,
      "loss": 0.2398,
      "step": 298
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6826909757757187,
      "learning_rate": 9.998239728557399e-06,
      "loss": 0.2417,
      "step": 299
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.167854846062482,
      "learning_rate": 9.998184293676606e-06,
      "loss": 0.6326,
      "step": 300
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.119365337839217,
      "learning_rate": 9.998127999549988e-06,
      "loss": 0.2756,
      "step": 301
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5746889604542103,
      "learning_rate": 9.998070846187225e-06,
      "loss": 0.2027,
      "step": 302
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8528843945538818,
      "learning_rate": 9.99801283359814e-06,
      "loss": 0.2168,
      "step": 303
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9346015589512269,
      "learning_rate": 9.997953961792708e-06,
      "loss": 0.2481,
      "step": 304
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.72298118923922,
      "learning_rate": 9.997894230781048e-06,
      "loss": 0.6999,
      "step": 305
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7910255770916688,
      "learning_rate": 9.99783364057343e-06,
      "loss": 0.2319,
      "step": 306
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8713470700613266,
      "learning_rate": 9.997772191180269e-06,
      "loss": 0.2817,
      "step": 307
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.607849143622752,
      "learning_rate": 9.997709882612128e-06,
      "loss": 0.2095,
      "step": 308
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8415118572911802,
      "learning_rate": 9.99764671487972e-06,
      "loss": 0.231,
      "step": 309
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5443999925008234,
      "learning_rate": 9.997582687993905e-06,
      "loss": 0.2387,
      "step": 310
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7479969338175332,
      "learning_rate": 9.99751780196569e-06,
      "loss": 0.2733,
      "step": 311
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5839123906268797,
      "learning_rate": 9.997452056806226e-06,
      "loss": 0.182,
      "step": 312
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.743290949583812,
      "learning_rate": 9.99738545252682e-06,
      "loss": 0.7219,
      "step": 313
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.948926141329891,
      "learning_rate": 9.99731798913892e-06,
      "loss": 0.5588,
      "step": 314
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8753336800986329,
      "learning_rate": 9.997249666654122e-06,
      "loss": 0.243,
      "step": 315
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8084246131051833,
      "learning_rate": 9.997180485084175e-06,
      "loss": 0.2508,
      "step": 316
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5342935817745358,
      "learning_rate": 9.99711044444097e-06,
      "loss": 0.2226,
      "step": 317
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7444271904847302,
      "learning_rate": 9.997039544736547e-06,
      "loss": 0.2165,
      "step": 318
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7380610873995055,
      "learning_rate": 9.996967785983097e-06,
      "loss": 0.2421,
      "step": 319
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8007179329882057,
      "learning_rate": 9.996895168192954e-06,
      "loss": 0.2408,
      "step": 320
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8404328661802543,
      "learning_rate": 9.996821691378603e-06,
      "loss": 0.1588,
      "step": 321
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6552110519213554,
      "learning_rate": 9.996747355552675e-06,
      "loss": 0.2015,
      "step": 322
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.874292043749998,
      "learning_rate": 9.99667216072795e-06,
      "loss": 0.2795,
      "step": 323
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6883933657602626,
      "learning_rate": 9.996596106917353e-06,
      "loss": 0.1792,
      "step": 324
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6321079462909487,
      "learning_rate": 9.99651919413396e-06,
      "loss": 0.2042,
      "step": 325
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8758681165099163,
      "learning_rate": 9.996441422390994e-06,
      "loss": 0.2284,
      "step": 326
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7096732675865267,
      "learning_rate": 9.996362791701822e-06,
      "loss": 0.2463,
      "step": 327
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7521281507383506,
      "learning_rate": 9.996283302079965e-06,
      "loss": 0.1998,
      "step": 328
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8475229620357405,
      "learning_rate": 9.996202953539085e-06,
      "loss": 0.2615,
      "step": 329
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7036144422250832,
      "learning_rate": 9.996121746092996e-06,
      "loss": 0.2355,
      "step": 330
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.687151243352997,
      "learning_rate": 9.99603967975566e-06,
      "loss": 0.622,
      "step": 331
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9877859939173192,
      "learning_rate": 9.995956754541185e-06,
      "loss": 0.2384,
      "step": 332
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.853602285541162,
      "learning_rate": 9.995872970463824e-06,
      "loss": 0.2591,
      "step": 333
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.355921203610223,
      "learning_rate": 9.995788327537983e-06,
      "loss": 0.6523,
      "step": 334
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7400880967514203,
      "learning_rate": 9.995702825778213e-06,
      "loss": 0.2497,
      "step": 335
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.961121256458503,
      "learning_rate": 9.995616465199209e-06,
      "loss": 0.5828,
      "step": 336
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7169258399252405,
      "learning_rate": 9.995529245815824e-06,
      "loss": 0.1894,
      "step": 337
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6114282008916527,
      "learning_rate": 9.995441167643048e-06,
      "loss": 0.2563,
      "step": 338
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7003908627481015,
      "learning_rate": 9.995352230696021e-06,
      "loss": 0.1994,
      "step": 339
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6279152663643415,
      "learning_rate": 9.995262434990036e-06,
      "loss": 0.2562,
      "step": 340
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5683475552209623,
      "learning_rate": 9.995171780540528e-06,
      "loss": 0.2243,
      "step": 341
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8008811499578228,
      "learning_rate": 9.995080267363082e-06,
      "loss": 0.2499,
      "step": 342
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.623355692877003,
      "learning_rate": 9.994987895473431e-06,
      "loss": 0.2326,
      "step": 343
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.55131499587847,
      "learning_rate": 9.994894664887453e-06,
      "loss": 0.5141,
      "step": 344
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6176094463575743,
      "learning_rate": 9.994800575621176e-06,
      "loss": 0.2812,
      "step": 345
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.812003520679854,
      "learning_rate": 9.994705627690777e-06,
      "loss": 0.2431,
      "step": 346
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8716195646668083,
      "learning_rate": 9.994609821112576e-06,
      "loss": 0.2333,
      "step": 347
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9186588802877804,
      "learning_rate": 9.994513155903042e-06,
      "loss": 0.2672,
      "step": 348
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8253015460415536,
      "learning_rate": 9.994415632078797e-06,
      "loss": 0.2409,
      "step": 349
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5762077712985119,
      "learning_rate": 9.994317249656607e-06,
      "loss": 0.2439,
      "step": 350
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.890913221545547,
      "learning_rate": 9.994218008653381e-06,
      "loss": 0.238,
      "step": 351
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.516510656514113,
      "learning_rate": 9.994117909086179e-06,
      "loss": 0.2194,
      "step": 352
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.389868759889487,
      "learning_rate": 9.994016950972214e-06,
      "loss": 0.5813,
      "step": 353
    },
    {
      "epoch": 0.05,
      "grad_norm": 17.559355922634747,
      "learning_rate": 9.99391513432884e-06,
      "loss": 0.5567,
      "step": 354
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.817552323172319,
      "learning_rate": 9.993812459173557e-06,
      "loss": 0.2492,
      "step": 355
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.455642348146528,
      "learning_rate": 9.993708925524022e-06,
      "loss": 0.6038,
      "step": 356
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.834432779875858,
      "learning_rate": 9.993604533398029e-06,
      "loss": 0.2584,
      "step": 357
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6328952778931827,
      "learning_rate": 9.993499282813528e-06,
      "loss": 0.2228,
      "step": 358
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.661731264233014,
      "learning_rate": 9.993393173788608e-06,
      "loss": 0.2064,
      "step": 359
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.818491429596564,
      "learning_rate": 9.993286206341515e-06,
      "loss": 0.2866,
      "step": 360
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6400089740209856,
      "learning_rate": 9.993178380490636e-06,
      "loss": 0.2257,
      "step": 361
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7405559757057736,
      "learning_rate": 9.993069696254506e-06,
      "loss": 0.2457,
      "step": 362
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7239442186650715,
      "learning_rate": 9.992960153651812e-06,
      "loss": 0.2264,
      "step": 363
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6199905786485465,
      "learning_rate": 9.992849752701384e-06,
      "loss": 0.2502,
      "step": 364
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6079631043860492,
      "learning_rate": 9.9927384934222e-06,
      "loss": 0.2117,
      "step": 365
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6552775455419944,
      "learning_rate": 9.99262637583339e-06,
      "loss": 0.2322,
      "step": 366
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8564863159167198,
      "learning_rate": 9.992513399954225e-06,
      "loss": 0.2759,
      "step": 367
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.833309076741993,
      "learning_rate": 9.99239956580413e-06,
      "loss": 0.7979,
      "step": 368
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.335078070041813,
      "learning_rate": 9.99228487340267e-06,
      "loss": 0.5283,
      "step": 369
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.95478417770924,
      "learning_rate": 9.992169322769568e-06,
      "loss": 0.7819,
      "step": 370
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.234201308873233,
      "learning_rate": 9.992052913924683e-06,
      "loss": 0.7247,
      "step": 371
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.275775438996168,
      "learning_rate": 9.991935646888031e-06,
      "loss": 0.3029,
      "step": 372
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8660549503167803,
      "learning_rate": 9.991817521679769e-06,
      "loss": 0.2275,
      "step": 373
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.348267346365922,
      "learning_rate": 9.991698538320205e-06,
      "loss": 0.6452,
      "step": 374
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.659562534676429,
      "learning_rate": 9.991578696829793e-06,
      "loss": 0.2443,
      "step": 375
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.19633102833578,
      "learning_rate": 9.991457997229136e-06,
      "loss": 0.5225,
      "step": 376
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.671570560628976,
      "learning_rate": 9.991336439538983e-06,
      "loss": 0.2422,
      "step": 377
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.2352387351201,
      "learning_rate": 9.991214023780232e-06,
      "loss": 0.5632,
      "step": 378
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5614666193107307,
      "learning_rate": 9.991090749973926e-06,
      "loss": 0.2129,
      "step": 379
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6332829557316741,
      "learning_rate": 9.99096661814126e-06,
      "loss": 0.2701,
      "step": 380
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5839969345021678,
      "learning_rate": 9.990841628303571e-06,
      "loss": 0.2524,
      "step": 381
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6303735916541966,
      "learning_rate": 9.990715780482348e-06,
      "loss": 0.2188,
      "step": 382
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7263788629052044,
      "learning_rate": 9.990589074699225e-06,
      "loss": 0.2884,
      "step": 383
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5594815872352896,
      "learning_rate": 9.990461510975983e-06,
      "loss": 0.2375,
      "step": 384
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.963283202244964,
      "learning_rate": 9.990333089334556e-06,
      "loss": 0.6585,
      "step": 385
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.040538126542633,
      "learning_rate": 9.990203809797014e-06,
      "loss": 0.5327,
      "step": 386
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9255707315771995,
      "learning_rate": 9.990073672385588e-06,
      "loss": 0.2728,
      "step": 387
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5791783801720203,
      "learning_rate": 9.989942677122648e-06,
      "loss": 0.2758,
      "step": 388
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5250168533350654,
      "learning_rate": 9.989810824030712e-06,
      "loss": 0.2027,
      "step": 389
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.635760353060768,
      "learning_rate": 9.989678113132451e-06,
      "loss": 0.2009,
      "step": 390
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.813384512614699,
      "learning_rate": 9.989544544450675e-06,
      "loss": 0.5301,
      "step": 391
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6395946018307472,
      "learning_rate": 9.989410118008348e-06,
      "loss": 0.2484,
      "step": 392
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8376206577716234,
      "learning_rate": 9.98927483382858e-06,
      "loss": 0.2781,
      "step": 393
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.970563245540003,
      "learning_rate": 9.989138691934628e-06,
      "loss": 0.7069,
      "step": 394
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7316291814016878,
      "learning_rate": 9.989001692349894e-06,
      "loss": 0.2568,
      "step": 395
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.635996403655164,
      "learning_rate": 9.988863835097934e-06,
      "loss": 0.5877,
      "step": 396
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8568937998836006,
      "learning_rate": 9.988725120202442e-06,
      "loss": 0.2609,
      "step": 397
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.2780932253193855,
      "learning_rate": 9.98858554768727e-06,
      "loss": 0.5997,
      "step": 398
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8383228431983276,
      "learning_rate": 9.988445117576408e-06,
      "loss": 0.2719,
      "step": 399
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7122310329319328,
      "learning_rate": 9.988303829894002e-06,
      "loss": 0.2705,
      "step": 400
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7796607205935646,
      "learning_rate": 9.988161684664336e-06,
      "loss": 0.2154,
      "step": 401
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9033654054762172,
      "learning_rate": 9.988018681911849e-06,
      "loss": 0.2769,
      "step": 402
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.25662979701234,
      "learning_rate": 9.987874821661124e-06,
      "loss": 0.578,
      "step": 403
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7137170843989922,
      "learning_rate": 9.987730103936895e-06,
      "loss": 0.2596,
      "step": 404
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5817136125444566,
      "learning_rate": 9.987584528764036e-06,
      "loss": 0.1575,
      "step": 405
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.470706623069368,
      "learning_rate": 9.987438096167577e-06,
      "loss": 0.2279,
      "step": 406
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6590214531563414,
      "learning_rate": 9.987290806172689e-06,
      "loss": 0.2307,
      "step": 407
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7259824839235343,
      "learning_rate": 9.987142658804692e-06,
      "loss": 0.2578,
      "step": 408
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.400359809994468,
      "learning_rate": 9.986993654089059e-06,
      "loss": 0.2186,
      "step": 409
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.712955938194739,
      "learning_rate": 9.986843792051402e-06,
      "loss": 0.2403,
      "step": 410
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.554447391349472,
      "learning_rate": 9.986693072717483e-06,
      "loss": 0.2464,
      "step": 411
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.93383099573809,
      "learning_rate": 9.986541496113215e-06,
      "loss": 0.5979,
      "step": 412
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8262005348683623,
      "learning_rate": 9.986389062264656e-06,
      "loss": 0.232,
      "step": 413
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6483960379118057,
      "learning_rate": 9.98623577119801e-06,
      "loss": 0.1864,
      "step": 414
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.202429656189494,
      "learning_rate": 9.98608162293963e-06,
      "loss": 0.675,
      "step": 415
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.847406226626909,
      "learning_rate": 9.985926617516012e-06,
      "loss": 0.3009,
      "step": 416
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7436987215787323,
      "learning_rate": 9.98577075495381e-06,
      "loss": 0.2356,
      "step": 417
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.625080649134119,
      "learning_rate": 9.985614035279813e-06,
      "loss": 0.2006,
      "step": 418
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6864224208141392,
      "learning_rate": 9.985456458520968e-06,
      "loss": 0.2382,
      "step": 419
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9620214445497024,
      "learning_rate": 9.98529802470436e-06,
      "loss": 0.242,
      "step": 420
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.602232195537267,
      "learning_rate": 9.985138733857225e-06,
      "loss": 0.2564,
      "step": 421
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5373818513807578,
      "learning_rate": 9.984978586006951e-06,
      "loss": 0.2122,
      "step": 422
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.777680274001366,
      "learning_rate": 9.984817581181068e-06,
      "loss": 0.2695,
      "step": 423
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5621982226769977,
      "learning_rate": 9.984655719407252e-06,
      "loss": 0.2344,
      "step": 424
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6622123279298093,
      "learning_rate": 9.98449300071333e-06,
      "loss": 0.2266,
      "step": 425
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.664954040865086,
      "learning_rate": 9.98432942512728e-06,
      "loss": 0.5894,
      "step": 426
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7537719892674517,
      "learning_rate": 9.984164992677215e-06,
      "loss": 0.3046,
      "step": 427
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.632439769416365,
      "learning_rate": 9.983999703391408e-06,
      "loss": 0.7596,
      "step": 428
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8098535250715984,
      "learning_rate": 9.98383355729827e-06,
      "loss": 0.3012,
      "step": 429
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5733400637676733,
      "learning_rate": 9.983666554426367e-06,
      "loss": 0.2191,
      "step": 430
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6662733343358092,
      "learning_rate": 9.98349869480441e-06,
      "loss": 0.2018,
      "step": 431
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.36631421067763,
      "learning_rate": 9.983329978461252e-06,
      "loss": 0.5276,
      "step": 432
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6803728803318794,
      "learning_rate": 9.9831604054259e-06,
      "loss": 0.212,
      "step": 433
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6375660936724619,
      "learning_rate": 9.982989975727502e-06,
      "loss": 0.2443,
      "step": 434
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.048726927702341,
      "learning_rate": 9.982818689395362e-06,
      "loss": 0.3169,
      "step": 435
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.585137694234173,
      "learning_rate": 9.982646546458922e-06,
      "loss": 0.2098,
      "step": 436
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4257362027074867,
      "learning_rate": 9.982473546947777e-06,
      "loss": 0.1935,
      "step": 437
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.741512766814629,
      "learning_rate": 9.982299690891668e-06,
      "loss": 0.2347,
      "step": 438
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.856975603436903,
      "learning_rate": 9.982124978320482e-06,
      "loss": 0.242,
      "step": 439
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.596538204015094,
      "learning_rate": 9.981949409264256e-06,
      "loss": 0.2289,
      "step": 440
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6439762514884262,
      "learning_rate": 9.981772983753169e-06,
      "loss": 0.2354,
      "step": 441
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.352948532035535,
      "learning_rate": 9.981595701817552e-06,
      "loss": 0.6304,
      "step": 442
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5359481521396647,
      "learning_rate": 9.981417563487884e-06,
      "loss": 0.1685,
      "step": 443
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.545638297895529,
      "learning_rate": 9.981238568794787e-06,
      "loss": 0.2051,
      "step": 444
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8577056213439953,
      "learning_rate": 9.981058717769031e-06,
      "loss": 0.2369,
      "step": 445
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7971600957649143,
      "learning_rate": 9.980878010441539e-06,
      "loss": 0.2427,
      "step": 446
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7229362087396198,
      "learning_rate": 9.980696446843371e-06,
      "loss": 0.2284,
      "step": 447
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6647454872611889,
      "learning_rate": 9.980514027005743e-06,
      "loss": 0.1802,
      "step": 448
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7604744010030937,
      "learning_rate": 9.980330750960014e-06,
      "loss": 0.2547,
      "step": 449
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7852443427934885,
      "learning_rate": 9.980146618737694e-06,
      "loss": 0.1989,
      "step": 450
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6672674407327284,
      "learning_rate": 9.979961630370435e-06,
      "loss": 0.2315,
      "step": 451
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5176960346456279,
      "learning_rate": 9.979775785890037e-06,
      "loss": 0.2385,
      "step": 452
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.657546809171371,
      "learning_rate": 9.979589085328454e-06,
      "loss": 0.223,
      "step": 453
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5303093673573211,
      "learning_rate": 9.979401528717777e-06,
      "loss": 0.2532,
      "step": 454
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.659857149085432,
      "learning_rate": 9.979213116090251e-06,
      "loss": 0.2515,
      "step": 455
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6873139880112082,
      "learning_rate": 9.979023847478268e-06,
      "loss": 0.2953,
      "step": 456
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.292515452424244,
      "learning_rate": 9.978833722914362e-06,
      "loss": 0.5223,
      "step": 457
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7874432250005288,
      "learning_rate": 9.97864274243122e-06,
      "loss": 0.2822,
      "step": 458
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5608351942930296,
      "learning_rate": 9.978450906061673e-06,
      "loss": 0.2214,
      "step": 459
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6243720412251763,
      "learning_rate": 9.978258213838701e-06,
      "loss": 0.2004,
      "step": 460
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.575892520180479,
      "learning_rate": 9.978064665795429e-06,
      "loss": 0.2803,
      "step": 461
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.667964998008004,
      "learning_rate": 9.977870261965128e-06,
      "loss": 0.2233,
      "step": 462
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.669940506213279,
      "learning_rate": 9.977675002381222e-06,
      "loss": 0.2588,
      "step": 463
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6016771450702958,
      "learning_rate": 9.977478887077277e-06,
      "loss": 0.2488,
      "step": 464
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.563070409121154,
      "learning_rate": 9.977281916087008e-06,
      "loss": 0.6625,
      "step": 465
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6753895358797057,
      "learning_rate": 9.977084089444276e-06,
      "loss": 0.1486,
      "step": 466
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6265252649347042,
      "learning_rate": 9.976885407183088e-06,
      "loss": 0.2375,
      "step": 467
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7533248591533241,
      "learning_rate": 9.976685869337603e-06,
      "loss": 0.2975,
      "step": 468
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.521339428825348,
      "learning_rate": 9.976485475942123e-06,
      "loss": 0.2068,
      "step": 469
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.429314504243305,
      "learning_rate": 9.976284227031097e-06,
      "loss": 0.4024,
      "step": 470
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.558729636612331,
      "learning_rate": 9.976082122639122e-06,
      "loss": 0.2161,
      "step": 471
    },
    {
      "epoch": 0.06,
      "grad_norm": 12.906402414991943,
      "learning_rate": 9.975879162800942e-06,
      "loss": 0.6319,
      "step": 472
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7559141264518867,
      "learning_rate": 9.975675347551449e-06,
      "loss": 0.2586,
      "step": 473
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.582182302617656,
      "learning_rate": 9.975470676925681e-06,
      "loss": 0.5672,
      "step": 474
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5566532192030131,
      "learning_rate": 9.975265150958822e-06,
      "loss": 0.2017,
      "step": 475
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7167627137826833,
      "learning_rate": 9.975058769686205e-06,
      "loss": 0.2174,
      "step": 476
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6942997343002146,
      "learning_rate": 9.97485153314331e-06,
      "loss": 0.2291,
      "step": 477
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6831518486413768,
      "learning_rate": 9.974643441365765e-06,
      "loss": 0.2853,
      "step": 478
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7187526432620863,
      "learning_rate": 9.97443449438934e-06,
      "loss": 0.2179,
      "step": 479
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6974058347990475,
      "learning_rate": 9.974224692249956e-06,
      "loss": 0.2675,
      "step": 480
    },
    {
      "epoch": 0.06,
      "grad_norm": 10.230417227106765,
      "learning_rate": 9.974014034983684e-06,
      "loss": 0.5662,
      "step": 481
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.909770372110493,
      "learning_rate": 9.973802522626731e-06,
      "loss": 0.2727,
      "step": 482
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6954119880140075,
      "learning_rate": 9.973590155215467e-06,
      "loss": 0.2414,
      "step": 483
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4693080775005698,
      "learning_rate": 9.973376932786396e-06,
      "loss": 0.2239,
      "step": 484
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.556897403253936,
      "learning_rate": 9.973162855376173e-06,
      "loss": 0.6073,
      "step": 485
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7243986950324641,
      "learning_rate": 9.972947923021602e-06,
      "loss": 0.2878,
      "step": 486
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5495583860221074,
      "learning_rate": 9.972732135759631e-06,
      "loss": 0.2327,
      "step": 487
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5433731498937793,
      "learning_rate": 9.972515493627357e-06,
      "loss": 0.1802,
      "step": 488
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6469014151120231,
      "learning_rate": 9.972297996662023e-06,
      "loss": 0.2677,
      "step": 489
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.691666131104068,
      "learning_rate": 9.972079644901021e-06,
      "loss": 0.2369,
      "step": 490
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7213745199657553,
      "learning_rate": 9.971860438381886e-06,
      "loss": 0.2278,
      "step": 491
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6740234123733575,
      "learning_rate": 9.971640377142301e-06,
      "loss": 0.266,
      "step": 492
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5564930192400246,
      "learning_rate": 9.971419461220102e-06,
      "loss": 0.2417,
      "step": 493
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6629309817799125,
      "learning_rate": 9.971197690653263e-06,
      "loss": 0.3001,
      "step": 494
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.94167317288281,
      "learning_rate": 9.970975065479909e-06,
      "loss": 0.2617,
      "step": 495
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6416209699004836,
      "learning_rate": 9.970751585738315e-06,
      "loss": 0.2332,
      "step": 496
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.650438348975666,
      "learning_rate": 9.970527251466895e-06,
      "loss": 0.7029,
      "step": 497
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8708029572605218,
      "learning_rate": 9.970302062704218e-06,
      "loss": 0.2912,
      "step": 498
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7562086762173517,
      "learning_rate": 9.970076019488997e-06,
      "loss": 0.2907,
      "step": 499
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6743771212961882,
      "learning_rate": 9.969849121860089e-06,
      "loss": 0.2592,
      "step": 500
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4360559378699882,
      "learning_rate": 9.969621369856502e-06,
      "loss": 0.1667,
      "step": 501
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.026366666431601,
      "learning_rate": 9.969392763517387e-06,
      "loss": 0.7516,
      "step": 502
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.525178304387076,
      "learning_rate": 9.969163302882048e-06,
      "loss": 0.2081,
      "step": 503
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.57796073303667,
      "learning_rate": 9.968932987989927e-06,
      "loss": 0.2606,
      "step": 504
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.684216405400672,
      "learning_rate": 9.968701818880624e-06,
      "loss": 0.2412,
      "step": 505
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6833586231837239,
      "learning_rate": 9.968469795593874e-06,
      "loss": 0.2505,
      "step": 506
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5314955965670223,
      "learning_rate": 9.968236918169567e-06,
      "loss": 0.2112,
      "step": 507
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.72484881800808,
      "learning_rate": 9.968003186647737e-06,
      "loss": 0.2785,
      "step": 508
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8713852236765587,
      "learning_rate": 9.967768601068566e-06,
      "loss": 0.2737,
      "step": 509
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5334030660515479,
      "learning_rate": 9.96753316147238e-06,
      "loss": 0.2462,
      "step": 510
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6851799984783564,
      "learning_rate": 9.967296867899656e-06,
      "loss": 0.2481,
      "step": 511
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.669846433004507,
      "learning_rate": 9.967059720391014e-06,
      "loss": 0.2716,
      "step": 512
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.698417495008913,
      "learning_rate": 9.966821718987222e-06,
      "loss": 0.2201,
      "step": 513
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6228187025362186,
      "learning_rate": 9.966582863729198e-06,
      "loss": 0.2538,
      "step": 514
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.592952935868426,
      "learning_rate": 9.966343154658002e-06,
      "loss": 0.2525,
      "step": 515
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6357807414588403,
      "learning_rate": 9.966102591814842e-06,
      "loss": 0.2343,
      "step": 516
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.329774322772522,
      "learning_rate": 9.965861175241075e-06,
      "loss": 0.6782,
      "step": 517
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.469387755436989,
      "learning_rate": 9.965618904978203e-06,
      "loss": 0.5803,
      "step": 518
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4103635208746614,
      "learning_rate": 9.965375781067874e-06,
      "loss": 0.2551,
      "step": 519
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5050451683262047,
      "learning_rate": 9.965131803551885e-06,
      "loss": 0.2421,
      "step": 520
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5923228419296174,
      "learning_rate": 9.96488697247218e-06,
      "loss": 0.212,
      "step": 521
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.727342723257293,
      "learning_rate": 9.964641287870844e-06,
      "loss": 0.2253,
      "step": 522
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.842169039801606,
      "learning_rate": 9.964394749790117e-06,
      "loss": 0.6811,
      "step": 523
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3369968676046182,
      "learning_rate": 9.964147358272379e-06,
      "loss": 0.1361,
      "step": 524
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5991503417829211,
      "learning_rate": 9.963899113360161e-06,
      "loss": 0.2352,
      "step": 525
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5632005439365482,
      "learning_rate": 9.96365001509614e-06,
      "loss": 0.2167,
      "step": 526
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.560156161422908,
      "learning_rate": 9.963400063523136e-06,
      "loss": 0.2129,
      "step": 527
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.192462598211521,
      "learning_rate": 9.963149258684122e-06,
      "loss": 0.5961,
      "step": 528
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.667129841258835,
      "learning_rate": 9.962897600622212e-06,
      "loss": 0.2828,
      "step": 529
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.747923425203638,
      "learning_rate": 9.962645089380672e-06,
      "loss": 0.5819,
      "step": 530
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7320652974484683,
      "learning_rate": 9.962391725002906e-06,
      "loss": 0.2333,
      "step": 531
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5801650325629488,
      "learning_rate": 9.962137507532474e-06,
      "loss": 0.1887,
      "step": 532
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4834713472879753,
      "learning_rate": 9.961882437013079e-06,
      "loss": 0.2102,
      "step": 533
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9321205808314295,
      "learning_rate": 9.961626513488572e-06,
      "loss": 0.2435,
      "step": 534
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9296174945344742,
      "learning_rate": 9.961369737002943e-06,
      "loss": 0.1966,
      "step": 535
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7085114063446287,
      "learning_rate": 9.961112107600343e-06,
      "loss": 0.2247,
      "step": 536
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4797647148396784,
      "learning_rate": 9.960853625325056e-06,
      "loss": 0.2174,
      "step": 537
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5093554573444663,
      "learning_rate": 9.960594290221519e-06,
      "loss": 0.2362,
      "step": 538
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.466966248453093,
      "learning_rate": 9.960334102334316e-06,
      "loss": 0.6088,
      "step": 539
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.003461631114323,
      "learning_rate": 9.960073061708176e-06,
      "loss": 0.6316,
      "step": 540
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7228814503195047,
      "learning_rate": 9.959811168387974e-06,
      "loss": 0.2945,
      "step": 541
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7983125830904985,
      "learning_rate": 9.959548422418734e-06,
      "loss": 0.2589,
      "step": 542
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6612849774349312,
      "learning_rate": 9.959284823845623e-06,
      "loss": 0.2593,
      "step": 543
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.045377396765115,
      "learning_rate": 9.959020372713959e-06,
      "loss": 0.8933,
      "step": 544
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5566222103913412,
      "learning_rate": 9.958755069069202e-06,
      "loss": 0.2205,
      "step": 545
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7040642554887317,
      "learning_rate": 9.958488912956961e-06,
      "loss": 0.2333,
      "step": 546
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.669870884898359,
      "learning_rate": 9.958221904422993e-06,
      "loss": 0.7903,
      "step": 547
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5234725023393179,
      "learning_rate": 9.9579540435132e-06,
      "loss": 0.1888,
      "step": 548
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7069829360504416,
      "learning_rate": 9.957685330273628e-06,
      "loss": 0.2514,
      "step": 549
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6396600896854683,
      "learning_rate": 9.957415764750474e-06,
      "loss": 0.2485,
      "step": 550
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6487387840871346,
      "learning_rate": 9.957145346990079e-06,
      "loss": 0.2511,
      "step": 551
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.646950002106126,
      "learning_rate": 9.95687407703893e-06,
      "loss": 0.5157,
      "step": 552
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7219027656992856,
      "learning_rate": 9.956601954943664e-06,
      "loss": 0.2722,
      "step": 553
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.81381278996118,
      "learning_rate": 9.956328980751057e-06,
      "loss": 0.5722,
      "step": 554
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.102499532827676,
      "learning_rate": 9.956055154508042e-06,
      "loss": 0.6748,
      "step": 555
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.941486103433965,
      "learning_rate": 9.955780476261689e-06,
      "loss": 0.2272,
      "step": 556
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.651436243851542,
      "learning_rate": 9.955504946059221e-06,
      "loss": 0.3088,
      "step": 557
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7854686629651233,
      "learning_rate": 9.955228563948003e-06,
      "loss": 0.2796,
      "step": 558
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6945339189428745,
      "learning_rate": 9.954951329975549e-06,
      "loss": 0.2314,
      "step": 559
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5940787380814525,
      "learning_rate": 9.954673244189518e-06,
      "loss": 0.2398,
      "step": 560
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.782815155561212,
      "learning_rate": 9.954394306637719e-06,
      "loss": 0.2503,
      "step": 561
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6166865747821855,
      "learning_rate": 9.9541145173681e-06,
      "loss": 0.2024,
      "step": 562
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.045656452676214,
      "learning_rate": 9.953833876428763e-06,
      "loss": 0.6469,
      "step": 563
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5868914629246105,
      "learning_rate": 9.953552383867953e-06,
      "loss": 0.2514,
      "step": 564
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.801863568290612,
      "learning_rate": 9.953270039734063e-06,
      "loss": 0.2563,
      "step": 565
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6605455098158908,
      "learning_rate": 9.952986844075628e-06,
      "loss": 0.2659,
      "step": 566
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.052487384072722,
      "learning_rate": 9.952702796941334e-06,
      "loss": 0.459,
      "step": 567
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5886727213182208,
      "learning_rate": 9.952417898380015e-06,
      "loss": 0.228,
      "step": 568
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4482267708414633,
      "learning_rate": 9.952132148440644e-06,
      "loss": 0.2221,
      "step": 569
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4665188891782803,
      "learning_rate": 9.951845547172347e-06,
      "loss": 0.2454,
      "step": 570
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.693298162502021,
      "learning_rate": 9.951558094624395e-06,
      "loss": 0.2737,
      "step": 571
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4826956750747202,
      "learning_rate": 9.951269790846204e-06,
      "loss": 0.2172,
      "step": 572
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6730098035722134,
      "learning_rate": 9.950980635887332e-06,
      "loss": 0.2927,
      "step": 573
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8999041249208097,
      "learning_rate": 9.950690629797494e-06,
      "loss": 0.2585,
      "step": 574
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6524621418634156,
      "learning_rate": 9.950399772626545e-06,
      "loss": 0.2345,
      "step": 575
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4172729455183433,
      "learning_rate": 9.950108064424482e-06,
      "loss": 0.2538,
      "step": 576
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5374651893220643,
      "learning_rate": 9.949815505241458e-06,
      "loss": 0.2366,
      "step": 577
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5888945754100752,
      "learning_rate": 9.949522095127765e-06,
      "loss": 0.2076,
      "step": 578
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6763336095767727,
      "learning_rate": 9.949227834133845e-06,
      "loss": 0.2337,
      "step": 579
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7673717235554736,
      "learning_rate": 9.948932722310283e-06,
      "loss": 0.2649,
      "step": 580
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9409042114184674,
      "learning_rate": 9.948636759707815e-06,
      "loss": 0.2597,
      "step": 581
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6148629468572844,
      "learning_rate": 9.948339946377316e-06,
      "loss": 0.2378,
      "step": 582
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8406148622873033,
      "learning_rate": 9.948042282369816e-06,
      "loss": 0.2702,
      "step": 583
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5979083494861022,
      "learning_rate": 9.947743767736485e-06,
      "loss": 0.2382,
      "step": 584
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5798537925500835,
      "learning_rate": 9.947444402528638e-06,
      "loss": 0.2277,
      "step": 585
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7699019416132222,
      "learning_rate": 9.947144186797747e-06,
      "loss": 0.2934,
      "step": 586
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8973506980408232,
      "learning_rate": 9.946843120595415e-06,
      "loss": 0.1929,
      "step": 587
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.142799579543763,
      "learning_rate": 9.946541203973402e-06,
      "loss": 0.7074,
      "step": 588
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.981287016848829,
      "learning_rate": 9.94623843698361e-06,
      "loss": 0.6793,
      "step": 589
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.832844979899832,
      "learning_rate": 9.945934819678092e-06,
      "loss": 0.6359,
      "step": 590
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8896111521034658,
      "learning_rate": 9.945630352109039e-06,
      "loss": 0.2968,
      "step": 591
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7373900341834845,
      "learning_rate": 9.945325034328791e-06,
      "loss": 0.2416,
      "step": 592
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.489171058559639,
      "learning_rate": 9.945018866389844e-06,
      "loss": 0.226,
      "step": 593
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5047807082334599,
      "learning_rate": 9.944711848344822e-06,
      "loss": 0.2468,
      "step": 594
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.560816215337684,
      "learning_rate": 9.94440398024651e-06,
      "loss": 0.2309,
      "step": 595
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.0042855472529,
      "learning_rate": 9.944095262147835e-06,
      "loss": 0.6598,
      "step": 596
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7137795712556374,
      "learning_rate": 9.943785694101865e-06,
      "loss": 0.2549,
      "step": 597
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.686518224870548,
      "learning_rate": 9.943475276161823e-06,
      "loss": 0.2251,
      "step": 598
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5341010410475493,
      "learning_rate": 9.943164008381072e-06,
      "loss": 0.2246,
      "step": 599
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6920378145992494,
      "learning_rate": 9.94285189081312e-06,
      "loss": 0.2557,
      "step": 600
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5508041469108993,
      "learning_rate": 9.942538923511627e-06,
      "loss": 0.215,
      "step": 601
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.6089330632218175,
      "learning_rate": 9.942225106530391e-06,
      "loss": 0.5752,
      "step": 602
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4988089190692178,
      "learning_rate": 9.941910439923367e-06,
      "loss": 0.2115,
      "step": 603
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5912633802760834,
      "learning_rate": 9.941594923744647e-06,
      "loss": 0.2509,
      "step": 604
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6836427811750905,
      "learning_rate": 9.941278558048471e-06,
      "loss": 0.2399,
      "step": 605
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5994727657613588,
      "learning_rate": 9.940961342889225e-06,
      "loss": 0.1927,
      "step": 606
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7959641054307327,
      "learning_rate": 9.940643278321447e-06,
      "loss": 0.2922,
      "step": 607
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.626752424588,
      "learning_rate": 9.94032436439981e-06,
      "loss": 0.6516,
      "step": 608
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8163894187197531,
      "learning_rate": 9.940004601179143e-06,
      "loss": 0.2683,
      "step": 609
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.771962983619006,
      "learning_rate": 9.939683988714415e-06,
      "loss": 0.2622,
      "step": 610
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6071389947376,
      "learning_rate": 9.939362527060743e-06,
      "loss": 0.2296,
      "step": 611
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9325642667952505,
      "learning_rate": 9.939040216273392e-06,
      "loss": 0.283,
      "step": 612
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.461260167201445,
      "learning_rate": 9.938717056407768e-06,
      "loss": 0.6665,
      "step": 613
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4994915381870415,
      "learning_rate": 9.938393047519428e-06,
      "loss": 0.1789,
      "step": 614
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4306336690501469,
      "learning_rate": 9.938068189664073e-06,
      "loss": 0.1802,
      "step": 615
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5773106311833553,
      "learning_rate": 9.937742482897549e-06,
      "loss": 0.257,
      "step": 616
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3633154357245916,
      "learning_rate": 9.937415927275848e-06,
      "loss": 0.1766,
      "step": 617
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.668801828724043,
      "learning_rate": 9.937088522855111e-06,
      "loss": 0.2297,
      "step": 618
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5731791312105399,
      "learning_rate": 9.936760269691621e-06,
      "loss": 0.2357,
      "step": 619
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.770634090955526,
      "learning_rate": 9.936431167841808e-06,
      "loss": 0.2545,
      "step": 620
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4685663854014859,
      "learning_rate": 9.93610121736225e-06,
      "loss": 0.2289,
      "step": 621
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9028555170717465,
      "learning_rate": 9.935770418309667e-06,
      "loss": 0.247,
      "step": 622
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4945081728774778,
      "learning_rate": 9.93543877074093e-06,
      "loss": 0.2366,
      "step": 623
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6527334849983435,
      "learning_rate": 9.935106274713053e-06,
      "loss": 0.2454,
      "step": 624
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6178345254941462,
      "learning_rate": 9.934772930283194e-06,
      "loss": 0.2763,
      "step": 625
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5794302424895938,
      "learning_rate": 9.934438737508658e-06,
      "loss": 0.2299,
      "step": 626
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5509501808528332,
      "learning_rate": 9.9341036964469e-06,
      "loss": 0.2318,
      "step": 627
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8488949671121544,
      "learning_rate": 9.933767807155515e-06,
      "loss": 0.2312,
      "step": 628
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5217717504200425,
      "learning_rate": 9.933431069692245e-06,
      "loss": 0.2011,
      "step": 629
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.272635941889189,
      "learning_rate": 9.933093484114983e-06,
      "loss": 0.6532,
      "step": 630
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9315496165120816,
      "learning_rate": 9.932755050481762e-06,
      "loss": 0.3025,
      "step": 631
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.776270097542846,
      "learning_rate": 9.93241576885076e-06,
      "loss": 0.2846,
      "step": 632
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8199974805418884,
      "learning_rate": 9.932075639280308e-06,
      "loss": 0.2248,
      "step": 633
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6980137008391476,
      "learning_rate": 9.931734661828876e-06,
      "loss": 0.2441,
      "step": 634
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7995118631508196,
      "learning_rate": 9.931392836555081e-06,
      "loss": 0.2364,
      "step": 635
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7770624537600288,
      "learning_rate": 9.931050163517688e-06,
      "loss": 0.2361,
      "step": 636
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.770456705968077,
      "learning_rate": 9.930706642775607e-06,
      "loss": 0.2714,
      "step": 637
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6028759722504868,
      "learning_rate": 9.930362274387892e-06,
      "loss": 0.2452,
      "step": 638
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6223187347309829,
      "learning_rate": 9.930017058413745e-06,
      "loss": 0.2103,
      "step": 639
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5707353917193756,
      "learning_rate": 9.92967099491251e-06,
      "loss": 0.1911,
      "step": 640
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4848186869851128,
      "learning_rate": 9.929324083943683e-06,
      "loss": 0.2673,
      "step": 641
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6732136629149015,
      "learning_rate": 9.9289763255669e-06,
      "loss": 0.2485,
      "step": 642
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7720416357093,
      "learning_rate": 9.928627719841945e-06,
      "loss": 0.2545,
      "step": 643
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.724326150638782,
      "learning_rate": 9.928278266828747e-06,
      "loss": 0.2749,
      "step": 644
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.512849891248241,
      "learning_rate": 9.927927966587383e-06,
      "loss": 0.1598,
      "step": 645
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.388980473879778,
      "learning_rate": 9.927576819178071e-06,
      "loss": 0.2025,
      "step": 646
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5967560271881123,
      "learning_rate": 9.927224824661179e-06,
      "loss": 0.2635,
      "step": 647
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4940822905880038,
      "learning_rate": 9.926871983097217e-06,
      "loss": 0.2051,
      "step": 648
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.611262518814622,
      "learning_rate": 9.926518294546846e-06,
      "loss": 0.2674,
      "step": 649
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6283137866023842,
      "learning_rate": 9.926163759070863e-06,
      "loss": 0.2901,
      "step": 650
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.632941992149547,
      "learning_rate": 9.925808376730225e-06,
      "loss": 0.5755,
      "step": 651
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6393607168806204,
      "learning_rate": 9.925452147586022e-06,
      "loss": 0.2187,
      "step": 652
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.892173844029097,
      "learning_rate": 9.925095071699491e-06,
      "loss": 0.7214,
      "step": 653
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.05656024960514,
      "learning_rate": 9.924737149132022e-06,
      "loss": 0.7127,
      "step": 654
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.726895273648809,
      "learning_rate": 9.924378379945145e-06,
      "loss": 0.2857,
      "step": 655
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6993785277621438,
      "learning_rate": 9.924018764200538e-06,
      "loss": 0.2639,
      "step": 656
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5566241721650984,
      "learning_rate": 9.923658301960018e-06,
      "loss": 0.2116,
      "step": 657
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.589318674272777,
      "learning_rate": 9.923296993285558e-06,
      "loss": 0.1826,
      "step": 658
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5225844379374347,
      "learning_rate": 9.922934838239269e-06,
      "loss": 0.2711,
      "step": 659
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5827182093644259,
      "learning_rate": 9.922571836883408e-06,
      "loss": 0.2009,
      "step": 660
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5622853583155631,
      "learning_rate": 9.922207989280383e-06,
      "loss": 0.2358,
      "step": 661
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.365434124491978,
      "learning_rate": 9.92184329549274e-06,
      "loss": 0.2405,
      "step": 662
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8548384288786246,
      "learning_rate": 9.921477755583178e-06,
      "loss": 0.2485,
      "step": 663
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4181161246286524,
      "learning_rate": 9.921111369614533e-06,
      "loss": 0.2192,
      "step": 664
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7037745660220036,
      "learning_rate": 9.920744137649793e-06,
      "loss": 0.2468,
      "step": 665
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6115904573591615,
      "learning_rate": 9.920376059752091e-06,
      "loss": 0.2397,
      "step": 666
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.996412881553388,
      "learning_rate": 9.920007135984701e-06,
      "loss": 0.7278,
      "step": 667
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7610468327648918,
      "learning_rate": 9.919637366411047e-06,
      "loss": 0.2673,
      "step": 668
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5715289549855223,
      "learning_rate": 9.919266751094698e-06,
      "loss": 0.2073,
      "step": 669
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.966829064224494,
      "learning_rate": 9.918895290099364e-06,
      "loss": 0.4394,
      "step": 670
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6293071925841174,
      "learning_rate": 9.918522983488905e-06,
      "loss": 0.2136,
      "step": 671
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.782364875125143,
      "learning_rate": 9.918149831327323e-06,
      "loss": 0.5842,
      "step": 672
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.515787449603775,
      "learning_rate": 9.917775833678772e-06,
      "loss": 0.2516,
      "step": 673
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.518606393761636,
      "learning_rate": 9.917400990607542e-06,
      "loss": 0.2251,
      "step": 674
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.937711538607205,
      "learning_rate": 9.917025302178074e-06,
      "loss": 0.6099,
      "step": 675
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.049201477440982,
      "learning_rate": 9.916648768454955e-06,
      "loss": 0.6968,
      "step": 676
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.954211451387822,
      "learning_rate": 9.916271389502911e-06,
      "loss": 0.5237,
      "step": 677
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4958225188004288,
      "learning_rate": 9.915893165386824e-06,
      "loss": 0.2084,
      "step": 678
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6980662054099436,
      "learning_rate": 9.915514096171711e-06,
      "loss": 0.2624,
      "step": 679
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.771116975919777,
      "learning_rate": 9.915134181922739e-06,
      "loss": 0.2492,
      "step": 680
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6076252074198587,
      "learning_rate": 9.91475342270522e-06,
      "loss": 0.2573,
      "step": 681
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.147612197345345,
      "learning_rate": 9.914371818584612e-06,
      "loss": 0.7556,
      "step": 682
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5420506117478068,
      "learning_rate": 9.913989369626515e-06,
      "loss": 0.2372,
      "step": 683
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.601405355934364,
      "learning_rate": 9.913606075896677e-06,
      "loss": 0.1974,
      "step": 684
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4667506544474116,
      "learning_rate": 9.913221937460991e-06,
      "loss": 0.2012,
      "step": 685
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6625558075111329,
      "learning_rate": 9.912836954385496e-06,
      "loss": 0.2302,
      "step": 686
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5610661724380053,
      "learning_rate": 9.912451126736374e-06,
      "loss": 0.2134,
      "step": 687
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.525454208264877,
      "learning_rate": 9.912064454579953e-06,
      "loss": 0.728,
      "step": 688
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5361051186482957,
      "learning_rate": 9.911676937982706e-06,
      "loss": 0.238,
      "step": 689
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.364933606919827,
      "learning_rate": 9.911288577011254e-06,
      "loss": 0.6148,
      "step": 690
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7937978714633869,
      "learning_rate": 9.910899371732358e-06,
      "loss": 0.2545,
      "step": 691
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.730565944752387,
      "learning_rate": 9.91050932221293e-06,
      "loss": 0.2576,
      "step": 692
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.820679322514803,
      "learning_rate": 9.910118428520019e-06,
      "loss": 0.2696,
      "step": 693
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.848617109269954,
      "learning_rate": 9.909726690720829e-06,
      "loss": 0.7507,
      "step": 694
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7434316876374785,
      "learning_rate": 9.9093341088827e-06,
      "loss": 0.2124,
      "step": 695
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4396206165949184,
      "learning_rate": 9.908940683073127e-06,
      "loss": 0.1876,
      "step": 696
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.436906146509697,
      "learning_rate": 9.90854641335974e-06,
      "loss": 0.626,
      "step": 697
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6931030966725613,
      "learning_rate": 9.908151299810319e-06,
      "loss": 0.2606,
      "step": 698
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6056658783213327,
      "learning_rate": 9.90775534249279e-06,
      "loss": 0.2744,
      "step": 699
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8472327619961735,
      "learning_rate": 9.907358541475223e-06,
      "loss": 0.2321,
      "step": 700
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5482826927525979,
      "learning_rate": 9.90696089682583e-06,
      "loss": 0.2928,
      "step": 701
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.328096092761234,
      "learning_rate": 9.906562408612973e-06,
      "loss": 0.6189,
      "step": 702
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.5651231529936314,
      "learning_rate": 9.906163076905158e-06,
      "loss": 0.4679,
      "step": 703
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.303383972689279,
      "learning_rate": 9.90576290177103e-06,
      "loss": 0.5973,
      "step": 704
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5203795948476473,
      "learning_rate": 9.905361883279388e-06,
      "loss": 0.2632,
      "step": 705
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7747943723975976,
      "learning_rate": 9.90496002149917e-06,
      "loss": 0.2348,
      "step": 706
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.3515255090488685,
      "learning_rate": 9.904557316499462e-06,
      "loss": 0.553,
      "step": 707
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.812910920448841,
      "learning_rate": 9.904153768349494e-06,
      "loss": 0.2552,
      "step": 708
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6157919372917329,
      "learning_rate": 9.903749377118637e-06,
      "loss": 0.2559,
      "step": 709
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.623182900911183,
      "learning_rate": 9.903344142876414e-06,
      "loss": 0.2753,
      "step": 710
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.628456406834472,
      "learning_rate": 9.902938065692488e-06,
      "loss": 0.2437,
      "step": 711
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7760513649134293,
      "learning_rate": 9.902531145636668e-06,
      "loss": 0.2933,
      "step": 712
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5236012165297268,
      "learning_rate": 9.902123382778911e-06,
      "loss": 0.2606,
      "step": 713
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.527396726668522,
      "learning_rate": 9.901714777189313e-06,
      "loss": 0.2502,
      "step": 714
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5385000553362822,
      "learning_rate": 9.90130532893812e-06,
      "loss": 0.2583,
      "step": 715
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.879618736710391,
      "learning_rate": 9.900895038095718e-06,
      "loss": 0.718,
      "step": 716
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6360547883417809,
      "learning_rate": 9.900483904732645e-06,
      "loss": 0.2217,
      "step": 717
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.792662580279206,
      "learning_rate": 9.900071928919577e-06,
      "loss": 0.3057,
      "step": 718
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4347323790246633,
      "learning_rate": 9.899659110727339e-06,
      "loss": 0.2044,
      "step": 719
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.692112495817601,
      "learning_rate": 9.899245450226897e-06,
      "loss": 0.3132,
      "step": 720
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.629785845923767,
      "learning_rate": 9.898830947489365e-06,
      "loss": 0.2582,
      "step": 721
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4469488236469454,
      "learning_rate": 9.898415602586e-06,
      "loss": 0.1844,
      "step": 722
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.134269080087288,
      "learning_rate": 9.89799941558821e-06,
      "loss": 0.2788,
      "step": 723
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7604763353707464,
      "learning_rate": 9.897582386567532e-06,
      "loss": 0.2463,
      "step": 724
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.474919020902522,
      "learning_rate": 9.897164515595667e-06,
      "loss": 0.6131,
      "step": 725
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8720240915441029,
      "learning_rate": 9.89674580274445e-06,
      "loss": 0.286,
      "step": 726
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5213774442569095,
      "learning_rate": 9.896326248085862e-06,
      "loss": 0.2348,
      "step": 727
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4788136875542297,
      "learning_rate": 9.895905851692026e-06,
      "loss": 0.1959,
      "step": 728
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5445240887812965,
      "learning_rate": 9.895484613635217e-06,
      "loss": 0.2079,
      "step": 729
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.702927655251176,
      "learning_rate": 9.89506253398785e-06,
      "loss": 0.5498,
      "step": 730
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8058708430664583,
      "learning_rate": 9.894639612822486e-06,
      "loss": 0.2534,
      "step": 731
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6952425681531327,
      "learning_rate": 9.894215850211825e-06,
      "loss": 0.2584,
      "step": 732
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.434980883976582,
      "learning_rate": 9.893791246228726e-06,
      "loss": 0.2324,
      "step": 733
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8947894346519283,
      "learning_rate": 9.893365800946176e-06,
      "loss": 0.2694,
      "step": 734
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8499179505929553,
      "learning_rate": 9.892939514437314e-06,
      "loss": 0.2092,
      "step": 735
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3804929950366884,
      "learning_rate": 9.892512386775429e-06,
      "loss": 0.2156,
      "step": 736
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.642537552907098,
      "learning_rate": 9.892084418033942e-06,
      "loss": 0.626,
      "step": 737
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6070729295284805,
      "learning_rate": 9.891655608286432e-06,
      "loss": 0.2203,
      "step": 738
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8739071772582057,
      "learning_rate": 9.891225957606613e-06,
      "loss": 0.2488,
      "step": 739
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4659669370463806,
      "learning_rate": 9.890795466068351e-06,
      "loss": 0.2089,
      "step": 740
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4239645149423243,
      "learning_rate": 9.890364133745646e-06,
      "loss": 0.1956,
      "step": 741
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.42710347247057,
      "learning_rate": 9.889931960712653e-06,
      "loss": 0.6545,
      "step": 742
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3413177000839556,
      "learning_rate": 9.88949894704367e-06,
      "loss": 0.1909,
      "step": 743
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.549361014253156,
      "learning_rate": 9.889065092813131e-06,
      "loss": 0.7725,
      "step": 744
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6854745019342228,
      "learning_rate": 9.888630398095623e-06,
      "loss": 0.2269,
      "step": 745
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6996701447208211,
      "learning_rate": 9.888194862965877e-06,
      "loss": 0.2498,
      "step": 746
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5337629583954788,
      "learning_rate": 9.887758487498765e-06,
      "loss": 0.2433,
      "step": 747
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5928547228969279,
      "learning_rate": 9.887321271769305e-06,
      "loss": 0.2347,
      "step": 748
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5451188259730213,
      "learning_rate": 9.88688321585266e-06,
      "loss": 0.2659,
      "step": 749
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6196276173409023,
      "learning_rate": 9.886444319824138e-06,
      "loss": 0.2824,
      "step": 750
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5327966488444615,
      "learning_rate": 9.886004583759187e-06,
      "loss": 0.235,
      "step": 751
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4648147445918396,
      "learning_rate": 9.885564007733406e-06,
      "loss": 0.2405,
      "step": 752
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6931532471396216,
      "learning_rate": 9.885122591822533e-06,
      "loss": 0.272,
      "step": 753
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.221441234294609,
      "learning_rate": 9.884680336102452e-06,
      "loss": 0.5262,
      "step": 754
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4815794496729398,
      "learning_rate": 9.884237240649195e-06,
      "loss": 0.2237,
      "step": 755
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9189108211619759,
      "learning_rate": 9.883793305538934e-06,
      "loss": 0.2387,
      "step": 756
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7202413700764891,
      "learning_rate": 9.883348530847985e-06,
      "loss": 0.311,
      "step": 757
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4971053991100145,
      "learning_rate": 9.88290291665281e-06,
      "loss": 0.232,
      "step": 758
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5310302402689158,
      "learning_rate": 9.882456463030017e-06,
      "loss": 0.236,
      "step": 759
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4497204417557312,
      "learning_rate": 9.882009170056354e-06,
      "loss": 0.1952,
      "step": 760
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7501835061751523,
      "learning_rate": 9.88156103780872e-06,
      "loss": 0.3036,
      "step": 761
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6453593934739126,
      "learning_rate": 9.881112066364151e-06,
      "loss": 0.2337,
      "step": 762
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7154689444986084,
      "learning_rate": 9.880662255799831e-06,
      "loss": 0.2358,
      "step": 763
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.401660227551079,
      "learning_rate": 9.880211606193089e-06,
      "loss": 0.5795,
      "step": 764
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.798012212532902,
      "learning_rate": 9.879760117621393e-06,
      "loss": 0.4085,
      "step": 765
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.645785653201969,
      "learning_rate": 9.879307790162364e-06,
      "loss": 0.2312,
      "step": 766
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.406247863247478,
      "learning_rate": 9.87885462389376e-06,
      "loss": 0.2269,
      "step": 767
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6610786259952939,
      "learning_rate": 9.878400618893483e-06,
      "loss": 0.3058,
      "step": 768
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3721485893544807,
      "learning_rate": 9.877945775239585e-06,
      "loss": 0.2082,
      "step": 769
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5723971168457274,
      "learning_rate": 9.87749009301026e-06,
      "loss": 0.2511,
      "step": 770
    },
    {
      "epoch": 0.1,
      "grad_norm": 12.539171296075876,
      "learning_rate": 9.877033572283842e-06,
      "loss": 0.7036,
      "step": 771
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.508447334613154,
      "learning_rate": 9.876576213138815e-06,
      "loss": 0.5491,
      "step": 772
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8231356985308775,
      "learning_rate": 9.876118015653804e-06,
      "loss": 0.2828,
      "step": 773
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5792772662872923,
      "learning_rate": 9.875658979907574e-06,
      "loss": 0.2425,
      "step": 774
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5617864332524358,
      "learning_rate": 9.875199105979043e-06,
      "loss": 0.2881,
      "step": 775
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.527413778018801,
      "learning_rate": 9.874738393947268e-06,
      "loss": 0.2276,
      "step": 776
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.609672306501683,
      "learning_rate": 9.87427684389145e-06,
      "loss": 0.2375,
      "step": 777
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8026276574539972,
      "learning_rate": 9.873814455890934e-06,
      "loss": 0.2772,
      "step": 778
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4293414119874757,
      "learning_rate": 9.873351230025212e-06,
      "loss": 0.2179,
      "step": 779
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.571981638691075,
      "learning_rate": 9.872887166373916e-06,
      "loss": 0.2079,
      "step": 780
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.226633449060357,
      "learning_rate": 9.872422265016827e-06,
      "loss": 0.2527,
      "step": 781
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.540208527236924,
      "learning_rate": 9.871956526033863e-06,
      "loss": 0.2473,
      "step": 782
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6278656921678318,
      "learning_rate": 9.871489949505091e-06,
      "loss": 0.2338,
      "step": 783
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6563407421736394,
      "learning_rate": 9.871022535510722e-06,
      "loss": 0.2454,
      "step": 784
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4226694142484215,
      "learning_rate": 9.870554284131111e-06,
      "loss": 0.2133,
      "step": 785
    },
    {
      "epoch": 0.1,
      "grad_norm": 10.641597426680496,
      "learning_rate": 9.870085195446752e-06,
      "loss": 0.5931,
      "step": 786
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4872202320100316,
      "learning_rate": 9.86961526953829e-06,
      "loss": 0.1727,
      "step": 787
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4578569786537643,
      "learning_rate": 9.86914450648651e-06,
      "loss": 0.188,
      "step": 788
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7065886527303935,
      "learning_rate": 9.868672906372341e-06,
      "loss": 0.2609,
      "step": 789
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.862518668776345,
      "learning_rate": 9.868200469276858e-06,
      "loss": 0.6653,
      "step": 790
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.556525446811141,
      "learning_rate": 9.867727195281275e-06,
      "loss": 0.2413,
      "step": 791
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7784474732030073,
      "learning_rate": 9.867253084466957e-06,
      "loss": 0.2857,
      "step": 792
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.60880351027115,
      "learning_rate": 9.866778136915408e-06,
      "loss": 0.2277,
      "step": 793
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5029885206802511,
      "learning_rate": 9.866302352708276e-06,
      "loss": 0.2264,
      "step": 794
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.596949900845371,
      "learning_rate": 9.865825731927356e-06,
      "loss": 0.2312,
      "step": 795
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.197910689205509,
      "learning_rate": 9.865348274654582e-06,
      "loss": 0.6782,
      "step": 796
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.098681312471882,
      "learning_rate": 9.864869980972036e-06,
      "loss": 0.5974,
      "step": 797
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5724048958001202,
      "learning_rate": 9.86439085096194e-06,
      "loss": 0.2351,
      "step": 798
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5302676514302356,
      "learning_rate": 9.863910884706667e-06,
      "loss": 0.2433,
      "step": 799
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5092758930965815,
      "learning_rate": 9.863430082288724e-06,
      "loss": 0.2227,
      "step": 800
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5028282577642684,
      "learning_rate": 9.862948443790768e-06,
      "loss": 0.1833,
      "step": 801
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5257653691440494,
      "learning_rate": 9.862465969295599e-06,
      "loss": 0.2643,
      "step": 802
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6189092720152625,
      "learning_rate": 9.861982658886158e-06,
      "loss": 0.2357,
      "step": 803
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7149404542675966,
      "learning_rate": 9.861498512645532e-06,
      "loss": 0.2613,
      "step": 804
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.547005740314361,
      "learning_rate": 9.861013530656956e-06,
      "loss": 0.2199,
      "step": 805
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5186843482138612,
      "learning_rate": 9.860527713003797e-06,
      "loss": 0.2557,
      "step": 806
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.515444882726058,
      "learning_rate": 9.86004105976958e-06,
      "loss": 0.2219,
      "step": 807
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6701676610887666,
      "learning_rate": 9.85955357103796e-06,
      "loss": 0.2916,
      "step": 808
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.929342380737433,
      "learning_rate": 9.859065246892744e-06,
      "loss": 0.6185,
      "step": 809
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6437679246340982,
      "learning_rate": 9.858576087417881e-06,
      "loss": 0.2255,
      "step": 810
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6142972469772776,
      "learning_rate": 9.858086092697464e-06,
      "loss": 0.2607,
      "step": 811
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5890341513730164,
      "learning_rate": 9.857595262815726e-06,
      "loss": 0.239,
      "step": 812
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.345967563614587,
      "learning_rate": 9.857103597857051e-06,
      "loss": 0.1771,
      "step": 813
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5163290184038432,
      "learning_rate": 9.856611097905956e-06,
      "loss": 0.1983,
      "step": 814
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7218822354086893,
      "learning_rate": 9.856117763047115e-06,
      "loss": 0.27,
      "step": 815
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6191387211686987,
      "learning_rate": 9.855623593365332e-06,
      "loss": 0.2105,
      "step": 816
    },
    {
      "epoch": 0.1,
      "grad_norm": 10.23924484412633,
      "learning_rate": 9.855128588945559e-06,
      "loss": 0.588,
      "step": 817
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3958907694554972,
      "learning_rate": 9.8546327498729e-06,
      "loss": 0.1722,
      "step": 818
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3873018578015397,
      "learning_rate": 9.854136076232587e-06,
      "loss": 0.1911,
      "step": 819
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6867436674418068,
      "learning_rate": 9.853638568110013e-06,
      "loss": 0.2048,
      "step": 820
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.491407977365309,
      "learning_rate": 9.853140225590698e-06,
      "loss": 0.5027,
      "step": 821
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.991426492441957,
      "learning_rate": 9.852641048760319e-06,
      "loss": 0.5912,
      "step": 822
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5257051860766309,
      "learning_rate": 9.852141037704683e-06,
      "loss": 0.2476,
      "step": 823
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5048168206195822,
      "learning_rate": 9.851640192509755e-06,
      "loss": 0.2016,
      "step": 824
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5661417585388566,
      "learning_rate": 9.85113851326163e-06,
      "loss": 0.2709,
      "step": 825
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.569487560698578,
      "learning_rate": 9.850636000046558e-06,
      "loss": 0.7548,
      "step": 826
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.614840639666413,
      "learning_rate": 9.850132652950923e-06,
      "loss": 0.2306,
      "step": 827
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.633172037146853,
      "learning_rate": 9.84962847206126e-06,
      "loss": 0.7061,
      "step": 828
    },
    {
      "epoch": 0.11,
      "grad_norm": 10.63856099748518,
      "learning_rate": 9.849123457464238e-06,
      "loss": 0.664,
      "step": 829
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.540873030985642,
      "learning_rate": 9.848617609246678e-06,
      "loss": 0.2276,
      "step": 830
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7020755321676508,
      "learning_rate": 9.84811092749554e-06,
      "loss": 0.2627,
      "step": 831
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.514100415433326,
      "learning_rate": 9.84760341229793e-06,
      "loss": 0.3174,
      "step": 832
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5144664172392628,
      "learning_rate": 9.847095063741097e-06,
      "loss": 0.205,
      "step": 833
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5533795719153936,
      "learning_rate": 9.84658588191243e-06,
      "loss": 0.2658,
      "step": 834
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.834841034193005,
      "learning_rate": 9.846075866899463e-06,
      "loss": 0.5467,
      "step": 835
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4955787943054684,
      "learning_rate": 9.845565018789873e-06,
      "loss": 0.2798,
      "step": 836
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5335781788185314,
      "learning_rate": 9.845053337671482e-06,
      "loss": 0.2269,
      "step": 837
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3115491564184094,
      "learning_rate": 9.844540823632254e-06,
      "loss": 0.2263,
      "step": 838
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.473471816112244,
      "learning_rate": 9.844027476760295e-06,
      "loss": 0.2146,
      "step": 839
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.733223538362547,
      "learning_rate": 9.843513297143856e-06,
      "loss": 0.6177,
      "step": 840
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4382935602464677,
      "learning_rate": 9.842998284871332e-06,
      "loss": 0.2197,
      "step": 841
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.732434643942212,
      "learning_rate": 9.842482440031256e-06,
      "loss": 0.6081,
      "step": 842
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5581008960075726,
      "learning_rate": 9.84196576271231e-06,
      "loss": 0.2372,
      "step": 843
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4842172702438654,
      "learning_rate": 9.841448253003319e-06,
      "loss": 0.2356,
      "step": 844
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.624683035754351,
      "learning_rate": 9.840929910993244e-06,
      "loss": 0.2489,
      "step": 845
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6250712812353134,
      "learning_rate": 9.840410736771196e-06,
      "loss": 0.2383,
      "step": 846
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.652654496608276,
      "learning_rate": 9.839890730426429e-06,
      "loss": 0.5482,
      "step": 847
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5748704857290212,
      "learning_rate": 9.839369892048337e-06,
      "loss": 0.247,
      "step": 848
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6645240232459553,
      "learning_rate": 9.838848221726455e-06,
      "loss": 0.2713,
      "step": 849
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4514269084353275,
      "learning_rate": 9.838325719550469e-06,
      "loss": 0.2163,
      "step": 850
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.466278683458454,
      "learning_rate": 9.8378023856102e-06,
      "loss": 0.2447,
      "step": 851
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7731206465962948,
      "learning_rate": 9.837278219995618e-06,
      "loss": 0.2396,
      "step": 852
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6251051424751664,
      "learning_rate": 9.836753222796831e-06,
      "loss": 0.2353,
      "step": 853
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3329932221796879,
      "learning_rate": 9.836227394104093e-06,
      "loss": 0.232,
      "step": 854
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6058070593893512,
      "learning_rate": 9.835700734007798e-06,
      "loss": 0.2671,
      "step": 855
    },
    {
      "epoch": 0.11,
      "grad_norm": 12.131637120905268,
      "learning_rate": 9.835173242598486e-06,
      "loss": 0.6236,
      "step": 856
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.370426244395862,
      "learning_rate": 9.834644919966842e-06,
      "loss": 0.1989,
      "step": 857
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6216518856222737,
      "learning_rate": 9.834115766203687e-06,
      "loss": 0.2582,
      "step": 858
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.97045749072954,
      "learning_rate": 9.833585781399989e-06,
      "loss": 0.7116,
      "step": 859
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.8952112842415945,
      "learning_rate": 9.83305496564686e-06,
      "loss": 0.7081,
      "step": 860
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5328388878054515,
      "learning_rate": 9.832523319035553e-06,
      "loss": 0.1995,
      "step": 861
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6075852985270176,
      "learning_rate": 9.831990841657465e-06,
      "loss": 0.2547,
      "step": 862
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.477895967586995,
      "learning_rate": 9.831457533604133e-06,
      "loss": 0.2601,
      "step": 863
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7860334372704338,
      "learning_rate": 9.83092339496724e-06,
      "loss": 0.222,
      "step": 864
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.644512923211377,
      "learning_rate": 9.83038842583861e-06,
      "loss": 0.6525,
      "step": 865
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5951611458768022,
      "learning_rate": 9.829852626310213e-06,
      "loss": 0.2297,
      "step": 866
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.473019782171749,
      "learning_rate": 9.829315996474156e-06,
      "loss": 0.2119,
      "step": 867
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.512593707683497,
      "learning_rate": 9.828778536422692e-06,
      "loss": 0.7494,
      "step": 868
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3170579563315143,
      "learning_rate": 9.828240246248217e-06,
      "loss": 0.2091,
      "step": 869
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4477755833965338,
      "learning_rate": 9.827701126043272e-06,
      "loss": 0.2126,
      "step": 870
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5666282920883194,
      "learning_rate": 9.827161175900535e-06,
      "loss": 0.2297,
      "step": 871
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6256887008096184,
      "learning_rate": 9.826620395912832e-06,
      "loss": 0.2399,
      "step": 872
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.613520879715766,
      "learning_rate": 9.826078786173126e-06,
      "loss": 0.2689,
      "step": 873
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.050525776106553,
      "learning_rate": 9.82553634677453e-06,
      "loss": 0.8386,
      "step": 874
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3692176168772354,
      "learning_rate": 9.82499307781029e-06,
      "loss": 0.1988,
      "step": 875
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5899142037592775,
      "learning_rate": 9.824448979373807e-06,
      "loss": 0.2688,
      "step": 876
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.615235948507568,
      "learning_rate": 9.823904051558613e-06,
      "loss": 0.2594,
      "step": 877
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.399123052080288,
      "learning_rate": 9.82335829445839e-06,
      "loss": 0.7852,
      "step": 878
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.442582914333988,
      "learning_rate": 9.822811708166958e-06,
      "loss": 0.2279,
      "step": 879
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.368643532355621,
      "learning_rate": 9.822264292778282e-06,
      "loss": 0.5232,
      "step": 880
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5336313897395017,
      "learning_rate": 9.821716048386472e-06,
      "loss": 0.2845,
      "step": 881
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.38456342982958,
      "learning_rate": 9.821166975085774e-06,
      "loss": 0.82,
      "step": 882
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4073549416307898,
      "learning_rate": 9.820617072970583e-06,
      "loss": 0.2094,
      "step": 883
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5619951566827421,
      "learning_rate": 9.820066342135428e-06,
      "loss": 0.2146,
      "step": 884
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8443171105016656,
      "learning_rate": 9.819514782674992e-06,
      "loss": 0.2671,
      "step": 885
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.546573194659899,
      "learning_rate": 9.818962394684094e-06,
      "loss": 0.1981,
      "step": 886
    },
    {
      "epoch": 0.11,
      "grad_norm": 16.32609694068637,
      "learning_rate": 9.818409178257691e-06,
      "loss": 0.5559,
      "step": 887
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4550044424752138,
      "learning_rate": 9.817855133490891e-06,
      "loss": 0.2062,
      "step": 888
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6077368240374879,
      "learning_rate": 9.817300260478942e-06,
      "loss": 0.2503,
      "step": 889
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5770234179508706,
      "learning_rate": 9.816744559317228e-06,
      "loss": 0.2858,
      "step": 890
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8210863969592725,
      "learning_rate": 9.816188030101287e-06,
      "loss": 0.2376,
      "step": 891
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6841750442471504,
      "learning_rate": 9.815630672926789e-06,
      "loss": 0.2751,
      "step": 892
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.484380997266658,
      "learning_rate": 9.81507248788955e-06,
      "loss": 0.2361,
      "step": 893
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5961351120675613,
      "learning_rate": 9.814513475085528e-06,
      "loss": 0.2203,
      "step": 894
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.40152141129811,
      "learning_rate": 9.813953634610827e-06,
      "loss": 0.256,
      "step": 895
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.620889155721481,
      "learning_rate": 9.813392966561688e-06,
      "loss": 0.222,
      "step": 896
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5214376745863236,
      "learning_rate": 9.812831471034495e-06,
      "loss": 0.2139,
      "step": 897
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6418912957321412,
      "learning_rate": 9.812269148125779e-06,
      "loss": 0.2574,
      "step": 898
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5515413704213419,
      "learning_rate": 9.811705997932206e-06,
      "loss": 0.1904,
      "step": 899
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3787823006495692,
      "learning_rate": 9.81114202055059e-06,
      "loss": 0.193,
      "step": 900
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.588430251682197,
      "learning_rate": 9.810577216077886e-06,
      "loss": 0.6714,
      "step": 901
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5216188314269417,
      "learning_rate": 9.810011584611189e-06,
      "loss": 0.2513,
      "step": 902
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6029519040208455,
      "learning_rate": 9.80944512624774e-06,
      "loss": 0.2545,
      "step": 903
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6722085082553133,
      "learning_rate": 9.808877841084915e-06,
      "loss": 0.2735,
      "step": 904
    },
    {
      "epoch": 0.12,
      "grad_norm": 10.741069409721307,
      "learning_rate": 9.808309729220241e-06,
      "loss": 0.5965,
      "step": 905
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.586401912988356,
      "learning_rate": 9.807740790751383e-06,
      "loss": 0.2177,
      "step": 906
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4483957245724999,
      "learning_rate": 9.807171025776145e-06,
      "loss": 0.2163,
      "step": 907
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5350464072086802,
      "learning_rate": 9.806600434392479e-06,
      "loss": 0.252,
      "step": 908
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7231564132891748,
      "learning_rate": 9.806029016698475e-06,
      "loss": 0.2755,
      "step": 909
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.378947168762495,
      "learning_rate": 9.805456772792367e-06,
      "loss": 0.1847,
      "step": 910
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.587590241998154,
      "learning_rate": 9.804883702772532e-06,
      "loss": 0.7138,
      "step": 911
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5625487474462048,
      "learning_rate": 9.804309806737482e-06,
      "loss": 0.2324,
      "step": 912
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.2501884594296415,
      "learning_rate": 9.803735084785884e-06,
      "loss": 0.4794,
      "step": 913
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6223689832826338,
      "learning_rate": 9.803159537016533e-06,
      "loss": 0.2224,
      "step": 914
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.3897595927554525,
      "learning_rate": 9.802583163528374e-06,
      "loss": 0.5329,
      "step": 915
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6033856432659646,
      "learning_rate": 9.802005964420493e-06,
      "loss": 0.2149,
      "step": 916
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.497407289935427,
      "learning_rate": 9.801427939792118e-06,
      "loss": 0.2523,
      "step": 917
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.688349081139293,
      "learning_rate": 9.800849089742617e-06,
      "loss": 0.5983,
      "step": 918
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.459235343356244,
      "learning_rate": 9.800269414371501e-06,
      "loss": 0.5158,
      "step": 919
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7382308011054384,
      "learning_rate": 9.799688913778423e-06,
      "loss": 0.266,
      "step": 920
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5153664023586395,
      "learning_rate": 9.799107588063178e-06,
      "loss": 0.2233,
      "step": 921
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.227499149220602,
      "learning_rate": 9.798525437325704e-06,
      "loss": 0.569,
      "step": 922
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5283174907241457,
      "learning_rate": 9.797942461666075e-06,
      "loss": 0.2163,
      "step": 923
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3728381155753568,
      "learning_rate": 9.797358661184517e-06,
      "loss": 0.2214,
      "step": 924
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4375057518454892,
      "learning_rate": 9.796774035981388e-06,
      "loss": 0.24,
      "step": 925
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6926151892894978,
      "learning_rate": 9.796188586157194e-06,
      "loss": 0.2674,
      "step": 926
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3958726916606918,
      "learning_rate": 9.79560231181258e-06,
      "loss": 0.2041,
      "step": 927
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.661356543847968,
      "learning_rate": 9.795015213048334e-06,
      "loss": 0.6693,
      "step": 928
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.308348003902374,
      "learning_rate": 9.794427289965383e-06,
      "loss": 0.5253,
      "step": 929
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5474522429458621,
      "learning_rate": 9.793838542664797e-06,
      "loss": 0.2968,
      "step": 930
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5754044112891212,
      "learning_rate": 9.793248971247792e-06,
      "loss": 0.2569,
      "step": 931
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.905017848683002,
      "learning_rate": 9.792658575815724e-06,
      "loss": 0.4829,
      "step": 932
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.506813795417511,
      "learning_rate": 9.792067356470083e-06,
      "loss": 0.195,
      "step": 933
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5796287657266856,
      "learning_rate": 9.791475313312508e-06,
      "loss": 0.2213,
      "step": 934
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.587828692840281,
      "learning_rate": 9.79088244644478e-06,
      "loss": 0.269,
      "step": 935
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5178434118176112,
      "learning_rate": 9.79028875596882e-06,
      "loss": 0.2834,
      "step": 936
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5780420593071292,
      "learning_rate": 9.789694241986687e-06,
      "loss": 0.2622,
      "step": 937
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.444783475897761,
      "learning_rate": 9.789098904600588e-06,
      "loss": 0.5285,
      "step": 938
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.991538771953783,
      "learning_rate": 9.788502743912866e-06,
      "loss": 0.5889,
      "step": 939
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.870264818877543,
      "learning_rate": 9.787905760026011e-06,
      "loss": 0.2398,
      "step": 940
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.544506223165339,
      "learning_rate": 9.78730795304265e-06,
      "loss": 0.2525,
      "step": 941
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6355148973027607,
      "learning_rate": 9.786709323065553e-06,
      "loss": 0.2093,
      "step": 942
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5804607095634884,
      "learning_rate": 9.786109870197629e-06,
      "loss": 0.28,
      "step": 943
    },
    {
      "epoch": 0.12,
      "grad_norm": 11.541287378192216,
      "learning_rate": 9.785509594541934e-06,
      "loss": 0.5836,
      "step": 944
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.56057592926112,
      "learning_rate": 9.78490849620166e-06,
      "loss": 0.2123,
      "step": 945
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.699434790402473,
      "learning_rate": 9.784306575280146e-06,
      "loss": 0.2426,
      "step": 946
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5693644922801973,
      "learning_rate": 9.783703831880867e-06,
      "loss": 0.1772,
      "step": 947
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4913217641840426,
      "learning_rate": 9.78310026610744e-06,
      "loss": 0.2769,
      "step": 948
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.210902382419579,
      "learning_rate": 9.78249587806363e-06,
      "loss": 0.5698,
      "step": 949
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4630297458088228,
      "learning_rate": 9.78189066785333e-06,
      "loss": 0.2218,
      "step": 950
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.063199285439943,
      "learning_rate": 9.781284635580593e-06,
      "loss": 0.7398,
      "step": 951
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5785778989075296,
      "learning_rate": 9.780677781349594e-06,
      "loss": 0.2403,
      "step": 952
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3960205174428302,
      "learning_rate": 9.780070105264665e-06,
      "loss": 0.1973,
      "step": 953
    },
    {
      "epoch": 0.12,
      "grad_norm": 10.992576239404627,
      "learning_rate": 9.779461607430268e-06,
      "loss": 0.5433,
      "step": 954
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.972627257663684,
      "learning_rate": 9.778852287951013e-06,
      "loss": 0.2817,
      "step": 955
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5801341628143994,
      "learning_rate": 9.77824214693165e-06,
      "loss": 0.2074,
      "step": 956
    },
    {
      "epoch": 0.12,
      "grad_norm": 12.312895739822144,
      "learning_rate": 9.777631184477065e-06,
      "loss": 0.678,
      "step": 957
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.32538172279063,
      "learning_rate": 9.777019400692292e-06,
      "loss": 0.7491,
      "step": 958
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5137711826404705,
      "learning_rate": 9.776406795682506e-06,
      "loss": 0.206,
      "step": 959
    },
    {
      "epoch": 0.12,
      "grad_norm": 11.19624004747375,
      "learning_rate": 9.775793369553019e-06,
      "loss": 0.4823,
      "step": 960
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7244829917729718,
      "learning_rate": 9.775179122409283e-06,
      "loss": 0.2477,
      "step": 961
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8027856696184619,
      "learning_rate": 9.7745640543569e-06,
      "loss": 0.2281,
      "step": 962
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5975044929160354,
      "learning_rate": 9.773948165501606e-06,
      "loss": 0.2407,
      "step": 963
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6951462635832668,
      "learning_rate": 9.773331455949275e-06,
      "loss": 0.2551,
      "step": 964
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6170413516158464,
      "learning_rate": 9.772713925805932e-06,
      "loss": 0.2778,
      "step": 965
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4656836380981473,
      "learning_rate": 9.772095575177733e-06,
      "loss": 0.2142,
      "step": 966
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.388116385048605,
      "learning_rate": 9.771476404170984e-06,
      "loss": 0.2167,
      "step": 967
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5277982255851064,
      "learning_rate": 9.770856412892122e-06,
      "loss": 0.2433,
      "step": 968
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6235879656298848,
      "learning_rate": 9.770235601447737e-06,
      "loss": 0.2583,
      "step": 969
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4984742127390174,
      "learning_rate": 9.769613969944549e-06,
      "loss": 0.229,
      "step": 970
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.588786948512916,
      "learning_rate": 9.768991518489427e-06,
      "loss": 0.2144,
      "step": 971
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6747389800471646,
      "learning_rate": 9.768368247189375e-06,
      "loss": 0.2303,
      "step": 972
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.457132619618757,
      "learning_rate": 9.767744156151542e-06,
      "loss": 0.2225,
      "step": 973
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.677416146797022,
      "learning_rate": 9.767119245483217e-06,
      "loss": 0.2652,
      "step": 974
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5137068956418132,
      "learning_rate": 9.766493515291827e-06,
      "loss": 0.228,
      "step": 975
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3640418409332795,
      "learning_rate": 9.765866965684944e-06,
      "loss": 0.2217,
      "step": 976
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7215447497317746,
      "learning_rate": 9.76523959677028e-06,
      "loss": 0.2481,
      "step": 977
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4401517967588748,
      "learning_rate": 9.764611408655687e-06,
      "loss": 0.2251,
      "step": 978
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.964274740542932,
      "learning_rate": 9.763982401449154e-06,
      "loss": 0.6551,
      "step": 979
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5312738296222064,
      "learning_rate": 9.763352575258817e-06,
      "loss": 0.2527,
      "step": 980
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5877974918152427,
      "learning_rate": 9.762721930192953e-06,
      "loss": 0.2517,
      "step": 981
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6853309349941876,
      "learning_rate": 9.762090466359975e-06,
      "loss": 0.2146,
      "step": 982
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6590179671825103,
      "learning_rate": 9.761458183868437e-06,
      "loss": 0.2514,
      "step": 983
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5421487038653998,
      "learning_rate": 9.760825082827039e-06,
      "loss": 0.2054,
      "step": 984
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.72808589744007,
      "learning_rate": 9.760191163344617e-06,
      "loss": 0.4219,
      "step": 985
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.94682077807742,
      "learning_rate": 9.75955642553015e-06,
      "loss": 0.619,
      "step": 986
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5832729909838419,
      "learning_rate": 9.758920869492755e-06,
      "loss": 0.2224,
      "step": 987
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3986890960587786,
      "learning_rate": 9.758284495341692e-06,
      "loss": 0.2247,
      "step": 988
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6377170875860998,
      "learning_rate": 9.757647303186361e-06,
      "loss": 0.2109,
      "step": 989
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8182619068185402,
      "learning_rate": 9.757009293136303e-06,
      "loss": 0.2746,
      "step": 990
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.641529178087155,
      "learning_rate": 9.756370465301202e-06,
      "loss": 0.22,
      "step": 991
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7153668786627965,
      "learning_rate": 9.755730819790876e-06,
      "loss": 0.2587,
      "step": 992
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.575840683035456,
      "learning_rate": 9.755090356715288e-06,
      "loss": 0.5301,
      "step": 993
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8170177453705956,
      "learning_rate": 9.754449076184544e-06,
      "loss": 0.2768,
      "step": 994
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5842907073502808,
      "learning_rate": 9.753806978308884e-06,
      "loss": 0.2467,
      "step": 995
    },
    {
      "epoch": 0.13,
      "grad_norm": 10.170094598033458,
      "learning_rate": 9.753164063198693e-06,
      "loss": 0.7024,
      "step": 996
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4712239389757151,
      "learning_rate": 9.752520330964497e-06,
      "loss": 0.183,
      "step": 997
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5636755566947191,
      "learning_rate": 9.75187578171696e-06,
      "loss": 0.2092,
      "step": 998
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5642423539592059,
      "learning_rate": 9.751230415566888e-06,
      "loss": 0.2739,
      "step": 999
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4696225277076729,
      "learning_rate": 9.750584232625226e-06,
      "loss": 0.2389,
      "step": 1000
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8014241236322668,
      "learning_rate": 9.749937233003062e-06,
      "loss": 0.2433,
      "step": 1001
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4782218049223648,
      "learning_rate": 9.749289416811622e-06,
      "loss": 0.2357,
      "step": 1002
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4462716258330974,
      "learning_rate": 9.74864078416227e-06,
      "loss": 0.2558,
      "step": 1003
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8409284751389985,
      "learning_rate": 9.747991335166518e-06,
      "loss": 0.2373,
      "step": 1004
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.218915353349342,
      "learning_rate": 9.747341069936013e-06,
      "loss": 0.6344,
      "step": 1005
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6880785628371244,
      "learning_rate": 9.74668998858254e-06,
      "loss": 0.2481,
      "step": 1006
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.551657455340695,
      "learning_rate": 9.746038091218032e-06,
      "loss": 0.2474,
      "step": 1007
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.749513727560512,
      "learning_rate": 9.745385377954555e-06,
      "loss": 0.6583,
      "step": 1008
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5349536071179815,
      "learning_rate": 9.744731848904316e-06,
      "loss": 0.2659,
      "step": 1009
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3918140896670135,
      "learning_rate": 9.74407750417967e-06,
      "loss": 0.2162,
      "step": 1010
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6160775253792208,
      "learning_rate": 9.743422343893099e-06,
      "loss": 0.2457,
      "step": 1011
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4822090306574778,
      "learning_rate": 9.742766368157239e-06,
      "loss": 0.1956,
      "step": 1012
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.704274091124098,
      "learning_rate": 9.742109577084857e-06,
      "loss": 0.6066,
      "step": 1013
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.290709211777096,
      "learning_rate": 9.741451970788863e-06,
      "loss": 0.2126,
      "step": 1014
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6938540969843376,
      "learning_rate": 9.740793549382308e-06,
      "loss": 0.231,
      "step": 1015
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5070263574592861,
      "learning_rate": 9.740134312978382e-06,
      "loss": 0.2436,
      "step": 1016
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6599644144774435,
      "learning_rate": 9.739474261690415e-06,
      "loss": 0.2484,
      "step": 1017
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4099311941669508,
      "learning_rate": 9.73881339563188e-06,
      "loss": 0.2075,
      "step": 1018
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4728679017743378,
      "learning_rate": 9.738151714916382e-06,
      "loss": 0.2459,
      "step": 1019
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.577987673079364,
      "learning_rate": 9.737489219657676e-06,
      "loss": 0.2076,
      "step": 1020
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.6890606672822255,
      "learning_rate": 9.73682590996965e-06,
      "loss": 0.6687,
      "step": 1021
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.28634010243354,
      "learning_rate": 9.736161785966339e-06,
      "loss": 0.5858,
      "step": 1022
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5312262136754955,
      "learning_rate": 9.735496847761908e-06,
      "loss": 0.2631,
      "step": 1023
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4731868705260995,
      "learning_rate": 9.734831095470671e-06,
      "loss": 0.218,
      "step": 1024
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4752602129214527,
      "learning_rate": 9.73416452920708e-06,
      "loss": 0.2266,
      "step": 1025
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.950153546040504,
      "learning_rate": 9.73349714908572e-06,
      "loss": 0.6819,
      "step": 1026
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.7167418130762835,
      "learning_rate": 9.732828955221328e-06,
      "loss": 0.7574,
      "step": 1027
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4843843068805904,
      "learning_rate": 9.732159947728767e-06,
      "loss": 0.2319,
      "step": 1028
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4830632712128824,
      "learning_rate": 9.731490126723053e-06,
      "loss": 0.2247,
      "step": 1029
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5878581179497495,
      "learning_rate": 9.730819492319336e-06,
      "loss": 0.2459,
      "step": 1030
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.425536626367204,
      "learning_rate": 9.730148044632902e-06,
      "loss": 0.7779,
      "step": 1031
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7073594376964238,
      "learning_rate": 9.729475783779182e-06,
      "loss": 0.2144,
      "step": 1032
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3893125456362143,
      "learning_rate": 9.728802709873747e-06,
      "loss": 0.1963,
      "step": 1033
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.504484407562703,
      "learning_rate": 9.728128823032305e-06,
      "loss": 0.2166,
      "step": 1034
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5859351101865589,
      "learning_rate": 9.727454123370705e-06,
      "loss": 0.2332,
      "step": 1035
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5462987273080802,
      "learning_rate": 9.726778611004936e-06,
      "loss": 0.2804,
      "step": 1036
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.330321865104465,
      "learning_rate": 9.726102286051126e-06,
      "loss": 0.6288,
      "step": 1037
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.706508992224581,
      "learning_rate": 9.725425148625544e-06,
      "loss": 0.236,
      "step": 1038
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5207471257402296,
      "learning_rate": 9.724747198844597e-06,
      "loss": 0.2086,
      "step": 1039
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5422230496016192,
      "learning_rate": 9.724068436824831e-06,
      "loss": 0.202,
      "step": 1040
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5657074007126985,
      "learning_rate": 9.723388862682936e-06,
      "loss": 0.2284,
      "step": 1041
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5682896541717324,
      "learning_rate": 9.722708476535736e-06,
      "loss": 0.2344,
      "step": 1042
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6417614973472312,
      "learning_rate": 9.722027278500201e-06,
      "loss": 0.2208,
      "step": 1043
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6030434500369601,
      "learning_rate": 9.721345268693433e-06,
      "loss": 0.2849,
      "step": 1044
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5635142865405036,
      "learning_rate": 9.720662447232679e-06,
      "loss": 0.2409,
      "step": 1045
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3782940852068748,
      "learning_rate": 9.719978814235323e-06,
      "loss": 0.1944,
      "step": 1046
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5893675745954297,
      "learning_rate": 9.71929436981889e-06,
      "loss": 0.2616,
      "step": 1047
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6061793180505095,
      "learning_rate": 9.718609114101045e-06,
      "loss": 0.2421,
      "step": 1048
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7252962324134506,
      "learning_rate": 9.717923047199591e-06,
      "loss": 0.2763,
      "step": 1049
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7299192969276562,
      "learning_rate": 9.71723616923247e-06,
      "loss": 0.2408,
      "step": 1050
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.443092152963464,
      "learning_rate": 9.716548480317763e-06,
      "loss": 0.2367,
      "step": 1051
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.582305215859814,
      "learning_rate": 9.715859980573694e-06,
      "loss": 0.1713,
      "step": 1052
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.074418370023519,
      "learning_rate": 9.715170670118625e-06,
      "loss": 0.6592,
      "step": 1053
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4616187661960147,
      "learning_rate": 9.714480549071053e-06,
      "loss": 0.2367,
      "step": 1054
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.29295923614098,
      "learning_rate": 9.713789617549621e-06,
      "loss": 0.175,
      "step": 1055
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5851096891256211,
      "learning_rate": 9.713097875673105e-06,
      "loss": 0.2334,
      "step": 1056
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4294625139588029,
      "learning_rate": 9.712405323560427e-06,
      "loss": 0.2292,
      "step": 1057
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.9814079531068876,
      "learning_rate": 9.711711961330644e-06,
      "loss": 0.6674,
      "step": 1058
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.649945110776656,
      "learning_rate": 9.711017789102948e-06,
      "loss": 0.2473,
      "step": 1059
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7844491630651613,
      "learning_rate": 9.710322806996682e-06,
      "loss": 0.3318,
      "step": 1060
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5865412004691313,
      "learning_rate": 9.70962701513132e-06,
      "loss": 0.247,
      "step": 1061
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9010909304753274,
      "learning_rate": 9.708930413626473e-06,
      "loss": 0.2765,
      "step": 1062
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6140557567730915,
      "learning_rate": 9.708233002601897e-06,
      "loss": 0.2034,
      "step": 1063
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6777782553793659,
      "learning_rate": 9.707534782177487e-06,
      "loss": 0.2407,
      "step": 1064
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4447463936203575,
      "learning_rate": 9.706835752473273e-06,
      "loss": 0.1942,
      "step": 1065
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5904855076536064,
      "learning_rate": 9.706135913609426e-06,
      "loss": 0.2482,
      "step": 1066
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5565779344076525,
      "learning_rate": 9.70543526570626e-06,
      "loss": 0.2386,
      "step": 1067
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.638805858785219,
      "learning_rate": 9.704733808884219e-06,
      "loss": 0.2474,
      "step": 1068
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.521352306674572,
      "learning_rate": 9.704031543263893e-06,
      "loss": 0.2182,
      "step": 1069
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5191981036687148,
      "learning_rate": 9.703328468966016e-06,
      "loss": 0.2352,
      "step": 1070
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2802672442882705,
      "learning_rate": 9.702624586111446e-06,
      "loss": 0.1642,
      "step": 1071
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6259663111041731,
      "learning_rate": 9.701919894821192e-06,
      "loss": 0.2579,
      "step": 1072
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.499975221108699,
      "learning_rate": 9.7012143952164e-06,
      "loss": 0.2171,
      "step": 1073
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5093652491400118,
      "learning_rate": 9.70050808741835e-06,
      "loss": 0.2303,
      "step": 1074
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8190013001181713,
      "learning_rate": 9.69980097154847e-06,
      "loss": 0.2557,
      "step": 1075
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4930488134184756,
      "learning_rate": 9.699093047728317e-06,
      "loss": 0.1923,
      "step": 1076
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.572568586874858,
      "learning_rate": 9.698384316079592e-06,
      "loss": 0.2464,
      "step": 1077
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.83744853439046,
      "learning_rate": 9.697674776724135e-06,
      "loss": 0.6255,
      "step": 1078
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7358267437286419,
      "learning_rate": 9.696964429783923e-06,
      "loss": 0.2335,
      "step": 1079
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.543338655424188,
      "learning_rate": 9.696253275381074e-06,
      "loss": 0.2088,
      "step": 1080
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.61782394728835,
      "learning_rate": 9.695541313637845e-06,
      "loss": 0.2099,
      "step": 1081
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.455271149835965,
      "learning_rate": 9.694828544676626e-06,
      "loss": 0.2637,
      "step": 1082
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.773904598230549,
      "learning_rate": 9.694114968619955e-06,
      "loss": 0.284,
      "step": 1083
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8436696505008552,
      "learning_rate": 9.693400585590502e-06,
      "loss": 0.2383,
      "step": 1084
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3236805554706357,
      "learning_rate": 9.692685395711077e-06,
      "loss": 0.1727,
      "step": 1085
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7045866194217478,
      "learning_rate": 9.69196939910463e-06,
      "loss": 0.2228,
      "step": 1086
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.695680983740636,
      "learning_rate": 9.69125259589425e-06,
      "loss": 0.2088,
      "step": 1087
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7605200086425312,
      "learning_rate": 9.690534986203164e-06,
      "loss": 0.2395,
      "step": 1088
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6598974041439092,
      "learning_rate": 9.689816570154735e-06,
      "loss": 0.2578,
      "step": 1089
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.473472710526562,
      "learning_rate": 9.68909734787247e-06,
      "loss": 0.18,
      "step": 1090
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.038113730116026,
      "learning_rate": 9.688377319480008e-06,
      "loss": 0.5669,
      "step": 1091
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5268618623394272,
      "learning_rate": 9.687656485101134e-06,
      "loss": 0.2555,
      "step": 1092
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.67577952676944,
      "learning_rate": 9.686934844859766e-06,
      "loss": 0.2112,
      "step": 1093
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.227209574823274,
      "learning_rate": 9.686212398879963e-06,
      "loss": 0.3854,
      "step": 1094
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3479969205272042,
      "learning_rate": 9.68548914728592e-06,
      "loss": 0.2363,
      "step": 1095
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6165498325082985,
      "learning_rate": 9.684765090201975e-06,
      "loss": 0.265,
      "step": 1096
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.375287974620047,
      "learning_rate": 9.684040227752601e-06,
      "loss": 0.2342,
      "step": 1097
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5104520787563516,
      "learning_rate": 9.683314560062409e-06,
      "loss": 0.2718,
      "step": 1098
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5025634434889137,
      "learning_rate": 9.68258808725615e-06,
      "loss": 0.2589,
      "step": 1099
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3976167526060723,
      "learning_rate": 9.681860809458713e-06,
      "loss": 0.1898,
      "step": 1100
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3782292069172053,
      "learning_rate": 9.681132726795128e-06,
      "loss": 0.2085,
      "step": 1101
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4593100706180406,
      "learning_rate": 9.680403839390558e-06,
      "loss": 0.2415,
      "step": 1102
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.860967796789328,
      "learning_rate": 9.679674147370308e-06,
      "loss": 0.651,
      "step": 1103
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6279660989657334,
      "learning_rate": 9.67894365085982e-06,
      "loss": 0.2532,
      "step": 1104
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.411598442895208,
      "learning_rate": 9.678212349984677e-06,
      "loss": 0.2429,
      "step": 1105
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.688920199748811,
      "learning_rate": 9.677480244870597e-06,
      "loss": 0.2636,
      "step": 1106
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7548497721717222,
      "learning_rate": 9.676747335643435e-06,
      "loss": 0.2376,
      "step": 1107
    },
    {
      "epoch": 0.14,
      "grad_norm": 18.08624254270899,
      "learning_rate": 9.676013622429187e-06,
      "loss": 0.579,
      "step": 1108
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.81112739653609,
      "learning_rate": 9.675279105353991e-06,
      "loss": 0.6587,
      "step": 1109
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7510757847156224,
      "learning_rate": 9.674543784544118e-06,
      "loss": 0.2697,
      "step": 1110
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6493045604766288,
      "learning_rate": 9.673807660125974e-06,
      "loss": 0.2914,
      "step": 1111
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7124780412485667,
      "learning_rate": 9.673070732226109e-06,
      "loss": 0.2292,
      "step": 1112
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.522983440415001,
      "learning_rate": 9.67233300097121e-06,
      "loss": 0.2055,
      "step": 1113
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.60182614424477,
      "learning_rate": 9.671594466488104e-06,
      "loss": 0.2346,
      "step": 1114
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.45886535604066,
      "learning_rate": 9.670855128903752e-06,
      "loss": 0.5746,
      "step": 1115
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7181699357578317,
      "learning_rate": 9.670114988345252e-06,
      "loss": 0.2222,
      "step": 1116
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2870506021389176,
      "learning_rate": 9.669374044939846e-06,
      "loss": 0.1959,
      "step": 1117
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4977340757733402,
      "learning_rate": 9.668632298814908e-06,
      "loss": 0.2522,
      "step": 1118
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.648233732183026,
      "learning_rate": 9.667889750097956e-06,
      "loss": 0.2538,
      "step": 1119
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.492454150813821,
      "learning_rate": 9.66714639891664e-06,
      "loss": 0.2103,
      "step": 1120
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.479198444921142,
      "learning_rate": 9.666402245398753e-06,
      "loss": 0.238,
      "step": 1121
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4141356935153284,
      "learning_rate": 9.665657289672222e-06,
      "loss": 0.2762,
      "step": 1122
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5140083818579142,
      "learning_rate": 9.664911531865115e-06,
      "loss": 0.2306,
      "step": 1123
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6829485712294032,
      "learning_rate": 9.664164972105634e-06,
      "loss": 0.2506,
      "step": 1124
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.3204184914942605,
      "learning_rate": 9.663417610522124e-06,
      "loss": 0.598,
      "step": 1125
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3910844766516302,
      "learning_rate": 9.66266944724306e-06,
      "loss": 0.1957,
      "step": 1126
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4682327310276921,
      "learning_rate": 9.661920482397069e-06,
      "loss": 0.2626,
      "step": 1127
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.615711928315387,
      "learning_rate": 9.661170716112897e-06,
      "loss": 0.248,
      "step": 1128
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5130560146500562,
      "learning_rate": 9.660420148519444e-06,
      "loss": 0.2234,
      "step": 1129
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.534871978766478,
      "learning_rate": 9.65966877974574e-06,
      "loss": 0.2536,
      "step": 1130
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6657423550198023,
      "learning_rate": 9.658916609920951e-06,
      "loss": 0.2122,
      "step": 1131
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7368448529961253,
      "learning_rate": 9.65816363917439e-06,
      "loss": 0.2478,
      "step": 1132
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.529160372607677,
      "learning_rate": 9.657409867635494e-06,
      "loss": 0.2398,
      "step": 1133
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.736040313900354,
      "learning_rate": 9.656655295433849e-06,
      "loss": 0.5764,
      "step": 1134
    },
    {
      "epoch": 0.15,
      "grad_norm": 23.884611679329915,
      "learning_rate": 9.655899922699174e-06,
      "loss": 0.4974,
      "step": 1135
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.71607794929385,
      "learning_rate": 9.655143749561326e-06,
      "loss": 0.5574,
      "step": 1136
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.013628491158259,
      "learning_rate": 9.654386776150301e-06,
      "loss": 0.3786,
      "step": 1137
    },
    {
      "epoch": 0.15,
      "grad_norm": 16.031996103075684,
      "learning_rate": 9.653629002596232e-06,
      "loss": 0.6124,
      "step": 1138
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7544891503154743,
      "learning_rate": 9.652870429029386e-06,
      "loss": 0.2534,
      "step": 1139
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6977381993944847,
      "learning_rate": 9.652111055580175e-06,
      "loss": 0.225,
      "step": 1140
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6331038691735211,
      "learning_rate": 9.65135088237914e-06,
      "loss": 0.2212,
      "step": 1141
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4104237081257334,
      "learning_rate": 9.650589909556964e-06,
      "loss": 0.2328,
      "step": 1142
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4431717219873108,
      "learning_rate": 9.649828137244471e-06,
      "loss": 0.2157,
      "step": 1143
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5956201859594996,
      "learning_rate": 9.649065565572615e-06,
      "loss": 0.285,
      "step": 1144
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.656341324995696,
      "learning_rate": 9.64830219467249e-06,
      "loss": 0.2342,
      "step": 1145
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3786438629705209,
      "learning_rate": 9.647538024675331e-06,
      "loss": 0.2256,
      "step": 1146
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4349575340515126,
      "learning_rate": 9.646773055712508e-06,
      "loss": 0.2035,
      "step": 1147
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4868408366364758,
      "learning_rate": 9.646007287915524e-06,
      "loss": 0.2441,
      "step": 1148
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4584559940772495,
      "learning_rate": 9.64524072141603e-06,
      "loss": 0.2654,
      "step": 1149
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6301482451828921,
      "learning_rate": 9.6444733563458e-06,
      "loss": 0.2577,
      "step": 1150
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4399902501576818,
      "learning_rate": 9.643705192836758e-06,
      "loss": 0.2495,
      "step": 1151
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5819377528494007,
      "learning_rate": 9.64293623102096e-06,
      "loss": 0.2158,
      "step": 1152
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5635179644561659,
      "learning_rate": 9.642166471030596e-06,
      "loss": 0.2714,
      "step": 1153
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4574776698871839,
      "learning_rate": 9.641395912998e-06,
      "loss": 0.2132,
      "step": 1154
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5781431646781359,
      "learning_rate": 9.64062455705564e-06,
      "loss": 0.2653,
      "step": 1155
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.633952407915657,
      "learning_rate": 9.639852403336118e-06,
      "loss": 0.2534,
      "step": 1156
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5604085784360928,
      "learning_rate": 9.63907945197218e-06,
      "loss": 0.2359,
      "step": 1157
    },
    {
      "epoch": 0.15,
      "grad_norm": 43.270945664415486,
      "learning_rate": 9.638305703096702e-06,
      "loss": 0.6411,
      "step": 1158
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.476792422207033,
      "learning_rate": 9.637531156842702e-06,
      "loss": 0.2262,
      "step": 1159
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6540438715462515,
      "learning_rate": 9.636755813343334e-06,
      "loss": 0.2452,
      "step": 1160
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5078020422715674,
      "learning_rate": 9.635979672731888e-06,
      "loss": 0.244,
      "step": 1161
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6746580440174696,
      "learning_rate": 9.635202735141792e-06,
      "loss": 0.2485,
      "step": 1162
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.471999348864232,
      "learning_rate": 9.634425000706607e-06,
      "loss": 0.791,
      "step": 1163
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5903980070612376,
      "learning_rate": 9.633646469560039e-06,
      "loss": 0.2237,
      "step": 1164
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5502155370467325,
      "learning_rate": 9.632867141835926e-06,
      "loss": 0.2554,
      "step": 1165
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.531426134518352,
      "learning_rate": 9.632087017668242e-06,
      "loss": 0.8039,
      "step": 1166
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6856117916815128,
      "learning_rate": 9.6313060971911e-06,
      "loss": 0.2636,
      "step": 1167
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.758221961311709,
      "learning_rate": 9.630524380538748e-06,
      "loss": 0.7284,
      "step": 1168
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6580730980472638,
      "learning_rate": 9.629741867845574e-06,
      "loss": 0.2683,
      "step": 1169
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6492536421469575,
      "learning_rate": 9.628958559246101e-06,
      "loss": 0.2214,
      "step": 1170
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5477840355847745,
      "learning_rate": 9.628174454874988e-06,
      "loss": 0.2033,
      "step": 1171
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.671353052771271,
      "learning_rate": 9.627389554867032e-06,
      "loss": 0.2859,
      "step": 1172
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6345217913232297,
      "learning_rate": 9.626603859357165e-06,
      "loss": 0.2303,
      "step": 1173
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.47669215457162,
      "learning_rate": 9.625817368480459e-06,
      "loss": 0.2236,
      "step": 1174
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.614364962263538,
      "learning_rate": 9.625030082372122e-06,
      "loss": 0.2835,
      "step": 1175
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4759238330483155,
      "learning_rate": 9.624242001167493e-06,
      "loss": 0.2058,
      "step": 1176
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3422964989367558,
      "learning_rate": 9.623453125002056e-06,
      "loss": 0.2019,
      "step": 1177
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.9232860348387115,
      "learning_rate": 9.622663454011429e-06,
      "loss": 0.6229,
      "step": 1178
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5286685895771748,
      "learning_rate": 9.621872988331362e-06,
      "loss": 0.2739,
      "step": 1179
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7006101341842628,
      "learning_rate": 9.621081728097747e-06,
      "loss": 0.2214,
      "step": 1180
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4683418641292802,
      "learning_rate": 9.620289673446611e-06,
      "loss": 0.2245,
      "step": 1181
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6095254602429423,
      "learning_rate": 9.619496824514118e-06,
      "loss": 0.2141,
      "step": 1182
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.608089327459364,
      "learning_rate": 9.618703181436566e-06,
      "loss": 0.2312,
      "step": 1183
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.160860269846557,
      "learning_rate": 9.617908744350392e-06,
      "loss": 0.5785,
      "step": 1184
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6040912487124463,
      "learning_rate": 9.61711351339217e-06,
      "loss": 0.1942,
      "step": 1185
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.872305974218926,
      "learning_rate": 9.61631748869861e-06,
      "loss": 0.4259,
      "step": 1186
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3697944396801938,
      "learning_rate": 9.615520670406555e-06,
      "loss": 0.206,
      "step": 1187
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5957911295002671,
      "learning_rate": 9.61472305865299e-06,
      "loss": 0.2214,
      "step": 1188
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4609579581761085,
      "learning_rate": 9.613924653575034e-06,
      "loss": 0.2231,
      "step": 1189
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5733520287398048,
      "learning_rate": 9.61312545530994e-06,
      "loss": 0.2424,
      "step": 1190
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.8248933678853625,
      "learning_rate": 9.612325463995099e-06,
      "loss": 0.5382,
      "step": 1191
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.90488538334309,
      "learning_rate": 9.61152467976804e-06,
      "loss": 0.6042,
      "step": 1192
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5780809265660944,
      "learning_rate": 9.610723102766429e-06,
      "loss": 0.2612,
      "step": 1193
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5948615155721828,
      "learning_rate": 9.609920733128064e-06,
      "loss": 0.2307,
      "step": 1194
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5205757085036982,
      "learning_rate": 9.609117570990882e-06,
      "loss": 0.2279,
      "step": 1195
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.973028834490256,
      "learning_rate": 9.608313616492954e-06,
      "loss": 0.588,
      "step": 1196
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5813411386167142,
      "learning_rate": 9.607508869772495e-06,
      "loss": 0.2994,
      "step": 1197
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.766784232559715,
      "learning_rate": 9.606703330967843e-06,
      "loss": 0.5768,
      "step": 1198
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.254342548963146,
      "learning_rate": 9.605897000217485e-06,
      "loss": 0.6229,
      "step": 1199
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4160856572717027,
      "learning_rate": 9.605089877660036e-06,
      "loss": 0.2254,
      "step": 1200
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.474705006992043,
      "learning_rate": 9.60428196343425e-06,
      "loss": 0.1769,
      "step": 1201
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.659855093229272,
      "learning_rate": 9.603473257679018e-06,
      "loss": 0.5422,
      "step": 1202
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3884289070859093,
      "learning_rate": 9.602663760533364e-06,
      "loss": 0.2358,
      "step": 1203
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5363495811492942,
      "learning_rate": 9.601853472136451e-06,
      "loss": 0.1978,
      "step": 1204
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5066552568136895,
      "learning_rate": 9.601042392627577e-06,
      "loss": 0.2137,
      "step": 1205
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.8006501451933135,
      "learning_rate": 9.600230522146176e-06,
      "loss": 0.6813,
      "step": 1206
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.538478283247136,
      "learning_rate": 9.599417860831818e-06,
      "loss": 0.225,
      "step": 1207
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.809278602820304,
      "learning_rate": 9.598604408824209e-06,
      "loss": 0.6372,
      "step": 1208
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.74885381316936,
      "learning_rate": 9.59779016626319e-06,
      "loss": 0.2743,
      "step": 1209
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.465352792533464,
      "learning_rate": 9.596975133288738e-06,
      "loss": 0.2109,
      "step": 1210
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.9304715332874745,
      "learning_rate": 9.596159310040968e-06,
      "loss": 0.7091,
      "step": 1211
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5923681208928318,
      "learning_rate": 9.595342696660133e-06,
      "loss": 0.2145,
      "step": 1212
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.489951372094076,
      "learning_rate": 9.594525293286611e-06,
      "loss": 0.1932,
      "step": 1213
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7586689023338602,
      "learning_rate": 9.593707100060927e-06,
      "loss": 0.2334,
      "step": 1214
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3114828951346622,
      "learning_rate": 9.59288811712374e-06,
      "loss": 0.1789,
      "step": 1215
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4887799258594492,
      "learning_rate": 9.592068344615837e-06,
      "loss": 0.2061,
      "step": 1216
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3046173887672954,
      "learning_rate": 9.591247782678153e-06,
      "loss": 0.2307,
      "step": 1217
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7848934369608367,
      "learning_rate": 9.590426431451748e-06,
      "loss": 0.3072,
      "step": 1218
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.48510922754613,
      "learning_rate": 9.58960429107782e-06,
      "loss": 0.1991,
      "step": 1219
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4199965492851605,
      "learning_rate": 9.58878136169771e-06,
      "loss": 0.2022,
      "step": 1220
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.057317137528475,
      "learning_rate": 9.587957643452886e-06,
      "loss": 0.7233,
      "step": 1221
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.393816112173648,
      "learning_rate": 9.587133136484953e-06,
      "loss": 0.2445,
      "step": 1222
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7582253697811865,
      "learning_rate": 9.586307840935658e-06,
      "loss": 0.2694,
      "step": 1223
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6842562730966855,
      "learning_rate": 9.585481756946875e-06,
      "loss": 0.2044,
      "step": 1224
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.706681446024937,
      "learning_rate": 9.584654884660618e-06,
      "loss": 0.2432,
      "step": 1225
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4833155215245384,
      "learning_rate": 9.583827224219035e-06,
      "loss": 0.2652,
      "step": 1226
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.661203971499847,
      "learning_rate": 9.582998775764414e-06,
      "loss": 0.7316,
      "step": 1227
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5841624217477865,
      "learning_rate": 9.582169539439173e-06,
      "loss": 0.2004,
      "step": 1228
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.7351864108723785,
      "learning_rate": 9.581339515385866e-06,
      "loss": 0.5766,
      "step": 1229
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.751660929706731,
      "learning_rate": 9.580508703747185e-06,
      "loss": 0.2467,
      "step": 1230
    },
    {
      "epoch": 0.16,
      "grad_norm": 9.619848052168381,
      "learning_rate": 9.579677104665957e-06,
      "loss": 0.628,
      "step": 1231
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5381178428018545,
      "learning_rate": 9.578844718285141e-06,
      "loss": 0.2416,
      "step": 1232
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5245655812354841,
      "learning_rate": 9.578011544747836e-06,
      "loss": 0.2021,
      "step": 1233
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4773624063461126,
      "learning_rate": 9.577177584197274e-06,
      "loss": 0.2216,
      "step": 1234
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5437452963425118,
      "learning_rate": 9.576342836776822e-06,
      "loss": 0.2076,
      "step": 1235
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5793829366383914,
      "learning_rate": 9.575507302629982e-06,
      "loss": 0.2643,
      "step": 1236
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5606836256043106,
      "learning_rate": 9.574670981900394e-06,
      "loss": 0.2403,
      "step": 1237
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4496384121029748,
      "learning_rate": 9.57383387473183e-06,
      "loss": 0.2269,
      "step": 1238
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4900611558740344,
      "learning_rate": 9.572995981268199e-06,
      "loss": 0.2551,
      "step": 1239
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.425202709960983,
      "learning_rate": 9.572157301653542e-06,
      "loss": 0.5496,
      "step": 1240
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4505246661191078,
      "learning_rate": 9.571317836032042e-06,
      "loss": 0.2196,
      "step": 1241
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4815989773798013,
      "learning_rate": 9.570477584548008e-06,
      "loss": 0.2208,
      "step": 1242
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.439060184609878,
      "learning_rate": 9.569636547345895e-06,
      "loss": 0.2435,
      "step": 1243
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.466473551136425,
      "learning_rate": 9.568794724570282e-06,
      "loss": 0.1714,
      "step": 1244
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.995623690744932,
      "learning_rate": 9.567952116365889e-06,
      "loss": 0.6222,
      "step": 1245
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3972677768634,
      "learning_rate": 9.567108722877572e-06,
      "loss": 0.2941,
      "step": 1246
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.500218106070521,
      "learning_rate": 9.566264544250319e-06,
      "loss": 0.2466,
      "step": 1247
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6797040450960212,
      "learning_rate": 9.565419580629254e-06,
      "loss": 0.2216,
      "step": 1248
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6754866367764445,
      "learning_rate": 9.564573832159638e-06,
      "loss": 0.2479,
      "step": 1249
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4701246047671166,
      "learning_rate": 9.56372729898686e-06,
      "loss": 0.2153,
      "step": 1250
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4239670614270612,
      "learning_rate": 9.562879981256455e-06,
      "loss": 0.2849,
      "step": 1251
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5486746264765519,
      "learning_rate": 9.562031879114082e-06,
      "loss": 0.2337,
      "step": 1252
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3408640291196363,
      "learning_rate": 9.561182992705541e-06,
      "loss": 0.1973,
      "step": 1253
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8170883713461086,
      "learning_rate": 9.560333322176767e-06,
      "loss": 0.2473,
      "step": 1254
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.602169433836393,
      "learning_rate": 9.559482867673825e-06,
      "loss": 0.5083,
      "step": 1255
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.513246324489037,
      "learning_rate": 9.558631629342922e-06,
      "loss": 0.2304,
      "step": 1256
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3841725982872413,
      "learning_rate": 9.557779607330393e-06,
      "loss": 0.1687,
      "step": 1257
    },
    {
      "epoch": 0.16,
      "grad_norm": 62.593047297059016,
      "learning_rate": 9.556926801782714e-06,
      "loss": 0.5803,
      "step": 1258
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5303558954832313,
      "learning_rate": 9.556073212846485e-06,
      "loss": 0.2393,
      "step": 1259
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4888407184684125,
      "learning_rate": 9.555218840668454e-06,
      "loss": 0.2331,
      "step": 1260
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6748405833776,
      "learning_rate": 9.554363685395496e-06,
      "loss": 0.2679,
      "step": 1261
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5293504053106608,
      "learning_rate": 9.553507747174622e-06,
      "loss": 0.2066,
      "step": 1262
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.8587979127137855,
      "learning_rate": 9.552651026152978e-06,
      "loss": 0.4566,
      "step": 1263
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.352097714363993,
      "learning_rate": 9.551793522477842e-06,
      "loss": 0.142,
      "step": 1264
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.717481782110893,
      "learning_rate": 9.55093523629663e-06,
      "loss": 0.6831,
      "step": 1265
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5412535203830917,
      "learning_rate": 9.550076167756892e-06,
      "loss": 0.2339,
      "step": 1266
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3251028620623677,
      "learning_rate": 9.549216317006313e-06,
      "loss": 0.2052,
      "step": 1267
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.589469896924288,
      "learning_rate": 9.548355684192712e-06,
      "loss": 0.2771,
      "step": 1268
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5361917652734332,
      "learning_rate": 9.547494269464037e-06,
      "loss": 0.2545,
      "step": 1269
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6704984070988609,
      "learning_rate": 9.546632072968379e-06,
      "loss": 0.2532,
      "step": 1270
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4733075776506683,
      "learning_rate": 9.545769094853958e-06,
      "loss": 0.2748,
      "step": 1271
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.617795765932214,
      "learning_rate": 9.54490533526913e-06,
      "loss": 0.2677,
      "step": 1272
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5209780168413567,
      "learning_rate": 9.544040794362389e-06,
      "loss": 0.2598,
      "step": 1273
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7398521027397302,
      "learning_rate": 9.543175472282353e-06,
      "loss": 0.2153,
      "step": 1274
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4635507315459644,
      "learning_rate": 9.542309369177785e-06,
      "loss": 0.2305,
      "step": 1275
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5019965869207264,
      "learning_rate": 9.541442485197577e-06,
      "loss": 0.2595,
      "step": 1276
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4775540475713593,
      "learning_rate": 9.540574820490759e-06,
      "loss": 0.2214,
      "step": 1277
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3136652022203765,
      "learning_rate": 9.539706375206487e-06,
      "loss": 0.1998,
      "step": 1278
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.732887957588465,
      "learning_rate": 9.538837149494065e-06,
      "loss": 0.7573,
      "step": 1279
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5382681379079113,
      "learning_rate": 9.537967143502915e-06,
      "loss": 0.2168,
      "step": 1280
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5938117283920155,
      "learning_rate": 9.537096357382606e-06,
      "loss": 0.219,
      "step": 1281
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4795084763396538,
      "learning_rate": 9.536224791282834e-06,
      "loss": 0.234,
      "step": 1282
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.456063435962597,
      "learning_rate": 9.53535244535343e-06,
      "loss": 0.2371,
      "step": 1283
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6893538110610695,
      "learning_rate": 9.534479319744366e-06,
      "loss": 0.2984,
      "step": 1284
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.55612774011548,
      "learning_rate": 9.533605414605736e-06,
      "loss": 0.2315,
      "step": 1285
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6579465625517313,
      "learning_rate": 9.532730730087779e-06,
      "loss": 0.2423,
      "step": 1286
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.648924053561711,
      "learning_rate": 9.53185526634086e-06,
      "loss": 0.2905,
      "step": 1287
    },
    {
      "epoch": 0.16,
      "grad_norm": 10.501576763086112,
      "learning_rate": 9.530979023515483e-06,
      "loss": 0.6118,
      "step": 1288
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.600396069648095,
      "learning_rate": 9.530102001762285e-06,
      "loss": 0.2469,
      "step": 1289
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.56497162763657,
      "learning_rate": 9.529224201232034e-06,
      "loss": 0.1811,
      "step": 1290
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5312925028565998,
      "learning_rate": 9.528345622075636e-06,
      "loss": 0.1849,
      "step": 1291
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4793295688859789,
      "learning_rate": 9.52746626444413e-06,
      "loss": 0.2022,
      "step": 1292
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5921775850206528,
      "learning_rate": 9.526586128488686e-06,
      "loss": 0.2728,
      "step": 1293
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6860202204597265,
      "learning_rate": 9.52570521436061e-06,
      "loss": 0.264,
      "step": 1294
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4619695463226428,
      "learning_rate": 9.52482352221134e-06,
      "loss": 0.2265,
      "step": 1295
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4288609008619053,
      "learning_rate": 9.52394105219245e-06,
      "loss": 0.2354,
      "step": 1296
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6181801416421695,
      "learning_rate": 9.523057804455648e-06,
      "loss": 0.2503,
      "step": 1297
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4942762207263316,
      "learning_rate": 9.522173779152773e-06,
      "loss": 0.1891,
      "step": 1298
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6979114142373892,
      "learning_rate": 9.5212889764358e-06,
      "loss": 0.249,
      "step": 1299
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4397465786321333,
      "learning_rate": 9.520403396456838e-06,
      "loss": 0.2226,
      "step": 1300
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.451029980933239,
      "learning_rate": 9.519517039368127e-06,
      "loss": 0.2549,
      "step": 1301
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5221946276455347,
      "learning_rate": 9.518629905322041e-06,
      "loss": 0.2353,
      "step": 1302
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5078336081293322,
      "learning_rate": 9.517741994471091e-06,
      "loss": 0.2287,
      "step": 1303
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.413349365077716,
      "learning_rate": 9.516853306967917e-06,
      "loss": 0.2303,
      "step": 1304
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5308158964441643,
      "learning_rate": 9.515963842965294e-06,
      "loss": 0.258,
      "step": 1305
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.112039808010687,
      "learning_rate": 9.515073602616135e-06,
      "loss": 0.6397,
      "step": 1306
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4639336269626773,
      "learning_rate": 9.51418258607348e-06,
      "loss": 0.1985,
      "step": 1307
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.651235655211813,
      "learning_rate": 9.513290793490506e-06,
      "loss": 0.6648,
      "step": 1308
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4708021584968265,
      "learning_rate": 9.512398225020523e-06,
      "loss": 0.2432,
      "step": 1309
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4183127134437568,
      "learning_rate": 9.511504880816971e-06,
      "loss": 0.2225,
      "step": 1310
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4492989772715374,
      "learning_rate": 9.510610761033427e-06,
      "loss": 0.2122,
      "step": 1311
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5920639702340986,
      "learning_rate": 9.509715865823605e-06,
      "loss": 0.2409,
      "step": 1312
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4792584543348728,
      "learning_rate": 9.508820195341343e-06,
      "loss": 0.2087,
      "step": 1313
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8369680307258807,
      "learning_rate": 9.507923749740619e-06,
      "loss": 0.2795,
      "step": 1314
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.341705835145092,
      "learning_rate": 9.50702652917554e-06,
      "loss": 0.6414,
      "step": 1315
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3637880199504182,
      "learning_rate": 9.506128533800353e-06,
      "loss": 0.1966,
      "step": 1316
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8078234149398473,
      "learning_rate": 9.505229763769432e-06,
      "loss": 0.22,
      "step": 1317
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.381454126550166,
      "learning_rate": 9.504330219237284e-06,
      "loss": 0.6742,
      "step": 1318
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4282456567932316,
      "learning_rate": 9.503429900358554e-06,
      "loss": 0.1972,
      "step": 1319
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4926282403285298,
      "learning_rate": 9.502528807288014e-06,
      "loss": 0.206,
      "step": 1320
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4855686647223705,
      "learning_rate": 9.501626940180574e-06,
      "loss": 0.214,
      "step": 1321
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6420017624135308,
      "learning_rate": 9.50072429919128e-06,
      "loss": 0.3031,
      "step": 1322
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.510749085934064,
      "learning_rate": 9.499820884475296e-06,
      "loss": 0.2498,
      "step": 1323
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.523219847450577,
      "learning_rate": 9.498916696187942e-06,
      "loss": 0.2539,
      "step": 1324
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.007127387288937,
      "learning_rate": 9.498011734484647e-06,
      "loss": 0.6365,
      "step": 1325
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5574619645033094,
      "learning_rate": 9.49710599952099e-06,
      "loss": 0.2196,
      "step": 1326
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4338149751791494,
      "learning_rate": 9.49619949145268e-06,
      "loss": 0.2025,
      "step": 1327
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3814382583571594,
      "learning_rate": 9.49529221043555e-06,
      "loss": 0.1911,
      "step": 1328
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4986907424847324,
      "learning_rate": 9.494384156625575e-06,
      "loss": 0.2038,
      "step": 1329
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5149401332535672,
      "learning_rate": 9.49347533017886e-06,
      "loss": 0.2393,
      "step": 1330
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3407841093225916,
      "learning_rate": 9.492565731251645e-06,
      "loss": 0.234,
      "step": 1331
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4905628626570806,
      "learning_rate": 9.491655360000298e-06,
      "loss": 0.2452,
      "step": 1332
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3800839911134655,
      "learning_rate": 9.490744216581323e-06,
      "loss": 0.1985,
      "step": 1333
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5166381039305716,
      "learning_rate": 9.489832301151354e-06,
      "loss": 0.2711,
      "step": 1334
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.147697788107952,
      "learning_rate": 9.488919613867162e-06,
      "loss": 0.6893,
      "step": 1335
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6612846527913374,
      "learning_rate": 9.48800615488565e-06,
      "loss": 0.3096,
      "step": 1336
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6877457814414836,
      "learning_rate": 9.487091924363852e-06,
      "loss": 0.2034,
      "step": 1337
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5194431664739807,
      "learning_rate": 9.486176922458929e-06,
      "loss": 0.2278,
      "step": 1338
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7977916744889912,
      "learning_rate": 9.485261149328189e-06,
      "loss": 0.206,
      "step": 1339
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5139294009003572,
      "learning_rate": 9.484344605129057e-06,
      "loss": 0.1964,
      "step": 1340
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.664749647003145,
      "learning_rate": 9.4834272900191e-06,
      "loss": 0.2503,
      "step": 1341
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3861632223685958,
      "learning_rate": 9.482509204156019e-06,
      "loss": 0.2056,
      "step": 1342
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.568118399173356,
      "learning_rate": 9.481590347697638e-06,
      "loss": 0.2603,
      "step": 1343
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5941911937764517,
      "learning_rate": 9.480670720801921e-06,
      "loss": 0.2339,
      "step": 1344
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6127396436245336,
      "learning_rate": 9.479750323626963e-06,
      "loss": 0.2453,
      "step": 1345
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3314780432372997,
      "learning_rate": 9.47882915633099e-06,
      "loss": 0.2355,
      "step": 1346
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4308253086888987,
      "learning_rate": 9.477907219072362e-06,
      "loss": 0.2059,
      "step": 1347
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5464797903158394,
      "learning_rate": 9.476984512009572e-06,
      "loss": 0.2721,
      "step": 1348
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.318403742243913,
      "learning_rate": 9.476061035301242e-06,
      "loss": 0.6199,
      "step": 1349
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6908974546846829,
      "learning_rate": 9.475136789106128e-06,
      "loss": 0.2118,
      "step": 1350
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.414003748459017,
      "learning_rate": 9.474211773583122e-06,
      "loss": 0.2368,
      "step": 1351
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3414150522347263,
      "learning_rate": 9.47328598889124e-06,
      "loss": 0.2559,
      "step": 1352
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5780364173182124,
      "learning_rate": 9.47235943518964e-06,
      "loss": 0.2547,
      "step": 1353
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3792110438717027,
      "learning_rate": 9.471432112637604e-06,
      "loss": 0.1903,
      "step": 1354
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.470013967221703,
      "learning_rate": 9.47050402139455e-06,
      "loss": 0.2214,
      "step": 1355
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5346564069524569,
      "learning_rate": 9.469575161620029e-06,
      "loss": 0.247,
      "step": 1356
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5427240001027898,
      "learning_rate": 9.468645533473721e-06,
      "loss": 0.2792,
      "step": 1357
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4919924880309399,
      "learning_rate": 9.467715137115442e-06,
      "loss": 0.2287,
      "step": 1358
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3382553324064512,
      "learning_rate": 9.466783972705138e-06,
      "loss": 0.2102,
      "step": 1359
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.379126795005938,
      "learning_rate": 9.465852040402883e-06,
      "loss": 0.7692,
      "step": 1360
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.530030680191017,
      "learning_rate": 9.46491934036889e-06,
      "loss": 0.2262,
      "step": 1361
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.810604418974676,
      "learning_rate": 9.463985872763501e-06,
      "loss": 0.7955,
      "step": 1362
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4417159335458853,
      "learning_rate": 9.463051637747191e-06,
      "loss": 0.1865,
      "step": 1363
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.352211541345928,
      "learning_rate": 9.462116635480562e-06,
      "loss": 0.6013,
      "step": 1364
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6261883869477791,
      "learning_rate": 9.461180866124357e-06,
      "loss": 0.2705,
      "step": 1365
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.467004997239236,
      "learning_rate": 9.46024432983944e-06,
      "loss": 0.2435,
      "step": 1366
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5931726600349407,
      "learning_rate": 9.459307026786817e-06,
      "loss": 0.2203,
      "step": 1367
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6321365400468026,
      "learning_rate": 9.458368957127616e-06,
      "loss": 0.2248,
      "step": 1368
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5774248212909254,
      "learning_rate": 9.45743012102311e-06,
      "loss": 0.2402,
      "step": 1369
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.550436754158137,
      "learning_rate": 9.456490518634688e-06,
      "loss": 0.2369,
      "step": 1370
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5741402415781947,
      "learning_rate": 9.455550150123884e-06,
      "loss": 0.2285,
      "step": 1371
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4151037368090316,
      "learning_rate": 9.454609015652355e-06,
      "loss": 0.2491,
      "step": 1372
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5556249027570033,
      "learning_rate": 9.453667115381893e-06,
      "loss": 0.2139,
      "step": 1373
    },
    {
      "epoch": 0.18,
      "grad_norm": 24.242152432811682,
      "learning_rate": 9.452724449474423e-06,
      "loss": 0.6064,
      "step": 1374
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6790643733446435,
      "learning_rate": 9.451781018092e-06,
      "loss": 0.2401,
      "step": 1375
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.041211028207774,
      "learning_rate": 9.45083682139681e-06,
      "loss": 0.7202,
      "step": 1376
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7880285217708065,
      "learning_rate": 9.449891859551172e-06,
      "loss": 0.2784,
      "step": 1377
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6039240956979144,
      "learning_rate": 9.448946132717536e-06,
      "loss": 0.2332,
      "step": 1378
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6469443907767378,
      "learning_rate": 9.447999641058486e-06,
      "loss": 0.2292,
      "step": 1379
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4533399046999647,
      "learning_rate": 9.447052384736729e-06,
      "loss": 0.2353,
      "step": 1380
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4871969920090922,
      "learning_rate": 9.446104363915112e-06,
      "loss": 0.2263,
      "step": 1381
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3958410259633964,
      "learning_rate": 9.445155578756612e-06,
      "loss": 0.2407,
      "step": 1382
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5034043459101294,
      "learning_rate": 9.444206029424334e-06,
      "loss": 0.2175,
      "step": 1383
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.483660690984251,
      "learning_rate": 9.443255716081522e-06,
      "loss": 0.2368,
      "step": 1384
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5667345948580198,
      "learning_rate": 9.442304638891538e-06,
      "loss": 0.2456,
      "step": 1385
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3746614746806467,
      "learning_rate": 9.441352798017886e-06,
      "loss": 0.2022,
      "step": 1386
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3646733843471897,
      "learning_rate": 9.440400193624202e-06,
      "loss": 0.1918,
      "step": 1387
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5292009116407772,
      "learning_rate": 9.439446825874248e-06,
      "loss": 0.2708,
      "step": 1388
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.225195286318258,
      "learning_rate": 9.438492694931916e-06,
      "loss": 0.594,
      "step": 1389
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5509452877009722,
      "learning_rate": 9.437537800961235e-06,
      "loss": 0.2539,
      "step": 1390
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.4409291932643695,
      "learning_rate": 9.436582144126362e-06,
      "loss": 0.6871,
      "step": 1391
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5650480668461877,
      "learning_rate": 9.435625724591584e-06,
      "loss": 0.2533,
      "step": 1392
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.573238444610548,
      "learning_rate": 9.434668542521323e-06,
      "loss": 0.2675,
      "step": 1393
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6736635478234227,
      "learning_rate": 9.433710598080128e-06,
      "loss": 0.2448,
      "step": 1394
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5469885218366042,
      "learning_rate": 9.432751891432681e-06,
      "loss": 0.2417,
      "step": 1395
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2502963856740192,
      "learning_rate": 9.431792422743795e-06,
      "loss": 0.1966,
      "step": 1396
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3535232992698771,
      "learning_rate": 9.430832192178413e-06,
      "loss": 0.2102,
      "step": 1397
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6570878315645066,
      "learning_rate": 9.429871199901614e-06,
      "loss": 0.2337,
      "step": 1398
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3801936568983943,
      "learning_rate": 9.428909446078597e-06,
      "loss": 0.1733,
      "step": 1399
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2978238157256916,
      "learning_rate": 9.427946930874704e-06,
      "loss": 0.1997,
      "step": 1400
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.84600068013697,
      "learning_rate": 9.426983654455399e-06,
      "loss": 0.653,
      "step": 1401
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6378804829869693,
      "learning_rate": 9.426019616986281e-06,
      "loss": 0.2629,
      "step": 1402
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.345982546820428,
      "learning_rate": 9.425054818633081e-06,
      "loss": 0.1923,
      "step": 1403
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4880946269293875,
      "learning_rate": 9.424089259561658e-06,
      "loss": 0.286,
      "step": 1404
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4094269203828091,
      "learning_rate": 9.423122939938003e-06,
      "loss": 0.217,
      "step": 1405
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4554198974678827,
      "learning_rate": 9.422155859928237e-06,
      "loss": 0.1979,
      "step": 1406
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5978978604095189,
      "learning_rate": 9.421188019698613e-06,
      "loss": 0.2411,
      "step": 1407
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3981501296043555,
      "learning_rate": 9.420219419415513e-06,
      "loss": 0.2223,
      "step": 1408
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5480660630809735,
      "learning_rate": 9.419250059245453e-06,
      "loss": 0.2202,
      "step": 1409
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2760680247679357,
      "learning_rate": 9.418279939355073e-06,
      "loss": 0.1869,
      "step": 1410
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8545946260423,
      "learning_rate": 9.417309059911152e-06,
      "loss": 0.2483,
      "step": 1411
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.558374572200328,
      "learning_rate": 9.416337421080594e-06,
      "loss": 0.2306,
      "step": 1412
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4363410482275596,
      "learning_rate": 9.415365023030433e-06,
      "loss": 0.2277,
      "step": 1413
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6320439638608994,
      "learning_rate": 9.41439186592784e-06,
      "loss": 0.2277,
      "step": 1414
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4776218697779169,
      "learning_rate": 9.413417949940107e-06,
      "loss": 0.2536,
      "step": 1415
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.622708927992692,
      "learning_rate": 9.412443275234663e-06,
      "loss": 0.2998,
      "step": 1416
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.480134500594811,
      "learning_rate": 9.411467841979069e-06,
      "loss": 0.1993,
      "step": 1417
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.9444783846845044,
      "learning_rate": 9.410491650341009e-06,
      "loss": 0.619,
      "step": 1418
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3903237901044379,
      "learning_rate": 9.409514700488304e-06,
      "loss": 0.2238,
      "step": 1419
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.198715519341588,
      "learning_rate": 9.408536992588903e-06,
      "loss": 0.676,
      "step": 1420
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.340339896605884,
      "learning_rate": 9.407558526810884e-06,
      "loss": 0.4429,
      "step": 1421
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5588017378697945,
      "learning_rate": 9.406579303322458e-06,
      "loss": 0.2173,
      "step": 1422
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4416530149484088,
      "learning_rate": 9.405599322291963e-06,
      "loss": 0.2483,
      "step": 1423
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6271866945745765,
      "learning_rate": 9.404618583887872e-06,
      "loss": 0.2917,
      "step": 1424
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.392078430105137,
      "learning_rate": 9.403637088278784e-06,
      "loss": 0.6685,
      "step": 1425
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.920480703551781,
      "learning_rate": 9.402654835633429e-06,
      "loss": 0.5966,
      "step": 1426
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4694883566926684,
      "learning_rate": 9.401671826120667e-06,
      "loss": 0.237,
      "step": 1427
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.921485253380387,
      "learning_rate": 9.400688059909489e-06,
      "loss": 0.5151,
      "step": 1428
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.685634463731508,
      "learning_rate": 9.399703537169017e-06,
      "loss": 0.2416,
      "step": 1429
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5524478330761402,
      "learning_rate": 9.398718258068502e-06,
      "loss": 0.2563,
      "step": 1430
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3939382362635977,
      "learning_rate": 9.397732222777323e-06,
      "loss": 0.2257,
      "step": 1431
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3202808078555026,
      "learning_rate": 9.396745431464993e-06,
      "loss": 0.2378,
      "step": 1432
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6306609517896444,
      "learning_rate": 9.395757884301152e-06,
      "loss": 0.2573,
      "step": 1433
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3091885523507527,
      "learning_rate": 9.394769581455569e-06,
      "loss": 0.2138,
      "step": 1434
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4430124782951714,
      "learning_rate": 9.393780523098148e-06,
      "loss": 0.2629,
      "step": 1435
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.7970764912326,
      "learning_rate": 9.392790709398916e-06,
      "loss": 0.8421,
      "step": 1436
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2246513027379304,
      "learning_rate": 9.391800140528038e-06,
      "loss": 0.172,
      "step": 1437
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5777023652138051,
      "learning_rate": 9.390808816655801e-06,
      "loss": 0.2618,
      "step": 1438
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8047631376586764,
      "learning_rate": 9.389816737952624e-06,
      "loss": 0.2986,
      "step": 1439
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.7742661057943305,
      "learning_rate": 9.388823904589062e-06,
      "loss": 0.4087,
      "step": 1440
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3937823168985626,
      "learning_rate": 9.387830316735789e-06,
      "loss": 0.1934,
      "step": 1441
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4298503882048745,
      "learning_rate": 9.386835974563616e-06,
      "loss": 0.2394,
      "step": 1442
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3645838002456734,
      "learning_rate": 9.385840878243482e-06,
      "loss": 0.1841,
      "step": 1443
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3355211102146707,
      "learning_rate": 9.384845027946458e-06,
      "loss": 0.2166,
      "step": 1444
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4100631575636184,
      "learning_rate": 9.38384842384374e-06,
      "loss": 0.1989,
      "step": 1445
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5137529013986846,
      "learning_rate": 9.382851066106655e-06,
      "loss": 0.2064,
      "step": 1446
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.886628746173691,
      "learning_rate": 9.381852954906662e-06,
      "loss": 0.665,
      "step": 1447
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3605945415039127,
      "learning_rate": 9.380854090415347e-06,
      "loss": 0.2379,
      "step": 1448
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4239200151067124,
      "learning_rate": 9.379854472804425e-06,
      "loss": 0.2223,
      "step": 1449
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3351573211134287,
      "learning_rate": 9.378854102245747e-06,
      "loss": 0.211,
      "step": 1450
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.484543826424525,
      "learning_rate": 9.377852978911282e-06,
      "loss": 0.1897,
      "step": 1451
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4749702147041803,
      "learning_rate": 9.376851102973139e-06,
      "loss": 0.1835,
      "step": 1452
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.776228169840929,
      "learning_rate": 9.375848474603549e-06,
      "loss": 0.2502,
      "step": 1453
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4869350818336633,
      "learning_rate": 9.374845093974879e-06,
      "loss": 0.2559,
      "step": 1454
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6953211718475107,
      "learning_rate": 9.373840961259619e-06,
      "loss": 0.2123,
      "step": 1455
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6422496314662434,
      "learning_rate": 9.372836076630391e-06,
      "loss": 0.25,
      "step": 1456
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.521545930027525,
      "learning_rate": 9.371830440259948e-06,
      "loss": 0.5631,
      "step": 1457
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7051980226521455,
      "learning_rate": 9.370824052321168e-06,
      "loss": 0.2558,
      "step": 1458
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4527083885282852,
      "learning_rate": 9.369816912987066e-06,
      "loss": 0.2197,
      "step": 1459
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5779780963102918,
      "learning_rate": 9.368809022430773e-06,
      "loss": 0.2415,
      "step": 1460
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3447088296543543,
      "learning_rate": 9.367800380825564e-06,
      "loss": 0.2109,
      "step": 1461
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5815483910542372,
      "learning_rate": 9.366790988344835e-06,
      "loss": 0.2709,
      "step": 1462
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4147430876752176,
      "learning_rate": 9.365780845162109e-06,
      "loss": 0.2369,
      "step": 1463
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5399443896043528,
      "learning_rate": 9.364769951451045e-06,
      "loss": 0.2365,
      "step": 1464
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.835737682152139,
      "learning_rate": 9.363758307385423e-06,
      "loss": 0.568,
      "step": 1465
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5125559245214302,
      "learning_rate": 9.36274591313916e-06,
      "loss": 0.23,
      "step": 1466
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.54549484906007,
      "learning_rate": 9.361732768886299e-06,
      "loss": 0.2584,
      "step": 1467
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4809957872256525,
      "learning_rate": 9.360718874801007e-06,
      "loss": 0.208,
      "step": 1468
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5284040292534338,
      "learning_rate": 9.359704231057587e-06,
      "loss": 0.2311,
      "step": 1469
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.411473748059378,
      "learning_rate": 9.358688837830469e-06,
      "loss": 0.2299,
      "step": 1470
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4503855679693156,
      "learning_rate": 9.357672695294208e-06,
      "loss": 0.2044,
      "step": 1471
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3970931656885266,
      "learning_rate": 9.356655803623493e-06,
      "loss": 0.2194,
      "step": 1472
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3963125803673355,
      "learning_rate": 9.355638162993139e-06,
      "loss": 0.2094,
      "step": 1473
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4754518240309573,
      "learning_rate": 9.354619773578088e-06,
      "loss": 0.2364,
      "step": 1474
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5262181155134742,
      "learning_rate": 9.353600635553418e-06,
      "loss": 0.2455,
      "step": 1475
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5429693484426572,
      "learning_rate": 9.352580749094325e-06,
      "loss": 0.2013,
      "step": 1476
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3161141005833452,
      "learning_rate": 9.351560114376142e-06,
      "loss": 0.2227,
      "step": 1477
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5141016840722317,
      "learning_rate": 9.350538731574329e-06,
      "loss": 0.2237,
      "step": 1478
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5502461634569584,
      "learning_rate": 9.349516600864473e-06,
      "loss": 0.2295,
      "step": 1479
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.390849516629579,
      "learning_rate": 9.348493722422288e-06,
      "loss": 0.2222,
      "step": 1480
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.498844828690649,
      "learning_rate": 9.347470096423618e-06,
      "loss": 0.6316,
      "step": 1481
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4966348258346573,
      "learning_rate": 9.346445723044441e-06,
      "loss": 0.2239,
      "step": 1482
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5171528721558565,
      "learning_rate": 9.345420602460856e-06,
      "loss": 0.2239,
      "step": 1483
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.611048982325185,
      "learning_rate": 9.344394734849092e-06,
      "loss": 0.2706,
      "step": 1484
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.4764918852694615,
      "learning_rate": 9.343368120385508e-06,
      "loss": 0.6073,
      "step": 1485
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4137847871790739,
      "learning_rate": 9.342340759246591e-06,
      "loss": 0.2347,
      "step": 1486
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4767279942396654,
      "learning_rate": 9.341312651608957e-06,
      "loss": 0.2431,
      "step": 1487
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5944652292771102,
      "learning_rate": 9.340283797649352e-06,
      "loss": 0.2682,
      "step": 1488
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6494795943156886,
      "learning_rate": 9.339254197544642e-06,
      "loss": 0.2629,
      "step": 1489
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5292870864492731,
      "learning_rate": 9.338223851471833e-06,
      "loss": 0.2418,
      "step": 1490
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5484164484309246,
      "learning_rate": 9.33719275960805e-06,
      "loss": 0.2163,
      "step": 1491
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4155297667888322,
      "learning_rate": 9.33616092213055e-06,
      "loss": 0.2387,
      "step": 1492
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3960425914479093,
      "learning_rate": 9.335128339216719e-06,
      "loss": 0.222,
      "step": 1493
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3958559922053175,
      "learning_rate": 9.33409501104407e-06,
      "loss": 0.1864,
      "step": 1494
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4012746175230952,
      "learning_rate": 9.333060937790243e-06,
      "loss": 0.2316,
      "step": 1495
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4176095220285463,
      "learning_rate": 9.332026119633007e-06,
      "loss": 0.2269,
      "step": 1496
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3869039418483269,
      "learning_rate": 9.33099055675026e-06,
      "loss": 0.2162,
      "step": 1497
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.45653874827404,
      "learning_rate": 9.329954249320028e-06,
      "loss": 0.5781,
      "step": 1498
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5098116539950137,
      "learning_rate": 9.328917197520461e-06,
      "loss": 0.1816,
      "step": 1499
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4792837716384593,
      "learning_rate": 9.327879401529846e-06,
      "loss": 0.2344,
      "step": 1500
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4914804737274145,
      "learning_rate": 9.326840861526587e-06,
      "loss": 0.233,
      "step": 1501
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.587965487698998,
      "learning_rate": 9.325801577689224e-06,
      "loss": 0.2704,
      "step": 1502
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3965081553692547,
      "learning_rate": 9.32476155019642e-06,
      "loss": 0.214,
      "step": 1503
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5037501939953315,
      "learning_rate": 9.323720779226969e-06,
      "loss": 0.2187,
      "step": 1504
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5806717831316401,
      "learning_rate": 9.322679264959789e-06,
      "loss": 0.2401,
      "step": 1505
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.54527952044549,
      "learning_rate": 9.321637007573932e-06,
      "loss": 0.632,
      "step": 1506
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.77494145708196,
      "learning_rate": 9.320594007248573e-06,
      "loss": 0.6175,
      "step": 1507
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3771781745001066,
      "learning_rate": 9.319550264163015e-06,
      "loss": 0.1893,
      "step": 1508
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4116441368131636,
      "learning_rate": 9.318505778496692e-06,
      "loss": 0.2315,
      "step": 1509
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5471666801504722,
      "learning_rate": 9.317460550429159e-06,
      "loss": 0.28,
      "step": 1510
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.380883959055194,
      "learning_rate": 9.316414580140105e-06,
      "loss": 0.5529,
      "step": 1511
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3432307110791593,
      "learning_rate": 9.315367867809346e-06,
      "loss": 0.1955,
      "step": 1512
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4058570169344113,
      "learning_rate": 9.314320413616821e-06,
      "loss": 0.2026,
      "step": 1513
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.340961171443198,
      "learning_rate": 9.313272217742604e-06,
      "loss": 0.6437,
      "step": 1514
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.602282931785258,
      "learning_rate": 9.312223280366886e-06,
      "loss": 0.2283,
      "step": 1515
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.510118710158088,
      "learning_rate": 9.311173601669996e-06,
      "loss": 0.2088,
      "step": 1516
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.137299853165342,
      "learning_rate": 9.310123181832382e-06,
      "loss": 0.6117,
      "step": 1517
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5443852089738435,
      "learning_rate": 9.309072021034629e-06,
      "loss": 0.205,
      "step": 1518
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3146607920268558,
      "learning_rate": 9.308020119457439e-06,
      "loss": 0.1815,
      "step": 1519
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4768548924061409,
      "learning_rate": 9.306967477281648e-06,
      "loss": 0.2182,
      "step": 1520
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4241774773970843,
      "learning_rate": 9.305914094688216e-06,
      "loss": 0.2266,
      "step": 1521
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5070947208639358,
      "learning_rate": 9.304859971858233e-06,
      "loss": 0.2536,
      "step": 1522
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4434681044221427,
      "learning_rate": 9.303805108972916e-06,
      "loss": 0.2249,
      "step": 1523
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4908728882920133,
      "learning_rate": 9.302749506213607e-06,
      "loss": 0.2684,
      "step": 1524
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.700547754339102,
      "learning_rate": 9.301693163761774e-06,
      "loss": 0.6818,
      "step": 1525
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2295647154526341,
      "learning_rate": 9.300636081799017e-06,
      "loss": 0.1793,
      "step": 1526
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.346557546129006,
      "learning_rate": 9.299578260507061e-06,
      "loss": 0.1858,
      "step": 1527
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5742415821098477,
      "learning_rate": 9.298519700067757e-06,
      "loss": 0.2683,
      "step": 1528
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5805493889141613,
      "learning_rate": 9.297460400663083e-06,
      "loss": 0.2167,
      "step": 1529
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5009315965108547,
      "learning_rate": 9.296400362475149e-06,
      "loss": 0.2458,
      "step": 1530
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3688091304552117,
      "learning_rate": 9.29533958568618e-06,
      "loss": 0.2198,
      "step": 1531
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.915658409696545,
      "learning_rate": 9.294278070478544e-06,
      "loss": 0.4802,
      "step": 1532
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.470764954229033,
      "learning_rate": 9.293215817034722e-06,
      "loss": 0.2292,
      "step": 1533
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.004583546404008,
      "learning_rate": 9.29215282553733e-06,
      "loss": 0.5682,
      "step": 1534
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5348302056288932,
      "learning_rate": 9.291089096169109e-06,
      "loss": 0.2727,
      "step": 1535
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6499889131823045,
      "learning_rate": 9.290024629112927e-06,
      "loss": 0.2277,
      "step": 1536
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.557123175894817,
      "learning_rate": 9.288959424551775e-06,
      "loss": 0.2001,
      "step": 1537
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6070841820018125,
      "learning_rate": 9.287893482668774e-06,
      "loss": 0.233,
      "step": 1538
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6586594676770667,
      "learning_rate": 9.286826803647178e-06,
      "loss": 0.284,
      "step": 1539
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3318800137192053,
      "learning_rate": 9.285759387670357e-06,
      "loss": 0.179,
      "step": 1540
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4296834403552743,
      "learning_rate": 9.28469123492181e-06,
      "loss": 0.2567,
      "step": 1541
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5207701617902445,
      "learning_rate": 9.283622345585169e-06,
      "loss": 0.2524,
      "step": 1542
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5534265039951263,
      "learning_rate": 9.282552719844186e-06,
      "loss": 0.3049,
      "step": 1543
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5814830272991194,
      "learning_rate": 9.281482357882743e-06,
      "loss": 0.2122,
      "step": 1544
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.747090764223827,
      "learning_rate": 9.280411259884847e-06,
      "loss": 0.6319,
      "step": 1545
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.455929882145764,
      "learning_rate": 9.279339426034634e-06,
      "loss": 0.2073,
      "step": 1546
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6213833069343417,
      "learning_rate": 9.278266856516363e-06,
      "loss": 0.2886,
      "step": 1547
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.487571687461687,
      "learning_rate": 9.27719355151442e-06,
      "loss": 0.2396,
      "step": 1548
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5400177378701192,
      "learning_rate": 9.276119511213321e-06,
      "loss": 0.2519,
      "step": 1549
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.2470437234772325,
      "learning_rate": 9.275044735797705e-06,
      "loss": 0.6206,
      "step": 1550
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5705277493814536,
      "learning_rate": 9.27396922545234e-06,
      "loss": 0.2346,
      "step": 1551
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5796580250901595,
      "learning_rate": 9.272892980362113e-06,
      "loss": 0.2535,
      "step": 1552
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3676348542212031,
      "learning_rate": 9.27181600071205e-06,
      "loss": 0.2166,
      "step": 1553
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4927594446347667,
      "learning_rate": 9.270738286687293e-06,
      "loss": 0.2226,
      "step": 1554
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3345949915670592,
      "learning_rate": 9.269659838473113e-06,
      "loss": 0.2358,
      "step": 1555
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.352894551729349,
      "learning_rate": 9.26858065625491e-06,
      "loss": 0.3811,
      "step": 1556
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4332890838418346,
      "learning_rate": 9.267500740218205e-06,
      "loss": 0.2053,
      "step": 1557
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4215855925359855,
      "learning_rate": 9.266420090548652e-06,
      "loss": 0.1728,
      "step": 1558
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.576816484875188,
      "learning_rate": 9.265338707432024e-06,
      "loss": 0.4525,
      "step": 1559
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4477220086341447,
      "learning_rate": 9.264256591054222e-06,
      "loss": 0.1841,
      "step": 1560
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4292362785095594,
      "learning_rate": 9.263173741601278e-06,
      "loss": 0.2204,
      "step": 1561
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3364845049010072,
      "learning_rate": 9.262090159259345e-06,
      "loss": 0.2101,
      "step": 1562
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4045837562447072,
      "learning_rate": 9.261005844214703e-06,
      "loss": 0.2591,
      "step": 1563
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4144891441526928,
      "learning_rate": 9.25992079665376e-06,
      "loss": 0.2067,
      "step": 1564
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6605267595983004,
      "learning_rate": 9.258835016763044e-06,
      "loss": 0.256,
      "step": 1565
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5417839503257402,
      "learning_rate": 9.257748504729218e-06,
      "loss": 0.245,
      "step": 1566
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4921484874311794,
      "learning_rate": 9.256661260739065e-06,
      "loss": 0.2308,
      "step": 1567
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.814002333764553,
      "learning_rate": 9.255573284979491e-06,
      "loss": 0.6194,
      "step": 1568
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5929693209468996,
      "learning_rate": 9.254484577637534e-06,
      "loss": 0.2518,
      "step": 1569
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.534066710388744,
      "learning_rate": 9.253395138900359e-06,
      "loss": 0.3021,
      "step": 1570
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.03174975051798,
      "learning_rate": 9.252304968955248e-06,
      "loss": 0.6913,
      "step": 1571
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3815721606166014,
      "learning_rate": 9.251214067989617e-06,
      "loss": 0.1988,
      "step": 1572
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4617601963086437,
      "learning_rate": 9.250122436191002e-06,
      "loss": 0.2376,
      "step": 1573
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5134871682870015,
      "learning_rate": 9.249030073747069e-06,
      "loss": 0.2383,
      "step": 1574
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.652436180490606,
      "learning_rate": 9.247936980845608e-06,
      "loss": 0.2158,
      "step": 1575
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5419883531160097,
      "learning_rate": 9.246843157674535e-06,
      "loss": 0.2074,
      "step": 1576
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.457905842320632,
      "learning_rate": 9.245748604421888e-06,
      "loss": 0.2379,
      "step": 1577
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.096645156340418,
      "learning_rate": 9.244653321275834e-06,
      "loss": 0.6049,
      "step": 1578
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.5539574785538806,
      "learning_rate": 9.243557308424667e-06,
      "loss": 0.6585,
      "step": 1579
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3387065250704742,
      "learning_rate": 9.242460566056803e-06,
      "loss": 0.2143,
      "step": 1580
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5583956360638012,
      "learning_rate": 9.241363094360785e-06,
      "loss": 0.2624,
      "step": 1581
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4143703635093123,
      "learning_rate": 9.240264893525281e-06,
      "loss": 0.2286,
      "step": 1582
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.862411729611918,
      "learning_rate": 9.239165963739084e-06,
      "loss": 0.568,
      "step": 1583
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2593038136158772,
      "learning_rate": 9.238066305191113e-06,
      "loss": 0.2129,
      "step": 1584
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.441928556530492,
      "learning_rate": 9.236965918070413e-06,
      "loss": 0.2403,
      "step": 1585
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4447057534805048,
      "learning_rate": 9.235864802566153e-06,
      "loss": 0.2012,
      "step": 1586
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.488284920470332,
      "learning_rate": 9.234762958867627e-06,
      "loss": 0.2232,
      "step": 1587
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.521729143309394,
      "learning_rate": 9.233660387164254e-06,
      "loss": 0.2383,
      "step": 1588
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.6777027319131803,
      "learning_rate": 9.23255708764558e-06,
      "loss": 0.5361,
      "step": 1589
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4314773048839375,
      "learning_rate": 9.231453060501274e-06,
      "loss": 0.2214,
      "step": 1590
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4865448848077683,
      "learning_rate": 9.230348305921131e-06,
      "loss": 0.202,
      "step": 1591
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6194263176916888,
      "learning_rate": 9.229242824095074e-06,
      "loss": 0.2172,
      "step": 1592
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2835779882903018,
      "learning_rate": 9.228136615213144e-06,
      "loss": 0.1928,
      "step": 1593
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.563281243457181,
      "learning_rate": 9.227029679465515e-06,
      "loss": 0.2733,
      "step": 1594
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3275483862688884,
      "learning_rate": 9.225922017042478e-06,
      "loss": 0.2135,
      "step": 1595
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.667228390042869,
      "learning_rate": 9.224813628134457e-06,
      "loss": 0.5978,
      "step": 1596
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3887356683607317,
      "learning_rate": 9.223704512931994e-06,
      "loss": 0.1867,
      "step": 1597
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.474696364508665,
      "learning_rate": 9.22259467162576e-06,
      "loss": 0.2378,
      "step": 1598
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4259952179170876,
      "learning_rate": 9.221484104406551e-06,
      "loss": 0.2504,
      "step": 1599
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3751095363721848,
      "learning_rate": 9.220372811465282e-06,
      "loss": 0.2298,
      "step": 1600
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4357074116622361,
      "learning_rate": 9.219260792993001e-06,
      "loss": 0.2107,
      "step": 1601
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5432041002445216,
      "learning_rate": 9.218148049180878e-06,
      "loss": 0.2645,
      "step": 1602
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5007792527844421,
      "learning_rate": 9.217034580220203e-06,
      "loss": 0.2,
      "step": 1603
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4708324613418262,
      "learning_rate": 9.215920386302393e-06,
      "loss": 0.2615,
      "step": 1604
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5819068365668694,
      "learning_rate": 9.214805467618996e-06,
      "loss": 0.2434,
      "step": 1605
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4826452416510287,
      "learning_rate": 9.213689824361678e-06,
      "loss": 0.2583,
      "step": 1606
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.515357273945478,
      "learning_rate": 9.212573456722227e-06,
      "loss": 0.2636,
      "step": 1607
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.841457726948671,
      "learning_rate": 9.211456364892566e-06,
      "loss": 0.6983,
      "step": 1608
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.278053578677179,
      "learning_rate": 9.210338549064728e-06,
      "loss": 0.2231,
      "step": 1609
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5031475833414536,
      "learning_rate": 9.209220009430886e-06,
      "loss": 0.2106,
      "step": 1610
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.515573772173353,
      "learning_rate": 9.208100746183327e-06,
      "loss": 0.5256,
      "step": 1611
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4126581521973889,
      "learning_rate": 9.206980759514464e-06,
      "loss": 0.2324,
      "step": 1612
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5225037545376272,
      "learning_rate": 9.205860049616837e-06,
      "loss": 0.2582,
      "step": 1613
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.008208596129366,
      "learning_rate": 9.20473861668311e-06,
      "loss": 0.6456,
      "step": 1614
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.608780241560669,
      "learning_rate": 9.203616460906068e-06,
      "loss": 0.2674,
      "step": 1615
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4440297186465725,
      "learning_rate": 9.202493582478625e-06,
      "loss": 0.241,
      "step": 1616
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.526887861589427,
      "learning_rate": 9.201369981593816e-06,
      "loss": 0.2228,
      "step": 1617
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.474861877346577,
      "learning_rate": 9.200245658444799e-06,
      "loss": 0.2234,
      "step": 1618
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.526398998866433,
      "learning_rate": 9.199120613224862e-06,
      "loss": 0.5742,
      "step": 1619
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4832864720498853,
      "learning_rate": 9.197994846127409e-06,
      "loss": 0.2259,
      "step": 1620
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6240508616051519,
      "learning_rate": 9.196868357345976e-06,
      "loss": 0.2186,
      "step": 1621
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3600757999052193,
      "learning_rate": 9.195741147074218e-06,
      "loss": 0.1982,
      "step": 1622
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.68024318753595,
      "learning_rate": 9.194613215505916e-06,
      "loss": 0.5705,
      "step": 1623
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6545810834201748,
      "learning_rate": 9.193484562834973e-06,
      "loss": 0.2831,
      "step": 1624
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4696006828674408,
      "learning_rate": 9.192355189255418e-06,
      "loss": 0.1959,
      "step": 1625
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.962184540786142,
      "learning_rate": 9.191225094961407e-06,
      "loss": 0.6522,
      "step": 1626
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5020183187755998,
      "learning_rate": 9.19009428014721e-06,
      "loss": 0.2068,
      "step": 1627
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.469135033107731,
      "learning_rate": 9.188962745007233e-06,
      "loss": 0.5591,
      "step": 1628
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5283635934883395,
      "learning_rate": 9.187830489735996e-06,
      "loss": 0.2334,
      "step": 1629
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5060701400422996,
      "learning_rate": 9.18669751452815e-06,
      "loss": 0.2518,
      "step": 1630
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5473392068145584,
      "learning_rate": 9.185563819578462e-06,
      "loss": 0.2488,
      "step": 1631
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6057955406942623,
      "learning_rate": 9.184429405081832e-06,
      "loss": 0.2529,
      "step": 1632
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5137355426002628,
      "learning_rate": 9.183294271233278e-06,
      "loss": 0.2209,
      "step": 1633
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4390023109736234,
      "learning_rate": 9.18215841822794e-06,
      "loss": 0.216,
      "step": 1634
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2774367567755425,
      "learning_rate": 9.181021846261088e-06,
      "loss": 0.1714,
      "step": 1635
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3911958110564755,
      "learning_rate": 9.17988455552811e-06,
      "loss": 0.2515,
      "step": 1636
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5699198178463594,
      "learning_rate": 9.178746546224517e-06,
      "loss": 0.2589,
      "step": 1637
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.802313868235618,
      "learning_rate": 9.177607818545951e-06,
      "loss": 0.5548,
      "step": 1638
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4940946121887626,
      "learning_rate": 9.176468372688168e-06,
      "loss": 0.2514,
      "step": 1639
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3749595792074554,
      "learning_rate": 9.175328208847056e-06,
      "loss": 0.1986,
      "step": 1640
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6943042286687702,
      "learning_rate": 9.17418732721862e-06,
      "loss": 0.2082,
      "step": 1641
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.979624528965953,
      "learning_rate": 9.17304572799899e-06,
      "loss": 0.5943,
      "step": 1642
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5814221462342855,
      "learning_rate": 9.171903411384422e-06,
      "loss": 0.2858,
      "step": 1643
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6537215301501216,
      "learning_rate": 9.170760377571293e-06,
      "loss": 0.2055,
      "step": 1644
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2309266964976135,
      "learning_rate": 9.169616626756103e-06,
      "loss": 0.1456,
      "step": 1645
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4113437521691694,
      "learning_rate": 9.168472159135477e-06,
      "loss": 0.2296,
      "step": 1646
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3014913584054797,
      "learning_rate": 9.167326974906161e-06,
      "loss": 0.2299,
      "step": 1647
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3161169675609818,
      "learning_rate": 9.166181074265027e-06,
      "loss": 0.1961,
      "step": 1648
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2766164548607286,
      "learning_rate": 9.165034457409066e-06,
      "loss": 0.1761,
      "step": 1649
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.468889219332883,
      "learning_rate": 9.163887124535398e-06,
      "loss": 0.1935,
      "step": 1650
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.764880775358446,
      "learning_rate": 9.162739075841263e-06,
      "loss": 0.7554,
      "step": 1651
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3840083946960475,
      "learning_rate": 9.16159031152402e-06,
      "loss": 0.215,
      "step": 1652
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.301367701489058,
      "learning_rate": 9.160440831781157e-06,
      "loss": 0.687,
      "step": 1653
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3214812040420576,
      "learning_rate": 9.159290636810284e-06,
      "loss": 0.1636,
      "step": 1654
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.808402615099616,
      "learning_rate": 9.158139726809132e-06,
      "loss": 0.2893,
      "step": 1655
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.789591375673088,
      "learning_rate": 9.156988101975555e-06,
      "loss": 0.6912,
      "step": 1656
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.22914297821385,
      "learning_rate": 9.155835762507532e-06,
      "loss": 0.6628,
      "step": 1657
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4217621983182793,
      "learning_rate": 9.154682708603162e-06,
      "loss": 0.24,
      "step": 1658
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6516656211815053,
      "learning_rate": 9.153528940460669e-06,
      "loss": 0.2969,
      "step": 1659
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7166325596088654,
      "learning_rate": 9.152374458278402e-06,
      "loss": 0.2924,
      "step": 1660
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4248099650329804,
      "learning_rate": 9.151219262254824e-06,
      "loss": 0.2264,
      "step": 1661
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4808174914905086,
      "learning_rate": 9.150063352588531e-06,
      "loss": 0.2124,
      "step": 1662
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.492816140118318,
      "learning_rate": 9.148906729478236e-06,
      "loss": 0.6387,
      "step": 1663
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3841756659057083,
      "learning_rate": 9.147749393122776e-06,
      "loss": 0.1813,
      "step": 1664
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3376682620384468,
      "learning_rate": 9.14659134372111e-06,
      "loss": 0.2392,
      "step": 1665
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5884693618029144,
      "learning_rate": 9.145432581472323e-06,
      "loss": 0.261,
      "step": 1666
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.52736904754684,
      "learning_rate": 9.144273106575614e-06,
      "loss": 0.2047,
      "step": 1667
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6635138710691346,
      "learning_rate": 9.143112919230314e-06,
      "loss": 0.2454,
      "step": 1668
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5360560941906007,
      "learning_rate": 9.141952019635874e-06,
      "loss": 0.2489,
      "step": 1669
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6002575580814953,
      "learning_rate": 9.140790407991862e-06,
      "loss": 0.2625,
      "step": 1670
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5700654093963524,
      "learning_rate": 9.139628084497975e-06,
      "loss": 0.2633,
      "step": 1671
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4582761181823638,
      "learning_rate": 9.138465049354031e-06,
      "loss": 0.2117,
      "step": 1672
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4774412186924963,
      "learning_rate": 9.137301302759968e-06,
      "loss": 0.202,
      "step": 1673
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5162234491692794,
      "learning_rate": 9.136136844915847e-06,
      "loss": 0.2687,
      "step": 1674
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6320881424554634,
      "learning_rate": 9.134971676021851e-06,
      "loss": 0.214,
      "step": 1675
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.462053457654976,
      "learning_rate": 9.133805796278288e-06,
      "loss": 0.1709,
      "step": 1676
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.399969052159625,
      "learning_rate": 9.132639205885584e-06,
      "loss": 0.2293,
      "step": 1677
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4939411325239766,
      "learning_rate": 9.131471905044294e-06,
      "loss": 0.1966,
      "step": 1678
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6332468857277167,
      "learning_rate": 9.130303893955084e-06,
      "loss": 0.2241,
      "step": 1679
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.094344149586073,
      "learning_rate": 9.129135172818754e-06,
      "loss": 0.7489,
      "step": 1680
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.545889464915774,
      "learning_rate": 9.127965741836218e-06,
      "loss": 0.5627,
      "step": 1681
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4977392883601615,
      "learning_rate": 9.126795601208516e-06,
      "loss": 0.2025,
      "step": 1682
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.478608992042656,
      "learning_rate": 9.125624751136809e-06,
      "loss": 0.2878,
      "step": 1683
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.500433688986048,
      "learning_rate": 9.124453191822376e-06,
      "loss": 0.2816,
      "step": 1684
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4452242354489349,
      "learning_rate": 9.123280923466627e-06,
      "loss": 0.2197,
      "step": 1685
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5714101232604034,
      "learning_rate": 9.122107946271086e-06,
      "loss": 0.2011,
      "step": 1686
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5324911355550852,
      "learning_rate": 9.1209342604374e-06,
      "loss": 0.2305,
      "step": 1687
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.797008526251933,
      "learning_rate": 9.119759866167342e-06,
      "loss": 0.2163,
      "step": 1688
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7370365347337815,
      "learning_rate": 9.118584763662803e-06,
      "loss": 0.2238,
      "step": 1689
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.662007417485374,
      "learning_rate": 9.117408953125794e-06,
      "loss": 0.6572,
      "step": 1690
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4566268203980643,
      "learning_rate": 9.116232434758456e-06,
      "loss": 0.2472,
      "step": 1691
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5952683091569144,
      "learning_rate": 9.115055208763042e-06,
      "loss": 0.2662,
      "step": 1692
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8819166291649219,
      "learning_rate": 9.113877275341932e-06,
      "loss": 0.2662,
      "step": 1693
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.484943830598999,
      "learning_rate": 9.112698634697625e-06,
      "loss": 0.2365,
      "step": 1694
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5142418368477786,
      "learning_rate": 9.111519287032745e-06,
      "loss": 0.2599,
      "step": 1695
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.280865535445386,
      "learning_rate": 9.110339232550034e-06,
      "loss": 0.2071,
      "step": 1696
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.473792020764006,
      "learning_rate": 9.109158471452358e-06,
      "loss": 0.2267,
      "step": 1697
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5714626663149631,
      "learning_rate": 9.107977003942703e-06,
      "loss": 0.2624,
      "step": 1698
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.321673514300932,
      "learning_rate": 9.106794830224179e-06,
      "loss": 0.5506,
      "step": 1699
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4464801167833885,
      "learning_rate": 9.105611950500012e-06,
      "loss": 0.2181,
      "step": 1700
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.6262161671383915,
      "learning_rate": 9.104428364973555e-06,
      "loss": 0.635,
      "step": 1701
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4346695115043593,
      "learning_rate": 9.10324407384828e-06,
      "loss": 0.2295,
      "step": 1702
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5946717884680248,
      "learning_rate": 9.102059077327779e-06,
      "loss": 0.2208,
      "step": 1703
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4235714086019953,
      "learning_rate": 9.100873375615767e-06,
      "loss": 0.2354,
      "step": 1704
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3241692041025177,
      "learning_rate": 9.099686968916081e-06,
      "loss": 0.2212,
      "step": 1705
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3998807815494634,
      "learning_rate": 9.098499857432677e-06,
      "loss": 0.1697,
      "step": 1706
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5595615787534098,
      "learning_rate": 9.097312041369634e-06,
      "loss": 0.252,
      "step": 1707
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6569340425466743,
      "learning_rate": 9.09612352093115e-06,
      "loss": 0.2722,
      "step": 1708
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.657932653107319,
      "learning_rate": 9.094934296321549e-06,
      "loss": 0.2457,
      "step": 1709
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4913546986092743,
      "learning_rate": 9.093744367745267e-06,
      "loss": 0.2211,
      "step": 1710
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6436930223058805,
      "learning_rate": 9.092553735406871e-06,
      "loss": 0.2352,
      "step": 1711
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5245997996238947,
      "learning_rate": 9.091362399511043e-06,
      "loss": 0.2613,
      "step": 1712
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3776000968264648,
      "learning_rate": 9.090170360262587e-06,
      "loss": 0.1633,
      "step": 1713
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.661207550096438,
      "learning_rate": 9.088977617866428e-06,
      "loss": 0.2238,
      "step": 1714
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3072492721014437,
      "learning_rate": 9.087784172527614e-06,
      "loss": 0.1952,
      "step": 1715
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4298818143211793,
      "learning_rate": 9.086590024451312e-06,
      "loss": 0.219,
      "step": 1716
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.1380089624674765,
      "learning_rate": 9.085395173842807e-06,
      "loss": 0.5475,
      "step": 1717
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4183351572277543,
      "learning_rate": 9.08419962090751e-06,
      "loss": 0.2102,
      "step": 1718
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4027360761041336,
      "learning_rate": 9.08300336585095e-06,
      "loss": 0.2103,
      "step": 1719
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5986245458603934,
      "learning_rate": 9.081806408878778e-06,
      "loss": 0.2537,
      "step": 1720
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.425123079653894,
      "learning_rate": 9.080608750196764e-06,
      "loss": 0.2169,
      "step": 1721
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6287703976054375,
      "learning_rate": 9.079410390010798e-06,
      "loss": 0.2138,
      "step": 1722
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4187939111090713,
      "learning_rate": 9.078211328526896e-06,
      "loss": 0.2363,
      "step": 1723
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.643544068624105,
      "learning_rate": 9.077011565951186e-06,
      "loss": 0.2355,
      "step": 1724
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2944142722704657,
      "learning_rate": 9.075811102489923e-06,
      "loss": 0.2437,
      "step": 1725
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5557262791513362,
      "learning_rate": 9.074609938349481e-06,
      "loss": 0.2217,
      "step": 1726
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.355502102429379,
      "learning_rate": 9.073408073736355e-06,
      "loss": 0.5966,
      "step": 1727
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3687400115521633,
      "learning_rate": 9.072205508857155e-06,
      "loss": 0.2089,
      "step": 1728
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.423053230510977,
      "learning_rate": 9.071002243918621e-06,
      "loss": 0.1989,
      "step": 1729
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.408488643664958,
      "learning_rate": 9.069798279127606e-06,
      "loss": 0.2155,
      "step": 1730
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5015734451265474,
      "learning_rate": 9.068593614691086e-06,
      "loss": 0.2562,
      "step": 1731
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5810370368716375,
      "learning_rate": 9.067388250816155e-06,
      "loss": 0.2135,
      "step": 1732
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3995136636050942,
      "learning_rate": 9.066182187710032e-06,
      "loss": 0.2204,
      "step": 1733
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4118521472116734,
      "learning_rate": 9.064975425580051e-06,
      "loss": 0.1899,
      "step": 1734
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.438486836080634,
      "learning_rate": 9.06376796463367e-06,
      "loss": 0.6314,
      "step": 1735
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5151089977847356,
      "learning_rate": 9.062559805078463e-06,
      "loss": 0.2102,
      "step": 1736
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4574057307332804,
      "learning_rate": 9.061350947122129e-06,
      "loss": 0.2385,
      "step": 1737
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5179960515198254,
      "learning_rate": 9.060141390972486e-06,
      "loss": 0.2451,
      "step": 1738
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.832007389248238,
      "learning_rate": 9.058931136837465e-06,
      "loss": 0.6404,
      "step": 1739
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.471270068359665,
      "learning_rate": 9.05772018492513e-06,
      "loss": 0.1918,
      "step": 1740
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4946674104692128,
      "learning_rate": 9.056508535443652e-06,
      "loss": 0.2095,
      "step": 1741
    },
    {
      "epoch": 0.22,
      "grad_norm": 8.01612031113401,
      "learning_rate": 9.05529618860133e-06,
      "loss": 0.6925,
      "step": 1742
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4330073735537243,
      "learning_rate": 9.05408314460658e-06,
      "loss": 0.2112,
      "step": 1743
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5061691025973858,
      "learning_rate": 9.052869403667938e-06,
      "loss": 0.2243,
      "step": 1744
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3285890462048395,
      "learning_rate": 9.051654965994062e-06,
      "loss": 0.1707,
      "step": 1745
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4359799387243122,
      "learning_rate": 9.050439831793726e-06,
      "loss": 0.2208,
      "step": 1746
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3010194528974575,
      "learning_rate": 9.049224001275825e-06,
      "loss": 0.202,
      "step": 1747
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4059528169035729,
      "learning_rate": 9.048007474649377e-06,
      "loss": 0.2204,
      "step": 1748
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4570571684873699,
      "learning_rate": 9.046790252123514e-06,
      "loss": 0.2119,
      "step": 1749
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3627423862783274,
      "learning_rate": 9.045572333907495e-06,
      "loss": 0.2206,
      "step": 1750
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6279272199650794,
      "learning_rate": 9.04435372021069e-06,
      "loss": 0.2611,
      "step": 1751
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6019228796047098,
      "learning_rate": 9.043134411242593e-06,
      "loss": 0.1547,
      "step": 1752
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.622460474830794,
      "learning_rate": 9.041914407212821e-06,
      "loss": 0.2736,
      "step": 1753
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4480365183146262,
      "learning_rate": 9.040693708331103e-06,
      "loss": 0.2386,
      "step": 1754
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.823697950265952,
      "learning_rate": 9.039472314807293e-06,
      "loss": 0.5323,
      "step": 1755
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4241444929684122,
      "learning_rate": 9.038250226851362e-06,
      "loss": 0.2167,
      "step": 1756
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.786970419370749,
      "learning_rate": 9.037027444673402e-06,
      "loss": 0.5895,
      "step": 1757
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2481511573573947,
      "learning_rate": 9.035803968483625e-06,
      "loss": 0.1519,
      "step": 1758
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.440229008038068,
      "learning_rate": 9.034579798492356e-06,
      "loss": 0.6276,
      "step": 1759
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5171564357742091,
      "learning_rate": 9.033354934910049e-06,
      "loss": 0.2865,
      "step": 1760
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.662459056943299,
      "learning_rate": 9.032129377947267e-06,
      "loss": 0.1987,
      "step": 1761
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3978139633443432,
      "learning_rate": 9.030903127814704e-06,
      "loss": 0.1845,
      "step": 1762
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4627527907441966,
      "learning_rate": 9.029676184723161e-06,
      "loss": 0.2212,
      "step": 1763
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3095055354935703,
      "learning_rate": 9.028448548883566e-06,
      "loss": 0.1689,
      "step": 1764
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4011278706995878,
      "learning_rate": 9.027220220506964e-06,
      "loss": 0.2463,
      "step": 1765
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5360594300710577,
      "learning_rate": 9.025991199804518e-06,
      "loss": 0.2483,
      "step": 1766
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.369593605715067,
      "learning_rate": 9.024761486987512e-06,
      "loss": 0.471,
      "step": 1767
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5084454842687007,
      "learning_rate": 9.023531082267347e-06,
      "loss": 0.1941,
      "step": 1768
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.23426808959054,
      "learning_rate": 9.022299985855544e-06,
      "loss": 0.7804,
      "step": 1769
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5198886221506356,
      "learning_rate": 9.021068197963744e-06,
      "loss": 0.2547,
      "step": 1770
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.345297989898179,
      "learning_rate": 9.019835718803704e-06,
      "loss": 0.1763,
      "step": 1771
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.511987270916311,
      "learning_rate": 9.018602548587303e-06,
      "loss": 0.2202,
      "step": 1772
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5215696975389759,
      "learning_rate": 9.017368687526535e-06,
      "loss": 0.226,
      "step": 1773
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5080085972263138,
      "learning_rate": 9.016134135833517e-06,
      "loss": 0.2227,
      "step": 1774
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4859123682096926,
      "learning_rate": 9.01489889372048e-06,
      "loss": 0.2253,
      "step": 1775
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5960846385690377,
      "learning_rate": 9.01366296139978e-06,
      "loss": 0.2491,
      "step": 1776
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5125330574922506,
      "learning_rate": 9.012426339083887e-06,
      "loss": 0.2178,
      "step": 1777
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3160438230072058,
      "learning_rate": 9.01118902698539e-06,
      "loss": 0.2165,
      "step": 1778
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.328429769290293,
      "learning_rate": 9.009951025316998e-06,
      "loss": 0.6431,
      "step": 1779
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.367768959654276,
      "learning_rate": 9.008712334291536e-06,
      "loss": 0.6643,
      "step": 1780
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.535467308499508,
      "learning_rate": 9.007472954121952e-06,
      "loss": 0.6005,
      "step": 1781
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.547841175703792,
      "learning_rate": 9.006232885021309e-06,
      "loss": 0.2388,
      "step": 1782
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3812911307063387,
      "learning_rate": 9.004992127202787e-06,
      "loss": 0.223,
      "step": 1783
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3615296928364058,
      "learning_rate": 9.00375068087969e-06,
      "loss": 0.2297,
      "step": 1784
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4750621411114349,
      "learning_rate": 9.002508546265433e-06,
      "loss": 0.2536,
      "step": 1785
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.503738667742806,
      "learning_rate": 9.001265723573559e-06,
      "loss": 0.2319,
      "step": 1786
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4504242994282897,
      "learning_rate": 9.00002221301772e-06,
      "loss": 0.2368,
      "step": 1787
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4982014272392215,
      "learning_rate": 8.998778014811688e-06,
      "loss": 0.2295,
      "step": 1788
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4758416807910857,
      "learning_rate": 8.99753312916936e-06,
      "loss": 0.2148,
      "step": 1789
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4411700742672309,
      "learning_rate": 8.996287556304743e-06,
      "loss": 0.2357,
      "step": 1790
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3770557341225886,
      "learning_rate": 8.995041296431965e-06,
      "loss": 0.2212,
      "step": 1791
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5450387644921484,
      "learning_rate": 8.993794349765275e-06,
      "loss": 0.2715,
      "step": 1792
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3388861201512976,
      "learning_rate": 8.992546716519034e-06,
      "loss": 0.2266,
      "step": 1793
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.9806858371089424,
      "learning_rate": 8.99129839690773e-06,
      "loss": 0.5166,
      "step": 1794
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4282329679744876,
      "learning_rate": 8.99004939114596e-06,
      "loss": 0.2477,
      "step": 1795
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4238164441773724,
      "learning_rate": 8.988799699448442e-06,
      "loss": 0.199,
      "step": 1796
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.608498585768861,
      "learning_rate": 8.987549322030013e-06,
      "loss": 0.6469,
      "step": 1797
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5908983292536045,
      "learning_rate": 8.986298259105629e-06,
      "loss": 0.2905,
      "step": 1798
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.186047870261879,
      "learning_rate": 8.98504651089036e-06,
      "loss": 0.6854,
      "step": 1799
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.910149069913866,
      "learning_rate": 8.983794077599398e-06,
      "loss": 0.5503,
      "step": 1800
    },
    {
      "epoch": 0.23,
      "grad_norm": 12.707370994347693,
      "learning_rate": 8.98254095944805e-06,
      "loss": 0.6024,
      "step": 1801
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.44155252948506,
      "learning_rate": 8.981287156651741e-06,
      "loss": 0.211,
      "step": 1802
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3662582606035418,
      "learning_rate": 8.980032669426015e-06,
      "loss": 0.2473,
      "step": 1803
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4315381491471921,
      "learning_rate": 8.978777497986533e-06,
      "loss": 0.2401,
      "step": 1804
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4519690948515138,
      "learning_rate": 8.977521642549073e-06,
      "loss": 0.2187,
      "step": 1805
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3199229338332958,
      "learning_rate": 8.97626510332953e-06,
      "loss": 0.1738,
      "step": 1806
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5865148272540366,
      "learning_rate": 8.97500788054392e-06,
      "loss": 0.2451,
      "step": 1807
    },
    {
      "epoch": 0.23,
      "grad_norm": 9.439608557088182,
      "learning_rate": 8.973749974408375e-06,
      "loss": 0.6193,
      "step": 1808
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3911457486321601,
      "learning_rate": 8.972491385139138e-06,
      "loss": 0.2144,
      "step": 1809
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.500463038649748,
      "learning_rate": 8.971232112952581e-06,
      "loss": 0.2327,
      "step": 1810
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.401334600313825,
      "learning_rate": 8.969972158065185e-06,
      "loss": 0.2197,
      "step": 1811
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.305136703474746,
      "learning_rate": 8.968711520693551e-06,
      "loss": 0.5804,
      "step": 1812
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4430662276340802,
      "learning_rate": 8.967450201054397e-06,
      "loss": 0.2308,
      "step": 1813
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.410317838100431,
      "learning_rate": 8.96618819936456e-06,
      "loss": 0.2191,
      "step": 1814
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.393828869576145,
      "learning_rate": 8.964925515840991e-06,
      "loss": 0.2396,
      "step": 1815
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3668392414190373,
      "learning_rate": 8.963662150700761e-06,
      "loss": 0.2154,
      "step": 1816
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8509999114479767,
      "learning_rate": 8.962398104161055e-06,
      "loss": 0.2259,
      "step": 1817
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.332922738957645,
      "learning_rate": 8.96113337643918e-06,
      "loss": 0.184,
      "step": 1818
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.235085375773179,
      "learning_rate": 8.959867967752556e-06,
      "loss": 0.206,
      "step": 1819
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6049816118179148,
      "learning_rate": 8.95860187831872e-06,
      "loss": 0.2516,
      "step": 1820
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4745603061272454,
      "learning_rate": 8.957335108355332e-06,
      "loss": 0.2154,
      "step": 1821
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.860816639547082,
      "learning_rate": 8.956067658080158e-06,
      "loss": 0.5436,
      "step": 1822
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5455287431606184,
      "learning_rate": 8.954799527711094e-06,
      "loss": 0.2588,
      "step": 1823
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4858789324646253,
      "learning_rate": 8.953530717466143e-06,
      "loss": 0.2563,
      "step": 1824
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7192355464522713,
      "learning_rate": 8.952261227563427e-06,
      "loss": 0.2846,
      "step": 1825
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7116821742117028,
      "learning_rate": 8.950991058221187e-06,
      "loss": 0.2307,
      "step": 1826
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.555944545736356,
      "learning_rate": 8.949720209657781e-06,
      "loss": 0.211,
      "step": 1827
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5307378904184052,
      "learning_rate": 8.948448682091684e-06,
      "loss": 0.2304,
      "step": 1828
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6097341341839546,
      "learning_rate": 8.947176475741482e-06,
      "loss": 0.2457,
      "step": 1829
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5050658813622448,
      "learning_rate": 8.945903590825884e-06,
      "loss": 0.221,
      "step": 1830
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5510403445407908,
      "learning_rate": 8.944630027563714e-06,
      "loss": 0.2249,
      "step": 1831
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.511224905187395,
      "learning_rate": 8.943355786173913e-06,
      "loss": 0.2512,
      "step": 1832
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4099341473808376,
      "learning_rate": 8.942080866875538e-06,
      "loss": 0.2078,
      "step": 1833
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.781103301969708,
      "learning_rate": 8.94080526988776e-06,
      "loss": 0.6113,
      "step": 1834
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4474763200391199,
      "learning_rate": 8.93952899542987e-06,
      "loss": 0.2353,
      "step": 1835
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.493353071573445,
      "learning_rate": 8.938252043721275e-06,
      "loss": 0.2129,
      "step": 1836
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6287069967670724,
      "learning_rate": 8.936974414981498e-06,
      "loss": 0.2398,
      "step": 1837
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.667929332810169,
      "learning_rate": 8.935696109430178e-06,
      "loss": 0.2009,
      "step": 1838
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8507343947902144,
      "learning_rate": 8.93441712728707e-06,
      "loss": 0.2971,
      "step": 1839
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3190129466767806,
      "learning_rate": 8.933137468772047e-06,
      "loss": 0.1935,
      "step": 1840
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5027347447649542,
      "learning_rate": 8.931857134105093e-06,
      "loss": 0.2496,
      "step": 1841
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4091278272403103,
      "learning_rate": 8.930576123506318e-06,
      "loss": 0.2067,
      "step": 1842
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.332621272821163,
      "learning_rate": 8.929294437195942e-06,
      "loss": 0.2048,
      "step": 1843
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.471875279693911,
      "learning_rate": 8.928012075394297e-06,
      "loss": 0.2316,
      "step": 1844
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3448955633238435,
      "learning_rate": 8.92672903832184e-06,
      "loss": 0.1814,
      "step": 1845
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.787183450074027,
      "learning_rate": 8.925445326199138e-06,
      "loss": 0.7159,
      "step": 1846
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5266009426505274,
      "learning_rate": 8.924160939246877e-06,
      "loss": 0.232,
      "step": 1847
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6831774659372443,
      "learning_rate": 8.922875877685859e-06,
      "loss": 0.2044,
      "step": 1848
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3162366693451255,
      "learning_rate": 8.921590141737e-06,
      "loss": 0.2327,
      "step": 1849
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.375017707298363,
      "learning_rate": 8.920303731621332e-06,
      "loss": 0.625,
      "step": 1850
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4982584560249796,
      "learning_rate": 8.919016647560003e-06,
      "loss": 0.2355,
      "step": 1851
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4907350784311642,
      "learning_rate": 8.91772888977428e-06,
      "loss": 0.2169,
      "step": 1852
    },
    {
      "epoch": 0.24,
      "grad_norm": 9.075113283387287,
      "learning_rate": 8.916440458485542e-06,
      "loss": 0.6899,
      "step": 1853
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.10692170365634,
      "learning_rate": 8.915151353915288e-06,
      "loss": 0.52,
      "step": 1854
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.257852889235576,
      "learning_rate": 8.913861576285126e-06,
      "loss": 0.5467,
      "step": 1855
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.586052250765536,
      "learning_rate": 8.912571125816787e-06,
      "loss": 0.2745,
      "step": 1856
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4828467177795972,
      "learning_rate": 8.911280002732112e-06,
      "loss": 0.2213,
      "step": 1857
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.24595941021538,
      "learning_rate": 8.909988207253062e-06,
      "loss": 0.6555,
      "step": 1858
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3614941819166655,
      "learning_rate": 8.90869573960171e-06,
      "loss": 0.2119,
      "step": 1859
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4395142541310608,
      "learning_rate": 8.907402600000248e-06,
      "loss": 0.2309,
      "step": 1860
    },
    {
      "epoch": 0.24,
      "grad_norm": 16.183711763926116,
      "learning_rate": 8.906108788670979e-06,
      "loss": 0.6449,
      "step": 1861
    },
    {
      "epoch": 0.24,
      "grad_norm": 17.644422933711898,
      "learning_rate": 8.904814305836326e-06,
      "loss": 0.6332,
      "step": 1862
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.924430319910763,
      "learning_rate": 8.903519151718826e-06,
      "loss": 0.5674,
      "step": 1863
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.526893104350717,
      "learning_rate": 8.90222332654113e-06,
      "loss": 0.2153,
      "step": 1864
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.409466032756542,
      "learning_rate": 8.900926830526006e-06,
      "loss": 0.2475,
      "step": 1865
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5407545783046601,
      "learning_rate": 8.899629663896336e-06,
      "loss": 0.2556,
      "step": 1866
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2758838015795575,
      "learning_rate": 8.898331826875119e-06,
      "loss": 0.1975,
      "step": 1867
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3299962024034824,
      "learning_rate": 8.897033319685466e-06,
      "loss": 0.2176,
      "step": 1868
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6319013175483355,
      "learning_rate": 8.89573414255061e-06,
      "loss": 0.2206,
      "step": 1869
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4960670019525462,
      "learning_rate": 8.89443429569389e-06,
      "loss": 0.2213,
      "step": 1870
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5542017044563787,
      "learning_rate": 8.893133779338765e-06,
      "loss": 0.1697,
      "step": 1871
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.8688953385423455,
      "learning_rate": 8.89183259370881e-06,
      "loss": 0.624,
      "step": 1872
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5047538383536425,
      "learning_rate": 8.890530739027718e-06,
      "loss": 0.211,
      "step": 1873
    },
    {
      "epoch": 0.24,
      "grad_norm": 10.91591181077284,
      "learning_rate": 8.889228215519286e-06,
      "loss": 0.5729,
      "step": 1874
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6450192058018374,
      "learning_rate": 8.887925023407437e-06,
      "loss": 0.2395,
      "step": 1875
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.401317830319315,
      "learning_rate": 8.886621162916204e-06,
      "loss": 0.1737,
      "step": 1876
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4531899387693308,
      "learning_rate": 8.885316634269735e-06,
      "loss": 0.2278,
      "step": 1877
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5605736326157424,
      "learning_rate": 8.884011437692295e-06,
      "loss": 0.2352,
      "step": 1878
    },
    {
      "epoch": 0.24,
      "grad_norm": 22.35118361363606,
      "learning_rate": 8.882705573408258e-06,
      "loss": 0.6808,
      "step": 1879
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5191852285048602,
      "learning_rate": 8.881399041642125e-06,
      "loss": 0.2624,
      "step": 1880
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4966291588163427,
      "learning_rate": 8.880091842618498e-06,
      "loss": 0.2337,
      "step": 1881
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.460990060654416,
      "learning_rate": 8.878783976562102e-06,
      "loss": 0.2311,
      "step": 1882
    },
    {
      "epoch": 0.24,
      "grad_norm": 14.420220083504136,
      "learning_rate": 8.877475443697773e-06,
      "loss": 0.5483,
      "step": 1883
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3202097685331051,
      "learning_rate": 8.876166244250463e-06,
      "loss": 0.2351,
      "step": 1884
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6475151443160228,
      "learning_rate": 8.87485637844524e-06,
      "loss": 0.2502,
      "step": 1885
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5740182286701323,
      "learning_rate": 8.873545846507286e-06,
      "loss": 0.2221,
      "step": 1886
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4667567832802841,
      "learning_rate": 8.872234648661893e-06,
      "loss": 0.2178,
      "step": 1887
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5522318763764102,
      "learning_rate": 8.870922785134473e-06,
      "loss": 0.27,
      "step": 1888
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5855814755270792,
      "learning_rate": 8.869610256150552e-06,
      "loss": 0.239,
      "step": 1889
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3644694396902632,
      "learning_rate": 8.868297061935767e-06,
      "loss": 0.1867,
      "step": 1890
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.474142504175817,
      "learning_rate": 8.86698320271587e-06,
      "loss": 0.2229,
      "step": 1891
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3928555985162692,
      "learning_rate": 8.865668678716734e-06,
      "loss": 0.2212,
      "step": 1892
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4514561356685294,
      "learning_rate": 8.864353490164335e-06,
      "loss": 0.1948,
      "step": 1893
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.504562949360884,
      "learning_rate": 8.863037637284773e-06,
      "loss": 0.2025,
      "step": 1894
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5376315932259204,
      "learning_rate": 8.861721120304257e-06,
      "loss": 0.2235,
      "step": 1895
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4415029004566484,
      "learning_rate": 8.86040393944911e-06,
      "loss": 0.2133,
      "step": 1896
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4856173247236526,
      "learning_rate": 8.859086094945773e-06,
      "loss": 0.2518,
      "step": 1897
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5680172553754044,
      "learning_rate": 8.857767587020798e-06,
      "loss": 0.2856,
      "step": 1898
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.707025858653054,
      "learning_rate": 8.856448415900851e-06,
      "loss": 0.5522,
      "step": 1899
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4792440916772083,
      "learning_rate": 8.855128581812714e-06,
      "loss": 0.2138,
      "step": 1900
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4084346731521333,
      "learning_rate": 8.853808084983282e-06,
      "loss": 0.1722,
      "step": 1901
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5292225392398675,
      "learning_rate": 8.852486925639562e-06,
      "loss": 0.2491,
      "step": 1902
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5213349995161998,
      "learning_rate": 8.851165104008678e-06,
      "loss": 0.2629,
      "step": 1903
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7220787651746108,
      "learning_rate": 8.849842620317865e-06,
      "loss": 0.2763,
      "step": 1904
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4009579169565847,
      "learning_rate": 8.848519474794476e-06,
      "loss": 0.2447,
      "step": 1905
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.85300109959866,
      "learning_rate": 8.847195667665974e-06,
      "loss": 0.7866,
      "step": 1906
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.283212845742842,
      "learning_rate": 8.845871199159935e-06,
      "loss": 0.6318,
      "step": 1907
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.6512318093892855,
      "learning_rate": 8.844546069504051e-06,
      "loss": 0.6474,
      "step": 1908
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6470299342590151,
      "learning_rate": 8.843220278926128e-06,
      "loss": 0.2393,
      "step": 1909
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5743853124265836,
      "learning_rate": 8.841893827654087e-06,
      "loss": 0.2435,
      "step": 1910
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.355538339005477,
      "learning_rate": 8.840566715915955e-06,
      "loss": 0.2072,
      "step": 1911
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3623951732494404,
      "learning_rate": 8.839238943939883e-06,
      "loss": 0.1647,
      "step": 1912
    },
    {
      "epoch": 0.24,
      "grad_norm": 17.355764841637605,
      "learning_rate": 8.837910511954128e-06,
      "loss": 0.5768,
      "step": 1913
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3201066567158593,
      "learning_rate": 8.836581420187062e-06,
      "loss": 0.2023,
      "step": 1914
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.427500284452338,
      "learning_rate": 8.835251668867172e-06,
      "loss": 0.6825,
      "step": 1915
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6403859220580959,
      "learning_rate": 8.833921258223059e-06,
      "loss": 0.2525,
      "step": 1916
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6131340359178368,
      "learning_rate": 8.832590188483437e-06,
      "loss": 0.2781,
      "step": 1917
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5141809700618418,
      "learning_rate": 8.831258459877128e-06,
      "loss": 0.2498,
      "step": 1918
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4151832705166627,
      "learning_rate": 8.829926072633075e-06,
      "loss": 0.2137,
      "step": 1919
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6440876400654634,
      "learning_rate": 8.828593026980328e-06,
      "loss": 0.2053,
      "step": 1920
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5618157473452834,
      "learning_rate": 8.827259323148056e-06,
      "loss": 0.1888,
      "step": 1921
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.526855861098516,
      "learning_rate": 8.825924961365538e-06,
      "loss": 0.252,
      "step": 1922
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4771424166155036,
      "learning_rate": 8.824589941862164e-06,
      "loss": 0.2615,
      "step": 1923
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.365141690124064,
      "learning_rate": 8.82325426486744e-06,
      "loss": 0.2645,
      "step": 1924
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3661893184344258,
      "learning_rate": 8.821917930610987e-06,
      "loss": 0.243,
      "step": 1925
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.656591990125645,
      "learning_rate": 8.820580939322532e-06,
      "loss": 0.2568,
      "step": 1926
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5003446782391752,
      "learning_rate": 8.819243291231922e-06,
      "loss": 0.2301,
      "step": 1927
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.480506495464975,
      "learning_rate": 8.817904986569115e-06,
      "loss": 0.2118,
      "step": 1928
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.505868697842644,
      "learning_rate": 8.81656602556418e-06,
      "loss": 0.188,
      "step": 1929
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4714546780647433,
      "learning_rate": 8.815226408447298e-06,
      "loss": 0.2667,
      "step": 1930
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4551014966460072,
      "learning_rate": 8.813886135448766e-06,
      "loss": 0.2211,
      "step": 1931
    },
    {
      "epoch": 0.25,
      "grad_norm": 10.638052857340703,
      "learning_rate": 8.812545206798995e-06,
      "loss": 0.6657,
      "step": 1932
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4860330195484799,
      "learning_rate": 8.811203622728504e-06,
      "loss": 0.2158,
      "step": 1933
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.765937441903496,
      "learning_rate": 8.809861383467926e-06,
      "loss": 0.663,
      "step": 1934
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.533570846453728,
      "learning_rate": 8.808518489248009e-06,
      "loss": 0.2116,
      "step": 1935
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3089449779980384,
      "learning_rate": 8.807174940299613e-06,
      "loss": 0.1893,
      "step": 1936
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4535702088035085,
      "learning_rate": 8.805830736853705e-06,
      "loss": 0.1933,
      "step": 1937
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.438508704709437,
      "learning_rate": 8.804485879141375e-06,
      "loss": 0.2146,
      "step": 1938
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5237570686785449,
      "learning_rate": 8.803140367393815e-06,
      "loss": 0.2191,
      "step": 1939
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.325682684593533,
      "learning_rate": 8.801794201842337e-06,
      "loss": 0.2095,
      "step": 1940
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.34969965010456,
      "learning_rate": 8.800447382718362e-06,
      "loss": 0.4926,
      "step": 1941
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3549565025719548,
      "learning_rate": 8.799099910253424e-06,
      "loss": 0.1907,
      "step": 1942
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5224595973850275,
      "learning_rate": 8.797751784679167e-06,
      "loss": 0.255,
      "step": 1943
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5315261258939354,
      "learning_rate": 8.796403006227352e-06,
      "loss": 0.2414,
      "step": 1944
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5689451504982028,
      "learning_rate": 8.795053575129846e-06,
      "loss": 0.2061,
      "step": 1945
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.486406433784716,
      "learning_rate": 8.793703491618638e-06,
      "loss": 0.5941,
      "step": 1946
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3963117781397911,
      "learning_rate": 8.792352755925817e-06,
      "loss": 0.2196,
      "step": 1947
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.464492455743207,
      "learning_rate": 8.791001368283593e-06,
      "loss": 0.2228,
      "step": 1948
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6653388740566704,
      "learning_rate": 8.789649328924286e-06,
      "loss": 0.2426,
      "step": 1949
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5146811883982776,
      "learning_rate": 8.788296638080325e-06,
      "loss": 0.2463,
      "step": 1950
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.175351494807616,
      "learning_rate": 8.786943295984254e-06,
      "loss": 0.6662,
      "step": 1951
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5140954453644564,
      "learning_rate": 8.785589302868729e-06,
      "loss": 0.249,
      "step": 1952
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5549434696834732,
      "learning_rate": 8.784234658966514e-06,
      "loss": 0.2253,
      "step": 1953
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3832849843371957,
      "learning_rate": 8.782879364510494e-06,
      "loss": 0.255,
      "step": 1954
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6620821590343375,
      "learning_rate": 8.781523419733655e-06,
      "loss": 0.2658,
      "step": 1955
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.488205534138985,
      "learning_rate": 8.7801668248691e-06,
      "loss": 0.524,
      "step": 1956
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2502311317391879,
      "learning_rate": 8.778809580150044e-06,
      "loss": 0.2101,
      "step": 1957
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.57655134950929,
      "learning_rate": 8.777451685809817e-06,
      "loss": 0.2287,
      "step": 1958
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3795004437965357,
      "learning_rate": 8.77609314208185e-06,
      "loss": 0.2294,
      "step": 1959
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5664434421413354,
      "learning_rate": 8.774733949199696e-06,
      "loss": 0.2381,
      "step": 1960
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5347604685205087,
      "learning_rate": 8.773374107397017e-06,
      "loss": 0.2106,
      "step": 1961
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7218320706310821,
      "learning_rate": 8.772013616907584e-06,
      "loss": 0.2862,
      "step": 1962
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4205556774567398,
      "learning_rate": 8.77065247796528e-06,
      "loss": 0.2001,
      "step": 1963
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.573004363493946,
      "learning_rate": 8.769290690804104e-06,
      "loss": 0.6293,
      "step": 1964
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.995486834576877,
      "learning_rate": 8.767928255658158e-06,
      "loss": 0.5372,
      "step": 1965
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3898216232860967,
      "learning_rate": 8.766565172761666e-06,
      "loss": 0.1922,
      "step": 1966
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4959640288802327,
      "learning_rate": 8.765201442348953e-06,
      "loss": 0.2223,
      "step": 1967
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6926747293500433,
      "learning_rate": 8.763837064654464e-06,
      "loss": 0.2598,
      "step": 1968
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5200073129635032,
      "learning_rate": 8.762472039912748e-06,
      "loss": 0.2172,
      "step": 1969
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.831471757614098,
      "learning_rate": 8.76110636835847e-06,
      "loss": 0.6401,
      "step": 1970
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.476593400351325,
      "learning_rate": 8.759740050226406e-06,
      "loss": 0.2318,
      "step": 1971
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3792574539943547,
      "learning_rate": 8.758373085751439e-06,
      "loss": 0.2086,
      "step": 1972
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.475990515319464,
      "learning_rate": 8.75700547516857e-06,
      "loss": 0.1991,
      "step": 1973
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5097033929661612,
      "learning_rate": 8.755637218712902e-06,
      "loss": 0.2097,
      "step": 1974
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5684855179028991,
      "learning_rate": 8.75426831661966e-06,
      "loss": 0.2306,
      "step": 1975
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3335736186374934,
      "learning_rate": 8.75289876912417e-06,
      "loss": 0.2034,
      "step": 1976
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5986364325753948,
      "learning_rate": 8.751528576461873e-06,
      "loss": 0.2488,
      "step": 1977
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4483021604442816,
      "learning_rate": 8.750157738868323e-06,
      "loss": 0.2064,
      "step": 1978
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4322455234790528,
      "learning_rate": 8.748786256579182e-06,
      "loss": 0.2077,
      "step": 1979
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.448597758626003,
      "learning_rate": 8.747414129830225e-06,
      "loss": 0.2103,
      "step": 1980
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7042509819366682,
      "learning_rate": 8.746041358857334e-06,
      "loss": 0.2414,
      "step": 1981
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.400952946231927,
      "learning_rate": 8.744667943896507e-06,
      "loss": 0.2369,
      "step": 1982
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4483026525853377,
      "learning_rate": 8.743293885183847e-06,
      "loss": 0.2452,
      "step": 1983
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4795889491026084,
      "learning_rate": 8.741919182955573e-06,
      "loss": 0.2368,
      "step": 1984
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4698654119489043,
      "learning_rate": 8.740543837448012e-06,
      "loss": 0.2338,
      "step": 1985
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.62064114689659,
      "learning_rate": 8.739167848897601e-06,
      "loss": 0.6479,
      "step": 1986
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2940820286030115,
      "learning_rate": 8.737791217540887e-06,
      "loss": 0.2048,
      "step": 1987
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5089613067766403,
      "learning_rate": 8.736413943614533e-06,
      "loss": 0.1958,
      "step": 1988
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.577519323538163,
      "learning_rate": 8.735036027355304e-06,
      "loss": 0.6899,
      "step": 1989
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.420933469622048,
      "learning_rate": 8.733657469000081e-06,
      "loss": 0.2322,
      "step": 1990
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3261470642557103,
      "learning_rate": 8.732278268785856e-06,
      "loss": 0.2286,
      "step": 1991
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.621377582427676,
      "learning_rate": 8.730898426949728e-06,
      "loss": 0.2499,
      "step": 1992
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.27621215041224,
      "learning_rate": 8.729517943728909e-06,
      "loss": 0.662,
      "step": 1993
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2201200482153445,
      "learning_rate": 8.728136819360717e-06,
      "loss": 0.1889,
      "step": 1994
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4597011798729762,
      "learning_rate": 8.726755054082589e-06,
      "loss": 0.2176,
      "step": 1995
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4143629759235388,
      "learning_rate": 8.72537264813206e-06,
      "loss": 0.2104,
      "step": 1996
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3795929534200833,
      "learning_rate": 8.723989601746785e-06,
      "loss": 0.2154,
      "step": 1997
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5705224952954615,
      "learning_rate": 8.722605915164526e-06,
      "loss": 0.2595,
      "step": 1998
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3342547388343753,
      "learning_rate": 8.721221588623154e-06,
      "loss": 0.1806,
      "step": 1999
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4717844115396048,
      "learning_rate": 8.71983662236065e-06,
      "loss": 0.1872,
      "step": 2000
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.447083713785723,
      "learning_rate": 8.718451016615108e-06,
      "loss": 0.2546,
      "step": 2001
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7428055463590997,
      "learning_rate": 8.717064771624728e-06,
      "loss": 0.266,
      "step": 2002
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4380577285059004,
      "learning_rate": 8.715677887627822e-06,
      "loss": 0.2274,
      "step": 2003
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4771284195856265,
      "learning_rate": 8.71429036486281e-06,
      "loss": 0.2201,
      "step": 2004
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5520785635567342,
      "learning_rate": 8.712902203568226e-06,
      "loss": 0.2333,
      "step": 2005
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3903689080800108,
      "learning_rate": 8.711513403982708e-06,
      "loss": 0.225,
      "step": 2006
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3482662761820974,
      "learning_rate": 8.710123966345008e-06,
      "loss": 0.2141,
      "step": 2007
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.256337911535555,
      "learning_rate": 8.708733890893987e-06,
      "loss": 0.1849,
      "step": 2008
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2800483304405346,
      "learning_rate": 8.707343177868616e-06,
      "loss": 0.1848,
      "step": 2009
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4349213403338494,
      "learning_rate": 8.705951827507974e-06,
      "loss": 0.2408,
      "step": 2010
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3502175942207804,
      "learning_rate": 8.704559840051249e-06,
      "loss": 0.1796,
      "step": 2011
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3765900836223974,
      "learning_rate": 8.70316721573774e-06,
      "loss": 0.2161,
      "step": 2012
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.620372280433362,
      "learning_rate": 8.701773954806856e-06,
      "loss": 0.2491,
      "step": 2013
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3705026106039353,
      "learning_rate": 8.700380057498115e-06,
      "loss": 0.1917,
      "step": 2014
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4825605855547181,
      "learning_rate": 8.698985524051143e-06,
      "loss": 0.1925,
      "step": 2015
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4572043064299778,
      "learning_rate": 8.697590354705679e-06,
      "loss": 0.2318,
      "step": 2016
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4129104715105232,
      "learning_rate": 8.696194549701564e-06,
      "loss": 0.1886,
      "step": 2017
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.392529779392798,
      "learning_rate": 8.694798109278758e-06,
      "loss": 0.2402,
      "step": 2018
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5530132945791446,
      "learning_rate": 8.693401033677322e-06,
      "loss": 0.2575,
      "step": 2019
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4308226863813711,
      "learning_rate": 8.69200332313743e-06,
      "loss": 0.2219,
      "step": 2020
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6763320124827819,
      "learning_rate": 8.690604977899369e-06,
      "loss": 0.2524,
      "step": 2021
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.510672014256551,
      "learning_rate": 8.689205998203522e-06,
      "loss": 0.2805,
      "step": 2022
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4048315149992257,
      "learning_rate": 8.6878063842904e-06,
      "loss": 0.215,
      "step": 2023
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4509852318356677,
      "learning_rate": 8.686406136400604e-06,
      "loss": 0.1769,
      "step": 2024
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4291353831178197,
      "learning_rate": 8.685005254774859e-06,
      "loss": 0.2368,
      "step": 2025
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4908148357402082,
      "learning_rate": 8.68360373965399e-06,
      "loss": 0.2411,
      "step": 2026
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.790845576166617,
      "learning_rate": 8.682201591278934e-06,
      "loss": 0.5981,
      "step": 2027
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.955137648074938,
      "learning_rate": 8.680798809890737e-06,
      "loss": 0.5621,
      "step": 2028
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4370087595192635,
      "learning_rate": 8.679395395730552e-06,
      "loss": 0.2517,
      "step": 2029
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.325431142706046,
      "learning_rate": 8.677991349039644e-06,
      "loss": 0.2331,
      "step": 2030
    },
    {
      "epoch": 0.26,
      "grad_norm": 10.103270364080196,
      "learning_rate": 8.676586670059383e-06,
      "loss": 0.6392,
      "step": 2031
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6574337419869865,
      "learning_rate": 8.675181359031253e-06,
      "loss": 0.2555,
      "step": 2032
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4383963793516137,
      "learning_rate": 8.673775416196838e-06,
      "loss": 0.2367,
      "step": 2033
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6147481897088733,
      "learning_rate": 8.672368841797842e-06,
      "loss": 0.29,
      "step": 2034
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6577695149694007,
      "learning_rate": 8.670961636076067e-06,
      "loss": 0.281,
      "step": 2035
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.309956972248304,
      "learning_rate": 8.669553799273429e-06,
      "loss": 0.244,
      "step": 2036
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0766011746718487,
      "learning_rate": 8.668145331631953e-06,
      "loss": 0.2261,
      "step": 2037
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5322730920422882,
      "learning_rate": 8.666736233393769e-06,
      "loss": 0.2478,
      "step": 2038
    },
    {
      "epoch": 0.26,
      "grad_norm": 10.651667918515233,
      "learning_rate": 8.665326504801117e-06,
      "loss": 0.5916,
      "step": 2039
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5128367776579066,
      "learning_rate": 8.663916146096348e-06,
      "loss": 0.2459,
      "step": 2040
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.445956394266399,
      "learning_rate": 8.662505157521918e-06,
      "loss": 0.4427,
      "step": 2041
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7598837341156974,
      "learning_rate": 8.66109353932039e-06,
      "loss": 0.2485,
      "step": 2042
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3488147454066228,
      "learning_rate": 8.659681291734441e-06,
      "loss": 0.1953,
      "step": 2043
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4074768548446708,
      "learning_rate": 8.658268415006853e-06,
      "loss": 0.2128,
      "step": 2044
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.413761981779152,
      "learning_rate": 8.656854909380512e-06,
      "loss": 0.2465,
      "step": 2045
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.563947266114772,
      "learning_rate": 8.655440775098418e-06,
      "loss": 0.5703,
      "step": 2046
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3259435736885556,
      "learning_rate": 8.654026012403678e-06,
      "loss": 0.2259,
      "step": 2047
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4509655457287274,
      "learning_rate": 8.652610621539505e-06,
      "loss": 0.2267,
      "step": 2048
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3579663299669202,
      "learning_rate": 8.651194602749223e-06,
      "loss": 0.2124,
      "step": 2049
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.021231822565021,
      "learning_rate": 8.649777956276257e-06,
      "loss": 0.47,
      "step": 2050
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.69821475338391,
      "learning_rate": 8.648360682364153e-06,
      "loss": 0.5843,
      "step": 2051
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4176719688003816,
      "learning_rate": 8.646942781256548e-06,
      "loss": 0.211,
      "step": 2052
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3991113221589215,
      "learning_rate": 8.645524253197202e-06,
      "loss": 0.1975,
      "step": 2053
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2702851076258086,
      "learning_rate": 8.644105098429975e-06,
      "loss": 0.1588,
      "step": 2054
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3186962700442897,
      "learning_rate": 8.642685317198833e-06,
      "loss": 0.2053,
      "step": 2055
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.370297741574393,
      "learning_rate": 8.64126490974786e-06,
      "loss": 0.2283,
      "step": 2056
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4679967388904538,
      "learning_rate": 8.639843876321232e-06,
      "loss": 0.1945,
      "step": 2057
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.375993042613005,
      "learning_rate": 8.638422217163244e-06,
      "loss": 0.2471,
      "step": 2058
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.300299589101739,
      "learning_rate": 8.6369999325183e-06,
      "loss": 0.2287,
      "step": 2059
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.476972778861405,
      "learning_rate": 8.6355770226309e-06,
      "loss": 0.2386,
      "step": 2060
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4277252025524947,
      "learning_rate": 8.634153487745667e-06,
      "loss": 0.1849,
      "step": 2061
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5135146210681014,
      "learning_rate": 8.632729328107317e-06,
      "loss": 0.1951,
      "step": 2062
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3265649691064778,
      "learning_rate": 8.63130454396068e-06,
      "loss": 0.1716,
      "step": 2063
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6773279171729196,
      "learning_rate": 8.629879135550695e-06,
      "loss": 0.2151,
      "step": 2064
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6229208299088158,
      "learning_rate": 8.628453103122404e-06,
      "loss": 0.2285,
      "step": 2065
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.947830989839566,
      "learning_rate": 8.627026446920963e-06,
      "loss": 0.68,
      "step": 2066
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5403269004831668,
      "learning_rate": 8.625599167191627e-06,
      "loss": 0.1961,
      "step": 2067
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4336041494701384,
      "learning_rate": 8.624171264179761e-06,
      "loss": 0.2221,
      "step": 2068
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.694645092989801,
      "learning_rate": 8.62274273813084e-06,
      "loss": 0.4827,
      "step": 2069
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3693227834655615,
      "learning_rate": 8.621313589290446e-06,
      "loss": 0.201,
      "step": 2070
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3948123804799821,
      "learning_rate": 8.619883817904262e-06,
      "loss": 0.2203,
      "step": 2071
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5374082620767753,
      "learning_rate": 8.618453424218085e-06,
      "loss": 0.242,
      "step": 2072
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.347192444963754,
      "learning_rate": 8.617022408477816e-06,
      "loss": 0.6137,
      "step": 2073
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5525372606828785,
      "learning_rate": 8.615590770929461e-06,
      "loss": 0.2412,
      "step": 2074
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.173399629796003,
      "learning_rate": 8.614158511819138e-06,
      "loss": 0.5652,
      "step": 2075
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.426110345132113,
      "learning_rate": 8.612725631393068e-06,
      "loss": 0.2518,
      "step": 2076
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5213863005971908,
      "learning_rate": 8.611292129897581e-06,
      "loss": 0.2064,
      "step": 2077
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4632748663579205,
      "learning_rate": 8.60985800757911e-06,
      "loss": 0.2469,
      "step": 2078
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.772872438298394,
      "learning_rate": 8.608423264684198e-06,
      "loss": 0.5583,
      "step": 2079
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.456436757773547,
      "learning_rate": 8.606987901459497e-06,
      "loss": 0.691,
      "step": 2080
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3106500141607096,
      "learning_rate": 8.605551918151755e-06,
      "loss": 0.1807,
      "step": 2081
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.321234821019563,
      "learning_rate": 8.604115315007844e-06,
      "loss": 0.221,
      "step": 2082
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3361285582707174,
      "learning_rate": 8.602678092274725e-06,
      "loss": 0.1998,
      "step": 2083
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4433247543821743,
      "learning_rate": 8.601240250199476e-06,
      "loss": 0.2301,
      "step": 2084
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5072693914764805,
      "learning_rate": 8.599801789029281e-06,
      "loss": 0.2371,
      "step": 2085
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.497938136685035,
      "learning_rate": 8.598362709011425e-06,
      "loss": 0.2201,
      "step": 2086
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3264400773630245,
      "learning_rate": 8.596923010393303e-06,
      "loss": 0.1999,
      "step": 2087
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3741557462554688,
      "learning_rate": 8.595482693422416e-06,
      "loss": 0.2077,
      "step": 2088
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4276869903977227,
      "learning_rate": 8.594041758346372e-06,
      "loss": 0.2159,
      "step": 2089
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5615499174942837,
      "learning_rate": 8.592600205412884e-06,
      "loss": 0.2175,
      "step": 2090
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4569159140059362,
      "learning_rate": 8.591158034869773e-06,
      "loss": 0.2286,
      "step": 2091
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.103486576624063,
      "learning_rate": 8.589715246964963e-06,
      "loss": 0.607,
      "step": 2092
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3925695691673312,
      "learning_rate": 8.588271841946485e-06,
      "loss": 0.1968,
      "step": 2093
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3757903185016458,
      "learning_rate": 8.58682782006248e-06,
      "loss": 0.2094,
      "step": 2094
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3300422910774512,
      "learning_rate": 8.585383181561191e-06,
      "loss": 0.1711,
      "step": 2095
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5001449300091676,
      "learning_rate": 8.583937926690967e-06,
      "loss": 0.2325,
      "step": 2096
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6790243397503761,
      "learning_rate": 8.582492055700264e-06,
      "loss": 0.2434,
      "step": 2097
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4122781197549399,
      "learning_rate": 8.581045568837647e-06,
      "loss": 0.297,
      "step": 2098
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2921829914249054,
      "learning_rate": 8.57959846635178e-06,
      "loss": 0.2117,
      "step": 2099
    },
    {
      "epoch": 0.27,
      "grad_norm": 12.086620414355348,
      "learning_rate": 8.578150748491438e-06,
      "loss": 0.5083,
      "step": 2100
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3871411691416513,
      "learning_rate": 8.576702415505501e-06,
      "loss": 0.2318,
      "step": 2101
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6005030777524596,
      "learning_rate": 8.575253467642954e-06,
      "loss": 0.2178,
      "step": 2102
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.466104621303711,
      "learning_rate": 8.57380390515289e-06,
      "loss": 0.2114,
      "step": 2103
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.199906297650748,
      "learning_rate": 8.5723537282845e-06,
      "loss": 0.186,
      "step": 2104
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5219573939639839,
      "learning_rate": 8.570902937287093e-06,
      "loss": 0.2783,
      "step": 2105
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.774624393377417,
      "learning_rate": 8.569451532410073e-06,
      "loss": 0.5385,
      "step": 2106
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.885067927211518,
      "learning_rate": 8.567999513902953e-06,
      "loss": 0.6868,
      "step": 2107
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4617417297870385,
      "learning_rate": 8.566546882015355e-06,
      "loss": 0.2275,
      "step": 2108
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3507943285096755,
      "learning_rate": 8.565093636996999e-06,
      "loss": 0.194,
      "step": 2109
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5599068621990893,
      "learning_rate": 8.563639779097717e-06,
      "loss": 0.2474,
      "step": 2110
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.350910376435295,
      "learning_rate": 8.562185308567443e-06,
      "loss": 0.5276,
      "step": 2111
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.438259538374095,
      "learning_rate": 8.56073022565622e-06,
      "loss": 0.2107,
      "step": 2112
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6792940975668007,
      "learning_rate": 8.55927453061419e-06,
      "loss": 0.2577,
      "step": 2113
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3497143091319155,
      "learning_rate": 8.557818223691607e-06,
      "loss": 0.2557,
      "step": 2114
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3495271073614963,
      "learning_rate": 8.556361305138825e-06,
      "loss": 0.214,
      "step": 2115
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.615064309121152,
      "learning_rate": 8.554903775206305e-06,
      "loss": 0.2442,
      "step": 2116
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3551161850581093,
      "learning_rate": 8.553445634144614e-06,
      "loss": 0.2254,
      "step": 2117
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4929026862594426,
      "learning_rate": 8.551986882204424e-06,
      "loss": 0.1755,
      "step": 2118
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3833317805757093,
      "learning_rate": 8.550527519636511e-06,
      "loss": 0.2087,
      "step": 2119
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4582381796693886,
      "learning_rate": 8.549067546691756e-06,
      "loss": 0.2139,
      "step": 2120
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4903101122624496,
      "learning_rate": 8.547606963621146e-06,
      "loss": 0.1904,
      "step": 2121
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4558546804150676,
      "learning_rate": 8.54614577067577e-06,
      "loss": 0.2107,
      "step": 2122
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5064859229930034,
      "learning_rate": 8.544683968106827e-06,
      "loss": 0.2479,
      "step": 2123
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5499054191792014,
      "learning_rate": 8.543221556165615e-06,
      "loss": 0.2151,
      "step": 2124
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4374170558576114,
      "learning_rate": 8.541758535103542e-06,
      "loss": 0.2244,
      "step": 2125
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.666976169729603,
      "learning_rate": 8.540294905172117e-06,
      "loss": 0.255,
      "step": 2126
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.281593476790365,
      "learning_rate": 8.538830666622955e-06,
      "loss": 0.63,
      "step": 2127
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5159369378655199,
      "learning_rate": 8.537365819707776e-06,
      "loss": 0.2419,
      "step": 2128
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4664730547363853,
      "learning_rate": 8.535900364678403e-06,
      "loss": 0.2484,
      "step": 2129
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.540030939590895,
      "learning_rate": 8.534434301786767e-06,
      "loss": 0.2255,
      "step": 2130
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.62909604508,
      "learning_rate": 8.532967631284898e-06,
      "loss": 0.24,
      "step": 2131
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6273677709005885,
      "learning_rate": 8.531500353424937e-06,
      "loss": 0.2448,
      "step": 2132
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6473794303941356,
      "learning_rate": 8.530032468459126e-06,
      "loss": 0.2172,
      "step": 2133
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3400118931149754,
      "learning_rate": 8.528563976639807e-06,
      "loss": 0.1969,
      "step": 2134
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.7202739046336895,
      "learning_rate": 8.527094878219435e-06,
      "loss": 0.6404,
      "step": 2135
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.1463744712915975,
      "learning_rate": 8.525625173450564e-06,
      "loss": 0.5194,
      "step": 2136
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.299348944811537,
      "learning_rate": 8.524154862585854e-06,
      "loss": 0.2203,
      "step": 2137
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.783928920036926,
      "learning_rate": 8.522683945878068e-06,
      "loss": 0.7387,
      "step": 2138
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.084573176307075,
      "learning_rate": 8.52121242358007e-06,
      "loss": 0.4995,
      "step": 2139
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.419621822835937,
      "learning_rate": 8.519740295944838e-06,
      "loss": 0.239,
      "step": 2140
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.36275599261139,
      "learning_rate": 8.518267563225443e-06,
      "loss": 0.2087,
      "step": 2141
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4173390714610348,
      "learning_rate": 8.51679422567507e-06,
      "loss": 0.2481,
      "step": 2142
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4620845909879778,
      "learning_rate": 8.515320283546996e-06,
      "loss": 0.2414,
      "step": 2143
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6109788756193009,
      "learning_rate": 8.513845737094613e-06,
      "loss": 0.2298,
      "step": 2144
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.382023732786991,
      "learning_rate": 8.51237058657141e-06,
      "loss": 0.2124,
      "step": 2145
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4030120797188177,
      "learning_rate": 8.510894832230988e-06,
      "loss": 0.1631,
      "step": 2146
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7724968343807552,
      "learning_rate": 8.50941847432704e-06,
      "loss": 0.2586,
      "step": 2147
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4367193172620807,
      "learning_rate": 8.507941513113372e-06,
      "loss": 0.2845,
      "step": 2148
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.522709513174078,
      "learning_rate": 8.506463948843891e-06,
      "loss": 0.1723,
      "step": 2149
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3741768372156753,
      "learning_rate": 8.504985781772606e-06,
      "loss": 0.2076,
      "step": 2150
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.222054103147704,
      "learning_rate": 8.503507012153632e-06,
      "loss": 0.1753,
      "step": 2151
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3872329235818233,
      "learning_rate": 8.502027640241188e-06,
      "loss": 0.205,
      "step": 2152
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4446803440195106,
      "learning_rate": 8.500547666289592e-06,
      "loss": 0.2171,
      "step": 2153
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3461269898778814,
      "learning_rate": 8.49906709055327e-06,
      "loss": 0.1896,
      "step": 2154
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4301161714512594,
      "learning_rate": 8.497585913286752e-06,
      "loss": 0.2621,
      "step": 2155
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4970060584242904,
      "learning_rate": 8.496104134744667e-06,
      "loss": 0.2571,
      "step": 2156
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.867898679032296,
      "learning_rate": 8.49462175518175e-06,
      "loss": 0.596,
      "step": 2157
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4149325245073137,
      "learning_rate": 8.493138774852842e-06,
      "loss": 0.2696,
      "step": 2158
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4215325349547043,
      "learning_rate": 8.49165519401288e-06,
      "loss": 0.1991,
      "step": 2159
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6480172436311378,
      "learning_rate": 8.490171012916915e-06,
      "loss": 0.2117,
      "step": 2160
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.456208257727299,
      "learning_rate": 8.488686231820089e-06,
      "loss": 0.1906,
      "step": 2161
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6579488690982689,
      "learning_rate": 8.487200850977657e-06,
      "loss": 0.1982,
      "step": 2162
    },
    {
      "epoch": 0.28,
      "grad_norm": 10.727261503841396,
      "learning_rate": 8.48571487064497e-06,
      "loss": 0.6981,
      "step": 2163
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.641045555952009,
      "learning_rate": 8.484228291077488e-06,
      "loss": 0.2424,
      "step": 2164
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.38741624033801,
      "learning_rate": 8.48274111253077e-06,
      "loss": 0.1799,
      "step": 2165
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3395936231193062,
      "learning_rate": 8.481253335260478e-06,
      "loss": 0.236,
      "step": 2166
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4995205851172673,
      "learning_rate": 8.479764959522381e-06,
      "loss": 0.2535,
      "step": 2167
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5055358535910093,
      "learning_rate": 8.478275985572346e-06,
      "loss": 0.1801,
      "step": 2168
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4658875691865885,
      "learning_rate": 8.476786413666346e-06,
      "loss": 0.1949,
      "step": 2169
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6493973910481883,
      "learning_rate": 8.475296244060454e-06,
      "loss": 0.2473,
      "step": 2170
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3942660263618134,
      "learning_rate": 8.473805477010848e-06,
      "loss": 0.212,
      "step": 2171
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4279349671455948,
      "learning_rate": 8.472314112773813e-06,
      "loss": 0.2043,
      "step": 2172
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4868504835909009,
      "learning_rate": 8.470822151605723e-06,
      "loss": 0.2638,
      "step": 2173
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4626140320765466,
      "learning_rate": 8.469329593763069e-06,
      "loss": 0.2075,
      "step": 2174
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2229000513220996,
      "learning_rate": 8.467836439502439e-06,
      "loss": 0.1341,
      "step": 2175
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5517080972149568,
      "learning_rate": 8.46634268908052e-06,
      "loss": 0.2029,
      "step": 2176
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.332650498512615,
      "learning_rate": 8.46484834275411e-06,
      "loss": 0.5717,
      "step": 2177
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.155102727033506,
      "learning_rate": 8.463353400780101e-06,
      "loss": 0.4978,
      "step": 2178
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.548951081623538,
      "learning_rate": 8.461857863415493e-06,
      "loss": 0.755,
      "step": 2179
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4744207089792087,
      "learning_rate": 8.460361730917384e-06,
      "loss": 0.261,
      "step": 2180
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5451918450893,
      "learning_rate": 8.458865003542975e-06,
      "loss": 0.2516,
      "step": 2181
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6665297358503994,
      "learning_rate": 8.457367681549577e-06,
      "loss": 0.2649,
      "step": 2182
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.528715270701386,
      "learning_rate": 8.455869765194592e-06,
      "loss": 0.2499,
      "step": 2183
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4131831949392504,
      "learning_rate": 8.45437125473553e-06,
      "loss": 0.1964,
      "step": 2184
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.382824307673467,
      "learning_rate": 8.452872150430002e-06,
      "loss": 0.1915,
      "step": 2185
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5893933563015712,
      "learning_rate": 8.451372452535724e-06,
      "loss": 0.2032,
      "step": 2186
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4210927699673748,
      "learning_rate": 8.44987216131051e-06,
      "loss": 0.2029,
      "step": 2187
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.309938766461874,
      "learning_rate": 8.448371277012275e-06,
      "loss": 0.2085,
      "step": 2188
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.58785675462853,
      "learning_rate": 8.446869799899042e-06,
      "loss": 0.6122,
      "step": 2189
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5296562018363076,
      "learning_rate": 8.44536773022893e-06,
      "loss": 0.2306,
      "step": 2190
    },
    {
      "epoch": 0.28,
      "grad_norm": 25.853224939522622,
      "learning_rate": 8.443865068260164e-06,
      "loss": 0.6863,
      "step": 2191
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6007869564138133,
      "learning_rate": 8.442361814251069e-06,
      "loss": 0.2293,
      "step": 2192
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.871658985399957,
      "learning_rate": 8.440857968460068e-06,
      "loss": 0.7444,
      "step": 2193
    },
    {
      "epoch": 0.28,
      "grad_norm": 9.557734654960557,
      "learning_rate": 8.439353531145695e-06,
      "loss": 0.6901,
      "step": 2194
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3878719866585807,
      "learning_rate": 8.437848502566576e-06,
      "loss": 0.2087,
      "step": 2195
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.456205606125253,
      "learning_rate": 8.436342882981445e-06,
      "loss": 0.2262,
      "step": 2196
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.416548943229904,
      "learning_rate": 8.434836672649134e-06,
      "loss": 0.1702,
      "step": 2197
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6052787202548273,
      "learning_rate": 8.433329871828582e-06,
      "loss": 0.2582,
      "step": 2198
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.415101295078164,
      "learning_rate": 8.431822480778818e-06,
      "loss": 0.2275,
      "step": 2199
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.451466469054131,
      "learning_rate": 8.430314499758986e-06,
      "loss": 0.2327,
      "step": 2200
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5515138134717101,
      "learning_rate": 8.428805929028327e-06,
      "loss": 0.237,
      "step": 2201
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3171751153316988,
      "learning_rate": 8.427296768846176e-06,
      "loss": 0.2082,
      "step": 2202
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4226746510561301,
      "learning_rate": 8.425787019471979e-06,
      "loss": 0.2272,
      "step": 2203
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.583125573547979,
      "learning_rate": 8.424276681165276e-06,
      "loss": 0.2308,
      "step": 2204
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.665150044164671,
      "learning_rate": 8.422765754185716e-06,
      "loss": 0.2584,
      "step": 2205
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.221827943282218,
      "learning_rate": 8.421254238793041e-06,
      "loss": 0.1589,
      "step": 2206
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3700317449583017,
      "learning_rate": 8.419742135247099e-06,
      "loss": 0.1834,
      "step": 2207
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.558831332508253,
      "learning_rate": 8.418229443807842e-06,
      "loss": 0.2267,
      "step": 2208
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.345259752122592,
      "learning_rate": 8.416716164735315e-06,
      "loss": 0.2521,
      "step": 2209
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.437150562651553,
      "learning_rate": 8.415202298289668e-06,
      "loss": 0.1747,
      "step": 2210
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.428510653866958,
      "learning_rate": 8.413687844731155e-06,
      "loss": 0.2481,
      "step": 2211
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7345697903036177,
      "learning_rate": 8.412172804320127e-06,
      "loss": 0.2852,
      "step": 2212
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7174131323772899,
      "learning_rate": 8.410657177317035e-06,
      "loss": 0.2579,
      "step": 2213
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5251116802935059,
      "learning_rate": 8.409140963982436e-06,
      "loss": 0.2293,
      "step": 2214
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3513313117519037,
      "learning_rate": 8.407624164576982e-06,
      "loss": 0.2317,
      "step": 2215
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5550645811610275,
      "learning_rate": 8.406106779361429e-06,
      "loss": 0.2556,
      "step": 2216
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.047969492954411,
      "learning_rate": 8.404588808596635e-06,
      "loss": 0.595,
      "step": 2217
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4633845821234335,
      "learning_rate": 8.403070252543555e-06,
      "loss": 0.194,
      "step": 2218
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3552574960906942,
      "learning_rate": 8.401551111463246e-06,
      "loss": 0.1881,
      "step": 2219
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3883042616357968,
      "learning_rate": 8.400031385616868e-06,
      "loss": 0.1878,
      "step": 2220
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5055889482008384,
      "learning_rate": 8.398511075265677e-06,
      "loss": 0.1971,
      "step": 2221
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.420113044254241,
      "learning_rate": 8.396990180671034e-06,
      "loss": 0.2247,
      "step": 2222
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4229878156502795,
      "learning_rate": 8.395468702094399e-06,
      "loss": 0.26,
      "step": 2223
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3717429133868952,
      "learning_rate": 8.393946639797328e-06,
      "loss": 0.2296,
      "step": 2224
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.307904206035647,
      "learning_rate": 8.392423994041486e-06,
      "loss": 0.7761,
      "step": 2225
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6608724950969072,
      "learning_rate": 8.39090076508863e-06,
      "loss": 0.2721,
      "step": 2226
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.494608028606583,
      "learning_rate": 8.389376953200622e-06,
      "loss": 0.2074,
      "step": 2227
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3764189136941172,
      "learning_rate": 8.387852558639422e-06,
      "loss": 0.2261,
      "step": 2228
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.206774772757921,
      "learning_rate": 8.386327581667091e-06,
      "loss": 0.5707,
      "step": 2229
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4323315205830631,
      "learning_rate": 8.384802022545793e-06,
      "loss": 0.2132,
      "step": 2230
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4623165296236047,
      "learning_rate": 8.383275881537786e-06,
      "loss": 0.193,
      "step": 2231
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3948103268780465,
      "learning_rate": 8.381749158905433e-06,
      "loss": 0.2164,
      "step": 2232
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3829214070368963,
      "learning_rate": 8.380221854911195e-06,
      "loss": 0.2036,
      "step": 2233
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6223133789217659,
      "learning_rate": 8.378693969817633e-06,
      "loss": 0.2242,
      "step": 2234
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4894229997040043,
      "learning_rate": 8.37716550388741e-06,
      "loss": 0.1979,
      "step": 2235
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.977601503808312,
      "learning_rate": 8.375636457383282e-06,
      "loss": 0.5751,
      "step": 2236
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3494811832711007,
      "learning_rate": 8.374106830568117e-06,
      "loss": 0.145,
      "step": 2237
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5071019606622726,
      "learning_rate": 8.372576623704872e-06,
      "loss": 0.2415,
      "step": 2238
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.381747025748785,
      "learning_rate": 8.371045837056603e-06,
      "loss": 0.1862,
      "step": 2239
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.8487457331333927,
      "learning_rate": 8.369514470886478e-06,
      "loss": 0.5764,
      "step": 2240
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2843183106778895,
      "learning_rate": 8.36798252545775e-06,
      "loss": 0.1828,
      "step": 2241
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5146646165728683,
      "learning_rate": 8.366450001033784e-06,
      "loss": 0.2007,
      "step": 2242
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.8217720939681374,
      "learning_rate": 8.364916897878033e-06,
      "loss": 0.5092,
      "step": 2243
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5189059777350238,
      "learning_rate": 8.363383216254058e-06,
      "loss": 0.2413,
      "step": 2244
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4328127556557884,
      "learning_rate": 8.361848956425516e-06,
      "loss": 0.2079,
      "step": 2245
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5827114190236833,
      "learning_rate": 8.360314118656165e-06,
      "loss": 0.2345,
      "step": 2246
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3223924219532113,
      "learning_rate": 8.358778703209862e-06,
      "loss": 0.1815,
      "step": 2247
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5105111519603096,
      "learning_rate": 8.357242710350561e-06,
      "loss": 0.2786,
      "step": 2248
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.490355147985618,
      "learning_rate": 8.355706140342317e-06,
      "loss": 0.2248,
      "step": 2249
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.596080253501593,
      "learning_rate": 8.354168993449285e-06,
      "loss": 0.5109,
      "step": 2250
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.4520147044937755,
      "learning_rate": 8.352631269935719e-06,
      "loss": 0.5263,
      "step": 2251
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5501352712547873,
      "learning_rate": 8.35109297006597e-06,
      "loss": 0.2217,
      "step": 2252
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.24846638578177,
      "learning_rate": 8.349554094104491e-06,
      "loss": 0.2153,
      "step": 2253
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.345015475808715,
      "learning_rate": 8.348014642315831e-06,
      "loss": 0.2005,
      "step": 2254
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3768517599358492,
      "learning_rate": 8.346474614964642e-06,
      "loss": 0.2392,
      "step": 2255
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.378562699410823,
      "learning_rate": 8.34493401231567e-06,
      "loss": 0.1975,
      "step": 2256
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5094440447462922,
      "learning_rate": 8.34339283463376e-06,
      "loss": 0.2234,
      "step": 2257
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5552202776265465,
      "learning_rate": 8.341851082183868e-06,
      "loss": 0.2488,
      "step": 2258
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.250399558949848,
      "learning_rate": 8.340308755231027e-06,
      "loss": 0.2353,
      "step": 2259
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.426060361490256,
      "learning_rate": 8.338765854040391e-06,
      "loss": 0.2521,
      "step": 2260
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2260153762766566,
      "learning_rate": 8.337222378877196e-06,
      "loss": 0.1808,
      "step": 2261
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4103179998268491,
      "learning_rate": 8.33567833000679e-06,
      "loss": 0.2214,
      "step": 2262
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.437560444948256,
      "learning_rate": 8.334133707694603e-06,
      "loss": 0.523,
      "step": 2263
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.599453227370635,
      "learning_rate": 8.33258851220618e-06,
      "loss": 0.7699,
      "step": 2264
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.8441701637580445,
      "learning_rate": 8.33104274380716e-06,
      "loss": 0.6225,
      "step": 2265
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.219501659164563,
      "learning_rate": 8.329496402763275e-06,
      "loss": 0.1953,
      "step": 2266
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4141794499639115,
      "learning_rate": 8.327949489340359e-06,
      "loss": 0.2294,
      "step": 2267
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3584694809746536,
      "learning_rate": 8.326402003804344e-06,
      "loss": 0.2267,
      "step": 2268
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.536903843065588,
      "learning_rate": 8.324853946421261e-06,
      "loss": 0.2422,
      "step": 2269
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.523672880829997,
      "learning_rate": 8.323305317457241e-06,
      "loss": 0.2655,
      "step": 2270
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.355333895687507,
      "learning_rate": 8.32175611717851e-06,
      "loss": 0.2085,
      "step": 2271
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5320950785061442,
      "learning_rate": 8.320206345851393e-06,
      "loss": 0.2632,
      "step": 2272
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4954216842763441,
      "learning_rate": 8.318656003742314e-06,
      "loss": 0.2393,
      "step": 2273
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.539477432344007,
      "learning_rate": 8.317105091117795e-06,
      "loss": 0.2444,
      "step": 2274
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.357115022410198,
      "learning_rate": 8.315553608244453e-06,
      "loss": 0.5096,
      "step": 2275
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4021024790863401,
      "learning_rate": 8.314001555389014e-06,
      "loss": 0.2281,
      "step": 2276
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4896685385385742,
      "learning_rate": 8.312448932818284e-06,
      "loss": 0.2325,
      "step": 2277
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3454995986477307,
      "learning_rate": 8.310895740799181e-06,
      "loss": 0.1737,
      "step": 2278
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.807553958142435,
      "learning_rate": 8.30934197959872e-06,
      "loss": 0.6402,
      "step": 2279
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4349878924022084,
      "learning_rate": 8.307787649484005e-06,
      "loss": 0.2605,
      "step": 2280
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5187126041369097,
      "learning_rate": 8.306232750722248e-06,
      "loss": 0.2356,
      "step": 2281
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.309636036215572,
      "learning_rate": 8.30467728358075e-06,
      "loss": 0.1577,
      "step": 2282
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5065027481107343,
      "learning_rate": 8.303121248326917e-06,
      "loss": 0.2206,
      "step": 2283
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.930569232996268,
      "learning_rate": 8.301564645228249e-06,
      "loss": 0.7262,
      "step": 2284
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2693054844212517,
      "learning_rate": 8.300007474552343e-06,
      "loss": 0.2222,
      "step": 2285
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4019260238366678,
      "learning_rate": 8.298449736566894e-06,
      "loss": 0.2274,
      "step": 2286
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3164944834703698,
      "learning_rate": 8.296891431539696e-06,
      "loss": 0.2245,
      "step": 2287
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.407329547541226,
      "learning_rate": 8.29533255973864e-06,
      "loss": 0.1958,
      "step": 2288
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3694746936894022,
      "learning_rate": 8.293773121431717e-06,
      "loss": 0.2449,
      "step": 2289
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4632510474334859,
      "learning_rate": 8.292213116887008e-06,
      "loss": 0.2139,
      "step": 2290
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3346910533562155,
      "learning_rate": 8.290652546372698e-06,
      "loss": 0.2039,
      "step": 2291
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4545991925368194,
      "learning_rate": 8.289091410157067e-06,
      "loss": 0.1967,
      "step": 2292
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.5660308813648385,
      "learning_rate": 8.28752970850849e-06,
      "loss": 0.4753,
      "step": 2293
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3656005759668501,
      "learning_rate": 8.285967441695445e-06,
      "loss": 0.2556,
      "step": 2294
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2534938561444704,
      "learning_rate": 8.284404609986505e-06,
      "loss": 0.2013,
      "step": 2295
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3722603601182821,
      "learning_rate": 8.282841213650334e-06,
      "loss": 0.1954,
      "step": 2296
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.60502250401749,
      "learning_rate": 8.281277252955699e-06,
      "loss": 0.2792,
      "step": 2297
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.221768127069745,
      "learning_rate": 8.279712728171468e-06,
      "loss": 0.1794,
      "step": 2298
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3024911029598638,
      "learning_rate": 8.278147639566596e-06,
      "loss": 0.2403,
      "step": 2299
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.913473919435227,
      "learning_rate": 8.27658198741014e-06,
      "loss": 0.6438,
      "step": 2300
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4911725885020097,
      "learning_rate": 8.275015771971255e-06,
      "loss": 0.2393,
      "step": 2301
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.195512632524997,
      "learning_rate": 8.273448993519194e-06,
      "loss": 0.5917,
      "step": 2302
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.789343842752345,
      "learning_rate": 8.2718816523233e-06,
      "loss": 0.6632,
      "step": 2303
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5198490532826479,
      "learning_rate": 8.270313748653018e-06,
      "loss": 0.2045,
      "step": 2304
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4959688144094765,
      "learning_rate": 8.268745282777893e-06,
      "loss": 0.2434,
      "step": 2305
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5129341296906291,
      "learning_rate": 8.267176254967556e-06,
      "loss": 0.2585,
      "step": 2306
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4354559147683754,
      "learning_rate": 8.265606665491746e-06,
      "loss": 0.1953,
      "step": 2307
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4580128329461,
      "learning_rate": 8.264036514620292e-06,
      "loss": 0.262,
      "step": 2308
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.243081761980143,
      "learning_rate": 8.262465802623122e-06,
      "loss": 0.6502,
      "step": 2309
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3960607127216011,
      "learning_rate": 8.260894529770258e-06,
      "loss": 0.1952,
      "step": 2310
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5012416793323498,
      "learning_rate": 8.25932269633182e-06,
      "loss": 0.2563,
      "step": 2311
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6111124289881258,
      "learning_rate": 8.257750302578027e-06,
      "loss": 0.2268,
      "step": 2312
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.9767705757678495,
      "learning_rate": 8.256177348779188e-06,
      "loss": 0.6009,
      "step": 2313
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4459138149753985,
      "learning_rate": 8.254603835205715e-06,
      "loss": 0.2427,
      "step": 2314
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.6376505055987804,
      "learning_rate": 8.253029762128111e-06,
      "loss": 0.358,
      "step": 2315
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2345179347788737,
      "learning_rate": 8.25145512981698e-06,
      "loss": 0.1867,
      "step": 2316
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4903979304115083,
      "learning_rate": 8.249879938543017e-06,
      "loss": 0.2305,
      "step": 2317
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3275593319001198,
      "learning_rate": 8.248304188577018e-06,
      "loss": 0.1983,
      "step": 2318
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.096509162597984,
      "learning_rate": 8.24672788018987e-06,
      "loss": 0.609,
      "step": 2319
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3787078259177767,
      "learning_rate": 8.245151013652561e-06,
      "loss": 0.2079,
      "step": 2320
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3393804930128135,
      "learning_rate": 8.24357358923617e-06,
      "loss": 0.2223,
      "step": 2321
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.479897679346653,
      "learning_rate": 8.241995607211878e-06,
      "loss": 0.2403,
      "step": 2322
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.710855332966207,
      "learning_rate": 8.240417067850957e-06,
      "loss": 0.7741,
      "step": 2323
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.004331140859312,
      "learning_rate": 8.238837971424776e-06,
      "loss": 0.5798,
      "step": 2324
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5662930631380267,
      "learning_rate": 8.2372583182048e-06,
      "loss": 0.2535,
      "step": 2325
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.951437573996603,
      "learning_rate": 8.235678108462589e-06,
      "loss": 0.7093,
      "step": 2326
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4309339312061193,
      "learning_rate": 8.2340973424698e-06,
      "loss": 0.217,
      "step": 2327
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2814118913640808,
      "learning_rate": 8.232516020498184e-06,
      "loss": 0.189,
      "step": 2328
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4380253733467336,
      "learning_rate": 8.230934142819588e-06,
      "loss": 0.2053,
      "step": 2329
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4111616460817207,
      "learning_rate": 8.229351709705961e-06,
      "loss": 0.2091,
      "step": 2330
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2510683687487665,
      "learning_rate": 8.227768721429334e-06,
      "loss": 0.2072,
      "step": 2331
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4980354563606866,
      "learning_rate": 8.226185178261846e-06,
      "loss": 0.2455,
      "step": 2332
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.450592768864916,
      "learning_rate": 8.224601080475723e-06,
      "loss": 0.6457,
      "step": 2333
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4264462641301348,
      "learning_rate": 8.223016428343294e-06,
      "loss": 0.2387,
      "step": 2334
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4907480691567376,
      "learning_rate": 8.221431222136976e-06,
      "loss": 0.2236,
      "step": 2335
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.817491098903967,
      "learning_rate": 8.219845462129284e-06,
      "loss": 0.5532,
      "step": 2336
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5092200787086116,
      "learning_rate": 8.218259148592828e-06,
      "loss": 0.199,
      "step": 2337
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3394959811426905,
      "learning_rate": 8.216672281800317e-06,
      "loss": 0.1948,
      "step": 2338
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.170549477226737,
      "learning_rate": 8.21508486202455e-06,
      "loss": 0.2116,
      "step": 2339
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4743223981682785,
      "learning_rate": 8.213496889538422e-06,
      "loss": 0.2455,
      "step": 2340
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5789310757792128,
      "learning_rate": 8.211908364614924e-06,
      "loss": 0.1975,
      "step": 2341
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4171925321703405,
      "learning_rate": 8.210319287527143e-06,
      "loss": 0.2489,
      "step": 2342
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5022227502006114,
      "learning_rate": 8.20872965854826e-06,
      "loss": 0.2116,
      "step": 2343
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3633624625694158,
      "learning_rate": 8.207139477951549e-06,
      "loss": 0.2028,
      "step": 2344
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.247432411408961,
      "learning_rate": 8.205548746010383e-06,
      "loss": 0.4651,
      "step": 2345
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5810466676385275,
      "learning_rate": 8.203957462998225e-06,
      "loss": 0.2228,
      "step": 2346
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2899841030965438,
      "learning_rate": 8.202365629188634e-06,
      "loss": 0.185,
      "step": 2347
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6415408174793817,
      "learning_rate": 8.200773244855267e-06,
      "loss": 0.1966,
      "step": 2348
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.840769996183703,
      "learning_rate": 8.199180310271873e-06,
      "loss": 0.5714,
      "step": 2349
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3881683378645409,
      "learning_rate": 8.197586825712295e-06,
      "loss": 0.2271,
      "step": 2350
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.263198062700988,
      "learning_rate": 8.195992791450475e-06,
      "loss": 0.2089,
      "step": 2351
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.337694045304786,
      "learning_rate": 8.19439820776044e-06,
      "loss": 0.2173,
      "step": 2352
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.393471041586846,
      "learning_rate": 8.19280307491632e-06,
      "loss": 0.5477,
      "step": 2353
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2023775614506973,
      "learning_rate": 8.19120739319234e-06,
      "loss": 0.2006,
      "step": 2354
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3764060555584032,
      "learning_rate": 8.189611162862811e-06,
      "loss": 0.1744,
      "step": 2355
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5796278107110473,
      "learning_rate": 8.188014384202148e-06,
      "loss": 0.1872,
      "step": 2356
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.901517063026991,
      "learning_rate": 8.186417057484851e-06,
      "loss": 0.4286,
      "step": 2357
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6806198994267343,
      "learning_rate": 8.184819182985524e-06,
      "loss": 0.2189,
      "step": 2358
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1682889131276943,
      "learning_rate": 8.183220760978858e-06,
      "loss": 0.1622,
      "step": 2359
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4119302553751327,
      "learning_rate": 8.18162179173964e-06,
      "loss": 0.2312,
      "step": 2360
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5400366051768342,
      "learning_rate": 8.18002227554275e-06,
      "loss": 0.217,
      "step": 2361
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3826299777785998,
      "learning_rate": 8.178422212663166e-06,
      "loss": 0.1936,
      "step": 2362
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.365189279632913,
      "learning_rate": 8.176821603375955e-06,
      "loss": 0.5089,
      "step": 2363
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.423732257246069,
      "learning_rate": 8.175220447956282e-06,
      "loss": 0.184,
      "step": 2364
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3915265995930508,
      "learning_rate": 8.173618746679406e-06,
      "loss": 0.1715,
      "step": 2365
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.428043584237966,
      "learning_rate": 8.172016499820672e-06,
      "loss": 0.2086,
      "step": 2366
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3229866053664237,
      "learning_rate": 8.170413707655532e-06,
      "loss": 0.2291,
      "step": 2367
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5260192870521683,
      "learning_rate": 8.168810370459519e-06,
      "loss": 0.2494,
      "step": 2368
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.543377057339912,
      "learning_rate": 8.167206488508268e-06,
      "loss": 0.3094,
      "step": 2369
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.44141659626777,
      "learning_rate": 8.165602062077502e-06,
      "loss": 0.6018,
      "step": 2370
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.488077172553359,
      "learning_rate": 8.163997091443046e-06,
      "loss": 0.4255,
      "step": 2371
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.707429957685848,
      "learning_rate": 8.162391576880808e-06,
      "loss": 0.5509,
      "step": 2372
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2634348635752266,
      "learning_rate": 8.160785518666795e-06,
      "loss": 0.2077,
      "step": 2373
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.439556366562617,
      "learning_rate": 8.159178917077112e-06,
      "loss": 0.2344,
      "step": 2374
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.422228475022409,
      "learning_rate": 8.157571772387947e-06,
      "loss": 0.1973,
      "step": 2375
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.83081769422166,
      "learning_rate": 8.155964084875587e-06,
      "loss": 0.5721,
      "step": 2376
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4699847652782785,
      "learning_rate": 8.154355854816416e-06,
      "loss": 0.192,
      "step": 2377
    },
    {
      "epoch": 0.3,
      "grad_norm": 13.37809372859587,
      "learning_rate": 8.152747082486905e-06,
      "loss": 0.6684,
      "step": 2378
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.345631129646679,
      "learning_rate": 8.15113776816362e-06,
      "loss": 0.5472,
      "step": 2379
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6047813217460074,
      "learning_rate": 8.14952791212322e-06,
      "loss": 0.2196,
      "step": 2380
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4080963006478093,
      "learning_rate": 8.147917514642462e-06,
      "loss": 0.2327,
      "step": 2381
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.932806998932996,
      "learning_rate": 8.146306575998188e-06,
      "loss": 0.8234,
      "step": 2382
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.466859967795837,
      "learning_rate": 8.14469509646734e-06,
      "loss": 0.2738,
      "step": 2383
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4354274475118438,
      "learning_rate": 8.143083076326947e-06,
      "loss": 0.2471,
      "step": 2384
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.864284803410115,
      "learning_rate": 8.141470515854137e-06,
      "loss": 0.6527,
      "step": 2385
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5731597393119312,
      "learning_rate": 8.139857415326125e-06,
      "loss": 0.2462,
      "step": 2386
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.741165406386393,
      "learning_rate": 8.138243775020222e-06,
      "loss": 0.5807,
      "step": 2387
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.310764836612727,
      "learning_rate": 8.136629595213834e-06,
      "loss": 0.2272,
      "step": 2388
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4073864682723392,
      "learning_rate": 8.135014876184454e-06,
      "loss": 0.2581,
      "step": 2389
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4145667238564759,
      "learning_rate": 8.133399618209675e-06,
      "loss": 0.1547,
      "step": 2390
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2421281023668302,
      "learning_rate": 8.131783821567175e-06,
      "loss": 0.1647,
      "step": 2391
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.305303580383811,
      "learning_rate": 8.13016748653473e-06,
      "loss": 0.2203,
      "step": 2392
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4461985507151522,
      "learning_rate": 8.128550613390205e-06,
      "loss": 0.241,
      "step": 2393
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2925597289884578,
      "learning_rate": 8.12693320241156e-06,
      "loss": 0.1397,
      "step": 2394
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4744121960636922,
      "learning_rate": 8.12531525387685e-06,
      "loss": 0.2113,
      "step": 2395
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3206885697287143,
      "learning_rate": 8.123696768064212e-06,
      "loss": 0.1475,
      "step": 2396
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.387053319806316,
      "learning_rate": 8.122077745251888e-06,
      "loss": 0.2143,
      "step": 2397
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.10647483716276,
      "learning_rate": 8.120458185718206e-06,
      "loss": 0.2617,
      "step": 2398
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3249151655947329,
      "learning_rate": 8.118838089741585e-06,
      "loss": 0.1824,
      "step": 2399
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5298040923392333,
      "learning_rate": 8.117217457600541e-06,
      "loss": 0.2237,
      "step": 2400
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2864185430226562,
      "learning_rate": 8.11559628957368e-06,
      "loss": 0.1929,
      "step": 2401
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.028800324432651,
      "learning_rate": 8.113974585939694e-06,
      "loss": 0.628,
      "step": 2402
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.374399842132973,
      "learning_rate": 8.112352346977378e-06,
      "loss": 0.6022,
      "step": 2403
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.742499973726386,
      "learning_rate": 8.110729572965613e-06,
      "loss": 0.6105,
      "step": 2404
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5831014852982999,
      "learning_rate": 8.109106264183369e-06,
      "loss": 0.2543,
      "step": 2405
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.29690220996709,
      "learning_rate": 8.107482420909719e-06,
      "loss": 0.6855,
      "step": 2406
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3974244980771509,
      "learning_rate": 8.105858043423811e-06,
      "loss": 0.1818,
      "step": 2407
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7925532934979573,
      "learning_rate": 8.104233132004902e-06,
      "loss": 0.2235,
      "step": 2408
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4062363557793691,
      "learning_rate": 8.10260768693233e-06,
      "loss": 0.2108,
      "step": 2409
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.49254094443231,
      "learning_rate": 8.100981708485527e-06,
      "loss": 0.2142,
      "step": 2410
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4958856813667842,
      "learning_rate": 8.09935519694402e-06,
      "loss": 0.2493,
      "step": 2411
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6074389054209715,
      "learning_rate": 8.09772815258742e-06,
      "loss": 0.2283,
      "step": 2412
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5458441051980718,
      "learning_rate": 8.096100575695443e-06,
      "loss": 0.206,
      "step": 2413
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3272424245219687,
      "learning_rate": 8.094472466547882e-06,
      "loss": 0.187,
      "step": 2414
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.527961866405052,
      "learning_rate": 8.09284382542463e-06,
      "loss": 0.2477,
      "step": 2415
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4153311665852046,
      "learning_rate": 8.09121465260567e-06,
      "loss": 0.2292,
      "step": 2416
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.479529654222503,
      "learning_rate": 8.089584948371074e-06,
      "loss": 0.2498,
      "step": 2417
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2494024468562352,
      "learning_rate": 8.087954713001007e-06,
      "loss": 0.1756,
      "step": 2418
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5051857285861303,
      "learning_rate": 8.086323946775727e-06,
      "loss": 0.2513,
      "step": 2419
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6397706820877453,
      "learning_rate": 8.084692649975583e-06,
      "loss": 0.2436,
      "step": 2420
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3817200313683697,
      "learning_rate": 8.083060822881008e-06,
      "loss": 0.2049,
      "step": 2421
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1489487109107022,
      "learning_rate": 8.081428465772539e-06,
      "loss": 0.1936,
      "step": 2422
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1482301904664467,
      "learning_rate": 8.079795578930792e-06,
      "loss": 0.1945,
      "step": 2423
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2671129101369947,
      "learning_rate": 8.07816216263648e-06,
      "loss": 0.1999,
      "step": 2424
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4934255301458161,
      "learning_rate": 8.076528217170408e-06,
      "loss": 0.2501,
      "step": 2425
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.188800832100828,
      "learning_rate": 8.07489374281347e-06,
      "loss": 0.5066,
      "step": 2426
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3726796524767049,
      "learning_rate": 8.07325873984665e-06,
      "loss": 0.2177,
      "step": 2427
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.280821975918923,
      "learning_rate": 8.071623208551023e-06,
      "loss": 0.7315,
      "step": 2428
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6969576407666538,
      "learning_rate": 8.069987149207759e-06,
      "loss": 0.2642,
      "step": 2429
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.403972018879964,
      "learning_rate": 8.068350562098113e-06,
      "loss": 0.2156,
      "step": 2430
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4721834505250866,
      "learning_rate": 8.066713447503434e-06,
      "loss": 0.2188,
      "step": 2431
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4438057902134929,
      "learning_rate": 8.065075805705161e-06,
      "loss": 0.1842,
      "step": 2432
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4628041236969762,
      "learning_rate": 8.063437636984824e-06,
      "loss": 0.2311,
      "step": 2433
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.129806986530985,
      "learning_rate": 8.061798941624041e-06,
      "loss": 0.6445,
      "step": 2434
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3640051214536622,
      "learning_rate": 8.060159719904526e-06,
      "loss": 0.2446,
      "step": 2435
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5829735966907441,
      "learning_rate": 8.058519972108078e-06,
      "loss": 0.2465,
      "step": 2436
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2631497413302222,
      "learning_rate": 8.05687969851659e-06,
      "loss": 0.2096,
      "step": 2437
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4065076660369589,
      "learning_rate": 8.055238899412046e-06,
      "loss": 0.2622,
      "step": 2438
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3559331955301017,
      "learning_rate": 8.053597575076513e-06,
      "loss": 0.1702,
      "step": 2439
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.3502070860563755,
      "learning_rate": 8.051955725792155e-06,
      "loss": 0.6939,
      "step": 2440
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.339587477977398,
      "learning_rate": 8.050313351841229e-06,
      "loss": 0.2353,
      "step": 2441
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4381276928215785,
      "learning_rate": 8.048670453506074e-06,
      "loss": 0.2652,
      "step": 2442
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6075720291080648,
      "learning_rate": 8.047027031069126e-06,
      "loss": 0.248,
      "step": 2443
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4083106264444571,
      "learning_rate": 8.045383084812907e-06,
      "loss": 0.197,
      "step": 2444
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5843394833906044,
      "learning_rate": 8.04373861502003e-06,
      "loss": 0.2188,
      "step": 2445
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.5910168239972435,
      "learning_rate": 8.0420936219732e-06,
      "loss": 0.713,
      "step": 2446
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.280297873611833,
      "learning_rate": 8.040448105955209e-06,
      "loss": 0.5885,
      "step": 2447
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.135982103954657,
      "learning_rate": 8.03880206724894e-06,
      "loss": 0.6024,
      "step": 2448
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3965913060827118,
      "learning_rate": 8.037155506137367e-06,
      "loss": 0.2265,
      "step": 2449
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2638697252752737,
      "learning_rate": 8.035508422903554e-06,
      "loss": 0.1884,
      "step": 2450
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.3021555448151054,
      "learning_rate": 8.033860817830651e-06,
      "loss": 0.3072,
      "step": 2451
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2537210155281615,
      "learning_rate": 8.0322126912019e-06,
      "loss": 0.2008,
      "step": 2452
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5949356804278476,
      "learning_rate": 8.03056404330064e-06,
      "loss": 0.2197,
      "step": 2453
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5761209451287153,
      "learning_rate": 8.028914874410284e-06,
      "loss": 0.22,
      "step": 2454
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4286956354741476,
      "learning_rate": 8.027265184814349e-06,
      "loss": 0.2283,
      "step": 2455
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3130230943768109,
      "learning_rate": 8.025614974796432e-06,
      "loss": 0.1994,
      "step": 2456
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5262508710411224,
      "learning_rate": 8.023964244640225e-06,
      "loss": 0.2402,
      "step": 2457
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3396331072921988,
      "learning_rate": 8.022312994629508e-06,
      "loss": 0.227,
      "step": 2458
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.36169374257355,
      "learning_rate": 8.020661225048149e-06,
      "loss": 0.2177,
      "step": 2459
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.553834703718132,
      "learning_rate": 8.019008936180108e-06,
      "loss": 0.1934,
      "step": 2460
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5261428598913969,
      "learning_rate": 8.017356128309432e-06,
      "loss": 0.2487,
      "step": 2461
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5251332198348557,
      "learning_rate": 8.015702801720255e-06,
      "loss": 0.2767,
      "step": 2462
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2332697747690757,
      "learning_rate": 8.014048956696807e-06,
      "loss": 0.1824,
      "step": 2463
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3267112713410583,
      "learning_rate": 8.012394593523403e-06,
      "loss": 0.1772,
      "step": 2464
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5613258059802342,
      "learning_rate": 8.010739712484443e-06,
      "loss": 0.2525,
      "step": 2465
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4775904373490294,
      "learning_rate": 8.009084313864424e-06,
      "loss": 0.1879,
      "step": 2466
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2831340480347473,
      "learning_rate": 8.007428397947926e-06,
      "loss": 0.1871,
      "step": 2467
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.516794180938716,
      "learning_rate": 8.005771965019622e-06,
      "loss": 0.1915,
      "step": 2468
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.55801971023651,
      "learning_rate": 8.004115015364273e-06,
      "loss": 0.2746,
      "step": 2469
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8188252863344185,
      "learning_rate": 8.002457549266725e-06,
      "loss": 0.2728,
      "step": 2470
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4888235705440023,
      "learning_rate": 8.000799567011916e-06,
      "loss": 0.2735,
      "step": 2471
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.483907656046454,
      "learning_rate": 7.999141068884873e-06,
      "loss": 0.206,
      "step": 2472
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7796911821610109,
      "learning_rate": 7.997482055170712e-06,
      "loss": 0.2959,
      "step": 2473
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3615216463713002,
      "learning_rate": 7.995822526154636e-06,
      "loss": 0.2352,
      "step": 2474
    },
    {
      "epoch": 0.32,
      "grad_norm": 26.520555923768583,
      "learning_rate": 7.994162482121936e-06,
      "loss": 0.5396,
      "step": 2475
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4910626729783651,
      "learning_rate": 7.992501923357996e-06,
      "loss": 0.2313,
      "step": 2476
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.9535386348017,
      "learning_rate": 7.990840850148283e-06,
      "loss": 0.6061,
      "step": 2477
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4337967280948054,
      "learning_rate": 7.989179262778356e-06,
      "loss": 0.2387,
      "step": 2478
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.422394229385491,
      "learning_rate": 7.987517161533858e-06,
      "loss": 0.6149,
      "step": 2479
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4445491945884117,
      "learning_rate": 7.985854546700526e-06,
      "loss": 0.2365,
      "step": 2480
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4519167864696834,
      "learning_rate": 7.984191418564183e-06,
      "loss": 0.2281,
      "step": 2481
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.325774575097178,
      "learning_rate": 7.982527777410738e-06,
      "loss": 0.1607,
      "step": 2482
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3925868791755964,
      "learning_rate": 7.980863623526195e-06,
      "loss": 0.1652,
      "step": 2483
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5055195248002058,
      "learning_rate": 7.979198957196634e-06,
      "loss": 0.2187,
      "step": 2484
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3557362765943948,
      "learning_rate": 7.977533778708237e-06,
      "loss": 0.1823,
      "step": 2485
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.333239829236841,
      "learning_rate": 7.975868088347263e-06,
      "loss": 0.2097,
      "step": 2486
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3931060457154185,
      "learning_rate": 7.974201886400066e-06,
      "loss": 0.1928,
      "step": 2487
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3235322926697906,
      "learning_rate": 7.972535173153087e-06,
      "loss": 0.2242,
      "step": 2488
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3286344702105541,
      "learning_rate": 7.970867948892849e-06,
      "loss": 0.2289,
      "step": 2489
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4424601918449431,
      "learning_rate": 7.96920021390597e-06,
      "loss": 0.2352,
      "step": 2490
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4649481854063777,
      "learning_rate": 7.967531968479152e-06,
      "loss": 0.2002,
      "step": 2491
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.3573245706254635,
      "learning_rate": 7.965863212899185e-06,
      "loss": 0.5579,
      "step": 2492
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5189920398179653,
      "learning_rate": 7.96419394745295e-06,
      "loss": 0.2365,
      "step": 2493
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4752330756327043,
      "learning_rate": 7.962524172427412e-06,
      "loss": 0.2256,
      "step": 2494
    },
    {
      "epoch": 0.32,
      "grad_norm": 12.926959000937511,
      "learning_rate": 7.960853888109624e-06,
      "loss": 0.643,
      "step": 2495
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4514213788979693,
      "learning_rate": 7.959183094786726e-06,
      "loss": 0.2334,
      "step": 2496
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4754325486713666,
      "learning_rate": 7.957511792745948e-06,
      "loss": 0.2202,
      "step": 2497
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.376474111951715,
      "learning_rate": 7.955839982274608e-06,
      "loss": 0.2091,
      "step": 2498
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4689979466248557,
      "learning_rate": 7.954167663660108e-06,
      "loss": 0.2276,
      "step": 2499
    },
    {
      "epoch": 0.32,
      "grad_norm": 15.299986642208728,
      "learning_rate": 7.952494837189937e-06,
      "loss": 0.6568,
      "step": 2500
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5978217816636828,
      "learning_rate": 7.950821503151678e-06,
      "loss": 0.2547,
      "step": 2501
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.399727595471494,
      "learning_rate": 7.949147661832992e-06,
      "loss": 0.1893,
      "step": 2502
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3394795576525274,
      "learning_rate": 7.947473313521635e-06,
      "loss": 0.2018,
      "step": 2503
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4146504447352992,
      "learning_rate": 7.945798458505442e-06,
      "loss": 0.2423,
      "step": 2504
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4154662485909866,
      "learning_rate": 7.944123097072348e-06,
      "loss": 0.2239,
      "step": 2505
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3222794682936243,
      "learning_rate": 7.942447229510359e-06,
      "loss": 0.1854,
      "step": 2506
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.399077598333623,
      "learning_rate": 7.94077085610758e-06,
      "loss": 0.2263,
      "step": 2507
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.391602059507044,
      "learning_rate": 7.9390939771522e-06,
      "loss": 0.2016,
      "step": 2508
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.32398216261596,
      "learning_rate": 7.93741659293249e-06,
      "loss": 0.6889,
      "step": 2509
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.143402992838027,
      "learning_rate": 7.935738703736817e-06,
      "loss": 0.573,
      "step": 2510
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.82310829094551,
      "learning_rate": 7.934060309853627e-06,
      "loss": 0.6806,
      "step": 2511
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4712118361311866,
      "learning_rate": 7.932381411571453e-06,
      "loss": 0.272,
      "step": 2512
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6580286565452371,
      "learning_rate": 7.930702009178919e-06,
      "loss": 0.2501,
      "step": 2513
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6374683604465994,
      "learning_rate": 7.929022102964736e-06,
      "loss": 0.2494,
      "step": 2514
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.635141715820675,
      "learning_rate": 7.927341693217697e-06,
      "loss": 0.7192,
      "step": 2515
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6603583718125314,
      "learning_rate": 7.925660780226683e-06,
      "loss": 0.2126,
      "step": 2516
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5401590378709087,
      "learning_rate": 7.923979364280664e-06,
      "loss": 0.2358,
      "step": 2517
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.224932074244137,
      "learning_rate": 7.922297445668695e-06,
      "loss": 0.467,
      "step": 2518
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.527773242980428,
      "learning_rate": 7.920615024679917e-06,
      "loss": 0.6924,
      "step": 2519
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3739720958360455,
      "learning_rate": 7.918932101603556e-06,
      "loss": 0.1513,
      "step": 2520
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5137651979977067,
      "learning_rate": 7.917248676728929e-06,
      "loss": 0.2702,
      "step": 2521
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4489087789194115,
      "learning_rate": 7.915564750345436e-06,
      "loss": 0.2001,
      "step": 2522
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.476006622071491,
      "learning_rate": 7.91388032274256e-06,
      "loss": 0.2234,
      "step": 2523
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.684654247922967,
      "learning_rate": 7.912195394209877e-06,
      "loss": 0.5681,
      "step": 2524
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3159330210706581,
      "learning_rate": 7.910509965037045e-06,
      "loss": 0.1689,
      "step": 2525
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5174625248123128,
      "learning_rate": 7.90882403551381e-06,
      "loss": 0.269,
      "step": 2526
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.22340314789877,
      "learning_rate": 7.907137605929998e-06,
      "loss": 0.2024,
      "step": 2527
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.760209800735942,
      "learning_rate": 7.90545067657553e-06,
      "loss": 0.5776,
      "step": 2528
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.062679586482424,
      "learning_rate": 7.903763247740407e-06,
      "loss": 0.5436,
      "step": 2529
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2837217646444778,
      "learning_rate": 7.902075319714722e-06,
      "loss": 0.209,
      "step": 2530
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3899170679326083,
      "learning_rate": 7.900386892788644e-06,
      "loss": 0.2075,
      "step": 2531
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.839151300127567,
      "learning_rate": 7.898697967252433e-06,
      "loss": 0.3989,
      "step": 2532
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6280356461731993,
      "learning_rate": 7.897008543396438e-06,
      "loss": 0.187,
      "step": 2533
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3746445692227427,
      "learning_rate": 7.89531862151109e-06,
      "loss": 0.2061,
      "step": 2534
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5202902201414001,
      "learning_rate": 7.893628201886906e-06,
      "loss": 0.1977,
      "step": 2535
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5084470294743682,
      "learning_rate": 7.891937284814489e-06,
      "loss": 0.2718,
      "step": 2536
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3678449364973373,
      "learning_rate": 7.890245870584523e-06,
      "loss": 0.1756,
      "step": 2537
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.791680457400534,
      "learning_rate": 7.888553959487788e-06,
      "loss": 0.5715,
      "step": 2538
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.570854130824897,
      "learning_rate": 7.886861551815139e-06,
      "loss": 0.2302,
      "step": 2539
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3855414402809172,
      "learning_rate": 7.885168647857523e-06,
      "loss": 0.2036,
      "step": 2540
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2861567493410835,
      "learning_rate": 7.88347524790597e-06,
      "loss": 0.1864,
      "step": 2541
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.342036562923758,
      "learning_rate": 7.881781352251591e-06,
      "loss": 0.2403,
      "step": 2542
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4164247678969488,
      "learning_rate": 7.88008696118559e-06,
      "loss": 0.2466,
      "step": 2543
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.479825785437264,
      "learning_rate": 7.878392074999252e-06,
      "loss": 0.5818,
      "step": 2544
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.305512397786445,
      "learning_rate": 7.876696693983947e-06,
      "loss": 0.1567,
      "step": 2545
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5263080620958596,
      "learning_rate": 7.875000818431132e-06,
      "loss": 0.2177,
      "step": 2546
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.431310622372571,
      "learning_rate": 7.873304448632345e-06,
      "loss": 0.2416,
      "step": 2547
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.596476483143075,
      "learning_rate": 7.871607584879211e-06,
      "loss": 0.265,
      "step": 2548
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3562993684304048,
      "learning_rate": 7.869910227463447e-06,
      "loss": 0.2307,
      "step": 2549
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4037058966822884,
      "learning_rate": 7.86821237667684e-06,
      "loss": 0.2239,
      "step": 2550
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4825562495294455,
      "learning_rate": 7.866514032811276e-06,
      "loss": 0.2019,
      "step": 2551
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6899596530486023,
      "learning_rate": 7.864815196158718e-06,
      "loss": 0.2136,
      "step": 2552
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2880228754720184,
      "learning_rate": 7.863115867011217e-06,
      "loss": 0.1875,
      "step": 2553
    },
    {
      "epoch": 0.33,
      "grad_norm": 10.934698979450449,
      "learning_rate": 7.861416045660906e-06,
      "loss": 0.6665,
      "step": 2554
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4193744373104702,
      "learning_rate": 7.859715732400004e-06,
      "loss": 0.1848,
      "step": 2555
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1950885161363765,
      "learning_rate": 7.858014927520815e-06,
      "loss": 0.1881,
      "step": 2556
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5424281113311273,
      "learning_rate": 7.856313631315726e-06,
      "loss": 0.2204,
      "step": 2557
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5334198232798562,
      "learning_rate": 7.854611844077213e-06,
      "loss": 0.2304,
      "step": 2558
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4479128118603635,
      "learning_rate": 7.852909566097828e-06,
      "loss": 0.2674,
      "step": 2559
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4259457074056556,
      "learning_rate": 7.851206797670217e-06,
      "loss": 0.2419,
      "step": 2560
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2648239989487362,
      "learning_rate": 7.849503539087102e-06,
      "loss": 0.1908,
      "step": 2561
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3989450398556982,
      "learning_rate": 7.847799790641295e-06,
      "loss": 0.2217,
      "step": 2562
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5534494946758535,
      "learning_rate": 7.846095552625688e-06,
      "loss": 0.2439,
      "step": 2563
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3203731057482668,
      "learning_rate": 7.844390825333264e-06,
      "loss": 0.1906,
      "step": 2564
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.370967967021711,
      "learning_rate": 7.842685609057078e-06,
      "loss": 0.2185,
      "step": 2565
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.2922882893108385,
      "learning_rate": 7.840979904090282e-06,
      "loss": 0.4958,
      "step": 2566
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.7758690601698355,
      "learning_rate": 7.839273710726107e-06,
      "loss": 0.5732,
      "step": 2567
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.49797860952309,
      "learning_rate": 7.837567029257864e-06,
      "loss": 0.1766,
      "step": 2568
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.603752953662032,
      "learning_rate": 7.835859859978952e-06,
      "loss": 0.2391,
      "step": 2569
    },
    {
      "epoch": 0.33,
      "grad_norm": 10.361604700293,
      "learning_rate": 7.834152203182854e-06,
      "loss": 0.4888,
      "step": 2570
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.446654949575969,
      "learning_rate": 7.832444059163137e-06,
      "loss": 0.1995,
      "step": 2571
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3642809970332885,
      "learning_rate": 7.830735428213451e-06,
      "loss": 0.2094,
      "step": 2572
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4344484035170086,
      "learning_rate": 7.829026310627526e-06,
      "loss": 0.1804,
      "step": 2573
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4412902554220204,
      "learning_rate": 7.827316706699182e-06,
      "loss": 0.2227,
      "step": 2574
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.50448652686267,
      "learning_rate": 7.825606616722319e-06,
      "loss": 0.4699,
      "step": 2575
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4526950270120444,
      "learning_rate": 7.823896040990922e-06,
      "loss": 0.2174,
      "step": 2576
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3285037663412123,
      "learning_rate": 7.822184979799057e-06,
      "loss": 0.1898,
      "step": 2577
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4873033294974536,
      "learning_rate": 7.820473433440876e-06,
      "loss": 0.2301,
      "step": 2578
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.194003563762566,
      "learning_rate": 7.818761402210616e-06,
      "loss": 0.5014,
      "step": 2579
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5432666575547265,
      "learning_rate": 7.817048886402591e-06,
      "loss": 0.2306,
      "step": 2580
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6238149568581273,
      "learning_rate": 7.815335886311205e-06,
      "loss": 0.2372,
      "step": 2581
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4317229266102578,
      "learning_rate": 7.813622402230943e-06,
      "loss": 0.2107,
      "step": 2582
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4529312245711825,
      "learning_rate": 7.811908434456372e-06,
      "loss": 0.2945,
      "step": 2583
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.456665422048814,
      "learning_rate": 7.810193983282142e-06,
      "loss": 0.2598,
      "step": 2584
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3810617408337882,
      "learning_rate": 7.808479049002988e-06,
      "loss": 0.2358,
      "step": 2585
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.425753540585918,
      "learning_rate": 7.806763631913726e-06,
      "loss": 0.2232,
      "step": 2586
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4508042026496306,
      "learning_rate": 7.805047732309257e-06,
      "loss": 0.226,
      "step": 2587
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.1812232900757085,
      "learning_rate": 7.803331350484563e-06,
      "loss": 0.7128,
      "step": 2588
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3881504749990308,
      "learning_rate": 7.801614486734712e-06,
      "loss": 0.1904,
      "step": 2589
    },
    {
      "epoch": 0.33,
      "grad_norm": 15.542615269259425,
      "learning_rate": 7.799897141354854e-06,
      "loss": 0.4977,
      "step": 2590
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.607177155365164,
      "learning_rate": 7.798179314640214e-06,
      "loss": 0.5898,
      "step": 2591
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5686028069904463,
      "learning_rate": 7.79646100688611e-06,
      "loss": 0.257,
      "step": 2592
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.416805536622159,
      "learning_rate": 7.794742218387943e-06,
      "loss": 0.6852,
      "step": 2593
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.411206140682198,
      "learning_rate": 7.793022949441189e-06,
      "loss": 0.2095,
      "step": 2594
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3281912326081444,
      "learning_rate": 7.79130320034141e-06,
      "loss": 0.1821,
      "step": 2595
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.309334025839328,
      "learning_rate": 7.789582971384252e-06,
      "loss": 0.1771,
      "step": 2596
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4050014716211825,
      "learning_rate": 7.787862262865438e-06,
      "loss": 0.2064,
      "step": 2597
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3278115860613193,
      "learning_rate": 7.786141075080786e-06,
      "loss": 0.2133,
      "step": 2598
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6100095786516093,
      "learning_rate": 7.784419408326181e-06,
      "loss": 0.2019,
      "step": 2599
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.366961109027683,
      "learning_rate": 7.782697262897601e-06,
      "loss": 0.2132,
      "step": 2600
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4865974832011353,
      "learning_rate": 7.7809746390911e-06,
      "loss": 0.2631,
      "step": 2601
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.66313048493537,
      "learning_rate": 7.77925153720282e-06,
      "loss": 0.7536,
      "step": 2602
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.411112940219534,
      "learning_rate": 7.777527957528982e-06,
      "loss": 0.1697,
      "step": 2603
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4417854624962894,
      "learning_rate": 7.775803900365885e-06,
      "loss": 0.2373,
      "step": 2604
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3658453998878695,
      "learning_rate": 7.77407936600992e-06,
      "loss": 0.232,
      "step": 2605
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.381605562097311,
      "learning_rate": 7.77235435475755e-06,
      "loss": 0.256,
      "step": 2606
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9411441768152737,
      "learning_rate": 7.77062886690533e-06,
      "loss": 0.2466,
      "step": 2607
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6584046631234153,
      "learning_rate": 7.768902902749885e-06,
      "loss": 0.2046,
      "step": 2608
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6346191140115884,
      "learning_rate": 7.767176462587932e-06,
      "loss": 0.2415,
      "step": 2609
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3354945066588808,
      "learning_rate": 7.765449546716266e-06,
      "loss": 0.2103,
      "step": 2610
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3760337977514845,
      "learning_rate": 7.76372215543176e-06,
      "loss": 0.1995,
      "step": 2611
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4656072372661109,
      "learning_rate": 7.76199428903138e-06,
      "loss": 0.2322,
      "step": 2612
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2478910564600338,
      "learning_rate": 7.76026594781216e-06,
      "loss": 0.1681,
      "step": 2613
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.416210434267406,
      "learning_rate": 7.758537132071224e-06,
      "loss": 0.2091,
      "step": 2614
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4022320100259795,
      "learning_rate": 7.756807842105774e-06,
      "loss": 0.2368,
      "step": 2615
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5241633295438797,
      "learning_rate": 7.755078078213099e-06,
      "loss": 0.2431,
      "step": 2616
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4007270687402873,
      "learning_rate": 7.75334784069056e-06,
      "loss": 0.2374,
      "step": 2617
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.178853530730235,
      "learning_rate": 7.75161712983561e-06,
      "loss": 0.1758,
      "step": 2618
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3574804579048212,
      "learning_rate": 7.749885945945774e-06,
      "loss": 0.2234,
      "step": 2619
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.537115833100029,
      "learning_rate": 7.748154289318666e-06,
      "loss": 0.2569,
      "step": 2620
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4057720323526786,
      "learning_rate": 7.746422160251975e-06,
      "loss": 0.1888,
      "step": 2621
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2254120248503069,
      "learning_rate": 7.744689559043475e-06,
      "loss": 0.1345,
      "step": 2622
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3114053309376583,
      "learning_rate": 7.742956485991022e-06,
      "loss": 0.1711,
      "step": 2623
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3578783209511045,
      "learning_rate": 7.74122294139255e-06,
      "loss": 0.193,
      "step": 2624
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4716840680816814,
      "learning_rate": 7.739488925546074e-06,
      "loss": 0.23,
      "step": 2625
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.377411912661415,
      "learning_rate": 7.737754438749694e-06,
      "loss": 0.2325,
      "step": 2626
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4699454491966846,
      "learning_rate": 7.736019481301587e-06,
      "loss": 0.2498,
      "step": 2627
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.245060930139345,
      "learning_rate": 7.734284053500015e-06,
      "loss": 0.1735,
      "step": 2628
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3092015112765685,
      "learning_rate": 7.732548155643311e-06,
      "loss": 0.1735,
      "step": 2629
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.53833712195927,
      "learning_rate": 7.730811788029903e-06,
      "loss": 0.1889,
      "step": 2630
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4702839313345464,
      "learning_rate": 7.72907495095829e-06,
      "loss": 0.2319,
      "step": 2631
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4620517051954083,
      "learning_rate": 7.727337644727055e-06,
      "loss": 0.2333,
      "step": 2632
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6214810871039167,
      "learning_rate": 7.725599869634861e-06,
      "loss": 0.1912,
      "step": 2633
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5657643650189275,
      "learning_rate": 7.72386162598045e-06,
      "loss": 0.2406,
      "step": 2634
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5077497191623197,
      "learning_rate": 7.722122914062649e-06,
      "loss": 0.2176,
      "step": 2635
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3948562285636277,
      "learning_rate": 7.720383734180362e-06,
      "loss": 0.1716,
      "step": 2636
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.3204530471756595,
      "learning_rate": 7.718644086632575e-06,
      "loss": 0.4431,
      "step": 2637
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.538701300303574,
      "learning_rate": 7.71690397171835e-06,
      "loss": 0.5244,
      "step": 2638
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.748936495447719,
      "learning_rate": 7.715163389736834e-06,
      "loss": 0.6745,
      "step": 2639
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4503094812810278,
      "learning_rate": 7.713422340987258e-06,
      "loss": 0.2135,
      "step": 2640
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5679426828969927,
      "learning_rate": 7.711680825768922e-06,
      "loss": 0.2297,
      "step": 2641
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5369629808057694,
      "learning_rate": 7.709938844381217e-06,
      "loss": 0.2083,
      "step": 2642
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.331181315377564,
      "learning_rate": 7.708196397123607e-06,
      "loss": 0.183,
      "step": 2643
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3521559579339346,
      "learning_rate": 7.706453484295643e-06,
      "loss": 0.2151,
      "step": 2644
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5940986948889586,
      "learning_rate": 7.704710106196947e-06,
      "loss": 0.2313,
      "step": 2645
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3307036156765049,
      "learning_rate": 7.702966263127227e-06,
      "loss": 0.1888,
      "step": 2646
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3335387012191093,
      "learning_rate": 7.701221955386273e-06,
      "loss": 0.1828,
      "step": 2647
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5218279401689987,
      "learning_rate": 7.699477183273949e-06,
      "loss": 0.2484,
      "step": 2648
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2814774733822831,
      "learning_rate": 7.6977319470902e-06,
      "loss": 0.197,
      "step": 2649
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4733220260573816,
      "learning_rate": 7.695986247135058e-06,
      "loss": 0.2861,
      "step": 2650
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3919040748119555,
      "learning_rate": 7.694240083708623e-06,
      "loss": 0.2238,
      "step": 2651
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.653069717226565,
      "learning_rate": 7.692493457111085e-06,
      "loss": 0.5203,
      "step": 2652
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4610451914729607,
      "learning_rate": 7.690746367642707e-06,
      "loss": 0.244,
      "step": 2653
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5671912225721913,
      "learning_rate": 7.688998815603837e-06,
      "loss": 0.261,
      "step": 2654
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3946440795062627,
      "learning_rate": 7.687250801294893e-06,
      "loss": 0.1861,
      "step": 2655
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6081373200330822,
      "learning_rate": 7.685502325016384e-06,
      "loss": 0.225,
      "step": 2656
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2228740204358381,
      "learning_rate": 7.683753387068894e-06,
      "loss": 0.1585,
      "step": 2657
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4390968421924908,
      "learning_rate": 7.682003987753082e-06,
      "loss": 0.2186,
      "step": 2658
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4406654207228176,
      "learning_rate": 7.680254127369695e-06,
      "loss": 0.2155,
      "step": 2659
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.530377920870163,
      "learning_rate": 7.67850380621955e-06,
      "loss": 0.2042,
      "step": 2660
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4010587169188926,
      "learning_rate": 7.676753024603546e-06,
      "loss": 0.1636,
      "step": 2661
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7720881639530854,
      "learning_rate": 7.675001782822669e-06,
      "loss": 0.2516,
      "step": 2662
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4263823786097012,
      "learning_rate": 7.673250081177974e-06,
      "loss": 0.2185,
      "step": 2663
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.469409703059715,
      "learning_rate": 7.671497919970598e-06,
      "loss": 0.7189,
      "step": 2664
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.431741631260665,
      "learning_rate": 7.669745299501758e-06,
      "loss": 0.2295,
      "step": 2665
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4646541397646555,
      "learning_rate": 7.667992220072751e-06,
      "loss": 0.2528,
      "step": 2666
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.837415095255308,
      "learning_rate": 7.666238681984952e-06,
      "loss": 0.7176,
      "step": 2667
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7134321173994378,
      "learning_rate": 7.664484685539814e-06,
      "loss": 0.2513,
      "step": 2668
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.601514386056483,
      "learning_rate": 7.662730231038869e-06,
      "loss": 0.6302,
      "step": 2669
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4397936577092747,
      "learning_rate": 7.66097531878373e-06,
      "loss": 0.2145,
      "step": 2670
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.506832682131831,
      "learning_rate": 7.659219949076082e-06,
      "loss": 0.2209,
      "step": 2671
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5008758518779504,
      "learning_rate": 7.657464122217699e-06,
      "loss": 0.2485,
      "step": 2672
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3094197178855278,
      "learning_rate": 7.655707838510423e-06,
      "loss": 0.2208,
      "step": 2673
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.754915324213935,
      "learning_rate": 7.653951098256184e-06,
      "loss": 0.753,
      "step": 2674
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3557531074568498,
      "learning_rate": 7.652193901756983e-06,
      "loss": 0.2352,
      "step": 2675
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.881294972663594,
      "learning_rate": 7.650436249314902e-06,
      "loss": 0.6321,
      "step": 2676
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3282367081391808,
      "learning_rate": 7.648678141232107e-06,
      "loss": 0.2211,
      "step": 2677
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.063468706877666,
      "learning_rate": 7.646919577810831e-06,
      "loss": 0.537,
      "step": 2678
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.460285831737244,
      "learning_rate": 7.645160559353392e-06,
      "loss": 0.2359,
      "step": 2679
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3798733139069113,
      "learning_rate": 7.64340108616219e-06,
      "loss": 0.2157,
      "step": 2680
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2805890802985564,
      "learning_rate": 7.641641158539696e-06,
      "loss": 0.2186,
      "step": 2681
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4513547949492356,
      "learning_rate": 7.639880776788464e-06,
      "loss": 0.2031,
      "step": 2682
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3878014379935117,
      "learning_rate": 7.63811994121112e-06,
      "loss": 0.2149,
      "step": 2683
    },
    {
      "epoch": 0.34,
      "grad_norm": 10.714097977392534,
      "learning_rate": 7.636358652110375e-06,
      "loss": 0.7219,
      "step": 2684
    },
    {
      "epoch": 0.34,
      "grad_norm": 9.405418419203233,
      "learning_rate": 7.634596909789014e-06,
      "loss": 0.5306,
      "step": 2685
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3693559868385758,
      "learning_rate": 7.632834714549902e-06,
      "loss": 0.2291,
      "step": 2686
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3269534316007903,
      "learning_rate": 7.63107206669598e-06,
      "loss": 0.1844,
      "step": 2687
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4630674281928586,
      "learning_rate": 7.629308966530268e-06,
      "loss": 0.2423,
      "step": 2688
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.248521080293325,
      "learning_rate": 7.6275454143558614e-06,
      "loss": 0.6835,
      "step": 2689
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6453743637126694,
      "learning_rate": 7.6257814104759385e-06,
      "loss": 0.229,
      "step": 2690
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4600995894066782,
      "learning_rate": 7.62401695519375e-06,
      "loss": 0.2503,
      "step": 2691
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4613600986710076,
      "learning_rate": 7.6222520488126286e-06,
      "loss": 0.1927,
      "step": 2692
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8121162964773072,
      "learning_rate": 7.620486691635978e-06,
      "loss": 0.2283,
      "step": 2693
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4922622483318995,
      "learning_rate": 7.618720883967285e-06,
      "loss": 0.2267,
      "step": 2694
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5738748542199068,
      "learning_rate": 7.616954626110116e-06,
      "loss": 0.2399,
      "step": 2695
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3710526902616385,
      "learning_rate": 7.615187918368107e-06,
      "loss": 0.2002,
      "step": 2696
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3613664600453241,
      "learning_rate": 7.613420761044976e-06,
      "loss": 0.2073,
      "step": 2697
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3547392636207491,
      "learning_rate": 7.611653154444521e-06,
      "loss": 0.2182,
      "step": 2698
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4160903367935356,
      "learning_rate": 7.60988509887061e-06,
      "loss": 0.2272,
      "step": 2699
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2658459911018907,
      "learning_rate": 7.608116594627195e-06,
      "loss": 0.1931,
      "step": 2700
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.083127765843589,
      "learning_rate": 7.606347642018301e-06,
      "loss": 0.6955,
      "step": 2701
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2882462555550225,
      "learning_rate": 7.604578241348033e-06,
      "loss": 0.183,
      "step": 2702
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.9947141376645554,
      "learning_rate": 7.602808392920568e-06,
      "loss": 0.5842,
      "step": 2703
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.458283862685475,
      "learning_rate": 7.601038097040169e-06,
      "loss": 0.4933,
      "step": 2704
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.075532097732124,
      "learning_rate": 7.599267354011166e-06,
      "loss": 0.6639,
      "step": 2705
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3133478159371874,
      "learning_rate": 7.597496164137972e-06,
      "loss": 0.1823,
      "step": 2706
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4019146505129476,
      "learning_rate": 7.595724527725074e-06,
      "loss": 0.2321,
      "step": 2707
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.696965559353475,
      "learning_rate": 7.593952445077035e-06,
      "loss": 0.6769,
      "step": 2708
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3755513386854101,
      "learning_rate": 7.592179916498503e-06,
      "loss": 0.2439,
      "step": 2709
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4471602283022738,
      "learning_rate": 7.590406942294191e-06,
      "loss": 0.2756,
      "step": 2710
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5985708302477701,
      "learning_rate": 7.588633522768896e-06,
      "loss": 0.2584,
      "step": 2711
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.503537172743101,
      "learning_rate": 7.586859658227487e-06,
      "loss": 0.5756,
      "step": 2712
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.474873191559995,
      "learning_rate": 7.585085348974912e-06,
      "loss": 0.2308,
      "step": 2713
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4002000175286338,
      "learning_rate": 7.5833105953161986e-06,
      "loss": 0.252,
      "step": 2714
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.947748049713727,
      "learning_rate": 7.581535397556445e-06,
      "loss": 0.468,
      "step": 2715
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5194544614831045,
      "learning_rate": 7.579759756000829e-06,
      "loss": 0.212,
      "step": 2716
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2637453138063002,
      "learning_rate": 7.577983670954604e-06,
      "loss": 0.215,
      "step": 2717
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.804712392753616,
      "learning_rate": 7.5762071427231e-06,
      "loss": 0.4351,
      "step": 2718
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.39222824377265,
      "learning_rate": 7.5744301716117206e-06,
      "loss": 0.1793,
      "step": 2719
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4779029754113544,
      "learning_rate": 7.572652757925951e-06,
      "loss": 0.263,
      "step": 2720
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.353818866378113,
      "learning_rate": 7.570874901971347e-06,
      "loss": 0.5388,
      "step": 2721
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.595541235668491,
      "learning_rate": 7.569096604053543e-06,
      "loss": 0.561,
      "step": 2722
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5244229477350215,
      "learning_rate": 7.5673178644782475e-06,
      "loss": 0.1782,
      "step": 2723
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4369991949190817,
      "learning_rate": 7.565538683551251e-06,
      "loss": 0.197,
      "step": 2724
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4212862120104692,
      "learning_rate": 7.563759061578412e-06,
      "loss": 0.229,
      "step": 2725
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5236396648983148,
      "learning_rate": 7.561978998865667e-06,
      "loss": 0.2234,
      "step": 2726
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4632873269783049,
      "learning_rate": 7.560198495719032e-06,
      "loss": 0.2059,
      "step": 2727
    },
    {
      "epoch": 0.35,
      "grad_norm": 9.920586626438988,
      "learning_rate": 7.558417552444595e-06,
      "loss": 0.6092,
      "step": 2728
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4260978351261242,
      "learning_rate": 7.556636169348524e-06,
      "loss": 0.1986,
      "step": 2729
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3748723271152576,
      "learning_rate": 7.554854346737053e-06,
      "loss": 0.2312,
      "step": 2730
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.438455704610008,
      "learning_rate": 7.5530720849165e-06,
      "loss": 0.1796,
      "step": 2731
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.39550127050478,
      "learning_rate": 7.551289384193261e-06,
      "loss": 0.1932,
      "step": 2732
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.303539141016156,
      "learning_rate": 7.549506244873799e-06,
      "loss": 0.1985,
      "step": 2733
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.559373210658666,
      "learning_rate": 7.547722667264656e-06,
      "loss": 0.2418,
      "step": 2734
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4722854408166035,
      "learning_rate": 7.545938651672453e-06,
      "loss": 0.2404,
      "step": 2735
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.252499581901719,
      "learning_rate": 7.5441541984038795e-06,
      "loss": 0.2049,
      "step": 2736
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.240988611551327,
      "learning_rate": 7.542369307765705e-06,
      "loss": 0.6754,
      "step": 2737
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.327179076042934,
      "learning_rate": 7.540583980064772e-06,
      "loss": 0.1803,
      "step": 2738
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4908096306721388,
      "learning_rate": 7.538798215608001e-06,
      "loss": 0.23,
      "step": 2739
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.117655952027422,
      "learning_rate": 7.537012014702383e-06,
      "loss": 0.6627,
      "step": 2740
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.0282687427913455,
      "learning_rate": 7.535225377654989e-06,
      "loss": 0.5658,
      "step": 2741
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1314942826839671,
      "learning_rate": 7.533438304772962e-06,
      "loss": 0.1935,
      "step": 2742
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4050394599858107,
      "learning_rate": 7.531650796363521e-06,
      "loss": 0.1998,
      "step": 2743
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3097407433365313,
      "learning_rate": 7.529862852733958e-06,
      "loss": 0.2284,
      "step": 2744
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.365365159764974,
      "learning_rate": 7.52807447419164e-06,
      "loss": 0.2096,
      "step": 2745
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.089999572804573,
      "learning_rate": 7.526285661044015e-06,
      "loss": 0.6128,
      "step": 2746
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3973016593172622,
      "learning_rate": 7.524496413598596e-06,
      "loss": 0.2272,
      "step": 2747
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5307912826378012,
      "learning_rate": 7.522706732162976e-06,
      "loss": 0.1792,
      "step": 2748
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4782698368010347,
      "learning_rate": 7.520916617044826e-06,
      "loss": 0.1906,
      "step": 2749
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5208983210658196,
      "learning_rate": 7.519126068551882e-06,
      "loss": 0.2311,
      "step": 2750
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3988091474206104,
      "learning_rate": 7.517335086991963e-06,
      "loss": 0.227,
      "step": 2751
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3335915358391557,
      "learning_rate": 7.51554367267296e-06,
      "loss": 0.2352,
      "step": 2752
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.476700264220967,
      "learning_rate": 7.513751825902837e-06,
      "loss": 0.2508,
      "step": 2753
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5030071130509326,
      "learning_rate": 7.5119595469896325e-06,
      "loss": 0.2024,
      "step": 2754
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.325485317272901,
      "learning_rate": 7.51016683624146e-06,
      "loss": 0.1678,
      "step": 2755
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4557928471092363,
      "learning_rate": 7.50837369396651e-06,
      "loss": 0.2517,
      "step": 2756
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4479268760440995,
      "learning_rate": 7.506580120473042e-06,
      "loss": 0.1928,
      "step": 2757
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.671207434242252,
      "learning_rate": 7.504786116069391e-06,
      "loss": 0.6085,
      "step": 2758
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3259492029197673,
      "learning_rate": 7.502991681063969e-06,
      "loss": 0.1982,
      "step": 2759
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3592406503107124,
      "learning_rate": 7.5011968157652615e-06,
      "loss": 0.1861,
      "step": 2760
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.953998735739404,
      "learning_rate": 7.499401520481824e-06,
      "loss": 0.4779,
      "step": 2761
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.665672698468821,
      "learning_rate": 7.497605795522289e-06,
      "loss": 0.2341,
      "step": 2762
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.371365182556272,
      "learning_rate": 7.495809641195364e-06,
      "loss": 0.7601,
      "step": 2763
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4184489251307035,
      "learning_rate": 7.494013057809827e-06,
      "loss": 0.2136,
      "step": 2764
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.915653000633212,
      "learning_rate": 7.492216045674532e-06,
      "loss": 0.5734,
      "step": 2765
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4275370099949913,
      "learning_rate": 7.490418605098408e-06,
      "loss": 0.2588,
      "step": 2766
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4330445554095965,
      "learning_rate": 7.488620736390454e-06,
      "loss": 0.2428,
      "step": 2767
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.832996123885523,
      "learning_rate": 7.486822439859744e-06,
      "loss": 0.5718,
      "step": 2768
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6362915452198028,
      "learning_rate": 7.485023715815427e-06,
      "loss": 0.2384,
      "step": 2769
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.591257687399514,
      "learning_rate": 7.483224564566725e-06,
      "loss": 0.6325,
      "step": 2770
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.351416687425504,
      "learning_rate": 7.481424986422933e-06,
      "loss": 0.1495,
      "step": 2771
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4524084601693632,
      "learning_rate": 7.479624981693416e-06,
      "loss": 0.2257,
      "step": 2772
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4760185719430117,
      "learning_rate": 7.47782455068762e-06,
      "loss": 0.1815,
      "step": 2773
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4414235394001134,
      "learning_rate": 7.476023693715059e-06,
      "loss": 0.215,
      "step": 2774
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3778424959964561,
      "learning_rate": 7.47422241108532e-06,
      "loss": 0.2137,
      "step": 2775
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5283629998060417,
      "learning_rate": 7.4724207031080645e-06,
      "loss": 0.2088,
      "step": 2776
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3835951504420907,
      "learning_rate": 7.4706185700930266e-06,
      "loss": 0.2167,
      "step": 2777
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5351904253053505,
      "learning_rate": 7.4688160123500165e-06,
      "loss": 0.2075,
      "step": 2778
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.940713564528564,
      "learning_rate": 7.467013030188912e-06,
      "loss": 0.6337,
      "step": 2779
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.340732742409104,
      "learning_rate": 7.465209623919668e-06,
      "loss": 0.1881,
      "step": 2780
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4428780557676228,
      "learning_rate": 7.46340579385231e-06,
      "loss": 0.2193,
      "step": 2781
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.364016773197593,
      "learning_rate": 7.4616015402969375e-06,
      "loss": 0.2055,
      "step": 2782
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6018176477994348,
      "learning_rate": 7.459796863563723e-06,
      "loss": 0.2367,
      "step": 2783
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4571699003087584,
      "learning_rate": 7.457991763962912e-06,
      "loss": 0.1681,
      "step": 2784
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4599660691096217,
      "learning_rate": 7.456186241804821e-06,
      "loss": 0.1853,
      "step": 2785
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2946623496673941,
      "learning_rate": 7.45438029739984e-06,
      "loss": 0.248,
      "step": 2786
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4557890462373462,
      "learning_rate": 7.4525739310584326e-06,
      "loss": 0.2361,
      "step": 2787
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.609663019167578,
      "learning_rate": 7.450767143091133e-06,
      "loss": 0.2169,
      "step": 2788
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5783802284542365,
      "learning_rate": 7.448959933808552e-06,
      "loss": 0.1929,
      "step": 2789
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4056464992037305,
      "learning_rate": 7.447152303521366e-06,
      "loss": 0.2189,
      "step": 2790
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.28807760501585,
      "learning_rate": 7.445344252540327e-06,
      "loss": 0.4528,
      "step": 2791
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5795103951974225,
      "learning_rate": 7.443535781176265e-06,
      "loss": 0.2203,
      "step": 2792
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2134026035924173,
      "learning_rate": 7.441726889740075e-06,
      "loss": 0.1728,
      "step": 2793
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4311396912129968,
      "learning_rate": 7.439917578542723e-06,
      "loss": 0.2301,
      "step": 2794
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.950901045710248,
      "learning_rate": 7.438107847895256e-06,
      "loss": 0.6885,
      "step": 2795
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7340623730354474,
      "learning_rate": 7.436297698108783e-06,
      "loss": 0.2626,
      "step": 2796
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4062583545469245,
      "learning_rate": 7.434487129494494e-06,
      "loss": 0.195,
      "step": 2797
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2804721522433093,
      "learning_rate": 7.4326761423636415e-06,
      "loss": 0.1673,
      "step": 2798
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.882214538917467,
      "learning_rate": 7.430864737027561e-06,
      "loss": 0.3899,
      "step": 2799
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6929415965772412,
      "learning_rate": 7.429052913797649e-06,
      "loss": 0.2685,
      "step": 2800
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.930929025174639,
      "learning_rate": 7.4272406729853805e-06,
      "loss": 0.5714,
      "step": 2801
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.067654385949035,
      "learning_rate": 7.425428014902304e-06,
      "loss": 0.5442,
      "step": 2802
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.860566391212544,
      "learning_rate": 7.423614939860033e-06,
      "loss": 0.7409,
      "step": 2803
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6382973867458595,
      "learning_rate": 7.421801448170256e-06,
      "loss": 0.2518,
      "step": 2804
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.501848523360429,
      "learning_rate": 7.419987540144733e-06,
      "loss": 0.2864,
      "step": 2805
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.561976671499772,
      "learning_rate": 7.4181732160952995e-06,
      "loss": 0.221,
      "step": 2806
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5618461267900865,
      "learning_rate": 7.416358476333856e-06,
      "loss": 0.2068,
      "step": 2807
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.422803933157803,
      "learning_rate": 7.414543321172376e-06,
      "loss": 0.5435,
      "step": 2808
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4142662127032437,
      "learning_rate": 7.412727750922908e-06,
      "loss": 0.2174,
      "step": 2809
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7264565772103524,
      "learning_rate": 7.4109117658975685e-06,
      "loss": 0.1986,
      "step": 2810
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4677468219590661,
      "learning_rate": 7.409095366408547e-06,
      "loss": 0.2369,
      "step": 2811
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6361860228444824,
      "learning_rate": 7.407278552768103e-06,
      "loss": 0.2556,
      "step": 2812
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3945851517746732,
      "learning_rate": 7.40546132528857e-06,
      "loss": 0.1748,
      "step": 2813
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.608634955417371,
      "learning_rate": 7.403643684282347e-06,
      "loss": 0.2169,
      "step": 2814
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4815501424588104,
      "learning_rate": 7.40182563006191e-06,
      "loss": 0.1922,
      "step": 2815
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4567357481204577,
      "learning_rate": 7.4000071629398015e-06,
      "loss": 0.1922,
      "step": 2816
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4404363449192499,
      "learning_rate": 7.398188283228641e-06,
      "loss": 0.2114,
      "step": 2817
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.265015156339756,
      "learning_rate": 7.39636899124111e-06,
      "loss": 0.2062,
      "step": 2818
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4395681080330542,
      "learning_rate": 7.39454928728997e-06,
      "loss": 0.224,
      "step": 2819
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.412300556527173,
      "learning_rate": 7.392729171688047e-06,
      "loss": 0.2339,
      "step": 2820
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.017554716491845,
      "learning_rate": 7.39090864474824e-06,
      "loss": 0.605,
      "step": 2821
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.780830455721264,
      "learning_rate": 7.389087706783518e-06,
      "loss": 0.2529,
      "step": 2822
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5196084355327033,
      "learning_rate": 7.387266358106925e-06,
      "loss": 0.2396,
      "step": 2823
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4148504580728065,
      "learning_rate": 7.385444599031568e-06,
      "loss": 0.1938,
      "step": 2824
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.387203119204988,
      "learning_rate": 7.38362242987063e-06,
      "loss": 0.183,
      "step": 2825
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.462031368993109,
      "learning_rate": 7.381799850937363e-06,
      "loss": 0.2078,
      "step": 2826
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5690883106657205,
      "learning_rate": 7.379976862545091e-06,
      "loss": 0.2834,
      "step": 2827
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5857190510695434,
      "learning_rate": 7.378153465007203e-06,
      "loss": 0.2181,
      "step": 2828
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.2707618981309565,
      "learning_rate": 7.376329658637167e-06,
      "loss": 0.6035,
      "step": 2829
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3923659254940925,
      "learning_rate": 7.374505443748512e-06,
      "loss": 0.2559,
      "step": 2830
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.454243995485074,
      "learning_rate": 7.372680820654844e-06,
      "loss": 0.2125,
      "step": 2831
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2421571407234657,
      "learning_rate": 7.3708557896698375e-06,
      "loss": 0.1985,
      "step": 2832
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5401240033237662,
      "learning_rate": 7.3690303511072324e-06,
      "loss": 0.2465,
      "step": 2833
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.367529499070023,
      "learning_rate": 7.367204505280848e-06,
      "loss": 0.2106,
      "step": 2834
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4381441746248187,
      "learning_rate": 7.365378252504567e-06,
      "loss": 0.2366,
      "step": 2835
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5400984094839743,
      "learning_rate": 7.363551593092342e-06,
      "loss": 0.2294,
      "step": 2836
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4810800962751531,
      "learning_rate": 7.361724527358195e-06,
      "loss": 0.2354,
      "step": 2837
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2896464252201933,
      "learning_rate": 7.359897055616225e-06,
      "loss": 0.1843,
      "step": 2838
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.291341099016995,
      "learning_rate": 7.358069178180592e-06,
      "loss": 0.7087,
      "step": 2839
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2389059076023703,
      "learning_rate": 7.356240895365531e-06,
      "loss": 0.1853,
      "step": 2840
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2778509586662083,
      "learning_rate": 7.354412207485343e-06,
      "loss": 0.2213,
      "step": 2841
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5333100637704922,
      "learning_rate": 7.352583114854402e-06,
      "loss": 0.2223,
      "step": 2842
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.49857829766868,
      "learning_rate": 7.350753617787151e-06,
      "loss": 0.2393,
      "step": 2843
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1754879448652444,
      "learning_rate": 7.348923716598099e-06,
      "loss": 0.16,
      "step": 2844
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4250899183396057,
      "learning_rate": 7.347093411601831e-06,
      "loss": 0.2609,
      "step": 2845
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.434501884998321,
      "learning_rate": 7.3452627031129964e-06,
      "loss": 0.7473,
      "step": 2846
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3362019254147233,
      "learning_rate": 7.3434315914463125e-06,
      "loss": 0.1619,
      "step": 2847
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3908720875316378,
      "learning_rate": 7.341600076916571e-06,
      "loss": 0.2406,
      "step": 2848
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3950464448518118,
      "learning_rate": 7.33976815983863e-06,
      "loss": 0.2449,
      "step": 2849
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.859822031077082,
      "learning_rate": 7.3379358405274195e-06,
      "loss": 0.7238,
      "step": 2850
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6302617816062166,
      "learning_rate": 7.33610311929793e-06,
      "loss": 0.2324,
      "step": 2851
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.268613755035629,
      "learning_rate": 7.334269996465236e-06,
      "loss": 0.176,
      "step": 2852
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6168579802575362,
      "learning_rate": 7.332436472344468e-06,
      "loss": 0.2701,
      "step": 2853
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4629302779602265,
      "learning_rate": 7.330602547250828e-06,
      "loss": 0.2229,
      "step": 2854
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4115162081380954,
      "learning_rate": 7.3287682214995934e-06,
      "loss": 0.2256,
      "step": 2855
    },
    {
      "epoch": 0.37,
      "grad_norm": 10.647251000728362,
      "learning_rate": 7.326933495406103e-06,
      "loss": 0.6548,
      "step": 2856
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.589963746353602,
      "learning_rate": 7.32509836928577e-06,
      "loss": 0.2317,
      "step": 2857
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3694880153012539,
      "learning_rate": 7.323262843454071e-06,
      "loss": 0.1846,
      "step": 2858
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3985892408226426,
      "learning_rate": 7.321426918226557e-06,
      "loss": 0.226,
      "step": 2859
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5302645602113283,
      "learning_rate": 7.319590593918844e-06,
      "loss": 0.2376,
      "step": 2860
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4923061884799904,
      "learning_rate": 7.317753870846615e-06,
      "loss": 0.2126,
      "step": 2861
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.436804449876818,
      "learning_rate": 7.315916749325626e-06,
      "loss": 0.2297,
      "step": 2862
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4804308669497455,
      "learning_rate": 7.3140792296717004e-06,
      "loss": 0.2327,
      "step": 2863
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5387081861403404,
      "learning_rate": 7.312241312200727e-06,
      "loss": 0.2096,
      "step": 2864
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4284452489098873,
      "learning_rate": 7.310402997228667e-06,
      "loss": 0.2241,
      "step": 2865
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.518006175603209,
      "learning_rate": 7.308564285071547e-06,
      "loss": 0.2481,
      "step": 2866
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.408207025381417,
      "learning_rate": 7.306725176045464e-06,
      "loss": 0.1667,
      "step": 2867
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.972965978516835,
      "learning_rate": 7.304885670466581e-06,
      "loss": 0.5763,
      "step": 2868
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4997869296423005,
      "learning_rate": 7.3030457686511305e-06,
      "loss": 0.2143,
      "step": 2869
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4681635078402497,
      "learning_rate": 7.3012054709154124e-06,
      "loss": 0.2592,
      "step": 2870
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.492570370518196,
      "learning_rate": 7.299364777575797e-06,
      "loss": 0.2453,
      "step": 2871
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3266840452910387,
      "learning_rate": 7.29752368894872e-06,
      "loss": 0.2209,
      "step": 2872
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2491561089902607,
      "learning_rate": 7.295682205350685e-06,
      "loss": 0.1977,
      "step": 2873
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.307652208153508,
      "learning_rate": 7.293840327098265e-06,
      "loss": 0.6168,
      "step": 2874
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.128858131458839,
      "learning_rate": 7.291998054508102e-06,
      "loss": 0.6355,
      "step": 2875
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.490989359124275,
      "learning_rate": 7.2901553878969e-06,
      "loss": 0.2258,
      "step": 2876
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4643551711994907,
      "learning_rate": 7.288312327581439e-06,
      "loss": 0.2361,
      "step": 2877
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4929876933571973,
      "learning_rate": 7.286468873878559e-06,
      "loss": 0.2092,
      "step": 2878
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.296953892168786,
      "learning_rate": 7.2846250271051735e-06,
      "loss": 0.1949,
      "step": 2879
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4049432551490992,
      "learning_rate": 7.282780787578258e-06,
      "loss": 0.2166,
      "step": 2880
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.6346623691330535,
      "learning_rate": 7.280936155614864e-06,
      "loss": 0.5407,
      "step": 2881
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4002885627255013,
      "learning_rate": 7.2790911315321015e-06,
      "loss": 0.212,
      "step": 2882
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.142664368848583,
      "learning_rate": 7.2772457156471496e-06,
      "loss": 0.7813,
      "step": 2883
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4138263112647504,
      "learning_rate": 7.275399908277261e-06,
      "loss": 0.2059,
      "step": 2884
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.204718899985812,
      "learning_rate": 7.273553709739749e-06,
      "loss": 0.5223,
      "step": 2885
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2656108636156969,
      "learning_rate": 7.271707120351997e-06,
      "loss": 0.2031,
      "step": 2886
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.34346888179194,
      "learning_rate": 7.269860140431455e-06,
      "loss": 0.1957,
      "step": 2887
    },
    {
      "epoch": 0.37,
      "grad_norm": 10.955983850592729,
      "learning_rate": 7.268012770295641e-06,
      "loss": 0.7073,
      "step": 2888
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.305810540979301,
      "learning_rate": 7.266165010262138e-06,
      "loss": 0.7085,
      "step": 2889
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2560908038784353,
      "learning_rate": 7.264316860648598e-06,
      "loss": 0.1959,
      "step": 2890
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.806699036227768,
      "learning_rate": 7.26246832177274e-06,
      "loss": 0.4768,
      "step": 2891
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4770641779322609,
      "learning_rate": 7.2606193939523496e-06,
      "loss": 0.2082,
      "step": 2892
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4315333916502104,
      "learning_rate": 7.258770077505276e-06,
      "loss": 0.2226,
      "step": 2893
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2788695940219899,
      "learning_rate": 7.256920372749441e-06,
      "loss": 0.2015,
      "step": 2894
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.319601629530266,
      "learning_rate": 7.255070280002829e-06,
      "loss": 0.2118,
      "step": 2895
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4434080459371432,
      "learning_rate": 7.253219799583495e-06,
      "loss": 0.2375,
      "step": 2896
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5486108989136518,
      "learning_rate": 7.251368931809554e-06,
      "loss": 0.2416,
      "step": 2897
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.358916587623181,
      "learning_rate": 7.249517676999192e-06,
      "loss": 0.2321,
      "step": 2898
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4711015289830582,
      "learning_rate": 7.247666035470666e-06,
      "loss": 0.1985,
      "step": 2899
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.30872452761627,
      "learning_rate": 7.24581400754229e-06,
      "loss": 0.1882,
      "step": 2900
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4240395344930488,
      "learning_rate": 7.24396159353245e-06,
      "loss": 0.2185,
      "step": 2901
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4411321622907374,
      "learning_rate": 7.242108793759597e-06,
      "loss": 0.1896,
      "step": 2902
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4819719635393662,
      "learning_rate": 7.240255608542252e-06,
      "loss": 0.2994,
      "step": 2903
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4726332173784797,
      "learning_rate": 7.238402038198995e-06,
      "loss": 0.2101,
      "step": 2904
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6231807653693795,
      "learning_rate": 7.236548083048478e-06,
      "loss": 0.1859,
      "step": 2905
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.147168394879729,
      "learning_rate": 7.234693743409418e-06,
      "loss": 0.676,
      "step": 2906
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2805537793505926,
      "learning_rate": 7.232839019600595e-06,
      "loss": 0.1761,
      "step": 2907
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.43580705374696,
      "learning_rate": 7.230983911940861e-06,
      "loss": 0.5546,
      "step": 2908
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.392580816766615,
      "learning_rate": 7.229128420749127e-06,
      "loss": 0.2196,
      "step": 2909
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6423127613655304,
      "learning_rate": 7.227272546344377e-06,
      "loss": 0.1731,
      "step": 2910
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.648937177195767,
      "learning_rate": 7.225416289045655e-06,
      "loss": 0.2157,
      "step": 2911
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4701269239090604,
      "learning_rate": 7.2235596491720724e-06,
      "loss": 0.2434,
      "step": 2912
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4261360627408368,
      "learning_rate": 7.22170262704281e-06,
      "loss": 0.2401,
      "step": 2913
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.419412997748404,
      "learning_rate": 7.21984522297711e-06,
      "loss": 0.1779,
      "step": 2914
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3819303092954132,
      "learning_rate": 7.217987437294281e-06,
      "loss": 0.2376,
      "step": 2915
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5981014970669118,
      "learning_rate": 7.216129270313698e-06,
      "loss": 0.2612,
      "step": 2916
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5602475021500952,
      "learning_rate": 7.214270722354802e-06,
      "loss": 0.1812,
      "step": 2917
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.465174818902243,
      "learning_rate": 7.2124117937371e-06,
      "loss": 0.611,
      "step": 2918
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5463106709709415,
      "learning_rate": 7.21055248478016e-06,
      "loss": 0.1701,
      "step": 2919
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3745033384245742,
      "learning_rate": 7.208692795803622e-06,
      "loss": 0.2217,
      "step": 2920
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.537516906429164,
      "learning_rate": 7.206832727127186e-06,
      "loss": 0.1741,
      "step": 2921
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3484801702910338,
      "learning_rate": 7.204972279070623e-06,
      "loss": 0.2202,
      "step": 2922
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.340564241276077,
      "learning_rate": 7.203111451953761e-06,
      "loss": 0.2087,
      "step": 2923
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4799739793245217,
      "learning_rate": 7.201250246096501e-06,
      "loss": 0.2216,
      "step": 2924
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6099369576504234,
      "learning_rate": 7.1993886618188025e-06,
      "loss": 0.226,
      "step": 2925
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.531904104918205,
      "learning_rate": 7.1975266994406965e-06,
      "loss": 0.2136,
      "step": 2926
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3937790268657748,
      "learning_rate": 7.195664359282275e-06,
      "loss": 0.1673,
      "step": 2927
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5497881825652398,
      "learning_rate": 7.193801641663697e-06,
      "loss": 0.2139,
      "step": 2928
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.192726863464721,
      "learning_rate": 7.191938546905183e-06,
      "loss": 0.1687,
      "step": 2929
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3493887582717647,
      "learning_rate": 7.190075075327021e-06,
      "loss": 0.2231,
      "step": 2930
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5376860602063713,
      "learning_rate": 7.188211227249565e-06,
      "loss": 0.2149,
      "step": 2931
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3043938991908781,
      "learning_rate": 7.186347002993233e-06,
      "loss": 0.1976,
      "step": 2932
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4220047597537686,
      "learning_rate": 7.184482402878501e-06,
      "loss": 0.2019,
      "step": 2933
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.397231967661103,
      "learning_rate": 7.182617427225922e-06,
      "loss": 0.2524,
      "step": 2934
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.576907338486608,
      "learning_rate": 7.180752076356102e-06,
      "loss": 0.7109,
      "step": 2935
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.490627312555908,
      "learning_rate": 7.178886350589721e-06,
      "loss": 0.1814,
      "step": 2936
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.778711606508208,
      "learning_rate": 7.177020250247515e-06,
      "loss": 0.7132,
      "step": 2937
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.197684298371971,
      "learning_rate": 7.17515377565029e-06,
      "loss": 0.6018,
      "step": 2938
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4225522048454964,
      "learning_rate": 7.173286927118914e-06,
      "loss": 0.1973,
      "step": 2939
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6309387110115983,
      "learning_rate": 7.171419704974321e-06,
      "loss": 0.2325,
      "step": 2940
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5679296434450885,
      "learning_rate": 7.169552109537507e-06,
      "loss": 0.2535,
      "step": 2941
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.405737733883908,
      "learning_rate": 7.167684141129536e-06,
      "loss": 0.1868,
      "step": 2942
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4742165871897626,
      "learning_rate": 7.165815800071529e-06,
      "loss": 0.2411,
      "step": 2943
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4128378428161106,
      "learning_rate": 7.163947086684677e-06,
      "loss": 0.2038,
      "step": 2944
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3988994993307746,
      "learning_rate": 7.1620780012902356e-06,
      "loss": 0.1775,
      "step": 2945
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5389838797112583,
      "learning_rate": 7.160208544209521e-06,
      "loss": 0.2287,
      "step": 2946
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4167493950272778,
      "learning_rate": 7.158338715763912e-06,
      "loss": 0.2173,
      "step": 2947
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.820605425271475,
      "learning_rate": 7.156468516274859e-06,
      "loss": 0.6448,
      "step": 2948
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3740353560794942,
      "learning_rate": 7.154597946063867e-06,
      "loss": 0.1941,
      "step": 2949
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5930779277704554,
      "learning_rate": 7.152727005452511e-06,
      "loss": 0.2025,
      "step": 2950
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5597637662431074,
      "learning_rate": 7.1508556947624245e-06,
      "loss": 0.1964,
      "step": 2951
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.3847849531672525,
      "learning_rate": 7.14898401431531e-06,
      "loss": 0.4134,
      "step": 2952
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.441016742117594,
      "learning_rate": 7.14711196443293e-06,
      "loss": 0.205,
      "step": 2953
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.8293088221969,
      "learning_rate": 7.145239545437113e-06,
      "loss": 0.5073,
      "step": 2954
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.275065939673485,
      "learning_rate": 7.143366757649746e-06,
      "loss": 0.5815,
      "step": 2955
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5873411134829305,
      "learning_rate": 7.141493601392787e-06,
      "loss": 0.2179,
      "step": 2956
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.626250481159325,
      "learning_rate": 7.139620076988252e-06,
      "loss": 0.5507,
      "step": 2957
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.537759252140876,
      "learning_rate": 7.137746184758218e-06,
      "loss": 0.4848,
      "step": 2958
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4077388204790973,
      "learning_rate": 7.135871925024835e-06,
      "loss": 0.2151,
      "step": 2959
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2631488876620098,
      "learning_rate": 7.133997298110308e-06,
      "loss": 0.1805,
      "step": 2960
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.85766030672642,
      "learning_rate": 7.1321223043369034e-06,
      "loss": 0.5884,
      "step": 2961
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3803775076576228,
      "learning_rate": 7.130246944026958e-06,
      "loss": 0.2417,
      "step": 2962
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2878260503423817,
      "learning_rate": 7.128371217502868e-06,
      "loss": 0.201,
      "step": 2963
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3427398636607408,
      "learning_rate": 7.12649512508709e-06,
      "loss": 0.1952,
      "step": 2964
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.544353916646796,
      "learning_rate": 7.1246186671021475e-06,
      "loss": 0.2135,
      "step": 2965
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.571683488025093,
      "learning_rate": 7.122741843870626e-06,
      "loss": 0.2385,
      "step": 2966
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2670884249585501,
      "learning_rate": 7.120864655715172e-06,
      "loss": 0.2207,
      "step": 2967
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.223165156378947,
      "learning_rate": 7.118987102958498e-06,
      "loss": 0.6106,
      "step": 2968
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4650594120484117,
      "learning_rate": 7.117109185923374e-06,
      "loss": 0.217,
      "step": 2969
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5769533317681197,
      "learning_rate": 7.115230904932639e-06,
      "loss": 0.2279,
      "step": 2970
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4430966189776675,
      "learning_rate": 7.113352260309189e-06,
      "loss": 0.2305,
      "step": 2971
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5171806200107232,
      "learning_rate": 7.111473252375983e-06,
      "loss": 0.2565,
      "step": 2972
    },
    {
      "epoch": 0.38,
      "grad_norm": 9.497423345637213,
      "learning_rate": 7.109593881456048e-06,
      "loss": 0.6099,
      "step": 2973
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.676600251565631,
      "learning_rate": 7.10771414787247e-06,
      "loss": 0.5893,
      "step": 2974
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3091395055240558,
      "learning_rate": 7.105834051948395e-06,
      "loss": 0.1786,
      "step": 2975
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2920487413087889,
      "learning_rate": 7.1039535940070305e-06,
      "loss": 0.2099,
      "step": 2976
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.552121113581119,
      "learning_rate": 7.102072774371654e-06,
      "loss": 0.2348,
      "step": 2977
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6800726029192021,
      "learning_rate": 7.1001915933655994e-06,
      "loss": 0.2338,
      "step": 2978
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.691199265462293,
      "learning_rate": 7.098310051312261e-06,
      "loss": 0.7341,
      "step": 2979
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2754379449437256,
      "learning_rate": 7.096428148535101e-06,
      "loss": 0.1601,
      "step": 2980
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3119933243238004,
      "learning_rate": 7.094545885357636e-06,
      "loss": 0.2255,
      "step": 2981
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5791503097227015,
      "learning_rate": 7.092663262103452e-06,
      "loss": 0.2327,
      "step": 2982
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.35780259541,
      "learning_rate": 7.0907802790961925e-06,
      "loss": 0.6402,
      "step": 2983
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3632289173550467,
      "learning_rate": 7.088896936659566e-06,
      "loss": 0.2256,
      "step": 2984
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.526390269052163,
      "learning_rate": 7.087013235117339e-06,
      "loss": 0.252,
      "step": 2985
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5568051043658409,
      "learning_rate": 7.0851291747933415e-06,
      "loss": 0.2173,
      "step": 2986
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3407940709901482,
      "learning_rate": 7.083244756011466e-06,
      "loss": 0.1957,
      "step": 2987
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.684584757452097,
      "learning_rate": 7.081359979095667e-06,
      "loss": 0.6775,
      "step": 2988
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.9129750165789203,
      "learning_rate": 7.079474844369958e-06,
      "loss": 0.5527,
      "step": 2989
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3792911904293954,
      "learning_rate": 7.077589352158415e-06,
      "loss": 0.1945,
      "step": 2990
    },
    {
      "epoch": 0.38,
      "grad_norm": 11.518770591542133,
      "learning_rate": 7.075703502785178e-06,
      "loss": 0.5568,
      "step": 2991
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.18262979610826,
      "learning_rate": 7.073817296574444e-06,
      "loss": 0.17,
      "step": 2992
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2367096532508728,
      "learning_rate": 7.071930733850476e-06,
      "loss": 0.1899,
      "step": 2993
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2907060590503565,
      "learning_rate": 7.070043814937595e-06,
      "loss": 0.1897,
      "step": 2994
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.373341495973381,
      "learning_rate": 7.068156540160182e-06,
      "loss": 0.2342,
      "step": 2995
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.341509208768129,
      "learning_rate": 7.066268909842687e-06,
      "loss": 0.6142,
      "step": 2996
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5256542185657518,
      "learning_rate": 7.06438092430961e-06,
      "loss": 0.2248,
      "step": 2997
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.377040772289911,
      "learning_rate": 7.062492583885521e-06,
      "loss": 0.198,
      "step": 2998
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5208551093226503,
      "learning_rate": 7.060603888895046e-06,
      "loss": 0.2264,
      "step": 2999
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.413544537902766,
      "learning_rate": 7.058714839662874e-06,
      "loss": 0.4717,
      "step": 3000
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.726161246808738,
      "learning_rate": 7.056825436513754e-06,
      "loss": 0.2028,
      "step": 3001
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4605665402843957,
      "learning_rate": 7.054935679772497e-06,
      "loss": 0.2067,
      "step": 3002
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.462603044930853,
      "learning_rate": 7.053045569763973e-06,
      "loss": 0.2424,
      "step": 3003
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3011666212884563,
      "learning_rate": 7.051155106813114e-06,
      "loss": 0.214,
      "step": 3004
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5341871489966088,
      "learning_rate": 7.049264291244915e-06,
      "loss": 0.2233,
      "step": 3005
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.55024862305395,
      "learning_rate": 7.047373123384426e-06,
      "loss": 0.2109,
      "step": 3006
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.206606901523878,
      "learning_rate": 7.045481603556763e-06,
      "loss": 0.6095,
      "step": 3007
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3891898885223342,
      "learning_rate": 7.043589732087098e-06,
      "loss": 0.2415,
      "step": 3008
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.15559055720451,
      "learning_rate": 7.041697509300667e-06,
      "loss": 0.6222,
      "step": 3009
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3284263327634718,
      "learning_rate": 7.039804935522766e-06,
      "loss": 0.2141,
      "step": 3010
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3658123472402193,
      "learning_rate": 7.037912011078749e-06,
      "loss": 0.1992,
      "step": 3011
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5179055170978226,
      "learning_rate": 7.03601873629403e-06,
      "loss": 0.2227,
      "step": 3012
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.518491514762883,
      "learning_rate": 7.0341251114940864e-06,
      "loss": 0.2222,
      "step": 3013
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1821207748847236,
      "learning_rate": 7.032231137004457e-06,
      "loss": 0.1694,
      "step": 3014
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3455429717042409,
      "learning_rate": 7.030336813150734e-06,
      "loss": 0.2218,
      "step": 3015
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3841209650500135,
      "learning_rate": 7.028442140258576e-06,
      "loss": 0.2068,
      "step": 3016
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.499695742132793,
      "learning_rate": 7.026547118653697e-06,
      "loss": 0.5869,
      "step": 3017
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.418470296715018,
      "learning_rate": 7.024651748661875e-06,
      "loss": 0.2322,
      "step": 3018
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3448863296422382,
      "learning_rate": 7.022756030608946e-06,
      "loss": 0.2009,
      "step": 3019
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.448120519885295,
      "learning_rate": 7.0208599648208054e-06,
      "loss": 0.2331,
      "step": 3020
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6504415956539729,
      "learning_rate": 7.01896355162341e-06,
      "loss": 0.245,
      "step": 3021
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.579337805708123,
      "learning_rate": 7.017066791342773e-06,
      "loss": 0.5855,
      "step": 3022
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2597024095103768,
      "learning_rate": 7.01516968430497e-06,
      "loss": 0.1978,
      "step": 3023
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4210472963675365,
      "learning_rate": 7.013272230836139e-06,
      "loss": 0.2238,
      "step": 3024
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2200409906515242,
      "learning_rate": 7.01137443126247e-06,
      "loss": 0.1762,
      "step": 3025
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.863666501268973,
      "learning_rate": 7.009476285910218e-06,
      "loss": 0.6925,
      "step": 3026
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.161862647170255,
      "learning_rate": 7.007577795105697e-06,
      "loss": 0.1248,
      "step": 3027
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.163869277060889,
      "learning_rate": 7.005678959175279e-06,
      "loss": 0.5311,
      "step": 3028
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4429883371344414,
      "learning_rate": 7.003779778445398e-06,
      "loss": 0.2248,
      "step": 3029
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4395130532278555,
      "learning_rate": 7.001880253242541e-06,
      "loss": 0.2404,
      "step": 3030
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4808519689764903,
      "learning_rate": 6.999980383893261e-06,
      "loss": 0.1721,
      "step": 3031
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.582375708060941,
      "learning_rate": 6.998080170724167e-06,
      "loss": 0.266,
      "step": 3032
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3147557934811462,
      "learning_rate": 6.996179614061929e-06,
      "loss": 0.216,
      "step": 3033
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5510765141571474,
      "learning_rate": 6.9942787142332735e-06,
      "loss": 0.2294,
      "step": 3034
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3649782872551317,
      "learning_rate": 6.992377471564987e-06,
      "loss": 0.2167,
      "step": 3035
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5047309608013428,
      "learning_rate": 6.990475886383915e-06,
      "loss": 0.2092,
      "step": 3036
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4243150426669637,
      "learning_rate": 6.988573959016963e-06,
      "loss": 0.2248,
      "step": 3037
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3999036954869617,
      "learning_rate": 6.9866716897910945e-06,
      "loss": 0.1545,
      "step": 3038
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6192960669601624,
      "learning_rate": 6.984769079033331e-06,
      "loss": 0.2132,
      "step": 3039
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4583881544384962,
      "learning_rate": 6.982866127070753e-06,
      "loss": 0.2149,
      "step": 3040
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4839161566640744,
      "learning_rate": 6.9809628342305e-06,
      "loss": 0.2428,
      "step": 3041
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3455820591020902,
      "learning_rate": 6.9790592008397705e-06,
      "loss": 0.2065,
      "step": 3042
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.389013048421537,
      "learning_rate": 6.9771552272258226e-06,
      "loss": 0.2113,
      "step": 3043
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3961095204051523,
      "learning_rate": 6.975250913715968e-06,
      "loss": 0.2619,
      "step": 3044
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.488770292850396,
      "learning_rate": 6.973346260637583e-06,
      "loss": 0.5833,
      "step": 3045
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5849883854056441,
      "learning_rate": 6.9714412683181e-06,
      "loss": 0.2013,
      "step": 3046
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3894538827509686,
      "learning_rate": 6.969535937085006e-06,
      "loss": 0.2361,
      "step": 3047
    },
    {
      "epoch": 0.39,
      "grad_norm": 11.017133014131803,
      "learning_rate": 6.967630267265854e-06,
      "loss": 0.4834,
      "step": 3048
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3263153276444692,
      "learning_rate": 6.965724259188246e-06,
      "loss": 0.1744,
      "step": 3049
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.463719595950077,
      "learning_rate": 6.96381791317985e-06,
      "loss": 0.2138,
      "step": 3050
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3592870962697705,
      "learning_rate": 6.961911229568388e-06,
      "loss": 0.1926,
      "step": 3051
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4028022024067766,
      "learning_rate": 6.9600042086816424e-06,
      "loss": 0.2321,
      "step": 3052
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4193513577857462,
      "learning_rate": 6.958096850847451e-06,
      "loss": 0.2388,
      "step": 3053
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.756932458551607,
      "learning_rate": 6.956189156393709e-06,
      "loss": 0.21,
      "step": 3054
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.152498117588351,
      "learning_rate": 6.954281125648373e-06,
      "loss": 0.7041,
      "step": 3055
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2723956545348414,
      "learning_rate": 6.952372758939457e-06,
      "loss": 0.1542,
      "step": 3056
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4860848972925844,
      "learning_rate": 6.9504640565950295e-06,
      "loss": 0.177,
      "step": 3057
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6197012046535133,
      "learning_rate": 6.948555018943219e-06,
      "loss": 0.2826,
      "step": 3058
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3865473700617075,
      "learning_rate": 6.946645646312212e-06,
      "loss": 0.2142,
      "step": 3059
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.402013608468295,
      "learning_rate": 6.944735939030249e-06,
      "loss": 0.2259,
      "step": 3060
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3186743667090475,
      "learning_rate": 6.942825897425633e-06,
      "loss": 0.1943,
      "step": 3061
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5401416870355888,
      "learning_rate": 6.940915521826723e-06,
      "loss": 0.2019,
      "step": 3062
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3594971584919764,
      "learning_rate": 6.939004812561934e-06,
      "loss": 0.2007,
      "step": 3063
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3546656137392847,
      "learning_rate": 6.937093769959737e-06,
      "loss": 0.1747,
      "step": 3064
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2774765952772746,
      "learning_rate": 6.9351823943486654e-06,
      "loss": 0.1829,
      "step": 3065
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.508836026352071,
      "learning_rate": 6.933270686057308e-06,
      "loss": 0.2621,
      "step": 3066
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4302780502136259,
      "learning_rate": 6.931358645414304e-06,
      "loss": 0.2126,
      "step": 3067
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.487350880465935,
      "learning_rate": 6.929446272748361e-06,
      "loss": 0.1562,
      "step": 3068
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.314816106486569,
      "learning_rate": 6.927533568388232e-06,
      "loss": 0.2237,
      "step": 3069
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.615352510873598,
      "learning_rate": 6.92562053266274e-06,
      "loss": 0.2404,
      "step": 3070
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4826651240547002,
      "learning_rate": 6.923707165900753e-06,
      "loss": 0.2138,
      "step": 3071
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5040752906302433,
      "learning_rate": 6.921793468431204e-06,
      "loss": 0.2003,
      "step": 3072
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.15939717729052,
      "learning_rate": 6.9198794405830776e-06,
      "loss": 0.7707,
      "step": 3073
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6471624203277897,
      "learning_rate": 6.917965082685418e-06,
      "loss": 0.2718,
      "step": 3074
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.131456305416181,
      "learning_rate": 6.9160503950673276e-06,
      "loss": 0.7062,
      "step": 3075
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6054869162339758,
      "learning_rate": 6.914135378057959e-06,
      "loss": 0.2183,
      "step": 3076
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2747479420236134,
      "learning_rate": 6.912220031986531e-06,
      "loss": 0.2067,
      "step": 3077
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5028857669433167,
      "learning_rate": 6.910304357182308e-06,
      "loss": 0.2446,
      "step": 3078
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4127815371060164,
      "learning_rate": 6.908388353974622e-06,
      "loss": 0.1936,
      "step": 3079
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.455884581615142,
      "learning_rate": 6.906472022692854e-06,
      "loss": 0.2263,
      "step": 3080
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3927992597596408,
      "learning_rate": 6.904555363666443e-06,
      "loss": 0.1923,
      "step": 3081
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3350838160668996,
      "learning_rate": 6.902638377224886e-06,
      "loss": 0.2017,
      "step": 3082
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.109052939685196,
      "learning_rate": 6.900721063697733e-06,
      "loss": 0.652,
      "step": 3083
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.265498612351919,
      "learning_rate": 6.898803423414595e-06,
      "loss": 0.1671,
      "step": 3084
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.198130597022815,
      "learning_rate": 6.896885456705137e-06,
      "loss": 0.4373,
      "step": 3085
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.32722917015208,
      "learning_rate": 6.894967163899077e-06,
      "loss": 0.7021,
      "step": 3086
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3753058381671914,
      "learning_rate": 6.893048545326193e-06,
      "loss": 0.2596,
      "step": 3087
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4917712863435337,
      "learning_rate": 6.891129601316319e-06,
      "loss": 0.2639,
      "step": 3088
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.506385138572602,
      "learning_rate": 6.889210332199343e-06,
      "loss": 0.5579,
      "step": 3089
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5091387087328183,
      "learning_rate": 6.887290738305208e-06,
      "loss": 0.2469,
      "step": 3090
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4701125146276834,
      "learning_rate": 6.885370819963917e-06,
      "loss": 0.2687,
      "step": 3091
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4357597238082471,
      "learning_rate": 6.883450577505524e-06,
      "loss": 0.2533,
      "step": 3092
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3653113434406536,
      "learning_rate": 6.881530011260142e-06,
      "loss": 0.245,
      "step": 3093
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3255790045367317,
      "learning_rate": 6.879609121557938e-06,
      "loss": 0.1846,
      "step": 3094
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3982386583088982,
      "learning_rate": 6.877687908729137e-06,
      "loss": 0.2488,
      "step": 3095
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3792266252546292,
      "learning_rate": 6.875766373104016e-06,
      "loss": 0.2141,
      "step": 3096
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4746876695057858,
      "learning_rate": 6.873844515012909e-06,
      "loss": 0.2015,
      "step": 3097
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.389434664085572,
      "learning_rate": 6.871922334786206e-06,
      "loss": 0.2458,
      "step": 3098
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3742397819841277,
      "learning_rate": 6.8699998327543545e-06,
      "loss": 0.2351,
      "step": 3099
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4652781477109815,
      "learning_rate": 6.868077009247852e-06,
      "loss": 0.2667,
      "step": 3100
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2264154039351327,
      "learning_rate": 6.866153864597254e-06,
      "loss": 0.2046,
      "step": 3101
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5298203691673427,
      "learning_rate": 6.864230399133172e-06,
      "loss": 0.2682,
      "step": 3102
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3426656825183056,
      "learning_rate": 6.862306613186275e-06,
      "loss": 0.2123,
      "step": 3103
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.695500110814722,
      "learning_rate": 6.86038250708728e-06,
      "loss": 0.4956,
      "step": 3104
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.044117941505145,
      "learning_rate": 6.858458081166964e-06,
      "loss": 0.6066,
      "step": 3105
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4381898990507438,
      "learning_rate": 6.856533335756159e-06,
      "loss": 0.2061,
      "step": 3106
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.158577829516537,
      "learning_rate": 6.854608271185752e-06,
      "loss": 0.7151,
      "step": 3107
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4243446717784216,
      "learning_rate": 6.852682887786681e-06,
      "loss": 0.2357,
      "step": 3108
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.343059440904526,
      "learning_rate": 6.850757185889945e-06,
      "loss": 0.1922,
      "step": 3109
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6756894063280063,
      "learning_rate": 6.848831165826591e-06,
      "loss": 0.2559,
      "step": 3110
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4531836192100824,
      "learning_rate": 6.846904827927728e-06,
      "loss": 0.2097,
      "step": 3111
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.608781177673214,
      "learning_rate": 6.844978172524514e-06,
      "loss": 0.2403,
      "step": 3112
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3797771600262787,
      "learning_rate": 6.843051199948162e-06,
      "loss": 0.2099,
      "step": 3113
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3601895070936871,
      "learning_rate": 6.841123910529943e-06,
      "loss": 0.1802,
      "step": 3114
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.551577629465943,
      "learning_rate": 6.839196304601179e-06,
      "loss": 0.6415,
      "step": 3115
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3794178151546568,
      "learning_rate": 6.837268382493248e-06,
      "loss": 0.1897,
      "step": 3116
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5370226282687678,
      "learning_rate": 6.835340144537584e-06,
      "loss": 0.2209,
      "step": 3117
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4784488055035476,
      "learning_rate": 6.833411591065673e-06,
      "loss": 0.2242,
      "step": 3118
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4008616600021184,
      "learning_rate": 6.831482722409053e-06,
      "loss": 0.1934,
      "step": 3119
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4455248503153262,
      "learning_rate": 6.829553538899321e-06,
      "loss": 0.2196,
      "step": 3120
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.298208169357119,
      "learning_rate": 6.8276240408681245e-06,
      "loss": 0.1866,
      "step": 3121
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3925173386608278,
      "learning_rate": 6.825694228647169e-06,
      "loss": 0.2415,
      "step": 3122
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2785494241375408,
      "learning_rate": 6.823764102568211e-06,
      "loss": 0.1927,
      "step": 3123
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3742460175466726,
      "learning_rate": 6.82183366296306e-06,
      "loss": 0.2024,
      "step": 3124
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4518946949919893,
      "learning_rate": 6.819902910163582e-06,
      "loss": 0.1921,
      "step": 3125
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3986070826669836,
      "learning_rate": 6.817971844501695e-06,
      "loss": 0.2171,
      "step": 3126
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4844072714714769,
      "learning_rate": 6.8160404663093725e-06,
      "loss": 0.2621,
      "step": 3127
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.9790039745002765,
      "learning_rate": 6.814108775918642e-06,
      "loss": 0.5629,
      "step": 3128
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3765292534959275,
      "learning_rate": 6.812176773661579e-06,
      "loss": 0.2181,
      "step": 3129
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5605448310708185,
      "learning_rate": 6.810244459870322e-06,
      "loss": 0.2339,
      "step": 3130
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.372347817824501,
      "learning_rate": 6.808311834877057e-06,
      "loss": 0.208,
      "step": 3131
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6113880534089762,
      "learning_rate": 6.806378899014023e-06,
      "loss": 0.2435,
      "step": 3132
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3436091655442999,
      "learning_rate": 6.804445652613514e-06,
      "loss": 0.2043,
      "step": 3133
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4589130433435642,
      "learning_rate": 6.802512096007879e-06,
      "loss": 0.2022,
      "step": 3134
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.913579347696697,
      "learning_rate": 6.800578229529519e-06,
      "loss": 0.6273,
      "step": 3135
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.792143891091931,
      "learning_rate": 6.798644053510886e-06,
      "loss": 0.5058,
      "step": 3136
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3850612810794352,
      "learning_rate": 6.796709568284488e-06,
      "loss": 0.1887,
      "step": 3137
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5688042089920018,
      "learning_rate": 6.794774774182887e-06,
      "loss": 0.2495,
      "step": 3138
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.115768170650309,
      "learning_rate": 6.792839671538696e-06,
      "loss": 0.8728,
      "step": 3139
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4320440249911726,
      "learning_rate": 6.790904260684581e-06,
      "loss": 0.1909,
      "step": 3140
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.26871591739796,
      "learning_rate": 6.788968541953262e-06,
      "loss": 0.6133,
      "step": 3141
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4289356369915824,
      "learning_rate": 6.787032515677509e-06,
      "loss": 0.2266,
      "step": 3142
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3877774913392686,
      "learning_rate": 6.785096182190152e-06,
      "loss": 0.2219,
      "step": 3143
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.5937059498718344,
      "learning_rate": 6.783159541824065e-06,
      "loss": 0.4897,
      "step": 3144
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4662217979728,
      "learning_rate": 6.781222594912182e-06,
      "loss": 0.1727,
      "step": 3145
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3026911543580795,
      "learning_rate": 6.779285341787484e-06,
      "loss": 0.1863,
      "step": 3146
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.773335878078139,
      "learning_rate": 6.7773477827830085e-06,
      "loss": 0.6999,
      "step": 3147
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3210414983176262,
      "learning_rate": 6.775409918231843e-06,
      "loss": 0.1723,
      "step": 3148
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.354771924960799,
      "learning_rate": 6.773471748467133e-06,
      "loss": 0.1615,
      "step": 3149
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2898350939580385,
      "learning_rate": 6.771533273822069e-06,
      "loss": 0.1838,
      "step": 3150
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.412402368770026,
      "learning_rate": 6.769594494629898e-06,
      "loss": 0.1881,
      "step": 3151
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.3809312001304095,
      "learning_rate": 6.767655411223917e-06,
      "loss": 0.4966,
      "step": 3152
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3920298309965349,
      "learning_rate": 6.76571602393748e-06,
      "loss": 0.2268,
      "step": 3153
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3958690730550019,
      "learning_rate": 6.7637763331039885e-06,
      "loss": 0.2069,
      "step": 3154
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3943509592896897,
      "learning_rate": 6.761836339056896e-06,
      "loss": 0.197,
      "step": 3155
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3462267041451044,
      "learning_rate": 6.759896042129713e-06,
      "loss": 0.1902,
      "step": 3156
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3776341009021602,
      "learning_rate": 6.757955442655998e-06,
      "loss": 0.2235,
      "step": 3157
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4054692191362317,
      "learning_rate": 6.756014540969362e-06,
      "loss": 0.23,
      "step": 3158
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3853472444121937,
      "learning_rate": 6.754073337403469e-06,
      "loss": 0.2291,
      "step": 3159
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.333510200332379,
      "learning_rate": 6.752131832292036e-06,
      "loss": 0.1808,
      "step": 3160
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3743681009130184,
      "learning_rate": 6.750190025968827e-06,
      "loss": 0.2113,
      "step": 3161
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3168785454983494,
      "learning_rate": 6.748247918767662e-06,
      "loss": 0.2176,
      "step": 3162
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3197053430321142,
      "learning_rate": 6.746305511022414e-06,
      "loss": 0.1934,
      "step": 3163
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4107453517626996,
      "learning_rate": 6.744362803067003e-06,
      "loss": 0.2136,
      "step": 3164
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.513967380986132,
      "learning_rate": 6.742419795235403e-06,
      "loss": 0.4307,
      "step": 3165
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4942870676629403,
      "learning_rate": 6.74047648786164e-06,
      "loss": 0.2009,
      "step": 3166
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3762571898187221,
      "learning_rate": 6.738532881279794e-06,
      "loss": 0.2172,
      "step": 3167
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4096354697180264,
      "learning_rate": 6.73658897582399e-06,
      "loss": 0.1711,
      "step": 3168
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.44009905793136,
      "learning_rate": 6.734644771828407e-06,
      "loss": 0.2129,
      "step": 3169
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.833951566419995,
      "learning_rate": 6.7327002696272795e-06,
      "loss": 0.5544,
      "step": 3170
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.413585720978878,
      "learning_rate": 6.730755469554888e-06,
      "loss": 0.4744,
      "step": 3171
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.606889277422966,
      "learning_rate": 6.728810371945567e-06,
      "loss": 0.6156,
      "step": 3172
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.618076431898457,
      "learning_rate": 6.7268649771337e-06,
      "loss": 0.6277,
      "step": 3173
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4457208456230004,
      "learning_rate": 6.724919285453726e-06,
      "loss": 0.2131,
      "step": 3174
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.2569580842129815,
      "learning_rate": 6.722973297240128e-06,
      "loss": 0.5844,
      "step": 3175
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.413977220430034,
      "learning_rate": 6.721027012827444e-06,
      "loss": 0.1675,
      "step": 3176
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5368211613458442,
      "learning_rate": 6.719080432550269e-06,
      "loss": 0.2023,
      "step": 3177
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.360807181500246,
      "learning_rate": 6.7171335567432365e-06,
      "loss": 0.55,
      "step": 3178
    },
    {
      "epoch": 0.41,
      "grad_norm": 8.125169263125807,
      "learning_rate": 6.7151863857410375e-06,
      "loss": 0.6581,
      "step": 3179
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3485249038248672,
      "learning_rate": 6.713238919878415e-06,
      "loss": 0.1982,
      "step": 3180
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.403417443348172,
      "learning_rate": 6.711291159490162e-06,
      "loss": 0.1682,
      "step": 3181
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3181231058620084,
      "learning_rate": 6.709343104911119e-06,
      "loss": 0.1796,
      "step": 3182
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.7780326373453157,
      "learning_rate": 6.70739475647618e-06,
      "loss": 0.5008,
      "step": 3183
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3248180109335126,
      "learning_rate": 6.705446114520289e-06,
      "loss": 0.2115,
      "step": 3184
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.278525037182716,
      "learning_rate": 6.7034971793784385e-06,
      "loss": 0.6134,
      "step": 3185
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.342253470095391,
      "learning_rate": 6.701547951385675e-06,
      "loss": 0.1598,
      "step": 3186
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.510525381350606,
      "learning_rate": 6.699598430877092e-06,
      "loss": 0.2333,
      "step": 3187
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5122833763987764,
      "learning_rate": 6.697648618187836e-06,
      "loss": 0.209,
      "step": 3188
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6761442553719599,
      "learning_rate": 6.6956985136531015e-06,
      "loss": 0.2949,
      "step": 3189
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4218510967430031,
      "learning_rate": 6.693748117608134e-06,
      "loss": 0.2261,
      "step": 3190
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3301789352346514,
      "learning_rate": 6.69179743038823e-06,
      "loss": 0.1813,
      "step": 3191
    },
    {
      "epoch": 0.41,
      "grad_norm": 8.799755506406733,
      "learning_rate": 6.6898464523287354e-06,
      "loss": 0.6479,
      "step": 3192
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6088526671676662,
      "learning_rate": 6.687895183765043e-06,
      "loss": 0.2407,
      "step": 3193
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5034254306828165,
      "learning_rate": 6.685943625032602e-06,
      "loss": 0.2294,
      "step": 3194
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.795028046795902,
      "learning_rate": 6.683991776466907e-06,
      "loss": 0.4102,
      "step": 3195
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.611355085100681,
      "learning_rate": 6.682039638403503e-06,
      "loss": 0.5506,
      "step": 3196
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.674752669997002,
      "learning_rate": 6.6800872111779854e-06,
      "loss": 0.644,
      "step": 3197
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.5516591058762454,
      "learning_rate": 6.678134495125999e-06,
      "loss": 0.6153,
      "step": 3198
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.847241831334392,
      "learning_rate": 6.676181490583238e-06,
      "loss": 0.541,
      "step": 3199
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.372409908702769,
      "learning_rate": 6.674228197885448e-06,
      "loss": 0.1748,
      "step": 3200
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3913692716413426,
      "learning_rate": 6.67227461736842e-06,
      "loss": 0.2191,
      "step": 3201
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4747814360993585,
      "learning_rate": 6.670320749367998e-06,
      "loss": 0.2216,
      "step": 3202
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3360326545866086,
      "learning_rate": 6.668366594220076e-06,
      "loss": 0.2086,
      "step": 3203
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3689610735840756,
      "learning_rate": 6.666412152260595e-06,
      "loss": 0.1859,
      "step": 3204
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.402204659863944,
      "learning_rate": 6.664457423825547e-06,
      "loss": 0.2107,
      "step": 3205
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3139838751260084,
      "learning_rate": 6.662502409250971e-06,
      "loss": 0.1964,
      "step": 3206
    },
    {
      "epoch": 0.41,
      "grad_norm": 8.466952719989365,
      "learning_rate": 6.660547108872959e-06,
      "loss": 0.6318,
      "step": 3207
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4982555534591586,
      "learning_rate": 6.6585915230276445e-06,
      "loss": 0.264,
      "step": 3208
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4814328068975944,
      "learning_rate": 6.656635652051222e-06,
      "loss": 0.2339,
      "step": 3209
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.550135063102599,
      "learning_rate": 6.654679496279925e-06,
      "loss": 0.2501,
      "step": 3210
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2629755401797251,
      "learning_rate": 6.6527230560500415e-06,
      "loss": 0.2081,
      "step": 3211
    },
    {
      "epoch": 0.41,
      "grad_norm": 9.76140572480758,
      "learning_rate": 6.6507663316979025e-06,
      "loss": 0.4421,
      "step": 3212
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4438939791626624,
      "learning_rate": 6.648809323559895e-06,
      "loss": 0.2379,
      "step": 3213
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.865966752449598,
      "learning_rate": 6.64685203197245e-06,
      "loss": 0.6398,
      "step": 3214
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3406365826024018,
      "learning_rate": 6.644894457272048e-06,
      "loss": 0.1779,
      "step": 3215
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2741971043998654,
      "learning_rate": 6.64293659979522e-06,
      "loss": 0.1943,
      "step": 3216
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3297866600571475,
      "learning_rate": 6.640978459878543e-06,
      "loss": 0.2295,
      "step": 3217
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.334151739394777,
      "learning_rate": 6.639020037858647e-06,
      "loss": 0.2164,
      "step": 3218
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.1917474593553035,
      "learning_rate": 6.637061334072204e-06,
      "loss": 0.6917,
      "step": 3219
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3838043408980356,
      "learning_rate": 6.635102348855939e-06,
      "loss": 0.2143,
      "step": 3220
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.469505560580028,
      "learning_rate": 6.6331430825466245e-06,
      "loss": 0.2046,
      "step": 3221
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4778918853766834,
      "learning_rate": 6.631183535481082e-06,
      "loss": 0.2029,
      "step": 3222
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4713950199193084,
      "learning_rate": 6.629223707996177e-06,
      "loss": 0.2095,
      "step": 3223
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5762847271612968,
      "learning_rate": 6.627263600428833e-06,
      "loss": 0.2217,
      "step": 3224
    },
    {
      "epoch": 0.41,
      "grad_norm": 9.59794456134367,
      "learning_rate": 6.625303213116009e-06,
      "loss": 0.6046,
      "step": 3225
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8125121221922724,
      "learning_rate": 6.62334254639472e-06,
      "loss": 0.2497,
      "step": 3226
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5752921455462054,
      "learning_rate": 6.621381600602028e-06,
      "loss": 0.2675,
      "step": 3227
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.046803924951245,
      "learning_rate": 6.619420376075043e-06,
      "loss": 0.5874,
      "step": 3228
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.18124868118369,
      "learning_rate": 6.617458873150922e-06,
      "loss": 0.7096,
      "step": 3229
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6102736727492797,
      "learning_rate": 6.6154970921668685e-06,
      "loss": 0.2059,
      "step": 3230
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8315790598435184,
      "learning_rate": 6.613535033460135e-06,
      "loss": 0.2002,
      "step": 3231
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2174785857880266,
      "learning_rate": 6.611572697368026e-06,
      "loss": 0.1895,
      "step": 3232
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4622974495139587,
      "learning_rate": 6.6096100842278865e-06,
      "loss": 0.2358,
      "step": 3233
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4450491299802335,
      "learning_rate": 6.607647194377113e-06,
      "loss": 0.2254,
      "step": 3234
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6723957546133112,
      "learning_rate": 6.60568402815315e-06,
      "loss": 0.2647,
      "step": 3235
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.29754046296303,
      "learning_rate": 6.603720585893487e-06,
      "loss": 0.1897,
      "step": 3236
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.345785923344584,
      "learning_rate": 6.601756867935664e-06,
      "loss": 0.237,
      "step": 3237
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6436544313250425,
      "learning_rate": 6.599792874617267e-06,
      "loss": 0.1814,
      "step": 3238
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4918333095353256,
      "learning_rate": 6.597828606275928e-06,
      "loss": 0.1835,
      "step": 3239
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3016022889738021,
      "learning_rate": 6.595864063249326e-06,
      "loss": 0.194,
      "step": 3240
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.422728500372264,
      "learning_rate": 6.593899245875193e-06,
      "loss": 0.4245,
      "step": 3241
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4498705145274264,
      "learning_rate": 6.591934154491301e-06,
      "loss": 0.1916,
      "step": 3242
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5998540193564885,
      "learning_rate": 6.589968789435472e-06,
      "loss": 0.1853,
      "step": 3243
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1977395258458263,
      "learning_rate": 6.588003151045577e-06,
      "loss": 0.2116,
      "step": 3244
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.75672690662583,
      "learning_rate": 6.586037239659529e-06,
      "loss": 0.4971,
      "step": 3245
    },
    {
      "epoch": 0.42,
      "grad_norm": 10.57517192712837,
      "learning_rate": 6.584071055615293e-06,
      "loss": 0.6497,
      "step": 3246
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4148561799463764,
      "learning_rate": 6.582104599250878e-06,
      "loss": 0.2345,
      "step": 3247
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5862699720245645,
      "learning_rate": 6.580137870904342e-06,
      "loss": 0.2279,
      "step": 3248
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4130569144505483,
      "learning_rate": 6.578170870913787e-06,
      "loss": 0.1791,
      "step": 3249
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4778535431011133,
      "learning_rate": 6.576203599617363e-06,
      "loss": 0.1942,
      "step": 3250
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.291428849610353,
      "learning_rate": 6.574236057353268e-06,
      "loss": 0.1882,
      "step": 3251
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4224365304958273,
      "learning_rate": 6.572268244459745e-06,
      "loss": 0.1563,
      "step": 3252
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3378324974019058,
      "learning_rate": 6.570300161275084e-06,
      "loss": 0.2284,
      "step": 3253
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.426554179938511,
      "learning_rate": 6.568331808137619e-06,
      "loss": 0.239,
      "step": 3254
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.617884993131297,
      "learning_rate": 6.5663631853857355e-06,
      "loss": 0.4824,
      "step": 3255
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.273855485607262,
      "learning_rate": 6.564394293357861e-06,
      "loss": 0.1613,
      "step": 3256
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3793230514406178,
      "learning_rate": 6.562425132392473e-06,
      "loss": 0.17,
      "step": 3257
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4850855680757635,
      "learning_rate": 6.560455702828089e-06,
      "loss": 0.2241,
      "step": 3258
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3881093688299422,
      "learning_rate": 6.558486005003281e-06,
      "loss": 0.1828,
      "step": 3259
    },
    {
      "epoch": 0.42,
      "grad_norm": 11.052086552335835,
      "learning_rate": 6.556516039256662e-06,
      "loss": 0.6444,
      "step": 3260
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3527776497064004,
      "learning_rate": 6.554545805926891e-06,
      "loss": 0.1969,
      "step": 3261
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4441313806820235,
      "learning_rate": 6.552575305352672e-06,
      "loss": 0.2059,
      "step": 3262
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.110150437257004,
      "learning_rate": 6.55060453787276e-06,
      "loss": 0.5072,
      "step": 3263
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5062302601752409,
      "learning_rate": 6.548633503825953e-06,
      "loss": 0.1817,
      "step": 3264
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2200807542125274,
      "learning_rate": 6.546662203551092e-06,
      "loss": 0.1296,
      "step": 3265
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4277301924974983,
      "learning_rate": 6.544690637387068e-06,
      "loss": 0.1908,
      "step": 3266
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5331728099021036,
      "learning_rate": 6.5427188056728165e-06,
      "loss": 0.2351,
      "step": 3267
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.896249708232255,
      "learning_rate": 6.540746708747317e-06,
      "loss": 0.4804,
      "step": 3268
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6229716877196092,
      "learning_rate": 6.538774346949597e-06,
      "loss": 0.2315,
      "step": 3269
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.433403325244316,
      "learning_rate": 6.536801720618727e-06,
      "loss": 0.1901,
      "step": 3270
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3785163203847135,
      "learning_rate": 6.534828830093827e-06,
      "loss": 0.1844,
      "step": 3271
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.316096898348941,
      "learning_rate": 6.532855675714055e-06,
      "loss": 0.627,
      "step": 3272
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.472230809066448,
      "learning_rate": 6.5308822578186225e-06,
      "loss": 0.2195,
      "step": 3273
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3820831508085196,
      "learning_rate": 6.528908576746784e-06,
      "loss": 0.2274,
      "step": 3274
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5008468904804322,
      "learning_rate": 6.526934632837835e-06,
      "loss": 0.2205,
      "step": 3275
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3676805390351443,
      "learning_rate": 6.5249604264311216e-06,
      "loss": 0.1972,
      "step": 3276
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.513672899971019,
      "learning_rate": 6.522985957866032e-06,
      "loss": 0.212,
      "step": 3277
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3159448597514216,
      "learning_rate": 6.521011227482e-06,
      "loss": 0.1798,
      "step": 3278
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4134854221865698,
      "learning_rate": 6.519036235618505e-06,
      "loss": 0.2186,
      "step": 3279
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.350929878023977,
      "learning_rate": 6.517060982615071e-06,
      "loss": 0.2146,
      "step": 3280
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6302565016272361,
      "learning_rate": 6.5150854688112686e-06,
      "loss": 0.2169,
      "step": 3281
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4010458633762322,
      "learning_rate": 6.513109694546707e-06,
      "loss": 0.2374,
      "step": 3282
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4406108369012456,
      "learning_rate": 6.511133660161047e-06,
      "loss": 0.2557,
      "step": 3283
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3544968612049715,
      "learning_rate": 6.5091573659939945e-06,
      "loss": 0.2364,
      "step": 3284
    },
    {
      "epoch": 0.42,
      "grad_norm": 10.311745482072508,
      "learning_rate": 6.507180812385295e-06,
      "loss": 0.6267,
      "step": 3285
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2785802288600423,
      "learning_rate": 6.50520399967474e-06,
      "loss": 0.1834,
      "step": 3286
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3913869449798726,
      "learning_rate": 6.503226928202167e-06,
      "loss": 0.2222,
      "step": 3287
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3182558088846752,
      "learning_rate": 6.5012495983074605e-06,
      "loss": 0.1692,
      "step": 3288
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4100538991825335,
      "learning_rate": 6.499272010330543e-06,
      "loss": 0.2062,
      "step": 3289
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4246868008765128,
      "learning_rate": 6.497294164611385e-06,
      "loss": 0.1974,
      "step": 3290
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3253629955132367,
      "learning_rate": 6.495316061490003e-06,
      "loss": 0.2048,
      "step": 3291
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.472285381258049,
      "learning_rate": 6.493337701306454e-06,
      "loss": 0.2156,
      "step": 3292
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.6378084901977576,
      "learning_rate": 6.4913590844008436e-06,
      "loss": 0.6414,
      "step": 3293
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.255342752893374,
      "learning_rate": 6.489380211113316e-06,
      "loss": 0.521,
      "step": 3294
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2336309467863007,
      "learning_rate": 6.487401081784063e-06,
      "loss": 0.2047,
      "step": 3295
    },
    {
      "epoch": 0.42,
      "grad_norm": 9.515053754622922,
      "learning_rate": 6.485421696753321e-06,
      "loss": 0.704,
      "step": 3296
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.827620371325327,
      "learning_rate": 6.4834420563613685e-06,
      "loss": 0.6044,
      "step": 3297
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.343538953348213,
      "learning_rate": 6.481462160948531e-06,
      "loss": 0.4157,
      "step": 3298
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.46576868327346,
      "learning_rate": 6.479482010855172e-06,
      "loss": 0.5233,
      "step": 3299
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.019114347022988,
      "learning_rate": 6.477501606421703e-06,
      "loss": 0.6048,
      "step": 3300
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.467380544686869,
      "learning_rate": 6.4755209479885806e-06,
      "loss": 0.5636,
      "step": 3301
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.422517607912735,
      "learning_rate": 6.473540035896301e-06,
      "loss": 0.2047,
      "step": 3302
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.665827316218256,
      "learning_rate": 6.471558870485407e-06,
      "loss": 0.6428,
      "step": 3303
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.530045801425968,
      "learning_rate": 6.469577452096483e-06,
      "loss": 0.239,
      "step": 3304
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5162339249904022,
      "learning_rate": 6.467595781070158e-06,
      "loss": 0.23,
      "step": 3305
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.639021102259344,
      "learning_rate": 6.4656138577471036e-06,
      "loss": 0.2373,
      "step": 3306
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4334114514267646,
      "learning_rate": 6.463631682468038e-06,
      "loss": 0.1859,
      "step": 3307
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.387432019326831,
      "learning_rate": 6.461649255573716e-06,
      "loss": 0.1895,
      "step": 3308
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.433326888222274,
      "learning_rate": 6.459666577404944e-06,
      "loss": 0.1989,
      "step": 3309
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3242387686648132,
      "learning_rate": 6.457683648302565e-06,
      "loss": 0.1862,
      "step": 3310
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4535444469496746,
      "learning_rate": 6.455700468607469e-06,
      "loss": 0.2097,
      "step": 3311
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4561364467527178,
      "learning_rate": 6.453717038660584e-06,
      "loss": 0.2423,
      "step": 3312
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2169511733163239,
      "learning_rate": 6.451733358802889e-06,
      "loss": 0.1383,
      "step": 3313
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4072078172758489,
      "learning_rate": 6.449749429375398e-06,
      "loss": 0.2469,
      "step": 3314
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.983643870607816,
      "learning_rate": 6.4477652507191744e-06,
      "loss": 0.6035,
      "step": 3315
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5922226442958025,
      "learning_rate": 6.44578082317532e-06,
      "loss": 0.2116,
      "step": 3316
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4087033725348757,
      "learning_rate": 6.443796147084982e-06,
      "loss": 0.1973,
      "step": 3317
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.020476476378708,
      "learning_rate": 6.441811222789347e-06,
      "loss": 0.6377,
      "step": 3318
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.431597814560198,
      "learning_rate": 6.439826050629649e-06,
      "loss": 0.1714,
      "step": 3319
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3481033880927666,
      "learning_rate": 6.4378406309471605e-06,
      "loss": 0.1958,
      "step": 3320
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0832536274339704,
      "learning_rate": 6.435854964083199e-06,
      "loss": 0.1546,
      "step": 3321
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4154953252547373,
      "learning_rate": 6.433869050379122e-06,
      "loss": 0.2279,
      "step": 3322
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.896986087833669,
      "learning_rate": 6.431882890176334e-06,
      "loss": 0.5071,
      "step": 3323
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.559569997264347,
      "learning_rate": 6.429896483816277e-06,
      "loss": 0.5448,
      "step": 3324
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5192338690908869,
      "learning_rate": 6.427909831640438e-06,
      "loss": 0.2004,
      "step": 3325
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3346738456645229,
      "learning_rate": 6.425922933990344e-06,
      "loss": 0.2147,
      "step": 3326
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5352851692796494,
      "learning_rate": 6.423935791207568e-06,
      "loss": 0.198,
      "step": 3327
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2532237676701525,
      "learning_rate": 6.421948403633721e-06,
      "loss": 0.1798,
      "step": 3328
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4917622410532132,
      "learning_rate": 6.4199607716104605e-06,
      "loss": 0.2929,
      "step": 3329
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.367667379912283,
      "learning_rate": 6.41797289547948e-06,
      "loss": 0.2064,
      "step": 3330
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.283926498810585,
      "learning_rate": 6.415984775582521e-06,
      "loss": 0.7206,
      "step": 3331
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.382017851074051,
      "learning_rate": 6.413996412261363e-06,
      "loss": 0.1783,
      "step": 3332
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4941366485007337,
      "learning_rate": 6.412007805857828e-06,
      "loss": 0.1995,
      "step": 3333
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4289981516945711,
      "learning_rate": 6.410018956713784e-06,
      "loss": 0.2294,
      "step": 3334
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4239480566771971,
      "learning_rate": 6.408029865171135e-06,
      "loss": 0.2269,
      "step": 3335
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.304465773413706,
      "learning_rate": 6.406040531571828e-06,
      "loss": 0.1945,
      "step": 3336
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.505071529253479,
      "learning_rate": 6.404050956257853e-06,
      "loss": 0.2194,
      "step": 3337
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4303829692859409,
      "learning_rate": 6.402061139571243e-06,
      "loss": 0.2399,
      "step": 3338
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.962604401229011,
      "learning_rate": 6.400071081854068e-06,
      "loss": 0.5131,
      "step": 3339
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3806585793987936,
      "learning_rate": 6.398080783448443e-06,
      "loss": 0.2276,
      "step": 3340
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.917483402214313,
      "learning_rate": 6.396090244696523e-06,
      "loss": 0.6406,
      "step": 3341
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4425012996335658,
      "learning_rate": 6.394099465940505e-06,
      "loss": 0.2367,
      "step": 3342
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.236484981774122,
      "learning_rate": 6.3921084475226295e-06,
      "loss": 0.1685,
      "step": 3343
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4125772338248381,
      "learning_rate": 6.39011718978517e-06,
      "loss": 0.2215,
      "step": 3344
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.659972184359046,
      "learning_rate": 6.388125693070452e-06,
      "loss": 0.2533,
      "step": 3345
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4881962959409074,
      "learning_rate": 6.386133957720833e-06,
      "loss": 0.2088,
      "step": 3346
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3338136012078925,
      "learning_rate": 6.384141984078719e-06,
      "loss": 0.172,
      "step": 3347
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3385029265175987,
      "learning_rate": 6.38214977248655e-06,
      "loss": 0.1809,
      "step": 3348
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4635004757632781,
      "learning_rate": 6.380157323286813e-06,
      "loss": 0.2427,
      "step": 3349
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4219016456831786,
      "learning_rate": 6.378164636822033e-06,
      "loss": 0.2203,
      "step": 3350
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3201739322686714,
      "learning_rate": 6.376171713434771e-06,
      "loss": 0.2068,
      "step": 3351
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3148919736842175,
      "learning_rate": 6.3741785534676404e-06,
      "loss": 0.1872,
      "step": 3352
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3711307486295923,
      "learning_rate": 6.372185157263287e-06,
      "loss": 0.2375,
      "step": 3353
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3733895107609915,
      "learning_rate": 6.370191525164394e-06,
      "loss": 0.2073,
      "step": 3354
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3633980766146325,
      "learning_rate": 6.368197657513695e-06,
      "loss": 0.2119,
      "step": 3355
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4227791128132878,
      "learning_rate": 6.366203554653957e-06,
      "loss": 0.2344,
      "step": 3356
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3847005251754176,
      "learning_rate": 6.36420921692799e-06,
      "loss": 0.2015,
      "step": 3357
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.9939238457196495,
      "learning_rate": 6.362214644678641e-06,
      "loss": 0.4718,
      "step": 3358
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.35203911279669,
      "learning_rate": 6.360219838248806e-06,
      "loss": 0.5958,
      "step": 3359
    },
    {
      "epoch": 0.43,
      "grad_norm": 8.492357940109944,
      "learning_rate": 6.358224797981409e-06,
      "loss": 0.6299,
      "step": 3360
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2398628318765443,
      "learning_rate": 6.356229524219425e-06,
      "loss": 0.1793,
      "step": 3361
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6180375772531632,
      "learning_rate": 6.354234017305863e-06,
      "loss": 0.2336,
      "step": 3362
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3777251098078573,
      "learning_rate": 6.352238277583773e-06,
      "loss": 0.2197,
      "step": 3363
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.458572844718904,
      "learning_rate": 6.350242305396247e-06,
      "loss": 0.2247,
      "step": 3364
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4798308513562992,
      "learning_rate": 6.348246101086414e-06,
      "loss": 0.2311,
      "step": 3365
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3961160178712073,
      "learning_rate": 6.346249664997448e-06,
      "loss": 0.1808,
      "step": 3366
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.696349435220102,
      "learning_rate": 6.344252997472556e-06,
      "loss": 0.6595,
      "step": 3367
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.977185599146695,
      "learning_rate": 6.342256098854992e-06,
      "loss": 0.5254,
      "step": 3368
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4355002982393796,
      "learning_rate": 6.34025896948804e-06,
      "loss": 0.2008,
      "step": 3369
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.332852915938578,
      "learning_rate": 6.338261609715037e-06,
      "loss": 0.2008,
      "step": 3370
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3532725111698032,
      "learning_rate": 6.336264019879348e-06,
      "loss": 0.1877,
      "step": 3371
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.363078545395588,
      "learning_rate": 6.334266200324381e-06,
      "loss": 0.1561,
      "step": 3372
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.415628638503568,
      "learning_rate": 6.332268151393589e-06,
      "loss": 0.2319,
      "step": 3373
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5568155185402244,
      "learning_rate": 6.330269873430455e-06,
      "loss": 0.2285,
      "step": 3374
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4883149118925414,
      "learning_rate": 6.3282713667785086e-06,
      "loss": 0.2264,
      "step": 3375
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5164875849456865,
      "learning_rate": 6.326272631781314e-06,
      "loss": 0.2161,
      "step": 3376
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.2409313645760305,
      "learning_rate": 6.32427366878248e-06,
      "loss": 0.5283,
      "step": 3377
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.541008299383714,
      "learning_rate": 6.322274478125651e-06,
      "loss": 0.2701,
      "step": 3378
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2856570999846533,
      "learning_rate": 6.320275060154508e-06,
      "loss": 0.1643,
      "step": 3379
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6109711774226143,
      "learning_rate": 6.318275415212777e-06,
      "loss": 0.2711,
      "step": 3380
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5173841098167942,
      "learning_rate": 6.316275543644219e-06,
      "loss": 0.2127,
      "step": 3381
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.121701714121274,
      "learning_rate": 6.314275445792637e-06,
      "loss": 0.5781,
      "step": 3382
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.175749080936147,
      "learning_rate": 6.312275122001867e-06,
      "loss": 0.6987,
      "step": 3383
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.304474249655251,
      "learning_rate": 6.310274572615792e-06,
      "loss": 0.1827,
      "step": 3384
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.555573412677199,
      "learning_rate": 6.308273797978328e-06,
      "loss": 0.2019,
      "step": 3385
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.417314393674722,
      "learning_rate": 6.30627279843343e-06,
      "loss": 0.2094,
      "step": 3386
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3148520821690872,
      "learning_rate": 6.304271574325096e-06,
      "loss": 0.1817,
      "step": 3387
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4946646999840718,
      "learning_rate": 6.3022701259973565e-06,
      "loss": 0.2088,
      "step": 3388
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6895823272742974,
      "learning_rate": 6.300268453794287e-06,
      "loss": 0.2444,
      "step": 3389
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2473542808382259,
      "learning_rate": 6.298266558059995e-06,
      "loss": 0.1668,
      "step": 3390
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6162359713396743,
      "learning_rate": 6.296264439138631e-06,
      "loss": 0.2523,
      "step": 3391
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5971758472875377,
      "learning_rate": 6.294262097374383e-06,
      "loss": 0.2187,
      "step": 3392
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.775546587507935,
      "learning_rate": 6.292259533111474e-06,
      "loss": 0.5655,
      "step": 3393
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3033650615642094,
      "learning_rate": 6.2902567466941725e-06,
      "loss": 0.2076,
      "step": 3394
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.480818572211129,
      "learning_rate": 6.288253738466777e-06,
      "loss": 0.2177,
      "step": 3395
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.975405177036882,
      "learning_rate": 6.286250508773631e-06,
      "loss": 0.6726,
      "step": 3396
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1454154028206776,
      "learning_rate": 6.284247057959107e-06,
      "loss": 0.1662,
      "step": 3397
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5155624057492803,
      "learning_rate": 6.282243386367628e-06,
      "loss": 0.2341,
      "step": 3398
    },
    {
      "epoch": 0.44,
      "grad_norm": 19.16925720531021,
      "learning_rate": 6.280239494343647e-06,
      "loss": 0.5803,
      "step": 3399
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7077591417584044,
      "learning_rate": 6.278235382231654e-06,
      "loss": 0.1875,
      "step": 3400
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2729215331712846,
      "learning_rate": 6.27623105037618e-06,
      "loss": 0.2144,
      "step": 3401
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3814351515564758,
      "learning_rate": 6.274226499121793e-06,
      "loss": 0.2508,
      "step": 3402
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.572326800368013,
      "learning_rate": 6.272221728813099e-06,
      "loss": 0.1793,
      "step": 3403
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4199238507165826,
      "learning_rate": 6.270216739794741e-06,
      "loss": 0.2348,
      "step": 3404
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4628405189730769,
      "learning_rate": 6.2682115324114e-06,
      "loss": 0.2322,
      "step": 3405
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3510773619929344,
      "learning_rate": 6.266206107007793e-06,
      "loss": 0.1753,
      "step": 3406
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.512906951483142,
      "learning_rate": 6.264200463928679e-06,
      "loss": 0.6103,
      "step": 3407
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4987143613004998,
      "learning_rate": 6.2621946035188474e-06,
      "loss": 0.1908,
      "step": 3408
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3702702224557648,
      "learning_rate": 6.260188526123133e-06,
      "loss": 0.1985,
      "step": 3409
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4937681928891775,
      "learning_rate": 6.2581822320864e-06,
      "loss": 0.1797,
      "step": 3410
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2076183045440205,
      "learning_rate": 6.256175721753556e-06,
      "loss": 0.1798,
      "step": 3411
    },
    {
      "epoch": 0.44,
      "grad_norm": 11.62827490404163,
      "learning_rate": 6.254168995469541e-06,
      "loss": 0.531,
      "step": 3412
    },
    {
      "epoch": 0.44,
      "grad_norm": 8.260454378494906,
      "learning_rate": 6.252162053579338e-06,
      "loss": 0.5689,
      "step": 3413
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5083953271372315,
      "learning_rate": 6.250154896427962e-06,
      "loss": 0.1739,
      "step": 3414
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4843502838092026,
      "learning_rate": 6.2481475243604654e-06,
      "loss": 0.2069,
      "step": 3415
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.471710099562942,
      "learning_rate": 6.246139937721939e-06,
      "loss": 0.2525,
      "step": 3416
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.883534092685821,
      "learning_rate": 6.244132136857511e-06,
      "loss": 0.441,
      "step": 3417
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4758053002825156,
      "learning_rate": 6.242124122112347e-06,
      "loss": 0.2151,
      "step": 3418
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.19437449933456,
      "learning_rate": 6.240115893831644e-06,
      "loss": 0.4604,
      "step": 3419
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6779671178287225,
      "learning_rate": 6.238107452360643e-06,
      "loss": 0.1849,
      "step": 3420
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.986092493244726,
      "learning_rate": 6.236098798044619e-06,
      "loss": 0.6408,
      "step": 3421
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4304294474033472,
      "learning_rate": 6.2340899312288795e-06,
      "loss": 0.1749,
      "step": 3422
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.478385729948262,
      "learning_rate": 6.232080852258776e-06,
      "loss": 0.1839,
      "step": 3423
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.242638520650504,
      "learning_rate": 6.23007156147969e-06,
      "loss": 0.1586,
      "step": 3424
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5181312601485633,
      "learning_rate": 6.228062059237041e-06,
      "loss": 0.2699,
      "step": 3425
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.402969019131633,
      "learning_rate": 6.226052345876288e-06,
      "loss": 0.1968,
      "step": 3426
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.276626163928174,
      "learning_rate": 6.224042421742924e-06,
      "loss": 0.5462,
      "step": 3427
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4188210798374972,
      "learning_rate": 6.222032287182477e-06,
      "loss": 0.212,
      "step": 3428
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.2157238396251095,
      "learning_rate": 6.220021942540512e-06,
      "loss": 0.5514,
      "step": 3429
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3771064427697435,
      "learning_rate": 6.21801138816263e-06,
      "loss": 0.1871,
      "step": 3430
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4031086635425674,
      "learning_rate": 6.2160006243944726e-06,
      "loss": 0.2073,
      "step": 3431
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3493844083434086,
      "learning_rate": 6.213989651581711e-06,
      "loss": 0.1739,
      "step": 3432
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2485967450877031,
      "learning_rate": 6.211978470070052e-06,
      "loss": 0.1573,
      "step": 3433
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5530208519620217,
      "learning_rate": 6.209967080205244e-06,
      "loss": 0.2028,
      "step": 3434
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.551301179195029,
      "learning_rate": 6.20795548233307e-06,
      "loss": 0.4815,
      "step": 3435
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4342338839864128,
      "learning_rate": 6.205943676799344e-06,
      "loss": 0.1589,
      "step": 3436
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.707952226890874,
      "learning_rate": 6.203931663949918e-06,
      "loss": 0.525,
      "step": 3437
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2643611104785581,
      "learning_rate": 6.201919444130684e-06,
      "loss": 0.2172,
      "step": 3438
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.0660056685866515,
      "learning_rate": 6.199907017687562e-06,
      "loss": 0.6061,
      "step": 3439
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4245111528905448,
      "learning_rate": 6.197894384966513e-06,
      "loss": 0.2309,
      "step": 3440
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.569840029272576,
      "learning_rate": 6.195881546313533e-06,
      "loss": 0.2251,
      "step": 3441
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.405935170726231,
      "learning_rate": 6.193868502074651e-06,
      "loss": 0.2158,
      "step": 3442
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.451525486072896,
      "learning_rate": 6.191855252595933e-06,
      "loss": 0.1955,
      "step": 3443
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5500039476068526,
      "learning_rate": 6.189841798223479e-06,
      "loss": 0.2928,
      "step": 3444
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6903077300886347,
      "learning_rate": 6.1878281393034275e-06,
      "loss": 0.2272,
      "step": 3445
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4621580213520744,
      "learning_rate": 6.1858142761819484e-06,
      "loss": 0.2375,
      "step": 3446
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.597415633591744,
      "learning_rate": 6.1838002092052465e-06,
      "loss": 0.5897,
      "step": 3447
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3066418678789027,
      "learning_rate": 6.181785938719566e-06,
      "loss": 0.206,
      "step": 3448
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4526041078222622,
      "learning_rate": 6.179771465071182e-06,
      "loss": 0.1885,
      "step": 3449
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3060356961658275,
      "learning_rate": 6.177756788606406e-06,
      "loss": 0.1767,
      "step": 3450
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3607119264222352,
      "learning_rate": 6.175741909671584e-06,
      "loss": 0.2163,
      "step": 3451
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4805369785903384,
      "learning_rate": 6.173726828613098e-06,
      "loss": 0.2413,
      "step": 3452
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2957504711629892,
      "learning_rate": 6.171711545777363e-06,
      "loss": 0.1998,
      "step": 3453
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.449429363397936,
      "learning_rate": 6.169696061510831e-06,
      "loss": 0.2077,
      "step": 3454
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4504879659793681,
      "learning_rate": 6.167680376159983e-06,
      "loss": 0.2107,
      "step": 3455
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.557184414569972,
      "learning_rate": 6.165664490071343e-06,
      "loss": 0.1994,
      "step": 3456
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1882523948330128,
      "learning_rate": 6.163648403591462e-06,
      "loss": 0.1803,
      "step": 3457
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.362546434278665,
      "learning_rate": 6.161632117066932e-06,
      "loss": 0.1986,
      "step": 3458
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2769265632672266,
      "learning_rate": 6.1596156308443746e-06,
      "loss": 0.1794,
      "step": 3459
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3770366105053562,
      "learning_rate": 6.157598945270447e-06,
      "loss": 0.2242,
      "step": 3460
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4359206537614428,
      "learning_rate": 6.1555820606918384e-06,
      "loss": 0.2071,
      "step": 3461
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.8004848299269405,
      "learning_rate": 6.153564977455278e-06,
      "loss": 0.3992,
      "step": 3462
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4372615182017368,
      "learning_rate": 6.151547695907525e-06,
      "loss": 0.2282,
      "step": 3463
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5040101701914135,
      "learning_rate": 6.149530216395374e-06,
      "loss": 0.2064,
      "step": 3464
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.1535248443588735,
      "learning_rate": 6.1475125392656506e-06,
      "loss": 0.6333,
      "step": 3465
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2922312074314526,
      "learning_rate": 6.1454946648652204e-06,
      "loss": 0.2086,
      "step": 3466
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4586932798226817,
      "learning_rate": 6.143476593540976e-06,
      "loss": 0.1878,
      "step": 3467
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4872871434852712,
      "learning_rate": 6.1414583256398494e-06,
      "loss": 0.216,
      "step": 3468
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2508496658100363,
      "learning_rate": 6.139439861508804e-06,
      "loss": 0.1645,
      "step": 3469
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2656104759846125,
      "learning_rate": 6.137421201494837e-06,
      "loss": 0.1786,
      "step": 3470
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6780307651786635,
      "learning_rate": 6.135402345944979e-06,
      "loss": 0.221,
      "step": 3471
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5444518989750775,
      "learning_rate": 6.1333832952062945e-06,
      "loss": 0.2207,
      "step": 3472
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2534193514830114,
      "learning_rate": 6.1313640496258834e-06,
      "loss": 0.1684,
      "step": 3473
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.290280725162761,
      "learning_rate": 6.129344609550876e-06,
      "loss": 0.5595,
      "step": 3474
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.249219156182278,
      "learning_rate": 6.127324975328437e-06,
      "loss": 0.4241,
      "step": 3475
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.393265603193384,
      "learning_rate": 6.125305147305764e-06,
      "loss": 0.1843,
      "step": 3476
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4840706487944058,
      "learning_rate": 6.1232851258300944e-06,
      "loss": 0.2243,
      "step": 3477
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.242389090519873,
      "learning_rate": 6.121264911248688e-06,
      "loss": 0.3962,
      "step": 3478
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.520758533451468,
      "learning_rate": 6.1192445039088435e-06,
      "loss": 0.2308,
      "step": 3479
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.580048155262556,
      "learning_rate": 6.117223904157893e-06,
      "loss": 0.2413,
      "step": 3480
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.217811698409541,
      "learning_rate": 6.115203112343203e-06,
      "loss": 0.7279,
      "step": 3481
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4913584822080914,
      "learning_rate": 6.11318212881217e-06,
      "loss": 0.2106,
      "step": 3482
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3783713068953412,
      "learning_rate": 6.111160953912222e-06,
      "loss": 0.2366,
      "step": 3483
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6187664233992558,
      "learning_rate": 6.1091395879908255e-06,
      "loss": 0.1995,
      "step": 3484
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5030788309642515,
      "learning_rate": 6.107118031395475e-06,
      "loss": 0.2077,
      "step": 3485
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3659966292355485,
      "learning_rate": 6.1050962844737005e-06,
      "loss": 0.2043,
      "step": 3486
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.436432993487482,
      "learning_rate": 6.103074347573062e-06,
      "loss": 0.2096,
      "step": 3487
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3999201301585855,
      "learning_rate": 6.1010522210411575e-06,
      "loss": 0.1821,
      "step": 3488
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5943641631430183,
      "learning_rate": 6.0990299052256105e-06,
      "loss": 0.2197,
      "step": 3489
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.429589782257565,
      "learning_rate": 6.097007400474081e-06,
      "loss": 0.5862,
      "step": 3490
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4258698259556128,
      "learning_rate": 6.094984707134263e-06,
      "loss": 0.1983,
      "step": 3491
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.599725460273218,
      "learning_rate": 6.092961825553881e-06,
      "loss": 0.5739,
      "step": 3492
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5308431108357887,
      "learning_rate": 6.090938756080688e-06,
      "loss": 0.2067,
      "step": 3493
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.471306078727796,
      "learning_rate": 6.088915499062475e-06,
      "loss": 0.2115,
      "step": 3494
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3355841423036094,
      "learning_rate": 6.0868920548470654e-06,
      "loss": 0.2003,
      "step": 3495
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.279214726455282,
      "learning_rate": 6.084868423782312e-06,
      "loss": 0.477,
      "step": 3496
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3504746608583638,
      "learning_rate": 6.082844606216098e-06,
      "loss": 0.2068,
      "step": 3497
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3613000404964442,
      "learning_rate": 6.080820602496345e-06,
      "loss": 0.1758,
      "step": 3498
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.450520947125926,
      "learning_rate": 6.078796412970997e-06,
      "loss": 0.205,
      "step": 3499
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4880021120985116,
      "learning_rate": 6.076772037988042e-06,
      "loss": 0.1848,
      "step": 3500
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3716716071359152,
      "learning_rate": 6.07474747789549e-06,
      "loss": 0.2285,
      "step": 3501
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.468323068777984,
      "learning_rate": 6.072722733041387e-06,
      "loss": 0.2551,
      "step": 3502
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.57020687732641,
      "learning_rate": 6.07069780377381e-06,
      "loss": 0.1823,
      "step": 3503
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.513150347090133,
      "learning_rate": 6.068672690440868e-06,
      "loss": 0.4246,
      "step": 3504
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4964500005940837,
      "learning_rate": 6.066647393390701e-06,
      "loss": 0.1885,
      "step": 3505
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.382814548663258,
      "learning_rate": 6.064621912971483e-06,
      "loss": 0.228,
      "step": 3506
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3372416912747542,
      "learning_rate": 6.062596249531414e-06,
      "loss": 0.1966,
      "step": 3507
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4380133891018596,
      "learning_rate": 6.060570403418731e-06,
      "loss": 0.2006,
      "step": 3508
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3886812198074452,
      "learning_rate": 6.058544374981701e-06,
      "loss": 0.1913,
      "step": 3509
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2943694010494202,
      "learning_rate": 6.056518164568622e-06,
      "loss": 0.1557,
      "step": 3510
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3114723030342446,
      "learning_rate": 6.054491772527822e-06,
      "loss": 0.2302,
      "step": 3511
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5755501661682956,
      "learning_rate": 6.052465199207661e-06,
      "loss": 0.2548,
      "step": 3512
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.50393191976377,
      "learning_rate": 6.050438444956531e-06,
      "loss": 0.2762,
      "step": 3513
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.2419643964948985,
      "learning_rate": 6.048411510122855e-06,
      "loss": 0.4943,
      "step": 3514
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.295606391046375,
      "learning_rate": 6.046384395055086e-06,
      "loss": 0.1941,
      "step": 3515
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5137464121154103,
      "learning_rate": 6.04435710010171e-06,
      "loss": 0.2621,
      "step": 3516
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4439200632723455,
      "learning_rate": 6.042329625611239e-06,
      "loss": 0.1881,
      "step": 3517
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.30307099611612,
      "learning_rate": 6.040301971932223e-06,
      "loss": 0.1915,
      "step": 3518
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3632417270758477,
      "learning_rate": 6.038274139413238e-06,
      "loss": 0.2118,
      "step": 3519
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3200224477828568,
      "learning_rate": 6.036246128402892e-06,
      "loss": 0.191,
      "step": 3520
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3808778914719442,
      "learning_rate": 6.034217939249823e-06,
      "loss": 0.2173,
      "step": 3521
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6072292027351112,
      "learning_rate": 6.0321895723027e-06,
      "loss": 0.2022,
      "step": 3522
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3976892747880059,
      "learning_rate": 6.030161027910223e-06,
      "loss": 0.195,
      "step": 3523
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4855317153683216,
      "learning_rate": 6.028132306421124e-06,
      "loss": 0.1938,
      "step": 3524
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2329289310987874,
      "learning_rate": 6.026103408184162e-06,
      "loss": 0.189,
      "step": 3525
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4897235296113904,
      "learning_rate": 6.0240743335481265e-06,
      "loss": 0.1796,
      "step": 3526
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.346075688622765,
      "learning_rate": 6.0220450828618424e-06,
      "loss": 0.2079,
      "step": 3527
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.432545010160866,
      "learning_rate": 6.0200156564741606e-06,
      "loss": 0.2239,
      "step": 3528
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3318928082219819,
      "learning_rate": 6.017986054733962e-06,
      "loss": 0.2147,
      "step": 3529
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.151838432537424,
      "learning_rate": 6.0159562779901605e-06,
      "loss": 0.5089,
      "step": 3530
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.417170228103885,
      "learning_rate": 6.013926326591695e-06,
      "loss": 0.2577,
      "step": 3531
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.279783203759128,
      "learning_rate": 6.0118962008875395e-06,
      "loss": 0.1965,
      "step": 3532
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3589422900588461,
      "learning_rate": 6.009865901226697e-06,
      "loss": 0.167,
      "step": 3533
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4059495971515523,
      "learning_rate": 6.007835427958199e-06,
      "loss": 0.1963,
      "step": 3534
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.094694087076251,
      "learning_rate": 6.005804781431106e-06,
      "loss": 0.5252,
      "step": 3535
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.2577008564962675,
      "learning_rate": 6.0037739619945114e-06,
      "loss": 0.5713,
      "step": 3536
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4654645254038443,
      "learning_rate": 6.001742969997535e-06,
      "loss": 0.2133,
      "step": 3537
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4552842397810601,
      "learning_rate": 5.99971180578933e-06,
      "loss": 0.2119,
      "step": 3538
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3957261112180313,
      "learning_rate": 5.997680469719076e-06,
      "loss": 0.2267,
      "step": 3539
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5691467932050975,
      "learning_rate": 5.995648962135983e-06,
      "loss": 0.2258,
      "step": 3540
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3350213385473704,
      "learning_rate": 5.993617283389289e-06,
      "loss": 0.175,
      "step": 3541
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6974171219617253,
      "learning_rate": 5.991585433828267e-06,
      "loss": 0.2293,
      "step": 3542
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4977978007751764,
      "learning_rate": 5.9895534138022136e-06,
      "loss": 0.2368,
      "step": 3543
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4446774522353831,
      "learning_rate": 5.9875212236604564e-06,
      "loss": 0.2321,
      "step": 3544
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2832396314070156,
      "learning_rate": 5.985488863752351e-06,
      "loss": 0.1649,
      "step": 3545
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3269142889289725,
      "learning_rate": 5.983456334427286e-06,
      "loss": 0.1778,
      "step": 3546
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4006617618723605,
      "learning_rate": 5.9814236360346765e-06,
      "loss": 0.1873,
      "step": 3547
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4948786452219027,
      "learning_rate": 5.9793907689239675e-06,
      "loss": 0.2025,
      "step": 3548
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.290249783368535,
      "learning_rate": 5.97735773344463e-06,
      "loss": 0.2034,
      "step": 3549
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3128109295128976,
      "learning_rate": 5.975324529946166e-06,
      "loss": 0.1877,
      "step": 3550
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.519641498237452,
      "learning_rate": 5.973291158778109e-06,
      "loss": 0.5817,
      "step": 3551
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.249446502806921,
      "learning_rate": 5.97125762029002e-06,
      "loss": 0.1648,
      "step": 3552
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.30700000802388,
      "learning_rate": 5.969223914831485e-06,
      "loss": 0.1727,
      "step": 3553
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.508476226856945,
      "learning_rate": 5.967190042752123e-06,
      "loss": 0.6044,
      "step": 3554
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.193579070053018,
      "learning_rate": 5.965156004401581e-06,
      "loss": 0.1603,
      "step": 3555
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3608572399923782,
      "learning_rate": 5.9631218001295325e-06,
      "loss": 0.2405,
      "step": 3556
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6663231002257533,
      "learning_rate": 5.961087430285681e-06,
      "loss": 0.2755,
      "step": 3557
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.247368411103134,
      "learning_rate": 5.959052895219758e-06,
      "loss": 0.2018,
      "step": 3558
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.5823303817016985,
      "learning_rate": 5.957018195281523e-06,
      "loss": 0.5553,
      "step": 3559
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3138979580912888,
      "learning_rate": 5.954983330820767e-06,
      "loss": 0.1814,
      "step": 3560
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2076103178793238,
      "learning_rate": 5.9529483021873055e-06,
      "loss": 0.159,
      "step": 3561
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4530318754874025,
      "learning_rate": 5.950913109730983e-06,
      "loss": 0.2277,
      "step": 3562
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4786434522834602,
      "learning_rate": 5.948877753801673e-06,
      "loss": 0.2558,
      "step": 3563
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5681948201296545,
      "learning_rate": 5.946842234749275e-06,
      "loss": 0.2224,
      "step": 3564
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3899310826159583,
      "learning_rate": 5.944806552923722e-06,
      "loss": 0.1827,
      "step": 3565
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.533699257536805,
      "learning_rate": 5.942770708674969e-06,
      "loss": 0.2441,
      "step": 3566
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.95781789543949,
      "learning_rate": 5.940734702353002e-06,
      "loss": 0.7925,
      "step": 3567
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3090423038326697,
      "learning_rate": 5.938698534307833e-06,
      "loss": 0.218,
      "step": 3568
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3932754947349082,
      "learning_rate": 5.936662204889504e-06,
      "loss": 0.2383,
      "step": 3569
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4097111035147711,
      "learning_rate": 5.934625714448084e-06,
      "loss": 0.2143,
      "step": 3570
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4057764538993818,
      "learning_rate": 5.932589063333668e-06,
      "loss": 0.1761,
      "step": 3571
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5707043896048591,
      "learning_rate": 5.9305522518963795e-06,
      "loss": 0.2085,
      "step": 3572
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4115878553281374,
      "learning_rate": 5.928515280486372e-06,
      "loss": 0.2329,
      "step": 3573
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.350996977399288,
      "learning_rate": 5.9264781494538235e-06,
      "loss": 0.1561,
      "step": 3574
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3751980348580424,
      "learning_rate": 5.924440859148941e-06,
      "loss": 0.2098,
      "step": 3575
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.632918269626928,
      "learning_rate": 5.922403409921957e-06,
      "loss": 0.1988,
      "step": 3576
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.226271288927771,
      "learning_rate": 5.9203658021231335e-06,
      "loss": 0.6834,
      "step": 3577
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.330747292914036,
      "learning_rate": 5.918328036102758e-06,
      "loss": 0.2187,
      "step": 3578
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.555454015196096,
      "learning_rate": 5.916290112211149e-06,
      "loss": 0.2196,
      "step": 3579
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4882961087612372,
      "learning_rate": 5.9142520307986455e-06,
      "loss": 0.1944,
      "step": 3580
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6200829947275592,
      "learning_rate": 5.91221379221562e-06,
      "loss": 0.1903,
      "step": 3581
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3339198842750593,
      "learning_rate": 5.910175396812468e-06,
      "loss": 0.1896,
      "step": 3582
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3627942184337658,
      "learning_rate": 5.908136844939612e-06,
      "loss": 0.2432,
      "step": 3583
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.699171094418912,
      "learning_rate": 5.906098136947506e-06,
      "loss": 0.5997,
      "step": 3584
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3576369819537186,
      "learning_rate": 5.904059273186627e-06,
      "loss": 0.1899,
      "step": 3585
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.502889157963305,
      "learning_rate": 5.9020202540074755e-06,
      "loss": 0.2255,
      "step": 3586
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5232251291780636,
      "learning_rate": 5.899981079760586e-06,
      "loss": 0.207,
      "step": 3587
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4873746166804407,
      "learning_rate": 5.897941750796517e-06,
      "loss": 0.2373,
      "step": 3588
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3607627884623188,
      "learning_rate": 5.895902267465851e-06,
      "loss": 0.1878,
      "step": 3589
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4504168904748191,
      "learning_rate": 5.893862630119197e-06,
      "loss": 0.2318,
      "step": 3590
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3032621825312107,
      "learning_rate": 5.891822839107195e-06,
      "loss": 0.1987,
      "step": 3591
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3680531327274434,
      "learning_rate": 5.8897828947805094e-06,
      "loss": 0.1515,
      "step": 3592
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2594488946327282,
      "learning_rate": 5.887742797489828e-06,
      "loss": 0.181,
      "step": 3593
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4920685817899693,
      "learning_rate": 5.8857025475858676e-06,
      "loss": 0.1947,
      "step": 3594
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.400662135106231,
      "learning_rate": 5.883662145419373e-06,
      "loss": 0.2097,
      "step": 3595
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.4432551376976,
      "learning_rate": 5.881621591341109e-06,
      "loss": 0.602,
      "step": 3596
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3473548224823915,
      "learning_rate": 5.879580885701874e-06,
      "loss": 0.1672,
      "step": 3597
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.342232532846938,
      "learning_rate": 5.877540028852489e-06,
      "loss": 0.1619,
      "step": 3598
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3376558890082415,
      "learning_rate": 5.875499021143799e-06,
      "loss": 0.1968,
      "step": 3599
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3407047871208764,
      "learning_rate": 5.873457862926677e-06,
      "loss": 0.1778,
      "step": 3600
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.805840780979361,
      "learning_rate": 5.871416554552021e-06,
      "loss": 0.7429,
      "step": 3601
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6893335157575944,
      "learning_rate": 5.869375096370759e-06,
      "loss": 0.1914,
      "step": 3602
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3822071585381979,
      "learning_rate": 5.86733348873384e-06,
      "loss": 0.213,
      "step": 3603
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.543364625532563,
      "learning_rate": 5.8652917319922374e-06,
      "loss": 0.2347,
      "step": 3604
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5035815714139322,
      "learning_rate": 5.863249826496955e-06,
      "loss": 0.1917,
      "step": 3605
    },
    {
      "epoch": 0.46,
      "grad_norm": 29.22982105863004,
      "learning_rate": 5.8612077725990206e-06,
      "loss": 0.7313,
      "step": 3606
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4374126773769147,
      "learning_rate": 5.859165570649485e-06,
      "loss": 0.2081,
      "step": 3607
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.823618106906577,
      "learning_rate": 5.857123220999429e-06,
      "loss": 0.6136,
      "step": 3608
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4803571056377944,
      "learning_rate": 5.855080723999954e-06,
      "loss": 0.2562,
      "step": 3609
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.299991307641927,
      "learning_rate": 5.853038080002189e-06,
      "loss": 0.1958,
      "step": 3610
    },
    {
      "epoch": 0.46,
      "grad_norm": 10.154060546777192,
      "learning_rate": 5.85099528935729e-06,
      "loss": 0.7084,
      "step": 3611
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4515821645940519,
      "learning_rate": 5.848952352416434e-06,
      "loss": 0.2295,
      "step": 3612
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.217766149737196,
      "learning_rate": 5.8469092695308274e-06,
      "loss": 0.1711,
      "step": 3613
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.549066089172687,
      "learning_rate": 5.844866041051699e-06,
      "loss": 0.2511,
      "step": 3614
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.002272525665449,
      "learning_rate": 5.8428226673303026e-06,
      "loss": 0.6321,
      "step": 3615
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3372064629606253,
      "learning_rate": 5.84077914871792e-06,
      "loss": 0.202,
      "step": 3616
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.255613147262601,
      "learning_rate": 5.838735485565855e-06,
      "loss": 0.5165,
      "step": 3617
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.583872468806402,
      "learning_rate": 5.8366916782254345e-06,
      "loss": 0.2371,
      "step": 3618
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6148519180576089,
      "learning_rate": 5.834647727048016e-06,
      "loss": 0.2337,
      "step": 3619
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2244068420413237,
      "learning_rate": 5.832603632384978e-06,
      "loss": 0.1781,
      "step": 3620
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3155450392125385,
      "learning_rate": 5.8305593945877236e-06,
      "loss": 0.1435,
      "step": 3621
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2974289162381452,
      "learning_rate": 5.828515014007678e-06,
      "loss": 0.2193,
      "step": 3622
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.921604913545427,
      "learning_rate": 5.826470490996299e-06,
      "loss": 0.5257,
      "step": 3623
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3850083451581323,
      "learning_rate": 5.82442582590506e-06,
      "loss": 0.199,
      "step": 3624
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5886391608129737,
      "learning_rate": 5.822381019085466e-06,
      "loss": 0.2071,
      "step": 3625
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4501512457049455,
      "learning_rate": 5.820336070889038e-06,
      "loss": 0.2125,
      "step": 3626
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3447668693653851,
      "learning_rate": 5.8182909816673316e-06,
      "loss": 0.2414,
      "step": 3627
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4055324061771255,
      "learning_rate": 5.816245751771917e-06,
      "loss": 0.2233,
      "step": 3628
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2510352379480403,
      "learning_rate": 5.814200381554397e-06,
      "loss": 0.1916,
      "step": 3629
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4268237696323582,
      "learning_rate": 5.81215487136639e-06,
      "loss": 0.2248,
      "step": 3630
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4650468582371667,
      "learning_rate": 5.810109221559548e-06,
      "loss": 0.2281,
      "step": 3631
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4380807414617771,
      "learning_rate": 5.808063432485538e-06,
      "loss": 0.1942,
      "step": 3632
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3609198728396885,
      "learning_rate": 5.806017504496055e-06,
      "loss": 0.1924,
      "step": 3633
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1832775170201768,
      "learning_rate": 5.803971437942819e-06,
      "loss": 0.1287,
      "step": 3634
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4732173562901085,
      "learning_rate": 5.801925233177574e-06,
      "loss": 0.2004,
      "step": 3635
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.578132424874161,
      "learning_rate": 5.799878890552083e-06,
      "loss": 0.1984,
      "step": 3636
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4322692353752704,
      "learning_rate": 5.79783241041814e-06,
      "loss": 0.2036,
      "step": 3637
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.455133832090002,
      "learning_rate": 5.795785793127554e-06,
      "loss": 0.2045,
      "step": 3638
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5520064886698268,
      "learning_rate": 5.793739039032166e-06,
      "loss": 0.1905,
      "step": 3639
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3720435649535945,
      "learning_rate": 5.791692148483834e-06,
      "loss": 0.1947,
      "step": 3640
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3971434441009376,
      "learning_rate": 5.789645121834445e-06,
      "loss": 0.1998,
      "step": 3641
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3874605319646738,
      "learning_rate": 5.7875979594359045e-06,
      "loss": 0.2035,
      "step": 3642
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.487904517376631,
      "learning_rate": 5.785550661640145e-06,
      "loss": 0.6123,
      "step": 3643
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3663043613869832,
      "learning_rate": 5.783503228799119e-06,
      "loss": 0.2157,
      "step": 3644
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3761245273603493,
      "learning_rate": 5.781455661264805e-06,
      "loss": 0.2026,
      "step": 3645
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3447889505117083,
      "learning_rate": 5.779407959389205e-06,
      "loss": 0.203,
      "step": 3646
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4584319531284675,
      "learning_rate": 5.777360123524338e-06,
      "loss": 0.2251,
      "step": 3647
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2026114042688576,
      "learning_rate": 5.775312154022256e-06,
      "loss": 0.1703,
      "step": 3648
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4773926860473887,
      "learning_rate": 5.773264051235026e-06,
      "loss": 0.2091,
      "step": 3649
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.455679128031975,
      "learning_rate": 5.771215815514741e-06,
      "loss": 0.2477,
      "step": 3650
    },
    {
      "epoch": 0.47,
      "grad_norm": 13.824347593662823,
      "learning_rate": 5.7691674472135175e-06,
      "loss": 0.6201,
      "step": 3651
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.49646295156964,
      "learning_rate": 5.767118946683491e-06,
      "loss": 0.1551,
      "step": 3652
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4159386084299326,
      "learning_rate": 5.765070314276826e-06,
      "loss": 0.1913,
      "step": 3653
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4680433463403528,
      "learning_rate": 5.763021550345703e-06,
      "loss": 0.2032,
      "step": 3654
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.599179903611691,
      "learning_rate": 5.760972655242332e-06,
      "loss": 0.7588,
      "step": 3655
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4104932539240786,
      "learning_rate": 5.758923629318938e-06,
      "loss": 0.2001,
      "step": 3656
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2526992678808215,
      "learning_rate": 5.756874472927775e-06,
      "loss": 0.1906,
      "step": 3657
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.440878329103373,
      "learning_rate": 5.7548251864211135e-06,
      "loss": 0.2234,
      "step": 3658
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3802131803752637,
      "learning_rate": 5.752775770151254e-06,
      "loss": 0.2738,
      "step": 3659
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5109077179949952,
      "learning_rate": 5.750726224470513e-06,
      "loss": 0.2156,
      "step": 3660
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3824369152565859,
      "learning_rate": 5.748676549731229e-06,
      "loss": 0.2023,
      "step": 3661
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.026131871241856,
      "learning_rate": 5.746626746285767e-06,
      "loss": 0.5378,
      "step": 3662
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4199820917339958,
      "learning_rate": 5.744576814486512e-06,
      "loss": 0.1955,
      "step": 3663
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5401652887844899,
      "learning_rate": 5.742526754685872e-06,
      "loss": 0.2325,
      "step": 3664
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3535736259718956,
      "learning_rate": 5.740476567236273e-06,
      "loss": 0.1943,
      "step": 3665
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.39038802241455,
      "learning_rate": 5.738426252490168e-06,
      "loss": 0.1924,
      "step": 3666
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3579301204210619,
      "learning_rate": 5.73637581080003e-06,
      "loss": 0.179,
      "step": 3667
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3368033630485028,
      "learning_rate": 5.734325242518355e-06,
      "loss": 0.1964,
      "step": 3668
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4277723618371148,
      "learning_rate": 5.732274547997655e-06,
      "loss": 0.1972,
      "step": 3669
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2783977876757693,
      "learning_rate": 5.730223727590472e-06,
      "loss": 0.194,
      "step": 3670
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3819089720556237,
      "learning_rate": 5.728172781649365e-06,
      "loss": 0.1999,
      "step": 3671
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2451828421291424,
      "learning_rate": 5.726121710526914e-06,
      "loss": 0.1742,
      "step": 3672
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4425278101767633,
      "learning_rate": 5.7240705145757244e-06,
      "loss": 0.2325,
      "step": 3673
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5968652869118958,
      "learning_rate": 5.722019194148419e-06,
      "loss": 0.233,
      "step": 3674
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.459639463314285,
      "learning_rate": 5.719967749597644e-06,
      "loss": 0.191,
      "step": 3675
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.449853384896813,
      "learning_rate": 5.717916181276065e-06,
      "loss": 0.2197,
      "step": 3676
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3698750993051938,
      "learning_rate": 5.715864489536373e-06,
      "loss": 0.2057,
      "step": 3677
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.375612386940072,
      "learning_rate": 5.713812674731276e-06,
      "loss": 0.2198,
      "step": 3678
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.382439834736986,
      "learning_rate": 5.711760737213504e-06,
      "loss": 0.2078,
      "step": 3679
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.328734438317488,
      "learning_rate": 5.70970867733581e-06,
      "loss": 0.1723,
      "step": 3680
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5119988607606185,
      "learning_rate": 5.707656495450969e-06,
      "loss": 0.2205,
      "step": 3681
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4448083087328547,
      "learning_rate": 5.70560419191177e-06,
      "loss": 0.1921,
      "step": 3682
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5734990821433372,
      "learning_rate": 5.7035517670710304e-06,
      "loss": 0.2149,
      "step": 3683
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.277827756057441,
      "learning_rate": 5.701499221281587e-06,
      "loss": 0.1668,
      "step": 3684
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3139652620610203,
      "learning_rate": 5.699446554896294e-06,
      "loss": 0.1931,
      "step": 3685
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2017373737104302,
      "learning_rate": 5.697393768268031e-06,
      "loss": 0.1332,
      "step": 3686
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.176420141303803,
      "learning_rate": 5.695340861749693e-06,
      "loss": 0.5699,
      "step": 3687
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3577436472568474,
      "learning_rate": 5.6932878356942005e-06,
      "loss": 0.2241,
      "step": 3688
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.104646380413394,
      "learning_rate": 5.691234690454491e-06,
      "loss": 0.6832,
      "step": 3689
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4793057434242318,
      "learning_rate": 5.6891814263835245e-06,
      "loss": 0.1666,
      "step": 3690
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4437605147185786,
      "learning_rate": 5.687128043834282e-06,
      "loss": 0.2026,
      "step": 3691
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2963746651206967,
      "learning_rate": 5.685074543159763e-06,
      "loss": 0.2221,
      "step": 3692
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5267588407555452,
      "learning_rate": 5.683020924712987e-06,
      "loss": 0.2245,
      "step": 3693
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2263255250979461,
      "learning_rate": 5.680967188846996e-06,
      "loss": 0.1669,
      "step": 3694
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3031702774096554,
      "learning_rate": 5.678913335914852e-06,
      "loss": 0.1618,
      "step": 3695
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4346971321485003,
      "learning_rate": 5.676859366269635e-06,
      "loss": 0.2042,
      "step": 3696
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4879025426969423,
      "learning_rate": 5.6748052802644445e-06,
      "loss": 0.1992,
      "step": 3697
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3697669699808248,
      "learning_rate": 5.672751078252403e-06,
      "loss": 0.2245,
      "step": 3698
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4503554466739972,
      "learning_rate": 5.670696760586654e-06,
      "loss": 0.1851,
      "step": 3699
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.068697338288556,
      "learning_rate": 5.668642327620356e-06,
      "loss": 0.4273,
      "step": 3700
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1706340858879842,
      "learning_rate": 5.66658777970669e-06,
      "loss": 0.1356,
      "step": 3701
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.549686041340553,
      "learning_rate": 5.664533117198856e-06,
      "loss": 0.2543,
      "step": 3702
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4063669123433549,
      "learning_rate": 5.662478340450076e-06,
      "loss": 0.24,
      "step": 3703
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4403593024182682,
      "learning_rate": 5.660423449813588e-06,
      "loss": 0.2466,
      "step": 3704
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2602427665384488,
      "learning_rate": 5.658368445642655e-06,
      "loss": 0.2095,
      "step": 3705
    },
    {
      "epoch": 0.47,
      "grad_norm": 8.330747587785247,
      "learning_rate": 5.656313328290552e-06,
      "loss": 0.5893,
      "step": 3706
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5122839319719652,
      "learning_rate": 5.65425809811058e-06,
      "loss": 0.2046,
      "step": 3707
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6213831993780499,
      "learning_rate": 5.652202755456055e-06,
      "loss": 0.229,
      "step": 3708
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.114097287170645,
      "learning_rate": 5.650147300680318e-06,
      "loss": 0.5683,
      "step": 3709
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4413858643369972,
      "learning_rate": 5.6480917341367205e-06,
      "loss": 0.2705,
      "step": 3710
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.531922403974835,
      "learning_rate": 5.646036056178641e-06,
      "loss": 0.2209,
      "step": 3711
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3578215700138017,
      "learning_rate": 5.643980267159474e-06,
      "loss": 0.1851,
      "step": 3712
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.328250052248455,
      "learning_rate": 5.641924367432634e-06,
      "loss": 0.5434,
      "step": 3713
    },
    {
      "epoch": 0.48,
      "grad_norm": 13.793486827123887,
      "learning_rate": 5.639868357351554e-06,
      "loss": 0.7025,
      "step": 3714
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2588307198445594,
      "learning_rate": 5.637812237269683e-06,
      "loss": 0.1569,
      "step": 3715
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.602957071567971,
      "learning_rate": 5.635756007540495e-06,
      "loss": 0.5632,
      "step": 3716
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.395099899212822,
      "learning_rate": 5.6336996685174795e-06,
      "loss": 0.2285,
      "step": 3717
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.615775647780837,
      "learning_rate": 5.631643220554146e-06,
      "loss": 0.5458,
      "step": 3718
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.717624414413211,
      "learning_rate": 5.629586664004018e-06,
      "loss": 0.2406,
      "step": 3719
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4824375042064677,
      "learning_rate": 5.6275299992206444e-06,
      "loss": 0.2222,
      "step": 3720
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5412424437741687,
      "learning_rate": 5.625473226557588e-06,
      "loss": 0.1919,
      "step": 3721
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.604946779652818,
      "learning_rate": 5.623416346368434e-06,
      "loss": 0.6296,
      "step": 3722
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4766252009949095,
      "learning_rate": 5.621359359006784e-06,
      "loss": 0.214,
      "step": 3723
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3960554357276813,
      "learning_rate": 5.6193022648262555e-06,
      "loss": 0.2568,
      "step": 3724
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.196288869244573,
      "learning_rate": 5.617245064180488e-06,
      "loss": 0.1767,
      "step": 3725
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3612247220447031,
      "learning_rate": 5.615187757423137e-06,
      "loss": 0.1845,
      "step": 3726
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.419850155255926,
      "learning_rate": 5.613130344907881e-06,
      "loss": 0.2178,
      "step": 3727
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.755613381623035,
      "learning_rate": 5.611072826988409e-06,
      "loss": 0.6085,
      "step": 3728
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.389888944303294,
      "learning_rate": 5.609015204018435e-06,
      "loss": 0.2089,
      "step": 3729
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4352476869436916,
      "learning_rate": 5.6069574763516866e-06,
      "loss": 0.1952,
      "step": 3730
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3256550815867942,
      "learning_rate": 5.604899644341911e-06,
      "loss": 0.1816,
      "step": 3731
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.616871986093345,
      "learning_rate": 5.602841708342876e-06,
      "loss": 0.2423,
      "step": 3732
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.254187021156376,
      "learning_rate": 5.60078366870836e-06,
      "loss": 0.564,
      "step": 3733
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.368204718405287,
      "learning_rate": 5.598725525792168e-06,
      "loss": 0.1714,
      "step": 3734
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.808372298335655,
      "learning_rate": 5.596667279948117e-06,
      "loss": 0.5579,
      "step": 3735
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4330809132846942,
      "learning_rate": 5.594608931530044e-06,
      "loss": 0.1964,
      "step": 3736
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4059784892214178,
      "learning_rate": 5.592550480891801e-06,
      "loss": 0.1884,
      "step": 3737
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.24935818463872,
      "learning_rate": 5.590491928387261e-06,
      "loss": 0.1782,
      "step": 3738
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3568328391584221,
      "learning_rate": 5.588433274370314e-06,
      "loss": 0.2333,
      "step": 3739
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.298161189071067,
      "learning_rate": 5.586374519194863e-06,
      "loss": 0.1981,
      "step": 3740
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3514759840473436,
      "learning_rate": 5.584315663214836e-06,
      "loss": 0.1894,
      "step": 3741
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2144368701460189,
      "learning_rate": 5.582256706784174e-06,
      "loss": 0.175,
      "step": 3742
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3843203147633243,
      "learning_rate": 5.580197650256832e-06,
      "loss": 0.1812,
      "step": 3743
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.464836875674126,
      "learning_rate": 5.578138493986788e-06,
      "loss": 0.2002,
      "step": 3744
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2795572310185912,
      "learning_rate": 5.576079238328035e-06,
      "loss": 0.2028,
      "step": 3745
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4540183359842642,
      "learning_rate": 5.574019883634582e-06,
      "loss": 0.2136,
      "step": 3746
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.482647189038951,
      "learning_rate": 5.571960430260457e-06,
      "loss": 0.6173,
      "step": 3747
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.231487334868479,
      "learning_rate": 5.569900878559704e-06,
      "loss": 0.1794,
      "step": 3748
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3073146459644356,
      "learning_rate": 5.567841228886381e-06,
      "loss": 0.2446,
      "step": 3749
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2497064174027375,
      "learning_rate": 5.56578148159457e-06,
      "loss": 0.175,
      "step": 3750
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6152931321669164,
      "learning_rate": 5.5637216370383615e-06,
      "loss": 0.1714,
      "step": 3751
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3867561263178252,
      "learning_rate": 5.561661695571869e-06,
      "loss": 0.22,
      "step": 3752
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4935694766588663,
      "learning_rate": 5.559601657549219e-06,
      "loss": 0.1957,
      "step": 3753
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3364636045950706,
      "learning_rate": 5.557541523324558e-06,
      "loss": 0.1735,
      "step": 3754
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.825581834104666,
      "learning_rate": 5.555481293252044e-06,
      "loss": 0.7141,
      "step": 3755
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4269603349445386,
      "learning_rate": 5.553420967685857e-06,
      "loss": 0.1712,
      "step": 3756
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5392866583704856,
      "learning_rate": 5.551360546980191e-06,
      "loss": 0.228,
      "step": 3757
    },
    {
      "epoch": 0.48,
      "grad_norm": 9.207695643502774,
      "learning_rate": 5.549300031489252e-06,
      "loss": 0.6085,
      "step": 3758
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.4516907667979,
      "learning_rate": 5.547239421567272e-06,
      "loss": 0.4745,
      "step": 3759
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3090457824531685,
      "learning_rate": 5.545178717568491e-06,
      "loss": 0.1426,
      "step": 3760
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.393201233701573,
      "learning_rate": 5.5431179198471674e-06,
      "loss": 0.1509,
      "step": 3761
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5185183739530326,
      "learning_rate": 5.5410570287575775e-06,
      "loss": 0.1555,
      "step": 3762
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5162100974658617,
      "learning_rate": 5.538996044654011e-06,
      "loss": 0.1893,
      "step": 3763
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.529994541890097,
      "learning_rate": 5.536934967890778e-06,
      "loss": 0.5487,
      "step": 3764
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.404678384429382,
      "learning_rate": 5.5348737988221964e-06,
      "loss": 0.176,
      "step": 3765
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5036836944997354,
      "learning_rate": 5.532812537802611e-06,
      "loss": 0.1983,
      "step": 3766
    },
    {
      "epoch": 0.48,
      "grad_norm": 38.50423226542573,
      "learning_rate": 5.530751185186372e-06,
      "loss": 0.7303,
      "step": 3767
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3170527580468785,
      "learning_rate": 5.528689741327851e-06,
      "loss": 0.1767,
      "step": 3768
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5168440312250504,
      "learning_rate": 5.526628206581434e-06,
      "loss": 0.2036,
      "step": 3769
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4782408430891465,
      "learning_rate": 5.5245665813015245e-06,
      "loss": 0.1944,
      "step": 3770
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3855662085613678,
      "learning_rate": 5.522504865842537e-06,
      "loss": 0.1935,
      "step": 3771
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4733802223887245,
      "learning_rate": 5.520443060558907e-06,
      "loss": 0.2104,
      "step": 3772
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.479257941126007,
      "learning_rate": 5.51838116580508e-06,
      "loss": 0.2221,
      "step": 3773
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4033047708784672,
      "learning_rate": 5.5163191819355225e-06,
      "loss": 0.1928,
      "step": 3774
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4363582216773414,
      "learning_rate": 5.514257109304712e-06,
      "loss": 0.2248,
      "step": 3775
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2448500364728252,
      "learning_rate": 5.51219494826714e-06,
      "loss": 0.1808,
      "step": 3776
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4453521005159635,
      "learning_rate": 5.51013269917732e-06,
      "loss": 0.1688,
      "step": 3777
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4456097164080675,
      "learning_rate": 5.5080703623897754e-06,
      "loss": 0.2035,
      "step": 3778
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5144969786669389,
      "learning_rate": 5.506007938259045e-06,
      "loss": 0.1987,
      "step": 3779
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4290343193817656,
      "learning_rate": 5.5039454271396856e-06,
      "loss": 0.2511,
      "step": 3780
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.408155846427746,
      "learning_rate": 5.501882829386262e-06,
      "loss": 0.2476,
      "step": 3781
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4067530150369238,
      "learning_rate": 5.499820145353364e-06,
      "loss": 0.162,
      "step": 3782
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5096694536916304,
      "learning_rate": 5.497757375395588e-06,
      "loss": 0.2339,
      "step": 3783
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.666540700495453,
      "learning_rate": 5.49569451986755e-06,
      "loss": 0.5834,
      "step": 3784
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4215646744654804,
      "learning_rate": 5.493631579123876e-06,
      "loss": 0.2303,
      "step": 3785
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4888108736692829,
      "learning_rate": 5.49156855351921e-06,
      "loss": 0.1972,
      "step": 3786
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3135625677107903,
      "learning_rate": 5.4895054434082115e-06,
      "loss": 0.2127,
      "step": 3787
    },
    {
      "epoch": 0.48,
      "grad_norm": 9.061575898115105,
      "learning_rate": 5.487442249145554e-06,
      "loss": 0.5133,
      "step": 3788
    },
    {
      "epoch": 0.48,
      "grad_norm": 9.165577760601693,
      "learning_rate": 5.485378971085921e-06,
      "loss": 0.5926,
      "step": 3789
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3249552935804176,
      "learning_rate": 5.4833156095840155e-06,
      "loss": 0.1477,
      "step": 3790
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3810908224094416,
      "learning_rate": 5.481252164994555e-06,
      "loss": 0.264,
      "step": 3791
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.5766347650952643,
      "learning_rate": 5.479188637672269e-06,
      "loss": 0.3821,
      "step": 3792
    },
    {
      "epoch": 0.49,
      "grad_norm": 10.519837989205685,
      "learning_rate": 5.4771250279719e-06,
      "loss": 0.6662,
      "step": 3793
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3185969919433334,
      "learning_rate": 5.475061336248208e-06,
      "loss": 0.177,
      "step": 3794
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6609926983563217,
      "learning_rate": 5.4729975628559626e-06,
      "loss": 0.2198,
      "step": 3795
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2898374811272681,
      "learning_rate": 5.470933708149955e-06,
      "loss": 0.1914,
      "step": 3796
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3956545422617226,
      "learning_rate": 5.468869772484982e-06,
      "loss": 0.2185,
      "step": 3797
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.586926766863504,
      "learning_rate": 5.4668057562158604e-06,
      "loss": 0.2697,
      "step": 3798
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6380600028215642,
      "learning_rate": 5.464741659697419e-06,
      "loss": 0.2254,
      "step": 3799
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.630997197614086,
      "learning_rate": 5.462677483284496e-06,
      "loss": 0.6406,
      "step": 3800
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.1940357779909965,
      "learning_rate": 5.46061322733195e-06,
      "loss": 0.7088,
      "step": 3801
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3539786389017454,
      "learning_rate": 5.458548892194652e-06,
      "loss": 0.197,
      "step": 3802
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.525361631439681,
      "learning_rate": 5.4564844782274826e-06,
      "loss": 0.566,
      "step": 3803
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4551972062371945,
      "learning_rate": 5.45441998578534e-06,
      "loss": 0.1917,
      "step": 3804
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3553936135034277,
      "learning_rate": 5.452355415223132e-06,
      "loss": 0.1713,
      "step": 3805
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4625452722753087,
      "learning_rate": 5.450290766895786e-06,
      "loss": 0.1935,
      "step": 3806
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5416299170389143,
      "learning_rate": 5.4482260411582365e-06,
      "loss": 0.1953,
      "step": 3807
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.828506092283911,
      "learning_rate": 5.446161238365433e-06,
      "loss": 0.6582,
      "step": 3808
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5989372729044748,
      "learning_rate": 5.44409635887234e-06,
      "loss": 0.2269,
      "step": 3809
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3821713515151401,
      "learning_rate": 5.4420314030339375e-06,
      "loss": 0.203,
      "step": 3810
    },
    {
      "epoch": 0.49,
      "grad_norm": 9.084924303118886,
      "learning_rate": 5.439966371205209e-06,
      "loss": 0.6311,
      "step": 3811
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6096331391769232,
      "learning_rate": 5.437901263741163e-06,
      "loss": 0.17,
      "step": 3812
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3235783455392927,
      "learning_rate": 5.435836080996812e-06,
      "loss": 0.1791,
      "step": 3813
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1988906567127278,
      "learning_rate": 5.433770823327187e-06,
      "loss": 0.179,
      "step": 3814
    },
    {
      "epoch": 0.49,
      "grad_norm": 9.896677972712396,
      "learning_rate": 5.431705491087327e-06,
      "loss": 0.5283,
      "step": 3815
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.364217699830095,
      "learning_rate": 5.42964008463229e-06,
      "loss": 0.2083,
      "step": 3816
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3834495661258823,
      "learning_rate": 5.427574604317142e-06,
      "loss": 0.2099,
      "step": 3817
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3629573991707886,
      "learning_rate": 5.4255090504969606e-06,
      "loss": 0.1502,
      "step": 3818
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3403032398396229,
      "learning_rate": 5.423443423526842e-06,
      "loss": 0.2161,
      "step": 3819
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4008079040262857,
      "learning_rate": 5.42137772376189e-06,
      "loss": 0.2197,
      "step": 3820
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4518734341989874,
      "learning_rate": 5.419311951557222e-06,
      "loss": 0.2032,
      "step": 3821
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4166629397033605,
      "learning_rate": 5.417246107267968e-06,
      "loss": 0.1975,
      "step": 3822
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5650290586488118,
      "learning_rate": 5.415180191249271e-06,
      "loss": 0.2487,
      "step": 3823
    },
    {
      "epoch": 0.49,
      "grad_norm": 10.210630774205674,
      "learning_rate": 5.413114203856287e-06,
      "loss": 0.7572,
      "step": 3824
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.316023284307556,
      "learning_rate": 5.411048145444182e-06,
      "loss": 0.2046,
      "step": 3825
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6073464670834465,
      "learning_rate": 5.408982016368134e-06,
      "loss": 0.1808,
      "step": 3826
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4117909391535497,
      "learning_rate": 5.406915816983337e-06,
      "loss": 0.216,
      "step": 3827
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5353676557305438,
      "learning_rate": 5.404849547644993e-06,
      "loss": 0.1999,
      "step": 3828
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.205970206324139,
      "learning_rate": 5.40278320870832e-06,
      "loss": 0.6573,
      "step": 3829
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5133900545490535,
      "learning_rate": 5.400716800528542e-06,
      "loss": 0.1902,
      "step": 3830
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.172386336040533,
      "learning_rate": 5.398650323460904e-06,
      "loss": 0.1514,
      "step": 3831
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.126007565145826,
      "learning_rate": 5.39658377786065e-06,
      "loss": 0.6331,
      "step": 3832
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5396764811706674,
      "learning_rate": 5.394517164083047e-06,
      "loss": 0.2129,
      "step": 3833
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4710113983571693,
      "learning_rate": 5.392450482483372e-06,
      "loss": 0.2291,
      "step": 3834
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4531032722345558,
      "learning_rate": 5.39038373341691e-06,
      "loss": 0.214,
      "step": 3835
    },
    {
      "epoch": 0.49,
      "grad_norm": 16.302624993411115,
      "learning_rate": 5.388316917238956e-06,
      "loss": 0.3713,
      "step": 3836
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3700505831648992,
      "learning_rate": 5.386250034304823e-06,
      "loss": 0.2202,
      "step": 3837
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3717565009892794,
      "learning_rate": 5.384183084969832e-06,
      "loss": 0.1459,
      "step": 3838
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.559654661615966,
      "learning_rate": 5.382116069589315e-06,
      "loss": 0.4857,
      "step": 3839
    },
    {
      "epoch": 0.49,
      "grad_norm": 10.50303431753912,
      "learning_rate": 5.380048988518613e-06,
      "loss": 0.4893,
      "step": 3840
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.524046396937437,
      "learning_rate": 5.3779818421130865e-06,
      "loss": 0.5892,
      "step": 3841
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4219198588515132,
      "learning_rate": 5.3759146307281e-06,
      "loss": 0.1938,
      "step": 3842
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3509165130982126,
      "learning_rate": 5.37384735471903e-06,
      "loss": 0.1673,
      "step": 3843
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.438143586057221,
      "learning_rate": 5.371780014441265e-06,
      "loss": 0.2277,
      "step": 3844
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.952302020291957,
      "learning_rate": 5.369712610250206e-06,
      "loss": 0.6123,
      "step": 3845
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.159844092404793,
      "learning_rate": 5.367645142501262e-06,
      "loss": 0.5764,
      "step": 3846
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5706923269184394,
      "learning_rate": 5.365577611549856e-06,
      "loss": 0.1759,
      "step": 3847
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2666263952342802,
      "learning_rate": 5.363510017751422e-06,
      "loss": 0.1975,
      "step": 3848
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.308969871328519,
      "learning_rate": 5.3614423614614016e-06,
      "loss": 0.1755,
      "step": 3849
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5628932578636394,
      "learning_rate": 5.359374643035248e-06,
      "loss": 0.2276,
      "step": 3850
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5035592077143956,
      "learning_rate": 5.357306862828427e-06,
      "loss": 0.1795,
      "step": 3851
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3259092803629833,
      "learning_rate": 5.355239021196416e-06,
      "loss": 0.1925,
      "step": 3852
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2100553473918074,
      "learning_rate": 5.353171118494698e-06,
      "loss": 0.1804,
      "step": 3853
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1562133621546646,
      "learning_rate": 5.351103155078768e-06,
      "loss": 0.1443,
      "step": 3854
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.471281829388111,
      "learning_rate": 5.349035131304138e-06,
      "loss": 0.1987,
      "step": 3855
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.269498591878193,
      "learning_rate": 5.346967047526321e-06,
      "loss": 0.1873,
      "step": 3856
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.57211709441967,
      "learning_rate": 5.344898904100848e-06,
      "loss": 0.1824,
      "step": 3857
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4845325291063018,
      "learning_rate": 5.342830701383254e-06,
      "loss": 0.1914,
      "step": 3858
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3878267745816038,
      "learning_rate": 5.340762439729088e-06,
      "loss": 0.2073,
      "step": 3859
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4336529038231234,
      "learning_rate": 5.338694119493908e-06,
      "loss": 0.2138,
      "step": 3860
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.388841167281707,
      "learning_rate": 5.336625741033283e-06,
      "loss": 0.1843,
      "step": 3861
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.276192117493417,
      "learning_rate": 5.33455730470279e-06,
      "loss": 0.1672,
      "step": 3862
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3112372597588724,
      "learning_rate": 5.332488810858017e-06,
      "loss": 0.2069,
      "step": 3863
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.5594729665084355,
      "learning_rate": 5.330420259854564e-06,
      "loss": 0.4573,
      "step": 3864
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4329153300492214,
      "learning_rate": 5.328351652048037e-06,
      "loss": 0.1883,
      "step": 3865
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5263942192510074,
      "learning_rate": 5.326282987794056e-06,
      "loss": 0.2061,
      "step": 3866
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.366908954335002,
      "learning_rate": 5.3242142674482456e-06,
      "loss": 0.189,
      "step": 3867
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4504737696201906,
      "learning_rate": 5.322145491366244e-06,
      "loss": 0.2013,
      "step": 3868
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4917065105514131,
      "learning_rate": 5.320076659903698e-06,
      "loss": 0.2441,
      "step": 3869
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4730268530746917,
      "learning_rate": 5.318007773416264e-06,
      "loss": 0.1986,
      "step": 3870
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2592444331636112,
      "learning_rate": 5.315938832259606e-06,
      "loss": 0.166,
      "step": 3871
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3817605154880213,
      "learning_rate": 5.3138698367894e-06,
      "loss": 0.2164,
      "step": 3872
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1817409232224871,
      "learning_rate": 5.311800787361332e-06,
      "loss": 0.1949,
      "step": 3873
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5156271142988804,
      "learning_rate": 5.309731684331093e-06,
      "loss": 0.2215,
      "step": 3874
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6679524857351604,
      "learning_rate": 5.307662528054387e-06,
      "loss": 0.2227,
      "step": 3875
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.325576980599023,
      "learning_rate": 5.305593318886925e-06,
      "loss": 0.1704,
      "step": 3876
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2125868022830673,
      "learning_rate": 5.303524057184429e-06,
      "loss": 0.1746,
      "step": 3877
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5056765376045371,
      "learning_rate": 5.301454743302629e-06,
      "loss": 0.2487,
      "step": 3878
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.110883441260615,
      "learning_rate": 5.299385377597265e-06,
      "loss": 0.697,
      "step": 3879
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.563202591543807,
      "learning_rate": 5.297315960424081e-06,
      "loss": 0.1891,
      "step": 3880
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4624238665612268,
      "learning_rate": 5.295246492138839e-06,
      "loss": 0.204,
      "step": 3881
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.404347118045868,
      "learning_rate": 5.293176973097303e-06,
      "loss": 0.1975,
      "step": 3882
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4758252495706583,
      "learning_rate": 5.2911074036552426e-06,
      "loss": 0.2378,
      "step": 3883
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3887928781189474,
      "learning_rate": 5.289037784168448e-06,
      "loss": 0.1938,
      "step": 3884
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5140326967582025,
      "learning_rate": 5.286968114992706e-06,
      "loss": 0.2039,
      "step": 3885
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.717098548452211,
      "learning_rate": 5.28489839648382e-06,
      "loss": 0.6803,
      "step": 3886
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.862219055802294,
      "learning_rate": 5.282828628997595e-06,
      "loss": 0.7245,
      "step": 3887
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.427085684976526,
      "learning_rate": 5.28075881288985e-06,
      "loss": 0.4468,
      "step": 3888
    },
    {
      "epoch": 0.5,
      "grad_norm": 8.263342329714982,
      "learning_rate": 5.2786889485164115e-06,
      "loss": 0.49,
      "step": 3889
    },
    {
      "epoch": 0.5,
      "grad_norm": 8.223156038780736,
      "learning_rate": 5.276619036233111e-06,
      "loss": 0.6256,
      "step": 3890
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5171188285841932,
      "learning_rate": 5.274549076395792e-06,
      "loss": 0.2269,
      "step": 3891
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4953156031733161,
      "learning_rate": 5.2724790693603025e-06,
      "loss": 0.228,
      "step": 3892
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.169781864926149,
      "learning_rate": 5.270409015482504e-06,
      "loss": 0.5914,
      "step": 3893
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5201581364017815,
      "learning_rate": 5.268338915118258e-06,
      "loss": 0.1763,
      "step": 3894
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.382979606694824,
      "learning_rate": 5.266268768623445e-06,
      "loss": 0.2267,
      "step": 3895
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3884893265824743,
      "learning_rate": 5.264198576353942e-06,
      "loss": 0.2443,
      "step": 3896
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4827893265976089,
      "learning_rate": 5.262128338665641e-06,
      "loss": 0.2189,
      "step": 3897
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4532636278905458,
      "learning_rate": 5.26005805591444e-06,
      "loss": 0.1827,
      "step": 3898
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.440166171311405,
      "learning_rate": 5.257987728456244e-06,
      "loss": 0.669,
      "step": 3899
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4248347148522673,
      "learning_rate": 5.255917356646968e-06,
      "loss": 0.2183,
      "step": 3900
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.293777851909419,
      "learning_rate": 5.2538469408425284e-06,
      "loss": 0.2105,
      "step": 3901
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3119738162734045,
      "learning_rate": 5.25177648139886e-06,
      "loss": 0.2008,
      "step": 3902
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3862826417576184,
      "learning_rate": 5.2497059786718955e-06,
      "loss": 0.1904,
      "step": 3903
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2941142611462824,
      "learning_rate": 5.247635433017576e-06,
      "loss": 0.2197,
      "step": 3904
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4301924537010862,
      "learning_rate": 5.245564844791857e-06,
      "loss": 0.2344,
      "step": 3905
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2957717844012944,
      "learning_rate": 5.243494214350693e-06,
      "loss": 0.1835,
      "step": 3906
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3262997155180032,
      "learning_rate": 5.2414235420500516e-06,
      "loss": 0.208,
      "step": 3907
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.289155796861749,
      "learning_rate": 5.2393528282459036e-06,
      "loss": 0.1983,
      "step": 3908
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3831366826930915,
      "learning_rate": 5.237282073294231e-06,
      "loss": 0.2253,
      "step": 3909
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4003682373808304,
      "learning_rate": 5.235211277551019e-06,
      "loss": 0.1817,
      "step": 3910
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.115938523158051,
      "learning_rate": 5.2331404413722595e-06,
      "loss": 0.4748,
      "step": 3911
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4306312275731445,
      "learning_rate": 5.231069565113957e-06,
      "loss": 0.2017,
      "step": 3912
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4328669770463636,
      "learning_rate": 5.2289986491321176e-06,
      "loss": 0.1953,
      "step": 3913
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3969676235204427,
      "learning_rate": 5.226927693782755e-06,
      "loss": 0.2256,
      "step": 3914
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2773270113776856,
      "learning_rate": 5.22485669942189e-06,
      "loss": 0.1869,
      "step": 3915
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.530787524865473,
      "learning_rate": 5.222785666405553e-06,
      "loss": 0.1879,
      "step": 3916
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4813248699708264,
      "learning_rate": 5.220714595089777e-06,
      "loss": 0.1917,
      "step": 3917
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.2786053398183626,
      "learning_rate": 5.218643485830604e-06,
      "loss": 0.4073,
      "step": 3918
    },
    {
      "epoch": 0.5,
      "grad_norm": 28.494755517218664,
      "learning_rate": 5.216572338984081e-06,
      "loss": 0.4959,
      "step": 3919
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5189680291615741,
      "learning_rate": 5.21450115490626e-06,
      "loss": 0.2366,
      "step": 3920
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.502386320324258,
      "learning_rate": 5.212429933953207e-06,
      "loss": 0.2228,
      "step": 3921
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3803474355322802,
      "learning_rate": 5.210358676480983e-06,
      "loss": 0.2238,
      "step": 3922
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.407384983388169,
      "learning_rate": 5.208287382845666e-06,
      "loss": 0.4533,
      "step": 3923
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.299236344950327,
      "learning_rate": 5.206216053403333e-06,
      "loss": 0.1736,
      "step": 3924
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.245332507734425,
      "learning_rate": 5.204144688510069e-06,
      "loss": 0.1743,
      "step": 3925
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.337864993982283,
      "learning_rate": 5.202073288521965e-06,
      "loss": 0.1858,
      "step": 3926
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4179795836228752,
      "learning_rate": 5.2000018537951226e-06,
      "loss": 0.1858,
      "step": 3927
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4022820496741204,
      "learning_rate": 5.197930384685642e-06,
      "loss": 0.1888,
      "step": 3928
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5864321706615385,
      "learning_rate": 5.195858881549633e-06,
      "loss": 0.2474,
      "step": 3929
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3418012179635737,
      "learning_rate": 5.19378734474321e-06,
      "loss": 0.1832,
      "step": 3930
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3522911071186543,
      "learning_rate": 5.1917157746225e-06,
      "loss": 0.208,
      "step": 3931
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.741331751656633,
      "learning_rate": 5.189644171543624e-06,
      "loss": 0.5223,
      "step": 3932
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4449067538941298,
      "learning_rate": 5.187572535862715e-06,
      "loss": 0.1923,
      "step": 3933
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.415392110277398,
      "learning_rate": 5.185500867935913e-06,
      "loss": 0.1851,
      "step": 3934
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4665551088667488,
      "learning_rate": 5.183429168119362e-06,
      "loss": 0.1948,
      "step": 3935
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3806629048538874,
      "learning_rate": 5.181357436769211e-06,
      "loss": 0.2191,
      "step": 3936
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.480179180725599,
      "learning_rate": 5.1792856742416145e-06,
      "loss": 0.5649,
      "step": 3937
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.079324473006053,
      "learning_rate": 5.17721388089273e-06,
      "loss": 0.6986,
      "step": 3938
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3599929946183917,
      "learning_rate": 5.175142057078727e-06,
      "loss": 0.2222,
      "step": 3939
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.80300219033621,
      "learning_rate": 5.173070203155772e-06,
      "loss": 0.6085,
      "step": 3940
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5739131674650804,
      "learning_rate": 5.170998319480044e-06,
      "loss": 0.205,
      "step": 3941
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.395253384690788,
      "learning_rate": 5.168926406407723e-06,
      "loss": 0.1895,
      "step": 3942
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3986652328296219,
      "learning_rate": 5.166854464294993e-06,
      "loss": 0.2229,
      "step": 3943
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3821033518804073,
      "learning_rate": 5.164782493498046e-06,
      "loss": 0.1844,
      "step": 3944
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5800103243491759,
      "learning_rate": 5.16271049437308e-06,
      "loss": 0.192,
      "step": 3945
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2924012439342416,
      "learning_rate": 5.160638467276293e-06,
      "loss": 0.2027,
      "step": 3946
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.518718156592295,
      "learning_rate": 5.15856641256389e-06,
      "loss": 0.2174,
      "step": 3947
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.811633898467364,
      "learning_rate": 5.156494330592084e-06,
      "loss": 0.6161,
      "step": 3948
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4915428749456658,
      "learning_rate": 5.154422221717087e-06,
      "loss": 0.2143,
      "step": 3949
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4184433857917744,
      "learning_rate": 5.152350086295121e-06,
      "loss": 0.1557,
      "step": 3950
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1960790785244584,
      "learning_rate": 5.150277924682408e-06,
      "loss": 0.1467,
      "step": 3951
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.3173534658227934,
      "learning_rate": 5.148205737235178e-06,
      "loss": 0.6866,
      "step": 3952
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4331463985524637,
      "learning_rate": 5.146133524309664e-06,
      "loss": 0.1984,
      "step": 3953
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3974532248173845,
      "learning_rate": 5.144061286262102e-06,
      "loss": 0.2226,
      "step": 3954
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3761582737834521,
      "learning_rate": 5.141989023448736e-06,
      "loss": 0.2071,
      "step": 3955
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3824929413070135,
      "learning_rate": 5.139916736225811e-06,
      "loss": 0.2155,
      "step": 3956
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3679375527639992,
      "learning_rate": 5.137844424949576e-06,
      "loss": 0.2001,
      "step": 3957
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3121926164154305,
      "learning_rate": 5.135772089976285e-06,
      "loss": 0.1838,
      "step": 3958
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4823574725898059,
      "learning_rate": 5.133699731662201e-06,
      "loss": 0.1978,
      "step": 3959
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.318111312110579,
      "learning_rate": 5.131627350363582e-06,
      "loss": 0.2162,
      "step": 3960
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.365723197484442,
      "learning_rate": 5.1295549464366944e-06,
      "loss": 0.2314,
      "step": 3961
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.1528389223665245,
      "learning_rate": 5.1274825202378085e-06,
      "loss": 0.7194,
      "step": 3962
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.620681751165282,
      "learning_rate": 5.125410072123203e-06,
      "loss": 0.38,
      "step": 3963
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4194207326419723,
      "learning_rate": 5.12333760244915e-06,
      "loss": 0.2294,
      "step": 3964
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.781972103262777,
      "learning_rate": 5.121265111571933e-06,
      "loss": 0.348,
      "step": 3965
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.619646180916202,
      "learning_rate": 5.119192599847838e-06,
      "loss": 0.3773,
      "step": 3966
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3156489608112876,
      "learning_rate": 5.1171200676331535e-06,
      "loss": 0.1829,
      "step": 3967
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4771656636837083,
      "learning_rate": 5.115047515284173e-06,
      "loss": 0.1973,
      "step": 3968
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2900104038565212,
      "learning_rate": 5.112974943157188e-06,
      "loss": 0.1986,
      "step": 3969
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.414744246384242,
      "learning_rate": 5.110902351608504e-06,
      "loss": 0.2006,
      "step": 3970
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5891795296449442,
      "learning_rate": 5.108829740994417e-06,
      "loss": 0.2548,
      "step": 3971
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2350221505988774,
      "learning_rate": 5.106757111671235e-06,
      "loss": 0.1913,
      "step": 3972
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6225830368794414,
      "learning_rate": 5.104684463995271e-06,
      "loss": 0.192,
      "step": 3973
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.1946463976239485,
      "learning_rate": 5.102611798322833e-06,
      "loss": 0.4858,
      "step": 3974
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.462886762023099,
      "learning_rate": 5.100539115010237e-06,
      "loss": 0.241,
      "step": 3975
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.468443402160405,
      "learning_rate": 5.098466414413801e-06,
      "loss": 0.2214,
      "step": 3976
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6067376544715262,
      "learning_rate": 5.096393696889848e-06,
      "loss": 0.2224,
      "step": 3977
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6059567482439487,
      "learning_rate": 5.094320962794701e-06,
      "loss": 0.2341,
      "step": 3978
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.79645563827723,
      "learning_rate": 5.092248212484686e-06,
      "loss": 0.6349,
      "step": 3979
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2234411121855266,
      "learning_rate": 5.090175446316133e-06,
      "loss": 0.2111,
      "step": 3980
    },
    {
      "epoch": 0.51,
      "grad_norm": 8.263409045003543,
      "learning_rate": 5.088102664645378e-06,
      "loss": 0.75,
      "step": 3981
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3020855000218243,
      "learning_rate": 5.086029867828752e-06,
      "loss": 0.1623,
      "step": 3982
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.623266702341692,
      "learning_rate": 5.083957056222595e-06,
      "loss": 0.1862,
      "step": 3983
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.422890154990019,
      "learning_rate": 5.081884230183247e-06,
      "loss": 0.2295,
      "step": 3984
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.426202094124763,
      "learning_rate": 5.079811390067049e-06,
      "loss": 0.2079,
      "step": 3985
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.242117651026643,
      "learning_rate": 5.07773853623035e-06,
      "loss": 0.5682,
      "step": 3986
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3481811566087931,
      "learning_rate": 5.075665669029494e-06,
      "loss": 0.1977,
      "step": 3987
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3240626772736908,
      "learning_rate": 5.073592788820832e-06,
      "loss": 0.2012,
      "step": 3988
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4025364319683369,
      "learning_rate": 5.071519895960717e-06,
      "loss": 0.2157,
      "step": 3989
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1732523223500346,
      "learning_rate": 5.069446990805501e-06,
      "loss": 0.1581,
      "step": 3990
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.998170861516725,
      "learning_rate": 5.067374073711545e-06,
      "loss": 0.6094,
      "step": 3991
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.693839195074981,
      "learning_rate": 5.065301145035204e-06,
      "loss": 0.4984,
      "step": 3992
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.494684950486891,
      "learning_rate": 5.063228205132838e-06,
      "loss": 0.2337,
      "step": 3993
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.457727350828059,
      "learning_rate": 5.061155254360811e-06,
      "loss": 0.2032,
      "step": 3994
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5567731327164722,
      "learning_rate": 5.059082293075487e-06,
      "loss": 0.2329,
      "step": 3995
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3117463119750479,
      "learning_rate": 5.057009321633233e-06,
      "loss": 0.1942,
      "step": 3996
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3022914267422514,
      "learning_rate": 5.054936340390414e-06,
      "loss": 0.1928,
      "step": 3997
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5988900394928836,
      "learning_rate": 5.052863349703402e-06,
      "loss": 0.2069,
      "step": 3998
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.952683731934008,
      "learning_rate": 5.050790349928566e-06,
      "loss": 0.5858,
      "step": 3999
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.141335678458154,
      "learning_rate": 5.048717341422282e-06,
      "loss": 0.5342,
      "step": 4000
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3948102101680475,
      "learning_rate": 5.0466443245409215e-06,
      "loss": 0.1831,
      "step": 4001
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4990852714599892,
      "learning_rate": 5.04457129964086e-06,
      "loss": 0.2582,
      "step": 4002
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.435945453175063,
      "learning_rate": 5.042498267078476e-06,
      "loss": 0.2351,
      "step": 4003
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.458382789953472,
      "learning_rate": 5.040425227210148e-06,
      "loss": 0.2312,
      "step": 4004
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2998619091882326,
      "learning_rate": 5.038352180392254e-06,
      "loss": 0.1978,
      "step": 4005
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3466120699257713,
      "learning_rate": 5.036279126981176e-06,
      "loss": 0.1713,
      "step": 4006
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3592136824950525,
      "learning_rate": 5.034206067333296e-06,
      "loss": 0.1612,
      "step": 4007
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3022232015530686,
      "learning_rate": 5.032133001804994e-06,
      "loss": 0.2058,
      "step": 4008
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2474166264897844,
      "learning_rate": 5.030059930752659e-06,
      "loss": 0.1872,
      "step": 4009
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5897378398038766,
      "learning_rate": 5.027986854532673e-06,
      "loss": 0.263,
      "step": 4010
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6723862108579888,
      "learning_rate": 5.025913773501421e-06,
      "loss": 0.2399,
      "step": 4011
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4535997828348923,
      "learning_rate": 5.023840688015291e-06,
      "loss": 0.2246,
      "step": 4012
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4505274448745151,
      "learning_rate": 5.021767598430672e-06,
      "loss": 0.2361,
      "step": 4013
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4984010271453936,
      "learning_rate": 5.019694505103949e-06,
      "loss": 0.1826,
      "step": 4014
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3892527078970023,
      "learning_rate": 5.017621408391513e-06,
      "loss": 0.2157,
      "step": 4015
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3218562260886833,
      "learning_rate": 5.015548308649752e-06,
      "loss": 0.1886,
      "step": 4016
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3454204580493814,
      "learning_rate": 5.013475206235056e-06,
      "loss": 0.194,
      "step": 4017
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4536881981581173,
      "learning_rate": 5.0114021015038185e-06,
      "loss": 0.1853,
      "step": 4018
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.404448860373234,
      "learning_rate": 5.0093289948124244e-06,
      "loss": 0.1919,
      "step": 4019
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1924523918529248,
      "learning_rate": 5.00725588651727e-06,
      "loss": 0.1545,
      "step": 4020
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4536247725386613,
      "learning_rate": 5.005182776974745e-06,
      "loss": 0.1879,
      "step": 4021
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.482206241137285,
      "learning_rate": 5.0031096665412385e-06,
      "loss": 0.2009,
      "step": 4022
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.56071710058989,
      "learning_rate": 5.0010365555731455e-06,
      "loss": 0.5332,
      "step": 4023
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.438976942284602,
      "learning_rate": 4.998963444426855e-06,
      "loss": 0.1836,
      "step": 4024
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5144774028122436,
      "learning_rate": 4.996890333458762e-06,
      "loss": 0.1912,
      "step": 4025
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4440422998027769,
      "learning_rate": 4.994817223025256e-06,
      "loss": 0.1953,
      "step": 4026
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7533505740228197,
      "learning_rate": 4.992744113482731e-06,
      "loss": 0.2847,
      "step": 4027
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.828303014046597,
      "learning_rate": 4.990671005187576e-06,
      "loss": 0.6561,
      "step": 4028
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.325271421747957,
      "learning_rate": 4.988597898496183e-06,
      "loss": 0.1979,
      "step": 4029
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8041129480328764,
      "learning_rate": 4.9865247937649445e-06,
      "loss": 0.2183,
      "step": 4030
    },
    {
      "epoch": 0.52,
      "grad_norm": 19.98058811568076,
      "learning_rate": 4.98445169135025e-06,
      "loss": 0.614,
      "step": 4031
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4512061386721553,
      "learning_rate": 4.982378591608489e-06,
      "loss": 0.1817,
      "step": 4032
    },
    {
      "epoch": 0.52,
      "grad_norm": 10.56877618158113,
      "learning_rate": 4.980305494896052e-06,
      "loss": 0.5447,
      "step": 4033
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1664188764291283,
      "learning_rate": 4.978232401569329e-06,
      "loss": 0.1601,
      "step": 4034
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4780139093216726,
      "learning_rate": 4.97615931198471e-06,
      "loss": 0.2136,
      "step": 4035
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2832619874119753,
      "learning_rate": 4.974086226498581e-06,
      "loss": 0.19,
      "step": 4036
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.771667742304106,
      "learning_rate": 4.972013145467329e-06,
      "loss": 0.6326,
      "step": 4037
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4233126000494363,
      "learning_rate": 4.969940069247343e-06,
      "loss": 0.2263,
      "step": 4038
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.274528708563254,
      "learning_rate": 4.9678669981950076e-06,
      "loss": 0.6857,
      "step": 4039
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4699362649626635,
      "learning_rate": 4.965793932666707e-06,
      "loss": 0.1695,
      "step": 4040
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.963650711705485,
      "learning_rate": 4.963720873018826e-06,
      "loss": 0.653,
      "step": 4041
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4929751968080776,
      "learning_rate": 4.961647819607749e-06,
      "loss": 0.2021,
      "step": 4042
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2104517809300455,
      "learning_rate": 4.959574772789853e-06,
      "loss": 0.1583,
      "step": 4043
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.470562315571343,
      "learning_rate": 4.957501732921524e-06,
      "loss": 0.2128,
      "step": 4044
    },
    {
      "epoch": 0.52,
      "grad_norm": 15.15529313680195,
      "learning_rate": 4.955428700359141e-06,
      "loss": 0.6983,
      "step": 4045
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.423073131213664,
      "learning_rate": 4.953355675459081e-06,
      "loss": 0.1992,
      "step": 4046
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3482609603331277,
      "learning_rate": 4.951282658577719e-06,
      "loss": 0.2019,
      "step": 4047
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2437490068548842,
      "learning_rate": 4.9492096500714346e-06,
      "loss": 0.1396,
      "step": 4048
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.518084792841237,
      "learning_rate": 4.947136650296601e-06,
      "loss": 0.6651,
      "step": 4049
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2426533438103347,
      "learning_rate": 4.945063659609588e-06,
      "loss": 0.178,
      "step": 4050
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2586523727086243,
      "learning_rate": 4.94299067836677e-06,
      "loss": 0.2158,
      "step": 4051
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.613680094235787,
      "learning_rate": 4.940917706924513e-06,
      "loss": 0.2017,
      "step": 4052
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.348439522425724,
      "learning_rate": 4.93884474563919e-06,
      "loss": 0.7244,
      "step": 4053
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.30292541955922,
      "learning_rate": 4.936771794867163e-06,
      "loss": 0.2016,
      "step": 4054
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4039780155733226,
      "learning_rate": 4.9346988549647974e-06,
      "loss": 0.2015,
      "step": 4055
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.461489640488889,
      "learning_rate": 4.9326259262884565e-06,
      "loss": 0.5872,
      "step": 4056
    },
    {
      "epoch": 0.52,
      "grad_norm": 8.207350774419115,
      "learning_rate": 4.9305530091945e-06,
      "loss": 0.6283,
      "step": 4057
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.535039123414706,
      "learning_rate": 4.928480104039285e-06,
      "loss": 0.2275,
      "step": 4058
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4644957479558207,
      "learning_rate": 4.92640721117917e-06,
      "loss": 0.2585,
      "step": 4059
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.245529910163482,
      "learning_rate": 4.924334330970509e-06,
      "loss": 0.1933,
      "step": 4060
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4454749610667732,
      "learning_rate": 4.922261463769652e-06,
      "loss": 0.2094,
      "step": 4061
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.553388826111865,
      "learning_rate": 4.920188609932951e-06,
      "loss": 0.2187,
      "step": 4062
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4565807018308763,
      "learning_rate": 4.918115769816754e-06,
      "loss": 0.1765,
      "step": 4063
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5501299212593196,
      "learning_rate": 4.9160429437774065e-06,
      "loss": 0.1994,
      "step": 4064
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4295015333668954,
      "learning_rate": 4.913970132171248e-06,
      "loss": 0.1573,
      "step": 4065
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.264916280103005,
      "learning_rate": 4.911897335354624e-06,
      "loss": 0.1787,
      "step": 4066
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.352310829039421,
      "learning_rate": 4.909824553683868e-06,
      "loss": 0.5617,
      "step": 4067
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2775683572569392,
      "learning_rate": 4.907751787515316e-06,
      "loss": 0.2115,
      "step": 4068
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2215005219199115,
      "learning_rate": 4.905679037205302e-06,
      "loss": 0.1815,
      "step": 4069
    },
    {
      "epoch": 0.52,
      "grad_norm": 8.180648385634765,
      "learning_rate": 4.903606303110153e-06,
      "loss": 0.6677,
      "step": 4070
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.904736246661793,
      "learning_rate": 4.9015335855862e-06,
      "loss": 0.6378,
      "step": 4071
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4055411135724916,
      "learning_rate": 4.899460884989764e-06,
      "loss": 0.2037,
      "step": 4072
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2323009027693188,
      "learning_rate": 4.897388201677169e-06,
      "loss": 0.1785,
      "step": 4073
    },
    {
      "epoch": 0.52,
      "grad_norm": 8.783767995913191,
      "learning_rate": 4.895315536004731e-06,
      "loss": 0.626,
      "step": 4074
    },
    {
      "epoch": 0.52,
      "grad_norm": 13.996884945164037,
      "learning_rate": 4.893242888328765e-06,
      "loss": 0.4402,
      "step": 4075
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5219874378136435,
      "learning_rate": 4.891170259005585e-06,
      "loss": 0.1932,
      "step": 4076
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.382941068165084,
      "learning_rate": 4.8890976483915e-06,
      "loss": 0.2031,
      "step": 4077
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6081122352826247,
      "learning_rate": 4.887025056842815e-06,
      "loss": 0.2168,
      "step": 4078
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4270187787037603,
      "learning_rate": 4.88495248471583e-06,
      "loss": 0.1826,
      "step": 4079
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.41679487418705,
      "learning_rate": 4.8828799323668465e-06,
      "loss": 0.5218,
      "step": 4080
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3308723273550238,
      "learning_rate": 4.880807400152162e-06,
      "loss": 0.1653,
      "step": 4081
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4133508811512794,
      "learning_rate": 4.878734888428068e-06,
      "loss": 0.1845,
      "step": 4082
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4023037717804485,
      "learning_rate": 4.876662397550851e-06,
      "loss": 0.2263,
      "step": 4083
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3769116019907939,
      "learning_rate": 4.874589927876799e-06,
      "loss": 0.2028,
      "step": 4084
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.980117676854406,
      "learning_rate": 4.872517479762192e-06,
      "loss": 0.4616,
      "step": 4085
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3303420614954367,
      "learning_rate": 4.870445053563307e-06,
      "loss": 0.1955,
      "step": 4086
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3787185101230155,
      "learning_rate": 4.868372649636421e-06,
      "loss": 0.1919,
      "step": 4087
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.056274532909217,
      "learning_rate": 4.866300268337802e-06,
      "loss": 0.6859,
      "step": 4088
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.894933857736531,
      "learning_rate": 4.864227910023714e-06,
      "loss": 0.5106,
      "step": 4089
    },
    {
      "epoch": 0.52,
      "grad_norm": 15.074261619614632,
      "learning_rate": 4.862155575050426e-06,
      "loss": 0.4224,
      "step": 4090
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5342646497085917,
      "learning_rate": 4.86008326377419e-06,
      "loss": 0.2232,
      "step": 4091
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2972783588333698,
      "learning_rate": 4.858010976551265e-06,
      "loss": 0.2075,
      "step": 4092
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.306949519706291,
      "learning_rate": 4.855938713737899e-06,
      "loss": 0.1741,
      "step": 4093
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.127699550771505,
      "learning_rate": 4.853866475690338e-06,
      "loss": 0.5816,
      "step": 4094
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2883921355070407,
      "learning_rate": 4.851794262764824e-06,
      "loss": 0.1599,
      "step": 4095
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.418200714801964,
      "learning_rate": 4.849722075317594e-06,
      "loss": 0.1681,
      "step": 4096
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3700876193507163,
      "learning_rate": 4.847649913704881e-06,
      "loss": 0.1792,
      "step": 4097
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.416127083563711,
      "learning_rate": 4.845577778282913e-06,
      "loss": 0.5606,
      "step": 4098
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.364955314844037,
      "learning_rate": 4.843505669407917e-06,
      "loss": 0.1759,
      "step": 4099
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4592796407882194,
      "learning_rate": 4.841433587436111e-06,
      "loss": 0.1936,
      "step": 4100
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3833953359606177,
      "learning_rate": 4.839361532723708e-06,
      "loss": 0.1769,
      "step": 4101
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3847108686193785,
      "learning_rate": 4.837289505626921e-06,
      "loss": 0.169,
      "step": 4102
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.342096213593133,
      "learning_rate": 4.835217506501955e-06,
      "loss": 0.574,
      "step": 4103
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4348026932155102,
      "learning_rate": 4.8331455357050084e-06,
      "loss": 0.2056,
      "step": 4104
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.671489680576698,
      "learning_rate": 4.831073593592279e-06,
      "loss": 0.1952,
      "step": 4105
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3548070912191743,
      "learning_rate": 4.829001680519958e-06,
      "loss": 0.2164,
      "step": 4106
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3890658151597766,
      "learning_rate": 4.826929796844228e-06,
      "loss": 0.1586,
      "step": 4107
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.47862680016271,
      "learning_rate": 4.824857942921274e-06,
      "loss": 0.5793,
      "step": 4108
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4113957451762662,
      "learning_rate": 4.822786119107271e-06,
      "loss": 0.1879,
      "step": 4109
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4380136675797808,
      "learning_rate": 4.820714325758388e-06,
      "loss": 0.2386,
      "step": 4110
    },
    {
      "epoch": 0.53,
      "grad_norm": 9.005796766811816,
      "learning_rate": 4.818642563230791e-06,
      "loss": 0.6138,
      "step": 4111
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.470097182549005,
      "learning_rate": 4.8165708318806385e-06,
      "loss": 0.5961,
      "step": 4112
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4988779191443438,
      "learning_rate": 4.814499132064088e-06,
      "loss": 0.229,
      "step": 4113
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6306708447197187,
      "learning_rate": 4.8124274641372875e-06,
      "loss": 0.213,
      "step": 4114
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.718269994306224,
      "learning_rate": 4.810355828456379e-06,
      "loss": 0.5403,
      "step": 4115
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4006345072621926,
      "learning_rate": 4.808284225377503e-06,
      "loss": 0.1901,
      "step": 4116
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.053645454802437,
      "learning_rate": 4.8062126552567896e-06,
      "loss": 0.5117,
      "step": 4117
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.405052860290166,
      "learning_rate": 4.804141118450369e-06,
      "loss": 0.3759,
      "step": 4118
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4307815042353238,
      "learning_rate": 4.80206961531436e-06,
      "loss": 0.1959,
      "step": 4119
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.35443605184593,
      "learning_rate": 4.799998146204879e-06,
      "loss": 0.2009,
      "step": 4120
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.911493710268533,
      "learning_rate": 4.797926711478037e-06,
      "loss": 0.5487,
      "step": 4121
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.316769823615401,
      "learning_rate": 4.7958553114899335e-06,
      "loss": 0.2028,
      "step": 4122
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.049351578542367,
      "learning_rate": 4.793783946596669e-06,
      "loss": 0.4911,
      "step": 4123
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.30614026790244,
      "learning_rate": 4.791712617154336e-06,
      "loss": 0.7331,
      "step": 4124
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.848731457816413,
      "learning_rate": 4.789641323519018e-06,
      "loss": 0.5851,
      "step": 4125
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3749335706677892,
      "learning_rate": 4.787570066046794e-06,
      "loss": 0.22,
      "step": 4126
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.741883560862228,
      "learning_rate": 4.785498845093739e-06,
      "loss": 0.2646,
      "step": 4127
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6552958210498598,
      "learning_rate": 4.78342766101592e-06,
      "loss": 0.1799,
      "step": 4128
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.4259195884657165,
      "learning_rate": 4.7813565141693965e-06,
      "loss": 0.6059,
      "step": 4129
    },
    {
      "epoch": 0.53,
      "grad_norm": 9.105849706431842,
      "learning_rate": 4.7792854049102234e-06,
      "loss": 0.7784,
      "step": 4130
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.076201968157865,
      "learning_rate": 4.7772143335944486e-06,
      "loss": 0.5299,
      "step": 4131
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.304389292550688,
      "learning_rate": 4.775143300578112e-06,
      "loss": 0.2173,
      "step": 4132
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4624429088746542,
      "learning_rate": 4.773072306217247e-06,
      "loss": 0.1706,
      "step": 4133
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.342831593233804,
      "learning_rate": 4.771001350867886e-06,
      "loss": 0.6365,
      "step": 4134
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3647042106624792,
      "learning_rate": 4.768930434886044e-06,
      "loss": 0.1597,
      "step": 4135
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3107451910554209,
      "learning_rate": 4.766859558627741e-06,
      "loss": 0.1847,
      "step": 4136
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2735912033812056,
      "learning_rate": 4.7647887224489834e-06,
      "loss": 0.1673,
      "step": 4137
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.2095430971853665,
      "learning_rate": 4.762717926705771e-06,
      "loss": 0.6604,
      "step": 4138
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.414296047593298,
      "learning_rate": 4.760647171754098e-06,
      "loss": 0.2241,
      "step": 4139
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4398675704664052,
      "learning_rate": 4.75857645794995e-06,
      "loss": 0.1779,
      "step": 4140
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.825349936415485,
      "learning_rate": 4.756505785649309e-06,
      "loss": 0.4595,
      "step": 4141
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.269802887584016,
      "learning_rate": 4.754435155208146e-06,
      "loss": 0.1844,
      "step": 4142
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3919430813590667,
      "learning_rate": 4.752364566982425e-06,
      "loss": 0.1559,
      "step": 4143
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4544791201460936,
      "learning_rate": 4.750294021328105e-06,
      "loss": 0.2032,
      "step": 4144
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5411952124303983,
      "learning_rate": 4.748223518601141e-06,
      "loss": 0.2056,
      "step": 4145
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4027551377409961,
      "learning_rate": 4.7461530591574715e-06,
      "loss": 0.2312,
      "step": 4146
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.394402394422781,
      "learning_rate": 4.744082643353034e-06,
      "loss": 0.18,
      "step": 4147
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4124782701029632,
      "learning_rate": 4.742012271543757e-06,
      "loss": 0.1856,
      "step": 4148
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3645301453837408,
      "learning_rate": 4.739941944085561e-06,
      "loss": 0.1726,
      "step": 4149
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3242662209462805,
      "learning_rate": 4.73787166133436e-06,
      "loss": 0.185,
      "step": 4150
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.347584042203452,
      "learning_rate": 4.735801423646059e-06,
      "loss": 0.1735,
      "step": 4151
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6694050429438345,
      "learning_rate": 4.733731231376557e-06,
      "loss": 0.2182,
      "step": 4152
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2112886678937191,
      "learning_rate": 4.731661084881744e-06,
      "loss": 0.1846,
      "step": 4153
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3631110725847384,
      "learning_rate": 4.729590984517498e-06,
      "loss": 0.1804,
      "step": 4154
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4882907116484745,
      "learning_rate": 4.727520930639698e-06,
      "loss": 0.1989,
      "step": 4155
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.453265777593563,
      "learning_rate": 4.7254509236042105e-06,
      "loss": 0.2759,
      "step": 4156
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.368539713421874,
      "learning_rate": 4.723380963766891e-06,
      "loss": 0.6329,
      "step": 4157
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.199156497206232,
      "learning_rate": 4.72131105148359e-06,
      "loss": 0.6545,
      "step": 4158
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4568865836330027,
      "learning_rate": 4.719241187110152e-06,
      "loss": 0.2085,
      "step": 4159
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.226462073119111,
      "learning_rate": 4.717171371002407e-06,
      "loss": 0.6471,
      "step": 4160
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.322632345325052,
      "learning_rate": 4.715101603516183e-06,
      "loss": 0.4514,
      "step": 4161
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.512513780620706,
      "learning_rate": 4.713031885007296e-06,
      "loss": 0.1462,
      "step": 4162
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3048859883952768,
      "learning_rate": 4.710962215831553e-06,
      "loss": 0.1769,
      "step": 4163
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.812890493823812,
      "learning_rate": 4.708892596344758e-06,
      "loss": 0.5358,
      "step": 4164
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4698741024699007,
      "learning_rate": 4.706823026902699e-06,
      "loss": 0.2334,
      "step": 4165
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.327836177699244,
      "learning_rate": 4.7047535078611626e-06,
      "loss": 0.2026,
      "step": 4166
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4593540329111958,
      "learning_rate": 4.70268403957592e-06,
      "loss": 0.2119,
      "step": 4167
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.317612906800678,
      "learning_rate": 4.700614622402737e-06,
      "loss": 0.2216,
      "step": 4168
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.089466275160961,
      "learning_rate": 4.698545256697373e-06,
      "loss": 0.5074,
      "step": 4169
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5205222970876695,
      "learning_rate": 4.696475942815573e-06,
      "loss": 0.2388,
      "step": 4170
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3013761009727378,
      "learning_rate": 4.6944066811130775e-06,
      "loss": 0.195,
      "step": 4171
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4913884185414776,
      "learning_rate": 4.692337471945614e-06,
      "loss": 0.218,
      "step": 4172
    },
    {
      "epoch": 0.53,
      "grad_norm": 8.066085513844946,
      "learning_rate": 4.690268315668908e-06,
      "loss": 0.5867,
      "step": 4173
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4291512697790625,
      "learning_rate": 4.688199212638669e-06,
      "loss": 0.2193,
      "step": 4174
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4628580183863185,
      "learning_rate": 4.6861301632106005e-06,
      "loss": 0.21,
      "step": 4175
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3884093100728137,
      "learning_rate": 4.684061167740395e-06,
      "loss": 0.2111,
      "step": 4176
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4990240024254442,
      "learning_rate": 4.6819922265837385e-06,
      "loss": 0.2435,
      "step": 4177
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.615571935879122,
      "learning_rate": 4.679923340096304e-06,
      "loss": 0.5107,
      "step": 4178
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5665669654291219,
      "learning_rate": 4.677854508633757e-06,
      "loss": 0.2236,
      "step": 4179
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2573168632310352,
      "learning_rate": 4.675785732551756e-06,
      "loss": 0.1752,
      "step": 4180
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4047971254173617,
      "learning_rate": 4.673717012205945e-06,
      "loss": 0.1936,
      "step": 4181
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.499856847976559,
      "learning_rate": 4.671648347951963e-06,
      "loss": 0.6413,
      "step": 4182
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.03669969324476,
      "learning_rate": 4.669579740145436e-06,
      "loss": 0.6317,
      "step": 4183
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3041177222473186,
      "learning_rate": 4.6675111891419835e-06,
      "loss": 0.1774,
      "step": 4184
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4177674969213068,
      "learning_rate": 4.665442695297212e-06,
      "loss": 0.2085,
      "step": 4185
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2419556039636817,
      "learning_rate": 4.663374258966718e-06,
      "loss": 0.1558,
      "step": 4186
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.667279334345029,
      "learning_rate": 4.661305880506094e-06,
      "loss": 0.2185,
      "step": 4187
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4126791126949478,
      "learning_rate": 4.659237560270914e-06,
      "loss": 0.1894,
      "step": 4188
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4332789383607523,
      "learning_rate": 4.657169298616748e-06,
      "loss": 0.2318,
      "step": 4189
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4984079720324868,
      "learning_rate": 4.655101095899154e-06,
      "loss": 0.2436,
      "step": 4190
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5165483042901853,
      "learning_rate": 4.653032952473678e-06,
      "loss": 0.1988,
      "step": 4191
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.624684049757558,
      "learning_rate": 4.650964868695863e-06,
      "loss": 0.2301,
      "step": 4192
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6469462505198926,
      "learning_rate": 4.648896844921233e-06,
      "loss": 0.2235,
      "step": 4193
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3311516965590708,
      "learning_rate": 4.6468288815053045e-06,
      "loss": 0.214,
      "step": 4194
    },
    {
      "epoch": 0.54,
      "grad_norm": 10.812499186908797,
      "learning_rate": 4.644760978803587e-06,
      "loss": 0.7594,
      "step": 4195
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4044239891002122,
      "learning_rate": 4.642693137171575e-06,
      "loss": 0.1915,
      "step": 4196
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2687862080926875,
      "learning_rate": 4.640625356964753e-06,
      "loss": 0.2022,
      "step": 4197
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4000869874872757,
      "learning_rate": 4.638557638538601e-06,
      "loss": 0.1769,
      "step": 4198
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.274448306138184,
      "learning_rate": 4.63648998224858e-06,
      "loss": 0.18,
      "step": 4199
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4346812792623087,
      "learning_rate": 4.634422388450144e-06,
      "loss": 0.2295,
      "step": 4200
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5103460696978135,
      "learning_rate": 4.632354857498738e-06,
      "loss": 0.1906,
      "step": 4201
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6644978840550064,
      "learning_rate": 4.6302873897497955e-06,
      "loss": 0.1949,
      "step": 4202
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5706306842195914,
      "learning_rate": 4.628219985558737e-06,
      "loss": 0.2089,
      "step": 4203
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3894994336656452,
      "learning_rate": 4.626152645280972e-06,
      "loss": 0.2208,
      "step": 4204
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2672146445035517,
      "learning_rate": 4.624085369271902e-06,
      "loss": 0.1302,
      "step": 4205
    },
    {
      "epoch": 0.54,
      "grad_norm": 8.445013836842735,
      "learning_rate": 4.622018157886915e-06,
      "loss": 0.7439,
      "step": 4206
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2527226838406715,
      "learning_rate": 4.6199510114813875e-06,
      "loss": 0.1817,
      "step": 4207
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4993635169633979,
      "learning_rate": 4.617883930410688e-06,
      "loss": 0.2207,
      "step": 4208
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.388291784737558,
      "learning_rate": 4.615816915030169e-06,
      "loss": 0.2003,
      "step": 4209
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.807617278115059,
      "learning_rate": 4.613749965695178e-06,
      "loss": 0.5585,
      "step": 4210
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.414665974374318,
      "learning_rate": 4.611683082761046e-06,
      "loss": 0.2128,
      "step": 4211
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.315707406064571,
      "learning_rate": 4.609616266583093e-06,
      "loss": 0.1434,
      "step": 4212
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.407572239655353,
      "learning_rate": 4.607549517516629e-06,
      "loss": 0.249,
      "step": 4213
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.725788023895319,
      "learning_rate": 4.605482835916954e-06,
      "loss": 0.4579,
      "step": 4214
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.935095559287942,
      "learning_rate": 4.603416222139352e-06,
      "loss": 0.6823,
      "step": 4215
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5940682633597734,
      "learning_rate": 4.6013496765391e-06,
      "loss": 0.2289,
      "step": 4216
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.546790517451567,
      "learning_rate": 4.59928319947146e-06,
      "loss": 0.6902,
      "step": 4217
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2829925123004926,
      "learning_rate": 4.597216791291681e-06,
      "loss": 0.2358,
      "step": 4218
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.845046042159102,
      "learning_rate": 4.595150452355006e-06,
      "loss": 0.4468,
      "step": 4219
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2879380607390223,
      "learning_rate": 4.593084183016664e-06,
      "loss": 0.1814,
      "step": 4220
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.828635214172637,
      "learning_rate": 4.5910179836318665e-06,
      "loss": 0.3945,
      "step": 4221
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4170879606411444,
      "learning_rate": 4.58895185455582e-06,
      "loss": 0.1816,
      "step": 4222
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3152769096584422,
      "learning_rate": 4.586885796143715e-06,
      "loss": 0.1668,
      "step": 4223
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2836090703916248,
      "learning_rate": 4.58481980875073e-06,
      "loss": 0.1694,
      "step": 4224
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3049068689319074,
      "learning_rate": 4.582753892732033e-06,
      "loss": 0.1961,
      "step": 4225
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4563411181824581,
      "learning_rate": 4.58068804844278e-06,
      "loss": 0.2202,
      "step": 4226
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5872653972065212,
      "learning_rate": 4.578622276238112e-06,
      "loss": 0.2266,
      "step": 4227
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3037335095330618,
      "learning_rate": 4.576556576473158e-06,
      "loss": 0.1908,
      "step": 4228
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3142216404378428,
      "learning_rate": 4.57449094950304e-06,
      "loss": 0.1727,
      "step": 4229
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3945563236268756,
      "learning_rate": 4.572425395682859e-06,
      "loss": 0.1933,
      "step": 4230
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.105325300025991,
      "learning_rate": 4.570359915367711e-06,
      "loss": 0.544,
      "step": 4231
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.682254445160143,
      "learning_rate": 4.568294508912674e-06,
      "loss": 0.2118,
      "step": 4232
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2561969194282523,
      "learning_rate": 4.566229176672815e-06,
      "loss": 0.1429,
      "step": 4233
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2792219761351349,
      "learning_rate": 4.5641639190031894e-06,
      "loss": 0.1914,
      "step": 4234
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6141172438524445,
      "learning_rate": 4.56209873625884e-06,
      "loss": 0.2253,
      "step": 4235
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4068987186253328,
      "learning_rate": 4.560033628794792e-06,
      "loss": 0.1899,
      "step": 4236
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.260773117261743,
      "learning_rate": 4.557968596966063e-06,
      "loss": 0.4485,
      "step": 4237
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5435469256394525,
      "learning_rate": 4.5559036411276596e-06,
      "loss": 0.2644,
      "step": 4238
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.469209014214999,
      "learning_rate": 4.553838761634569e-06,
      "loss": 0.1918,
      "step": 4239
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.2397272046378705,
      "learning_rate": 4.551773958841765e-06,
      "loss": 0.3693,
      "step": 4240
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3884501148550041,
      "learning_rate": 4.549709233104216e-06,
      "loss": 0.2128,
      "step": 4241
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.363673340722367,
      "learning_rate": 4.54764458477687e-06,
      "loss": 0.2054,
      "step": 4242
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4874861304253344,
      "learning_rate": 4.5455800142146626e-06,
      "loss": 0.2236,
      "step": 4243
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3647685281594404,
      "learning_rate": 4.54351552177252e-06,
      "loss": 0.1863,
      "step": 4244
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2263771761719562,
      "learning_rate": 4.541451107805351e-06,
      "loss": 0.1853,
      "step": 4245
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.574194204140475,
      "learning_rate": 4.53938677266805e-06,
      "loss": 0.1943,
      "step": 4246
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.198679448554334,
      "learning_rate": 4.537322516715505e-06,
      "loss": 0.1762,
      "step": 4247
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3086449667496114,
      "learning_rate": 4.535258340302583e-06,
      "loss": 0.1824,
      "step": 4248
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.334133425259196,
      "learning_rate": 4.53319424378414e-06,
      "loss": 0.5138,
      "step": 4249
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4021475209449556,
      "learning_rate": 4.531130227515019e-06,
      "loss": 0.1888,
      "step": 4250
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2502775315237504,
      "learning_rate": 4.529066291850047e-06,
      "loss": 0.1807,
      "step": 4251
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.30109611918384,
      "learning_rate": 4.527002437144039e-06,
      "loss": 0.1801,
      "step": 4252
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4743857879612492,
      "learning_rate": 4.524938663751796e-06,
      "loss": 0.2113,
      "step": 4253
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.281267589688021,
      "learning_rate": 4.522874972028103e-06,
      "loss": 0.6015,
      "step": 4254
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4368178354528311,
      "learning_rate": 4.520811362327732e-06,
      "loss": 0.2195,
      "step": 4255
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4061236100016246,
      "learning_rate": 4.518747835005445e-06,
      "loss": 0.2013,
      "step": 4256
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4285455086154388,
      "learning_rate": 4.516684390415985e-06,
      "loss": 0.1982,
      "step": 4257
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4084262240098977,
      "learning_rate": 4.51462102891408e-06,
      "loss": 0.1917,
      "step": 4258
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4537361973033827,
      "learning_rate": 4.512557750854448e-06,
      "loss": 0.2351,
      "step": 4259
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5572833653035332,
      "learning_rate": 4.51049455659179e-06,
      "loss": 0.1982,
      "step": 4260
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.371885671798101,
      "learning_rate": 4.508431446480792e-06,
      "loss": 0.2076,
      "step": 4261
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3462038978396225,
      "learning_rate": 4.506368420876127e-06,
      "loss": 0.1991,
      "step": 4262
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2269796632871834,
      "learning_rate": 4.504305480132454e-06,
      "loss": 0.17,
      "step": 4263
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4431434149056244,
      "learning_rate": 4.502242624604413e-06,
      "loss": 0.1983,
      "step": 4264
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2870332772395547,
      "learning_rate": 4.5001798546466365e-06,
      "loss": 0.1683,
      "step": 4265
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.493273088192751,
      "learning_rate": 4.498117170613738e-06,
      "loss": 0.2367,
      "step": 4266
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.351365919967509,
      "learning_rate": 4.496054572860317e-06,
      "loss": 0.5638,
      "step": 4267
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.988347661782494,
      "learning_rate": 4.493992061740956e-06,
      "loss": 0.4957,
      "step": 4268
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2405183514487694,
      "learning_rate": 4.491929637610225e-06,
      "loss": 0.1688,
      "step": 4269
    },
    {
      "epoch": 0.55,
      "grad_norm": 11.348711809453278,
      "learning_rate": 4.489867300822681e-06,
      "loss": 0.5118,
      "step": 4270
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.388933928147609,
      "learning_rate": 4.4878050517328625e-06,
      "loss": 0.5273,
      "step": 4271
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4382488352707794,
      "learning_rate": 4.485742890695292e-06,
      "loss": 0.2147,
      "step": 4272
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4401573578598201,
      "learning_rate": 4.483680818064481e-06,
      "loss": 0.1844,
      "step": 4273
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3483032972794398,
      "learning_rate": 4.481618834194921e-06,
      "loss": 0.1955,
      "step": 4274
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5182340074611982,
      "learning_rate": 4.479556939441095e-06,
      "loss": 0.2539,
      "step": 4275
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.831285348628483,
      "learning_rate": 4.477495134157464e-06,
      "loss": 0.5334,
      "step": 4276
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4836225171329938,
      "learning_rate": 4.475433418698477e-06,
      "loss": 0.1758,
      "step": 4277
    },
    {
      "epoch": 0.55,
      "grad_norm": 8.84647533571796,
      "learning_rate": 4.473371793418567e-06,
      "loss": 0.4973,
      "step": 4278
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3202396380873838,
      "learning_rate": 4.471310258672151e-06,
      "loss": 0.1744,
      "step": 4279
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3743772301217325,
      "learning_rate": 4.469248814813631e-06,
      "loss": 0.1653,
      "step": 4280
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3232804648753271,
      "learning_rate": 4.467187462197392e-06,
      "loss": 0.191,
      "step": 4281
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4993525368064187,
      "learning_rate": 4.465126201177804e-06,
      "loss": 0.2136,
      "step": 4282
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5192210580366579,
      "learning_rate": 4.463065032109224e-06,
      "loss": 0.1688,
      "step": 4283
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3282252756792703,
      "learning_rate": 4.461003955345989e-06,
      "loss": 0.146,
      "step": 4284
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5182220063232168,
      "learning_rate": 4.458942971242423e-06,
      "loss": 0.2047,
      "step": 4285
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3316952047753714,
      "learning_rate": 4.456882080152834e-06,
      "loss": 0.1648,
      "step": 4286
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5126471557876715,
      "learning_rate": 4.454821282431511e-06,
      "loss": 0.2012,
      "step": 4287
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5512801115963373,
      "learning_rate": 4.4527605784327295e-06,
      "loss": 0.1796,
      "step": 4288
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5770474563766008,
      "learning_rate": 4.450699968510749e-06,
      "loss": 0.2205,
      "step": 4289
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.402219183918579,
      "learning_rate": 4.448639453019812e-06,
      "loss": 0.1671,
      "step": 4290
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3255509242710013,
      "learning_rate": 4.446579032314145e-06,
      "loss": 0.1849,
      "step": 4291
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7493442653430844,
      "learning_rate": 4.444518706747956e-06,
      "loss": 0.1923,
      "step": 4292
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.0636907506018565,
      "learning_rate": 4.442458476675443e-06,
      "loss": 0.2108,
      "step": 4293
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2991293268636366,
      "learning_rate": 4.440398342450782e-06,
      "loss": 0.2097,
      "step": 4294
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.547638357298998,
      "learning_rate": 4.438338304428132e-06,
      "loss": 0.2155,
      "step": 4295
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3520271039943759,
      "learning_rate": 4.43627836296164e-06,
      "loss": 0.1883,
      "step": 4296
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.504315294739271,
      "learning_rate": 4.434218518405432e-06,
      "loss": 0.1936,
      "step": 4297
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2686867166519573,
      "learning_rate": 4.43215877111362e-06,
      "loss": 0.1841,
      "step": 4298
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.380048162292071,
      "learning_rate": 4.4300991214402986e-06,
      "loss": 0.2224,
      "step": 4299
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4261230172787025,
      "learning_rate": 4.428039569739544e-06,
      "loss": 0.1926,
      "step": 4300
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3328807184890967,
      "learning_rate": 4.425980116365419e-06,
      "loss": 0.1974,
      "step": 4301
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3416307795529197,
      "learning_rate": 4.423920761671965e-06,
      "loss": 0.2001,
      "step": 4302
    },
    {
      "epoch": 0.55,
      "grad_norm": 12.442098808113506,
      "learning_rate": 4.421861506013213e-06,
      "loss": 0.5697,
      "step": 4303
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.437603670486912,
      "learning_rate": 4.419802349743169e-06,
      "loss": 0.1549,
      "step": 4304
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3032927354941848,
      "learning_rate": 4.417743293215827e-06,
      "loss": 0.186,
      "step": 4305
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5459041279696268,
      "learning_rate": 4.415684336785165e-06,
      "loss": 0.2074,
      "step": 4306
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5268351097585142,
      "learning_rate": 4.413625480805138e-06,
      "loss": 0.2156,
      "step": 4307
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.480897160029066,
      "learning_rate": 4.411566725629688e-06,
      "loss": 0.2055,
      "step": 4308
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.437977053426803,
      "learning_rate": 4.409508071612741e-06,
      "loss": 0.1815,
      "step": 4309
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4903615998629942,
      "learning_rate": 4.407449519108203e-06,
      "loss": 0.1694,
      "step": 4310
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.280871621435785,
      "learning_rate": 4.405391068469958e-06,
      "loss": 0.1683,
      "step": 4311
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4319325283515307,
      "learning_rate": 4.403332720051884e-06,
      "loss": 0.1975,
      "step": 4312
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3697935030236692,
      "learning_rate": 4.401274474207833e-06,
      "loss": 0.1477,
      "step": 4313
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2424552331168202,
      "learning_rate": 4.3992163312916404e-06,
      "loss": 0.166,
      "step": 4314
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.463390935795949,
      "learning_rate": 4.397158291657125e-06,
      "loss": 0.2156,
      "step": 4315
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4379897167084752,
      "learning_rate": 4.39510035565809e-06,
      "loss": 0.1674,
      "step": 4316
    },
    {
      "epoch": 0.55,
      "grad_norm": 11.012599694665418,
      "learning_rate": 4.393042523648315e-06,
      "loss": 0.6118,
      "step": 4317
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4372396390267494,
      "learning_rate": 4.390984795981566e-06,
      "loss": 0.2305,
      "step": 4318
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4663388786991316,
      "learning_rate": 4.388927173011592e-06,
      "loss": 0.1791,
      "step": 4319
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4273416537888848,
      "learning_rate": 4.38686965509212e-06,
      "loss": 0.1967,
      "step": 4320
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.2069124765662975,
      "learning_rate": 4.384812242576863e-06,
      "loss": 0.5263,
      "step": 4321
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3498087408162776,
      "learning_rate": 4.382754935819514e-06,
      "loss": 0.1932,
      "step": 4322
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.406743043848111,
      "learning_rate": 4.380697735173745e-06,
      "loss": 0.1994,
      "step": 4323
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4467613995464264,
      "learning_rate": 4.378640640993218e-06,
      "loss": 0.1868,
      "step": 4324
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.643629983171337,
      "learning_rate": 4.376583653631567e-06,
      "loss": 0.5262,
      "step": 4325
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3916181464370805,
      "learning_rate": 4.374526773442413e-06,
      "loss": 0.1716,
      "step": 4326
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3408671640389198,
      "learning_rate": 4.372470000779357e-06,
      "loss": 0.1781,
      "step": 4327
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3412894811101153,
      "learning_rate": 4.370413335995985e-06,
      "loss": 0.1679,
      "step": 4328
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4001503470997398,
      "learning_rate": 4.368356779445856e-06,
      "loss": 0.2246,
      "step": 4329
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.581328259811922,
      "learning_rate": 4.36630033148252e-06,
      "loss": 0.2321,
      "step": 4330
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2459050664199531,
      "learning_rate": 4.364243992459506e-06,
      "loss": 0.1824,
      "step": 4331
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.932638288452758,
      "learning_rate": 4.362187762730319e-06,
      "loss": 0.7123,
      "step": 4332
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4670543167937435,
      "learning_rate": 4.360131642648449e-06,
      "loss": 0.2023,
      "step": 4333
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.528879723181518,
      "learning_rate": 4.358075632567368e-06,
      "loss": 0.2319,
      "step": 4334
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5485436269891297,
      "learning_rate": 4.356019732840528e-06,
      "loss": 0.2546,
      "step": 4335
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4191476345331715,
      "learning_rate": 4.3539639438213606e-06,
      "loss": 0.219,
      "step": 4336
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.492247547339455,
      "learning_rate": 4.351908265863282e-06,
      "loss": 0.1779,
      "step": 4337
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5477962893883213,
      "learning_rate": 4.349852699319686e-06,
      "loss": 0.2027,
      "step": 4338
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2808235653254727,
      "learning_rate": 4.347797244543945e-06,
      "loss": 0.1574,
      "step": 4339
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.935780997951666,
      "learning_rate": 4.34574190188942e-06,
      "loss": 0.5144,
      "step": 4340
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4610411217131083,
      "learning_rate": 4.343686671709449e-06,
      "loss": 0.2221,
      "step": 4341
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3838966809295583,
      "learning_rate": 4.341631554357347e-06,
      "loss": 0.1867,
      "step": 4342
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2450101132278784,
      "learning_rate": 4.339576550186413e-06,
      "loss": 0.2097,
      "step": 4343
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.882641770394771,
      "learning_rate": 4.3375216595499254e-06,
      "loss": 0.6086,
      "step": 4344
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.560497911253629,
      "learning_rate": 4.335466882801146e-06,
      "loss": 0.2212,
      "step": 4345
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3182269864765852,
      "learning_rate": 4.333412220293313e-06,
      "loss": 0.1372,
      "step": 4346
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1563726185094023,
      "learning_rate": 4.3313576723796464e-06,
      "loss": 0.1628,
      "step": 4347
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3868220090066512,
      "learning_rate": 4.329303239413346e-06,
      "loss": 0.2566,
      "step": 4348
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6089688138010017,
      "learning_rate": 4.327248921747597e-06,
      "loss": 0.2335,
      "step": 4349
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6114490383183606,
      "learning_rate": 4.325194719735557e-06,
      "loss": 0.2603,
      "step": 4350
    },
    {
      "epoch": 0.56,
      "grad_norm": 31.85345512684861,
      "learning_rate": 4.3231406337303665e-06,
      "loss": 0.7006,
      "step": 4351
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.450441463277651,
      "learning_rate": 4.321086664085149e-06,
      "loss": 0.251,
      "step": 4352
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3247331032918386,
      "learning_rate": 4.319032811153005e-06,
      "loss": 0.2076,
      "step": 4353
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3528724777980625,
      "learning_rate": 4.316979075287014e-06,
      "loss": 0.1842,
      "step": 4354
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.4026122934911,
      "learning_rate": 4.314925456840239e-06,
      "loss": 0.509,
      "step": 4355
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4641285893613518,
      "learning_rate": 4.3128719561657205e-06,
      "loss": 0.1912,
      "step": 4356
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3549310778202557,
      "learning_rate": 4.310818573616476e-06,
      "loss": 0.2233,
      "step": 4357
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2812005006352787,
      "learning_rate": 4.30876530954551e-06,
      "loss": 0.1953,
      "step": 4358
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.334667633715911,
      "learning_rate": 4.3067121643058e-06,
      "loss": 0.2133,
      "step": 4359
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.544275197788835,
      "learning_rate": 4.304659138250309e-06,
      "loss": 0.6338,
      "step": 4360
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2893093872879844,
      "learning_rate": 4.302606231731971e-06,
      "loss": 0.1539,
      "step": 4361
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3778475555871421,
      "learning_rate": 4.300553445103707e-06,
      "loss": 0.1788,
      "step": 4362
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.478933234995698,
      "learning_rate": 4.298500778718415e-06,
      "loss": 0.2058,
      "step": 4363
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2864261738352973,
      "learning_rate": 4.296448232928971e-06,
      "loss": 0.2034,
      "step": 4364
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4762541677740006,
      "learning_rate": 4.294395808088232e-06,
      "loss": 0.2166,
      "step": 4365
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.32490822695779,
      "learning_rate": 4.292343504549032e-06,
      "loss": 0.5555,
      "step": 4366
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3340030397191036,
      "learning_rate": 4.29029132266419e-06,
      "loss": 0.1741,
      "step": 4367
    },
    {
      "epoch": 0.56,
      "grad_norm": 9.145454559730199,
      "learning_rate": 4.288239262786497e-06,
      "loss": 0.4814,
      "step": 4368
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1754463633780432,
      "learning_rate": 4.286187325268726e-06,
      "loss": 0.192,
      "step": 4369
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3715968784939563,
      "learning_rate": 4.284135510463628e-06,
      "loss": 0.1945,
      "step": 4370
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2385227787767303,
      "learning_rate": 4.282083818723937e-06,
      "loss": 0.1848,
      "step": 4371
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2601756238421666,
      "learning_rate": 4.280032250402358e-06,
      "loss": 0.1588,
      "step": 4372
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2888938873575249,
      "learning_rate": 4.2779808058515825e-06,
      "loss": 0.1848,
      "step": 4373
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4421032806077687,
      "learning_rate": 4.275929485424278e-06,
      "loss": 0.2064,
      "step": 4374
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3444777946836524,
      "learning_rate": 4.2738782894730876e-06,
      "loss": 0.19,
      "step": 4375
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3174565264078233,
      "learning_rate": 4.271827218350636e-06,
      "loss": 0.1781,
      "step": 4376
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5536821402676242,
      "learning_rate": 4.269776272409529e-06,
      "loss": 0.2223,
      "step": 4377
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.698787602127358,
      "learning_rate": 4.2677254520023465e-06,
      "loss": 0.6719,
      "step": 4378
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.555707831200862,
      "learning_rate": 4.265674757481647e-06,
      "loss": 0.2209,
      "step": 4379
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4702599671452539,
      "learning_rate": 4.263624189199971e-06,
      "loss": 0.2097,
      "step": 4380
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3664683308204815,
      "learning_rate": 4.261573747509833e-06,
      "loss": 0.1499,
      "step": 4381
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3940018374969358,
      "learning_rate": 4.259523432763728e-06,
      "loss": 0.2448,
      "step": 4382
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4779487422112494,
      "learning_rate": 4.25747324531413e-06,
      "loss": 0.1983,
      "step": 4383
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.462405540956413,
      "learning_rate": 4.25542318551349e-06,
      "loss": 0.2018,
      "step": 4384
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4269840491726606,
      "learning_rate": 4.2533732537142335e-06,
      "loss": 0.1891,
      "step": 4385
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3436766706079777,
      "learning_rate": 4.2513234502687725e-06,
      "loss": 0.2015,
      "step": 4386
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3334820148624398,
      "learning_rate": 4.249273775529489e-06,
      "loss": 0.1952,
      "step": 4387
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6002966207764988,
      "learning_rate": 4.247224229848747e-06,
      "loss": 0.2402,
      "step": 4388
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4176425434433648,
      "learning_rate": 4.245174813578887e-06,
      "loss": 0.2093,
      "step": 4389
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4815276917564042,
      "learning_rate": 4.243125527072227e-06,
      "loss": 0.2456,
      "step": 4390
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.331358523960658,
      "learning_rate": 4.241076370681064e-06,
      "loss": 0.5365,
      "step": 4391
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3905921558358945,
      "learning_rate": 4.239027344757671e-06,
      "loss": 0.2132,
      "step": 4392
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3344032918146467,
      "learning_rate": 4.2369784496542986e-06,
      "loss": 0.166,
      "step": 4393
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4728455289443365,
      "learning_rate": 4.234929685723175e-06,
      "loss": 0.217,
      "step": 4394
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.455790642181059,
      "learning_rate": 4.2328810533165095e-06,
      "loss": 0.192,
      "step": 4395
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5877437042441165,
      "learning_rate": 4.230832552786485e-06,
      "loss": 0.2103,
      "step": 4396
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4896504690565868,
      "learning_rate": 4.2287841844852595e-06,
      "loss": 0.2061,
      "step": 4397
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4656629252866813,
      "learning_rate": 4.226735948764976e-06,
      "loss": 0.1918,
      "step": 4398
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.005161157894726,
      "learning_rate": 4.2246878459777465e-06,
      "loss": 0.4593,
      "step": 4399
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.742475003587761,
      "learning_rate": 4.222639876475663e-06,
      "loss": 0.7054,
      "step": 4400
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3170355077703506,
      "learning_rate": 4.220592040610798e-06,
      "loss": 0.1913,
      "step": 4401
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3501163964466192,
      "learning_rate": 4.218544338735197e-06,
      "loss": 0.2151,
      "step": 4402
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2928161130049316,
      "learning_rate": 4.216496771200881e-06,
      "loss": 0.1863,
      "step": 4403
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5369374018381408,
      "learning_rate": 4.214449338359856e-06,
      "loss": 0.2248,
      "step": 4404
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5298862119036407,
      "learning_rate": 4.2124020405640955e-06,
      "loss": 0.2381,
      "step": 4405
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3147101734085962,
      "learning_rate": 4.2103548781655555e-06,
      "loss": 0.1712,
      "step": 4406
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.626006722643224,
      "learning_rate": 4.2083078515161664e-06,
      "loss": 0.2098,
      "step": 4407
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4814853089158702,
      "learning_rate": 4.206260960967836e-06,
      "loss": 0.1747,
      "step": 4408
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3689643079601361,
      "learning_rate": 4.204214206872448e-06,
      "loss": 0.2234,
      "step": 4409
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3283437583051376,
      "learning_rate": 4.202167589581863e-06,
      "loss": 0.1871,
      "step": 4410
    },
    {
      "epoch": 0.56,
      "grad_norm": 8.149164342436707,
      "learning_rate": 4.200121109447919e-06,
      "loss": 0.6185,
      "step": 4411
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5917736070103565,
      "learning_rate": 4.198074766822429e-06,
      "loss": 0.1911,
      "step": 4412
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3794761840409944,
      "learning_rate": 4.196028562057181e-06,
      "loss": 0.1923,
      "step": 4413
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.7506297307104965,
      "learning_rate": 4.193982495503946e-06,
      "loss": 0.4926,
      "step": 4414
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3533812275352943,
      "learning_rate": 4.191936567514464e-06,
      "loss": 0.1763,
      "step": 4415
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.908707775238628,
      "learning_rate": 4.189890778440454e-06,
      "loss": 0.5934,
      "step": 4416
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3786913936438494,
      "learning_rate": 4.187845128633611e-06,
      "loss": 0.1623,
      "step": 4417
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5025014224110012,
      "learning_rate": 4.185799618445605e-06,
      "loss": 0.1895,
      "step": 4418
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.284357578535576,
      "learning_rate": 4.1837542482280845e-06,
      "loss": 0.8393,
      "step": 4419
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3042776420337059,
      "learning_rate": 4.181709018332672e-06,
      "loss": 0.1839,
      "step": 4420
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5476513123126503,
      "learning_rate": 4.179663929110964e-06,
      "loss": 0.2262,
      "step": 4421
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.739612928339002,
      "learning_rate": 4.177618980914536e-06,
      "loss": 0.5577,
      "step": 4422
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3155502250177724,
      "learning_rate": 4.1755741740949405e-06,
      "loss": 0.1619,
      "step": 4423
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5058371184281436,
      "learning_rate": 4.173529509003702e-06,
      "loss": 0.1755,
      "step": 4424
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2266652676318817,
      "learning_rate": 4.171484985992323e-06,
      "loss": 0.1488,
      "step": 4425
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.274087273178257,
      "learning_rate": 4.169440605412278e-06,
      "loss": 0.4097,
      "step": 4426
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.213737670958577,
      "learning_rate": 4.167396367615023e-06,
      "loss": 0.1687,
      "step": 4427
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.433082331987379,
      "learning_rate": 4.165352272951985e-06,
      "loss": 0.1973,
      "step": 4428
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.580801350959576,
      "learning_rate": 4.163308321774566e-06,
      "loss": 0.5301,
      "step": 4429
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6128843039316576,
      "learning_rate": 4.161264514434148e-06,
      "loss": 0.2138,
      "step": 4430
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.297270628926722,
      "learning_rate": 4.15922085128208e-06,
      "loss": 0.2164,
      "step": 4431
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5612597207044623,
      "learning_rate": 4.157177332669698e-06,
      "loss": 0.1829,
      "step": 4432
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.639010985962243,
      "learning_rate": 4.155133958948302e-06,
      "loss": 0.4502,
      "step": 4433
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3588818979191124,
      "learning_rate": 4.153090730469174e-06,
      "loss": 0.1916,
      "step": 4434
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5832290770232342,
      "learning_rate": 4.151047647583568e-06,
      "loss": 0.2099,
      "step": 4435
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3805785672045048,
      "learning_rate": 4.149004710642712e-06,
      "loss": 0.1556,
      "step": 4436
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3898582754709548,
      "learning_rate": 4.146961919997813e-06,
      "loss": 0.2078,
      "step": 4437
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.459773503935242,
      "learning_rate": 4.144919276000048e-06,
      "loss": 0.2065,
      "step": 4438
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3993865187767365,
      "learning_rate": 4.142876779000573e-06,
      "loss": 0.1687,
      "step": 4439
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3127339382229652,
      "learning_rate": 4.1408344293505154e-06,
      "loss": 0.1791,
      "step": 4440
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4537279038982718,
      "learning_rate": 4.13879222740098e-06,
      "loss": 0.1878,
      "step": 4441
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2804906061489025,
      "learning_rate": 4.136750173503046e-06,
      "loss": 0.1987,
      "step": 4442
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3370736763494215,
      "learning_rate": 4.134708268007764e-06,
      "loss": 0.1913,
      "step": 4443
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.378933238571422,
      "learning_rate": 4.132666511266162e-06,
      "loss": 0.1994,
      "step": 4444
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4840319360622243,
      "learning_rate": 4.130624903629242e-06,
      "loss": 0.2271,
      "step": 4445
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2759033887165716,
      "learning_rate": 4.12858344544798e-06,
      "loss": 0.1524,
      "step": 4446
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3494357524760305,
      "learning_rate": 4.126542137073325e-06,
      "loss": 0.1937,
      "step": 4447
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4067601544858277,
      "learning_rate": 4.124500978856204e-06,
      "loss": 0.197,
      "step": 4448
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2497832371152102,
      "learning_rate": 4.122459971147514e-06,
      "loss": 0.153,
      "step": 4449
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3366975284588838,
      "learning_rate": 4.120419114298127e-06,
      "loss": 0.1899,
      "step": 4450
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.5244535317401136,
      "learning_rate": 4.118378408658891e-06,
      "loss": 0.6721,
      "step": 4451
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5479830688824456,
      "learning_rate": 4.1163378545806286e-06,
      "loss": 0.2361,
      "step": 4452
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.052321576038291,
      "learning_rate": 4.114297452414133e-06,
      "loss": 0.4988,
      "step": 4453
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.537826347606105,
      "learning_rate": 4.112257202510173e-06,
      "loss": 0.2354,
      "step": 4454
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.788387632475543,
      "learning_rate": 4.110217105219492e-06,
      "loss": 0.526,
      "step": 4455
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3538608587852088,
      "learning_rate": 4.108177160892807e-06,
      "loss": 0.2085,
      "step": 4456
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2640721230888896,
      "learning_rate": 4.106137369880804e-06,
      "loss": 0.1624,
      "step": 4457
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3098691051170759,
      "learning_rate": 4.104097732534153e-06,
      "loss": 0.1889,
      "step": 4458
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.475129284610638,
      "learning_rate": 4.102058249203483e-06,
      "loss": 0.2139,
      "step": 4459
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3940817159380738,
      "learning_rate": 4.1000189202394144e-06,
      "loss": 0.1978,
      "step": 4460
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.965133682238179,
      "learning_rate": 4.097979745992526e-06,
      "loss": 0.4774,
      "step": 4461
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5394630737234072,
      "learning_rate": 4.095940726813375e-06,
      "loss": 0.1823,
      "step": 4462
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3585666634675946,
      "learning_rate": 4.093901863052495e-06,
      "loss": 0.1789,
      "step": 4463
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5496755179740083,
      "learning_rate": 4.091863155060389e-06,
      "loss": 0.2277,
      "step": 4464
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4060379594703598,
      "learning_rate": 4.0898246031875346e-06,
      "loss": 0.1837,
      "step": 4465
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2958144664517766,
      "learning_rate": 4.087786207784383e-06,
      "loss": 0.1734,
      "step": 4466
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.101493163691714,
      "learning_rate": 4.085747969201357e-06,
      "loss": 0.4481,
      "step": 4467
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4453181643148367,
      "learning_rate": 4.083709887788852e-06,
      "loss": 0.2056,
      "step": 4468
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4061208420322329,
      "learning_rate": 4.081671963897241e-06,
      "loss": 0.2038,
      "step": 4469
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3216977550653202,
      "learning_rate": 4.079634197876867e-06,
      "loss": 0.1824,
      "step": 4470
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3959125409044015,
      "learning_rate": 4.077596590078044e-06,
      "loss": 0.2073,
      "step": 4471
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.39002146551582,
      "learning_rate": 4.075559140851061e-06,
      "loss": 0.1776,
      "step": 4472
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.392343127568286,
      "learning_rate": 4.073521850546177e-06,
      "loss": 0.5154,
      "step": 4473
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.497774247415078,
      "learning_rate": 4.07148471951363e-06,
      "loss": 0.5984,
      "step": 4474
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.398691756909102,
      "learning_rate": 4.069447748103621e-06,
      "loss": 0.2042,
      "step": 4475
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6448332262371979,
      "learning_rate": 4.067410936666335e-06,
      "loss": 0.2345,
      "step": 4476
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5161892618320456,
      "learning_rate": 4.065374285551917e-06,
      "loss": 0.2023,
      "step": 4477
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5204691575255203,
      "learning_rate": 4.063337795110497e-06,
      "loss": 0.2272,
      "step": 4478
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.93668812693943,
      "learning_rate": 4.0613014656921675e-06,
      "loss": 0.4246,
      "step": 4479
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4139627409116406,
      "learning_rate": 4.059265297646999e-06,
      "loss": 0.2124,
      "step": 4480
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3697683276875725,
      "learning_rate": 4.057229291325032e-06,
      "loss": 0.2382,
      "step": 4481
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5081390441324596,
      "learning_rate": 4.05519344707628e-06,
      "loss": 0.1945,
      "step": 4482
    },
    {
      "epoch": 0.57,
      "grad_norm": 8.738755379390282,
      "learning_rate": 4.0531577652507256e-06,
      "loss": 0.5938,
      "step": 4483
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5307367989544212,
      "learning_rate": 4.05112224619833e-06,
      "loss": 0.2037,
      "step": 4484
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4216282533090923,
      "learning_rate": 4.049086890269021e-06,
      "loss": 0.1614,
      "step": 4485
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2425352558149092,
      "learning_rate": 4.047051697812697e-06,
      "loss": 0.1562,
      "step": 4486
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.6364729991507225,
      "learning_rate": 4.0450166691792335e-06,
      "loss": 0.3827,
      "step": 4487
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5233602705601772,
      "learning_rate": 4.042981804718478e-06,
      "loss": 0.1959,
      "step": 4488
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1679737413735043,
      "learning_rate": 4.040947104780244e-06,
      "loss": 0.181,
      "step": 4489
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.551418021012274,
      "learning_rate": 4.03891256971432e-06,
      "loss": 0.5347,
      "step": 4490
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5035701119766087,
      "learning_rate": 4.036878199870469e-06,
      "loss": 0.2141,
      "step": 4491
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2881632880227802,
      "learning_rate": 4.034843995598421e-06,
      "loss": 0.1834,
      "step": 4492
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2061302933835443,
      "learning_rate": 4.032809957247878e-06,
      "loss": 0.1381,
      "step": 4493
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2931628779560143,
      "learning_rate": 4.030776085168516e-06,
      "loss": 0.1973,
      "step": 4494
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.403907667643826,
      "learning_rate": 4.028742379709982e-06,
      "loss": 0.2051,
      "step": 4495
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.617843889024766,
      "learning_rate": 4.0267088412218906e-06,
      "loss": 0.189,
      "step": 4496
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3315275713729942,
      "learning_rate": 4.024675470053836e-06,
      "loss": 0.2027,
      "step": 4497
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.502219059500666,
      "learning_rate": 4.0226422665553724e-06,
      "loss": 0.1968,
      "step": 4498
    },
    {
      "epoch": 0.58,
      "grad_norm": 8.571386445694376,
      "learning_rate": 4.020609231076035e-06,
      "loss": 0.7396,
      "step": 4499
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.923161259095488,
      "learning_rate": 4.018576363965324e-06,
      "loss": 0.611,
      "step": 4500
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.89432985770311,
      "learning_rate": 4.0165436655727144e-06,
      "loss": 0.6047,
      "step": 4501
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4196882778939344,
      "learning_rate": 4.01451113624765e-06,
      "loss": 0.1961,
      "step": 4502
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.528794909225824,
      "learning_rate": 4.012478776339547e-06,
      "loss": 0.7011,
      "step": 4503
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3366972377870094,
      "learning_rate": 4.010446586197788e-06,
      "loss": 0.1951,
      "step": 4504
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4731451415751944,
      "learning_rate": 4.008414566171733e-06,
      "loss": 0.1919,
      "step": 4505
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.321018133848184,
      "learning_rate": 4.006382716610711e-06,
      "loss": 0.5857,
      "step": 4506
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3398574772083014,
      "learning_rate": 4.00435103786402e-06,
      "loss": 0.1932,
      "step": 4507
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.602401252168323,
      "learning_rate": 4.002319530280925e-06,
      "loss": 0.2196,
      "step": 4508
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6315061876622081,
      "learning_rate": 4.000288194210671e-06,
      "loss": 0.1783,
      "step": 4509
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.250257389376554,
      "learning_rate": 3.998257030002466e-06,
      "loss": 0.171,
      "step": 4510
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4617238322975408,
      "learning_rate": 3.996226038005491e-06,
      "loss": 0.2231,
      "step": 4511
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3803309288600387,
      "learning_rate": 3.994195218568896e-06,
      "loss": 0.2117,
      "step": 4512
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.242643616772911,
      "learning_rate": 3.992164572041805e-06,
      "loss": 0.1818,
      "step": 4513
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2672272484666356,
      "learning_rate": 3.990134098773304e-06,
      "loss": 0.1578,
      "step": 4514
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.473648355601473,
      "learning_rate": 3.988103799112461e-06,
      "loss": 0.1965,
      "step": 4515
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5808008504939228,
      "learning_rate": 3.986073673408306e-06,
      "loss": 0.2393,
      "step": 4516
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3928088896923487,
      "learning_rate": 3.984043722009842e-06,
      "loss": 0.1534,
      "step": 4517
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3669453388999566,
      "learning_rate": 3.98201394526604e-06,
      "loss": 0.2195,
      "step": 4518
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4866073402998417,
      "learning_rate": 3.97998434352584e-06,
      "loss": 0.2015,
      "step": 4519
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2618016497886706,
      "learning_rate": 3.977954917138158e-06,
      "loss": 0.158,
      "step": 4520
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2999010863031824,
      "learning_rate": 3.975925666451875e-06,
      "loss": 0.182,
      "step": 4521
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.824801086539275,
      "learning_rate": 3.973896591815841e-06,
      "loss": 0.477,
      "step": 4522
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3059903125037284,
      "learning_rate": 3.971867693578879e-06,
      "loss": 0.2051,
      "step": 4523
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3984275650950577,
      "learning_rate": 3.969838972089778e-06,
      "loss": 0.2137,
      "step": 4524
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4290913515731056,
      "learning_rate": 3.967810427697301e-06,
      "loss": 0.1706,
      "step": 4525
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4328036443191663,
      "learning_rate": 3.965782060750178e-06,
      "loss": 0.2385,
      "step": 4526
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3383486676668659,
      "learning_rate": 3.96375387159711e-06,
      "loss": 0.1883,
      "step": 4527
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3384851663636181,
      "learning_rate": 3.961725860586763e-06,
      "loss": 0.1615,
      "step": 4528
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3382293183117608,
      "learning_rate": 3.9596980280677775e-06,
      "loss": 0.1672,
      "step": 4529
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6575898019863697,
      "learning_rate": 3.957670374388762e-06,
      "loss": 0.1963,
      "step": 4530
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3597188360662278,
      "learning_rate": 3.955642899898293e-06,
      "loss": 0.2251,
      "step": 4531
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4359353704162692,
      "learning_rate": 3.953615604944915e-06,
      "loss": 0.1978,
      "step": 4532
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.501259424366729,
      "learning_rate": 3.9515884898771455e-06,
      "loss": 0.6307,
      "step": 4533
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2200443217264247,
      "learning_rate": 3.949561555043469e-06,
      "loss": 0.1333,
      "step": 4534
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.449433448710725,
      "learning_rate": 3.94753480079234e-06,
      "loss": 0.2135,
      "step": 4535
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5634755323100242,
      "learning_rate": 3.94550822747218e-06,
      "loss": 0.2506,
      "step": 4536
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.496424901348576,
      "learning_rate": 3.943481835431379e-06,
      "loss": 0.5574,
      "step": 4537
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2274896868924168,
      "learning_rate": 3.9414556250183e-06,
      "loss": 0.1575,
      "step": 4538
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3774140235713934,
      "learning_rate": 3.939429596581271e-06,
      "loss": 0.2047,
      "step": 4539
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.536062199318567,
      "learning_rate": 3.937403750468588e-06,
      "loss": 0.23,
      "step": 4540
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5275552156226002,
      "learning_rate": 3.935378087028521e-06,
      "loss": 0.2227,
      "step": 4541
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5048754687871904,
      "learning_rate": 3.9333526066093e-06,
      "loss": 0.2065,
      "step": 4542
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.503506390256207,
      "learning_rate": 3.931327309559133e-06,
      "loss": 0.2171,
      "step": 4543
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.57345393310056,
      "learning_rate": 3.9293021962261906e-06,
      "loss": 0.221,
      "step": 4544
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.386585934282675,
      "learning_rate": 3.927277266958614e-06,
      "loss": 0.2039,
      "step": 4545
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4518240017882449,
      "learning_rate": 3.925252522104512e-06,
      "loss": 0.1685,
      "step": 4546
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.483089640730275,
      "learning_rate": 3.923227962011959e-06,
      "loss": 0.5748,
      "step": 4547
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4798382676165427,
      "learning_rate": 3.9212035870290035e-06,
      "loss": 0.1725,
      "step": 4548
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4006198683630138,
      "learning_rate": 3.919179397503659e-06,
      "loss": 0.1954,
      "step": 4549
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.519414946164785,
      "learning_rate": 3.917155393783903e-06,
      "loss": 0.2171,
      "step": 4550
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2816607530142945,
      "learning_rate": 3.9151315762176885e-06,
      "loss": 0.2134,
      "step": 4551
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3633899588809104,
      "learning_rate": 3.9131079451529345e-06,
      "loss": 0.1914,
      "step": 4552
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4143354977748828,
      "learning_rate": 3.9110845009375255e-06,
      "loss": 0.2023,
      "step": 4553
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4586752284458318,
      "learning_rate": 3.909061243919313e-06,
      "loss": 0.2129,
      "step": 4554
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3157940039632576,
      "learning_rate": 3.907038174446122e-06,
      "loss": 0.1896,
      "step": 4555
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3996502519588443,
      "learning_rate": 3.905015292865738e-06,
      "loss": 0.1739,
      "step": 4556
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6022490586496385,
      "learning_rate": 3.902992599525921e-06,
      "loss": 0.2233,
      "step": 4557
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2568610294223648,
      "learning_rate": 3.900970094774391e-06,
      "loss": 0.2102,
      "step": 4558
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4560103652108225,
      "learning_rate": 3.898947778958845e-06,
      "loss": 0.2013,
      "step": 4559
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.773044882391762,
      "learning_rate": 3.8969256524269395e-06,
      "loss": 0.5355,
      "step": 4560
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2850644030998455,
      "learning_rate": 3.894903715526301e-06,
      "loss": 0.1712,
      "step": 4561
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.920741551713909,
      "learning_rate": 3.892881968604525e-06,
      "loss": 0.5812,
      "step": 4562
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3117896100654285,
      "learning_rate": 3.890860412009176e-06,
      "loss": 0.2051,
      "step": 4563
    },
    {
      "epoch": 0.58,
      "grad_norm": 8.259656919830064,
      "learning_rate": 3.888839046087779e-06,
      "loss": 0.5603,
      "step": 4564
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4636064382240674,
      "learning_rate": 3.886817871187832e-06,
      "loss": 0.1983,
      "step": 4565
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.343491738245714,
      "learning_rate": 3.8847968876567985e-06,
      "loss": 0.5333,
      "step": 4566
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.854370352657451,
      "learning_rate": 3.882776095842108e-06,
      "loss": 0.5514,
      "step": 4567
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4225045044983606,
      "learning_rate": 3.880755496091158e-06,
      "loss": 0.2099,
      "step": 4568
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5198312956131008,
      "learning_rate": 3.878735088751315e-06,
      "loss": 0.2011,
      "step": 4569
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.316455356950858,
      "learning_rate": 3.876714874169906e-06,
      "loss": 0.2024,
      "step": 4570
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4499787407170284,
      "learning_rate": 3.874694852694236e-06,
      "loss": 0.1869,
      "step": 4571
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3638145863144076,
      "learning_rate": 3.8726750246715635e-06,
      "loss": 0.1588,
      "step": 4572
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4324422490743811,
      "learning_rate": 3.870655390449126e-06,
      "loss": 0.2393,
      "step": 4573
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4661640669530616,
      "learning_rate": 3.868635950374119e-06,
      "loss": 0.2301,
      "step": 4574
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3515416461776175,
      "learning_rate": 3.866616704793706e-06,
      "loss": 0.2008,
      "step": 4575
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3865674323325912,
      "learning_rate": 3.864597654055024e-06,
      "loss": 0.1731,
      "step": 4576
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3482823802143777,
      "learning_rate": 3.862578798505166e-06,
      "loss": 0.2346,
      "step": 4577
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5977188008001393,
      "learning_rate": 3.860560138491199e-06,
      "loss": 0.1801,
      "step": 4578
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.505007740728695,
      "learning_rate": 3.858541674360151e-06,
      "loss": 0.1989,
      "step": 4579
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4483656914459537,
      "learning_rate": 3.856523406459025e-06,
      "loss": 0.2086,
      "step": 4580
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4087510893932473,
      "learning_rate": 3.854505335134781e-06,
      "loss": 0.2083,
      "step": 4581
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4057065468419707,
      "learning_rate": 3.852487460734351e-06,
      "loss": 0.2075,
      "step": 4582
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4158457874062897,
      "learning_rate": 3.850469783604628e-06,
      "loss": 0.1883,
      "step": 4583
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.455478755651067,
      "learning_rate": 3.848452304092477e-06,
      "loss": 0.2305,
      "step": 4584
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3701176945667313,
      "learning_rate": 3.846435022544724e-06,
      "loss": 0.1796,
      "step": 4585
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3967454342452479,
      "learning_rate": 3.844417939308163e-06,
      "loss": 0.183,
      "step": 4586
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.600884499087193,
      "learning_rate": 3.842401054729557e-06,
      "loss": 0.6261,
      "step": 4587
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4421599487281787,
      "learning_rate": 3.840384369155626e-06,
      "loss": 0.2002,
      "step": 4588
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4227020655534168,
      "learning_rate": 3.838367882933068e-06,
      "loss": 0.1702,
      "step": 4589
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.333122797919476,
      "learning_rate": 3.836351596408537e-06,
      "loss": 0.4737,
      "step": 4590
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.532807781891341,
      "learning_rate": 3.834335509928658e-06,
      "loss": 0.1982,
      "step": 4591
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3799998981033414,
      "learning_rate": 3.832319623840018e-06,
      "loss": 0.2187,
      "step": 4592
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3046554757764135,
      "learning_rate": 3.830303938489172e-06,
      "loss": 0.1799,
      "step": 4593
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6115009403057916,
      "learning_rate": 3.8282884542226385e-06,
      "loss": 0.209,
      "step": 4594
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4315759574400895,
      "learning_rate": 3.826273171386904e-06,
      "loss": 0.2018,
      "step": 4595
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.648327144641748,
      "learning_rate": 3.8242580903284186e-06,
      "loss": 0.6141,
      "step": 4596
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4504934559679004,
      "learning_rate": 3.822243211393596e-06,
      "loss": 0.2646,
      "step": 4597
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1639079134667158,
      "learning_rate": 3.820228534928818e-06,
      "loss": 0.1799,
      "step": 4598
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2247288333830828,
      "learning_rate": 3.818214061280435e-06,
      "loss": 0.1257,
      "step": 4599
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.402072201501487,
      "learning_rate": 3.816199790794754e-06,
      "loss": 0.1788,
      "step": 4600
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.157542724136976,
      "learning_rate": 3.8141857238180537e-06,
      "loss": 0.5145,
      "step": 4601
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1702949056968455,
      "learning_rate": 3.812171860696574e-06,
      "loss": 0.1234,
      "step": 4602
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.442329352342721,
      "learning_rate": 3.810158201776523e-06,
      "loss": 0.213,
      "step": 4603
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4256666403661278,
      "learning_rate": 3.808144747404069e-06,
      "loss": 0.176,
      "step": 4604
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4723884745422675,
      "learning_rate": 3.806131497925351e-06,
      "loss": 0.2192,
      "step": 4605
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3913577107934443,
      "learning_rate": 3.8041184536864694e-06,
      "loss": 0.1862,
      "step": 4606
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4480580151065932,
      "learning_rate": 3.802105615033488e-06,
      "loss": 0.2316,
      "step": 4607
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4341025377727314,
      "learning_rate": 3.8000929823124387e-06,
      "loss": 0.2292,
      "step": 4608
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.846760345041472,
      "learning_rate": 3.798080555869318e-06,
      "loss": 0.5184,
      "step": 4609
    },
    {
      "epoch": 0.59,
      "grad_norm": 11.42481585949183,
      "learning_rate": 3.796068336050083e-06,
      "loss": 0.6344,
      "step": 4610
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.250892857256506,
      "learning_rate": 3.7940563232006573e-06,
      "loss": 0.4814,
      "step": 4611
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.481175318242117,
      "learning_rate": 3.792044517666931e-06,
      "loss": 0.1638,
      "step": 4612
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3470275574986341,
      "learning_rate": 3.790032919794757e-06,
      "loss": 0.1654,
      "step": 4613
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4313409524553153,
      "learning_rate": 3.788021529929949e-06,
      "loss": 0.2221,
      "step": 4614
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3385029622403084,
      "learning_rate": 3.786010348418292e-06,
      "loss": 0.1781,
      "step": 4615
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.375759990318493,
      "learning_rate": 3.783999375605527e-06,
      "loss": 0.5893,
      "step": 4616
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5565329130044154,
      "learning_rate": 3.7819886118373694e-06,
      "loss": 0.2544,
      "step": 4617
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.806441103843029,
      "learning_rate": 3.77997805745949e-06,
      "loss": 0.6444,
      "step": 4618
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3965711837159591,
      "learning_rate": 3.7779677128175247e-06,
      "loss": 0.2272,
      "step": 4619
    },
    {
      "epoch": 0.59,
      "grad_norm": 9.401877818142802,
      "learning_rate": 3.775957578257078e-06,
      "loss": 0.4604,
      "step": 4620
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3783567692268428,
      "learning_rate": 3.7739476541237137e-06,
      "loss": 0.192,
      "step": 4621
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.358813223318587,
      "learning_rate": 3.7719379407629597e-06,
      "loss": 0.1648,
      "step": 4622
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.515483630374196,
      "learning_rate": 3.769928438520312e-06,
      "loss": 0.6291,
      "step": 4623
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3657681010478813,
      "learning_rate": 3.7679191477412263e-06,
      "loss": 0.2139,
      "step": 4624
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.6602790628455155,
      "learning_rate": 3.7659100687711205e-06,
      "loss": 0.5649,
      "step": 4625
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.999353730348896,
      "learning_rate": 3.7639012019553813e-06,
      "loss": 0.5976,
      "step": 4626
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4452997749802166,
      "learning_rate": 3.761892547639357e-06,
      "loss": 0.2301,
      "step": 4627
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5536009583790151,
      "learning_rate": 3.7598841061683566e-06,
      "loss": 0.1828,
      "step": 4628
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.966161266207984,
      "learning_rate": 3.7578758778876546e-06,
      "loss": 0.5742,
      "step": 4629
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3062413335699838,
      "learning_rate": 3.75586786314249e-06,
      "loss": 0.1522,
      "step": 4630
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5182669124995618,
      "learning_rate": 3.753860062278063e-06,
      "loss": 0.212,
      "step": 4631
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4386752385374362,
      "learning_rate": 3.7518524756395362e-06,
      "loss": 0.1722,
      "step": 4632
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4428109777072258,
      "learning_rate": 3.74984510357204e-06,
      "loss": 0.1523,
      "step": 4633
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.550149622344002,
      "learning_rate": 3.747837946420664e-06,
      "loss": 0.2138,
      "step": 4634
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2347999597128212,
      "learning_rate": 3.7458310045304586e-06,
      "loss": 0.1887,
      "step": 4635
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4245307282780537,
      "learning_rate": 3.7438242782464453e-06,
      "loss": 0.1642,
      "step": 4636
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2994434561685893,
      "learning_rate": 3.7418177679136005e-06,
      "loss": 0.1333,
      "step": 4637
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3821897570774595,
      "learning_rate": 3.7398114738768686e-06,
      "loss": 0.1932,
      "step": 4638
    },
    {
      "epoch": 0.59,
      "grad_norm": 9.33110583549079,
      "learning_rate": 3.7378053964811534e-06,
      "loss": 0.4177,
      "step": 4639
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.411283773395314,
      "learning_rate": 3.735799536071322e-06,
      "loss": 0.1915,
      "step": 4640
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5665774042641445,
      "learning_rate": 3.733793892992208e-06,
      "loss": 0.1737,
      "step": 4641
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.246163482585636,
      "learning_rate": 3.7317884675886025e-06,
      "loss": 0.1681,
      "step": 4642
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4144784713848304,
      "learning_rate": 3.7297832602052608e-06,
      "loss": 0.1935,
      "step": 4643
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5913643403691975,
      "learning_rate": 3.727778271186901e-06,
      "loss": 0.2075,
      "step": 4644
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6080469153991046,
      "learning_rate": 3.7257735008782076e-06,
      "loss": 0.2323,
      "step": 4645
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4582776834315876,
      "learning_rate": 3.7237689496238217e-06,
      "loss": 0.2255,
      "step": 4646
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.400700014688477,
      "learning_rate": 3.721764617768347e-06,
      "loss": 0.6223,
      "step": 4647
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.450621785070563,
      "learning_rate": 3.7197605056563545e-06,
      "loss": 0.1861,
      "step": 4648
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2799421557876016,
      "learning_rate": 3.7177566136323726e-06,
      "loss": 0.1583,
      "step": 4649
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3827753594289258,
      "learning_rate": 3.7157529420408933e-06,
      "loss": 0.2269,
      "step": 4650
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.868160090018921,
      "learning_rate": 3.7137494912263723e-06,
      "loss": 0.4919,
      "step": 4651
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5135859070602167,
      "learning_rate": 3.7117462615332254e-06,
      "loss": 0.228,
      "step": 4652
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.512320564574722,
      "learning_rate": 3.7097432533058288e-06,
      "loss": 0.2056,
      "step": 4653
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3058833663221732,
      "learning_rate": 3.7077404668885263e-06,
      "loss": 0.1706,
      "step": 4654
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.499650137140572,
      "learning_rate": 3.7057379026256185e-06,
      "loss": 0.2308,
      "step": 4655
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5243144041892105,
      "learning_rate": 3.7037355608613703e-06,
      "loss": 0.2101,
      "step": 4656
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5482471774288624,
      "learning_rate": 3.701733441940007e-06,
      "loss": 0.2295,
      "step": 4657
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.002668479098118,
      "learning_rate": 3.699731546205715e-06,
      "loss": 0.611,
      "step": 4658
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4264425564041927,
      "learning_rate": 3.6977298740026448e-06,
      "loss": 0.1709,
      "step": 4659
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.540768327493481,
      "learning_rate": 3.695728425674906e-06,
      "loss": 0.2088,
      "step": 4660
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.195708999451657,
      "learning_rate": 3.693727201566571e-06,
      "loss": 0.1616,
      "step": 4661
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4497494906632251,
      "learning_rate": 3.6917262020216727e-06,
      "loss": 0.2403,
      "step": 4662
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3711630281788822,
      "learning_rate": 3.6897254273842087e-06,
      "loss": 0.1616,
      "step": 4663
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.411325896837784,
      "learning_rate": 3.6877248779981332e-06,
      "loss": 0.5686,
      "step": 4664
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4655886203170425,
      "learning_rate": 3.685724554207365e-06,
      "loss": 0.2005,
      "step": 4665
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.017701295211014,
      "learning_rate": 3.6837244563557815e-06,
      "loss": 0.6497,
      "step": 4666
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2816528024359644,
      "learning_rate": 3.6817245847872253e-06,
      "loss": 0.177,
      "step": 4667
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.86186844110466,
      "learning_rate": 3.679724939845494e-06,
      "loss": 0.5149,
      "step": 4668
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5119347933872427,
      "learning_rate": 3.677725521874352e-06,
      "loss": 0.2199,
      "step": 4669
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.401842926226571,
      "learning_rate": 3.6757263312175216e-06,
      "loss": 0.1965,
      "step": 4670
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3127986244083745,
      "learning_rate": 3.673727368218687e-06,
      "loss": 0.1856,
      "step": 4671
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.702132750487339,
      "learning_rate": 3.671728633221493e-06,
      "loss": 0.201,
      "step": 4672
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4180552671529532,
      "learning_rate": 3.669730126569546e-06,
      "loss": 0.1803,
      "step": 4673
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.212201286433844,
      "learning_rate": 3.667731848606413e-06,
      "loss": 0.1443,
      "step": 4674
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.438619073932616,
      "learning_rate": 3.665733799675619e-06,
      "loss": 0.2123,
      "step": 4675
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2431728926845063,
      "learning_rate": 3.663735980120653e-06,
      "loss": 0.1698,
      "step": 4676
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.270616154745865,
      "learning_rate": 3.6617383902849645e-06,
      "loss": 0.1691,
      "step": 4677
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4839480796790563,
      "learning_rate": 3.6597410305119605e-06,
      "loss": 0.194,
      "step": 4678
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4620107846396457,
      "learning_rate": 3.6577439011450112e-06,
      "loss": 0.2019,
      "step": 4679
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.474720713107648,
      "learning_rate": 3.6557470025274453e-06,
      "loss": 0.1769,
      "step": 4680
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2271572713916858,
      "learning_rate": 3.6537503350025525e-06,
      "loss": 0.1855,
      "step": 4681
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4016105840241977,
      "learning_rate": 3.6517538989135866e-06,
      "loss": 0.2211,
      "step": 4682
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4301968762954314,
      "learning_rate": 3.649757694603754e-06,
      "loss": 0.2467,
      "step": 4683
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3391220210880368,
      "learning_rate": 3.647761722416229e-06,
      "loss": 0.1872,
      "step": 4684
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3571790829999388,
      "learning_rate": 3.645765982694139e-06,
      "loss": 0.2095,
      "step": 4685
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.446074021526168,
      "learning_rate": 3.643770475780576e-06,
      "loss": 0.2035,
      "step": 4686
    },
    {
      "epoch": 0.6,
      "grad_norm": 15.811928203593629,
      "learning_rate": 3.641775202018592e-06,
      "loss": 0.5962,
      "step": 4687
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.260792428042919,
      "learning_rate": 3.6397801617511965e-06,
      "loss": 0.2095,
      "step": 4688
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2224363527711402,
      "learning_rate": 3.6377853553213593e-06,
      "loss": 0.1914,
      "step": 4689
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5396593117798096,
      "learning_rate": 3.635790783072012e-06,
      "loss": 0.2311,
      "step": 4690
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4785288941284942,
      "learning_rate": 3.633796445346044e-06,
      "loss": 0.2023,
      "step": 4691
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3082611099415573,
      "learning_rate": 3.6318023424863057e-06,
      "loss": 0.2136,
      "step": 4692
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.104412735560672,
      "learning_rate": 3.6298084748356077e-06,
      "loss": 0.6779,
      "step": 4693
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5886062640755068,
      "learning_rate": 3.6278148427367154e-06,
      "loss": 0.2442,
      "step": 4694
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.068117294072581,
      "learning_rate": 3.6258214465323604e-06,
      "loss": 0.5806,
      "step": 4695
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8062065114643706,
      "learning_rate": 3.6238282865652304e-06,
      "loss": 0.2015,
      "step": 4696
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.432416203844863,
      "learning_rate": 3.62183536317797e-06,
      "loss": 0.2039,
      "step": 4697
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6448950502189177,
      "learning_rate": 3.6198426767131893e-06,
      "loss": 0.2404,
      "step": 4698
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2795396669655106,
      "learning_rate": 3.6178502275134507e-06,
      "loss": 0.1607,
      "step": 4699
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.438580459891171,
      "learning_rate": 3.6158580159212833e-06,
      "loss": 0.5979,
      "step": 4700
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4486746063478324,
      "learning_rate": 3.6138660422791673e-06,
      "loss": 0.1544,
      "step": 4701
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3047516357443583,
      "learning_rate": 3.6118743069295503e-06,
      "loss": 0.1552,
      "step": 4702
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3632451232595786,
      "learning_rate": 3.609882810214832e-06,
      "loss": 0.1854,
      "step": 4703
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4638152853755726,
      "learning_rate": 3.6078915524773726e-06,
      "loss": 0.1789,
      "step": 4704
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3364805138168272,
      "learning_rate": 3.605900534059496e-06,
      "loss": 0.2157,
      "step": 4705
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1888923818239092,
      "learning_rate": 3.603909755303479e-06,
      "loss": 0.1571,
      "step": 4706
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3688067261272832,
      "learning_rate": 3.6019192165515595e-06,
      "loss": 0.1954,
      "step": 4707
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4140139573693313,
      "learning_rate": 3.5999289181459346e-06,
      "loss": 0.1672,
      "step": 4708
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3803998343544934,
      "learning_rate": 3.597938860428758e-06,
      "loss": 0.2415,
      "step": 4709
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.441835565878244,
      "learning_rate": 3.5959490437421473e-06,
      "loss": 0.1877,
      "step": 4710
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3751809456283108,
      "learning_rate": 3.5939594684281736e-06,
      "loss": 0.1788,
      "step": 4711
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.606995004446463,
      "learning_rate": 3.591970134828866e-06,
      "loss": 0.2233,
      "step": 4712
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6219701382539304,
      "learning_rate": 3.589981043286217e-06,
      "loss": 0.2481,
      "step": 4713
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4308337942968126,
      "learning_rate": 3.587992194142173e-06,
      "loss": 0.2185,
      "step": 4714
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.885463264146404,
      "learning_rate": 3.586003587738639e-06,
      "loss": 0.5717,
      "step": 4715
    },
    {
      "epoch": 0.6,
      "grad_norm": 9.624032969778115,
      "learning_rate": 3.584015224417482e-06,
      "loss": 0.6052,
      "step": 4716
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8339506581908402,
      "learning_rate": 3.5820271045205227e-06,
      "loss": 0.2382,
      "step": 4717
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3350917473378165,
      "learning_rate": 3.580039228389541e-06,
      "loss": 0.1885,
      "step": 4718
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2999231492915324,
      "learning_rate": 3.578051596366279e-06,
      "loss": 0.1983,
      "step": 4719
    },
    {
      "epoch": 0.6,
      "grad_norm": 8.021376863208443,
      "learning_rate": 3.576064208792433e-06,
      "loss": 0.4336,
      "step": 4720
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5551419194961762,
      "learning_rate": 3.574077066009657e-06,
      "loss": 0.1981,
      "step": 4721
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3685575982627032,
      "learning_rate": 3.5720901683595633e-06,
      "loss": 0.2115,
      "step": 4722
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.114709760939068,
      "learning_rate": 3.570103516183724e-06,
      "loss": 0.71,
      "step": 4723
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5013330466790846,
      "learning_rate": 3.568117109823668e-06,
      "loss": 0.2059,
      "step": 4724
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6078236927204446,
      "learning_rate": 3.5661309496208785e-06,
      "loss": 0.1849,
      "step": 4725
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3379166987582618,
      "learning_rate": 3.564145035916803e-06,
      "loss": 0.1766,
      "step": 4726
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2667558876299774,
      "learning_rate": 3.56215936905284e-06,
      "loss": 0.1485,
      "step": 4727
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3594319984845957,
      "learning_rate": 3.5601739493703517e-06,
      "loss": 0.1927,
      "step": 4728
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5106554981435416,
      "learning_rate": 3.5581887772106536e-06,
      "loss": 0.2124,
      "step": 4729
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3510689938837455,
      "learning_rate": 3.5562038529150187e-06,
      "loss": 0.1418,
      "step": 4730
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4080780341967871,
      "learning_rate": 3.554219176824681e-06,
      "loss": 0.1899,
      "step": 4731
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3812123787173962,
      "learning_rate": 3.5522347492808272e-06,
      "loss": 0.1691,
      "step": 4732
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4573879569798465,
      "learning_rate": 3.5502505706246027e-06,
      "loss": 0.2164,
      "step": 4733
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2946876548143342,
      "learning_rate": 3.5482666411971134e-06,
      "loss": 0.1737,
      "step": 4734
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.545605871131214,
      "learning_rate": 3.5462829613394186e-06,
      "loss": 0.2176,
      "step": 4735
    },
    {
      "epoch": 0.61,
      "grad_norm": 9.279162972387478,
      "learning_rate": 3.544299531392533e-06,
      "loss": 0.6888,
      "step": 4736
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3262758850244563,
      "learning_rate": 3.5423163516974356e-06,
      "loss": 0.1894,
      "step": 4737
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6523000463330308,
      "learning_rate": 3.540333422595057e-06,
      "loss": 0.2173,
      "step": 4738
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.07380860811021,
      "learning_rate": 3.538350744426285e-06,
      "loss": 0.4397,
      "step": 4739
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3526754961413883,
      "learning_rate": 3.5363683175319637e-06,
      "loss": 0.2328,
      "step": 4740
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3940417656976654,
      "learning_rate": 3.5343861422528973e-06,
      "loss": 0.1747,
      "step": 4741
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5975005661429964,
      "learning_rate": 3.5324042189298445e-06,
      "loss": 0.2041,
      "step": 4742
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3980133418907261,
      "learning_rate": 3.5304225479035193e-06,
      "loss": 0.1945,
      "step": 4743
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3384195670483974,
      "learning_rate": 3.5284411295145954e-06,
      "loss": 0.2082,
      "step": 4744
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.368392020779055,
      "learning_rate": 3.526459964103701e-06,
      "loss": 0.6213,
      "step": 4745
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.324548815840477,
      "learning_rate": 3.5244790520114202e-06,
      "loss": 0.513,
      "step": 4746
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4128291020507633,
      "learning_rate": 3.5224983935782975e-06,
      "loss": 0.1949,
      "step": 4747
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2474886186645897,
      "learning_rate": 3.5205179891448294e-06,
      "loss": 0.1986,
      "step": 4748
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.518592927106811,
      "learning_rate": 3.518537839051471e-06,
      "loss": 0.2296,
      "step": 4749
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.352867540031532,
      "learning_rate": 3.5165579436386323e-06,
      "loss": 0.2147,
      "step": 4750
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3926528348270288,
      "learning_rate": 3.51457830324668e-06,
      "loss": 0.1951,
      "step": 4751
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3039581475270545,
      "learning_rate": 3.5125989182159393e-06,
      "loss": 0.1905,
      "step": 4752
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3846769724782073,
      "learning_rate": 3.5106197888866873e-06,
      "loss": 0.1627,
      "step": 4753
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2004343341005803,
      "learning_rate": 3.508640915599159e-06,
      "loss": 0.1609,
      "step": 4754
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.521484182956767,
      "learning_rate": 3.506662298693546e-06,
      "loss": 0.4227,
      "step": 4755
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4564540369656518,
      "learning_rate": 3.5046839385099977e-06,
      "loss": 0.1725,
      "step": 4756
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.315218693364636,
      "learning_rate": 3.502705835388616e-06,
      "loss": 0.1844,
      "step": 4757
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4373598037916384,
      "learning_rate": 3.500727989669458e-06,
      "loss": 0.1859,
      "step": 4758
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.0998854292751625,
      "learning_rate": 3.498750401692541e-06,
      "loss": 0.6128,
      "step": 4759
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4456972322063395,
      "learning_rate": 3.496773071797834e-06,
      "loss": 0.2122,
      "step": 4760
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4921370215626488,
      "learning_rate": 3.4947960003252614e-06,
      "loss": 0.2062,
      "step": 4761
    },
    {
      "epoch": 0.61,
      "grad_norm": 13.224657442978836,
      "learning_rate": 3.492819187614707e-06,
      "loss": 0.6,
      "step": 4762
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1793254808465403,
      "learning_rate": 3.4908426340060075e-06,
      "loss": 0.1408,
      "step": 4763
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.444196111148455,
      "learning_rate": 3.488866339838953e-06,
      "loss": 0.171,
      "step": 4764
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.458413945187006,
      "learning_rate": 3.4868903054532934e-06,
      "loss": 0.1845,
      "step": 4765
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3816144822742016,
      "learning_rate": 3.4849145311887335e-06,
      "loss": 0.1815,
      "step": 4766
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3468523602583973,
      "learning_rate": 3.4829390173849296e-06,
      "loss": 0.179,
      "step": 4767
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.4722195399940965,
      "learning_rate": 3.4809637643814965e-06,
      "loss": 0.7501,
      "step": 4768
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2842429081001716,
      "learning_rate": 3.478988772518001e-06,
      "loss": 0.1795,
      "step": 4769
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5350924907356733,
      "learning_rate": 3.47701404213397e-06,
      "loss": 0.2138,
      "step": 4770
    },
    {
      "epoch": 0.61,
      "grad_norm": 8.909763923745647,
      "learning_rate": 3.475039573568881e-06,
      "loss": 0.534,
      "step": 4771
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2269505803066845,
      "learning_rate": 3.4730653671621667e-06,
      "loss": 0.1644,
      "step": 4772
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.304994566052046,
      "learning_rate": 3.4710914232532167e-06,
      "loss": 0.1732,
      "step": 4773
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3557700756560565,
      "learning_rate": 3.4691177421813783e-06,
      "loss": 0.1675,
      "step": 4774
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4675551389333734,
      "learning_rate": 3.4671443242859465e-06,
      "loss": 0.2059,
      "step": 4775
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5149543871973334,
      "learning_rate": 3.465171169906175e-06,
      "loss": 0.2193,
      "step": 4776
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6097198636534238,
      "learning_rate": 3.4631982793812745e-06,
      "loss": 0.1927,
      "step": 4777
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.53711291507394,
      "learning_rate": 3.461225653050405e-06,
      "loss": 0.1894,
      "step": 4778
    },
    {
      "epoch": 0.61,
      "grad_norm": 10.087540356789015,
      "learning_rate": 3.4592532912526845e-06,
      "loss": 0.5183,
      "step": 4779
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2793908171659907,
      "learning_rate": 3.4572811943271856e-06,
      "loss": 0.2352,
      "step": 4780
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4455337614674584,
      "learning_rate": 3.4553093626129343e-06,
      "loss": 0.2016,
      "step": 4781
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5209556387971155,
      "learning_rate": 3.45333779644891e-06,
      "loss": 0.1719,
      "step": 4782
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.419939863645464,
      "learning_rate": 3.451366496174048e-06,
      "loss": 0.5693,
      "step": 4783
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4128506004667223,
      "learning_rate": 3.4493954621272407e-06,
      "loss": 0.1782,
      "step": 4784
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5063868124402227,
      "learning_rate": 3.447424694647329e-06,
      "loss": 0.182,
      "step": 4785
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4883373205086232,
      "learning_rate": 3.445454194073111e-06,
      "loss": 0.1989,
      "step": 4786
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4250981423979647,
      "learning_rate": 3.4434839607433396e-06,
      "loss": 0.1727,
      "step": 4787
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2860236712063058,
      "learning_rate": 3.4415139949967203e-06,
      "loss": 0.1661,
      "step": 4788
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.457132789693538,
      "learning_rate": 3.439544297171913e-06,
      "loss": 0.1617,
      "step": 4789
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4193202109886813,
      "learning_rate": 3.437574867607529e-06,
      "loss": 0.1673,
      "step": 4790
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.868316073489745,
      "learning_rate": 3.435605706642141e-06,
      "loss": 0.4018,
      "step": 4791
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2279488273513366,
      "learning_rate": 3.4336368146142653e-06,
      "loss": 0.1685,
      "step": 4792
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.480703769010855,
      "learning_rate": 3.4316681918623825e-06,
      "loss": 0.2138,
      "step": 4793
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.394915991341191,
      "learning_rate": 3.4296998387249175e-06,
      "loss": 0.1909,
      "step": 4794
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2387411840891784,
      "learning_rate": 3.427731755540256e-06,
      "loss": 0.1768,
      "step": 4795
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5390812325703485,
      "learning_rate": 3.425763942646733e-06,
      "loss": 0.2221,
      "step": 4796
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.263449827368543,
      "learning_rate": 3.423796400382637e-06,
      "loss": 0.1736,
      "step": 4797
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4424734448483014,
      "learning_rate": 3.421829129086215e-06,
      "loss": 0.1913,
      "step": 4798
    },
    {
      "epoch": 0.61,
      "grad_norm": 9.948392054802266,
      "learning_rate": 3.4198621290956603e-06,
      "loss": 0.5984,
      "step": 4799
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2530984079690273,
      "learning_rate": 3.4178954007491237e-06,
      "loss": 0.1647,
      "step": 4800
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.446332487346318,
      "learning_rate": 3.4159289443847077e-06,
      "loss": 0.2103,
      "step": 4801
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.8662300079584995,
      "learning_rate": 3.4139627603404724e-06,
      "loss": 0.3694,
      "step": 4802
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.184233080027498,
      "learning_rate": 3.4119968489544254e-06,
      "loss": 0.5965,
      "step": 4803
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.516384549822969,
      "learning_rate": 3.4100312105645283e-06,
      "loss": 0.6255,
      "step": 4804
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4977762530738294,
      "learning_rate": 3.4080658455087e-06,
      "loss": 0.2019,
      "step": 4805
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.34533593557825,
      "learning_rate": 3.4061007541248093e-06,
      "loss": 0.1784,
      "step": 4806
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3640983348637774,
      "learning_rate": 3.404135936750674e-06,
      "loss": 0.2209,
      "step": 4807
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.313943244950753,
      "learning_rate": 3.4021713937240748e-06,
      "loss": 0.1759,
      "step": 4808
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.829425279706251,
      "learning_rate": 3.4002071253827356e-06,
      "loss": 0.6815,
      "step": 4809
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3185861306003048,
      "learning_rate": 3.398243132064336e-06,
      "loss": 0.1769,
      "step": 4810
    },
    {
      "epoch": 0.62,
      "grad_norm": 13.143066576862582,
      "learning_rate": 3.3962794141065136e-06,
      "loss": 0.5929,
      "step": 4811
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.48190066337099,
      "learning_rate": 3.3943159718468503e-06,
      "loss": 0.2033,
      "step": 4812
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4173888027120891,
      "learning_rate": 3.392352805622888e-06,
      "loss": 0.1578,
      "step": 4813
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.537335961958579,
      "learning_rate": 3.3903899157721156e-06,
      "loss": 0.5153,
      "step": 4814
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4348192275739324,
      "learning_rate": 3.388427302631975e-06,
      "loss": 0.2112,
      "step": 4815
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1605306044239685,
      "learning_rate": 3.386464966539865e-06,
      "loss": 0.1391,
      "step": 4816
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.7337839021691863,
      "learning_rate": 3.3845029078331344e-06,
      "loss": 0.6252,
      "step": 4817
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.295291989421047,
      "learning_rate": 3.3825411268490803e-06,
      "loss": 0.181,
      "step": 4818
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.106032957668024,
      "learning_rate": 3.380579623924959e-06,
      "loss": 0.5503,
      "step": 4819
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4665062821528143,
      "learning_rate": 3.378618399397972e-06,
      "loss": 0.1909,
      "step": 4820
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4458306776282923,
      "learning_rate": 3.3766574536052808e-06,
      "loss": 0.2204,
      "step": 4821
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.053724495470401,
      "learning_rate": 3.374696786883992e-06,
      "loss": 0.5853,
      "step": 4822
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2048302564664883,
      "learning_rate": 3.3727363995711695e-06,
      "loss": 0.1654,
      "step": 4823
    },
    {
      "epoch": 0.62,
      "grad_norm": 14.547161508532636,
      "learning_rate": 3.3707762920038235e-06,
      "loss": 0.6368,
      "step": 4824
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4034236265275886,
      "learning_rate": 3.3688164645189198e-06,
      "loss": 0.1906,
      "step": 4825
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.521183918186148,
      "learning_rate": 3.366856917453377e-06,
      "loss": 0.569,
      "step": 4826
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2405630646429027,
      "learning_rate": 3.3648976511440636e-06,
      "loss": 0.1634,
      "step": 4827
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5730885617760613,
      "learning_rate": 3.3629386659277984e-06,
      "loss": 0.2322,
      "step": 4828
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3555994240926204,
      "learning_rate": 3.3609799621413554e-06,
      "loss": 0.1836,
      "step": 4829
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4708597675922939,
      "learning_rate": 3.359021540121457e-06,
      "loss": 0.2178,
      "step": 4830
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4017747547283033,
      "learning_rate": 3.3570634002047815e-06,
      "loss": 0.1591,
      "step": 4831
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.897970145161382,
      "learning_rate": 3.355105542727954e-06,
      "loss": 0.5117,
      "step": 4832
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3177416350428894,
      "learning_rate": 3.353147968027552e-06,
      "loss": 0.1643,
      "step": 4833
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.231661438320789,
      "learning_rate": 3.351190676440107e-06,
      "loss": 0.1788,
      "step": 4834
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4110550786249885,
      "learning_rate": 3.3492336683021e-06,
      "loss": 0.2196,
      "step": 4835
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2693518397170898,
      "learning_rate": 3.347276943949961e-06,
      "loss": 0.1438,
      "step": 4836
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4267609098426088,
      "learning_rate": 3.3453205037200766e-06,
      "loss": 0.1931,
      "step": 4837
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2581238422695098,
      "learning_rate": 3.3433643479487777e-06,
      "loss": 0.1411,
      "step": 4838
    },
    {
      "epoch": 0.62,
      "grad_norm": 8.270175668652136,
      "learning_rate": 3.3414084769723554e-06,
      "loss": 0.5566,
      "step": 4839
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.405301501128341,
      "learning_rate": 3.339452891127043e-06,
      "loss": 0.571,
      "step": 4840
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3460879041581515,
      "learning_rate": 3.33749759074903e-06,
      "loss": 0.1665,
      "step": 4841
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3320395517569972,
      "learning_rate": 3.3355425761744553e-06,
      "loss": 0.1904,
      "step": 4842
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.500075944418699,
      "learning_rate": 3.3335878477394058e-06,
      "loss": 0.2131,
      "step": 4843
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.36240238705773,
      "learning_rate": 3.331633405779926e-06,
      "loss": 0.1955,
      "step": 4844
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4203797824187276,
      "learning_rate": 3.3296792506320043e-06,
      "loss": 0.1593,
      "step": 4845
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.521859033957712,
      "learning_rate": 3.3277253826315824e-06,
      "loss": 0.2023,
      "step": 4846
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.503572360842384,
      "learning_rate": 3.325771802114555e-06,
      "loss": 0.5754,
      "step": 4847
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.975324741846765,
      "learning_rate": 3.323818509416763e-06,
      "loss": 0.6352,
      "step": 4848
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5769859944567202,
      "learning_rate": 3.321865504874002e-06,
      "loss": 0.2201,
      "step": 4849
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.600370215067961,
      "learning_rate": 3.3199127888220162e-06,
      "loss": 0.228,
      "step": 4850
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3528588509334505,
      "learning_rate": 3.317960361596498e-06,
      "loss": 0.1654,
      "step": 4851
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7364685007273453,
      "learning_rate": 3.3160082235330937e-06,
      "loss": 0.2314,
      "step": 4852
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3721684149841191,
      "learning_rate": 3.3140563749673994e-06,
      "loss": 0.1732,
      "step": 4853
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3755338564462423,
      "learning_rate": 3.3121048162349577e-06,
      "loss": 0.2014,
      "step": 4854
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4305876293225779,
      "learning_rate": 3.3101535476712675e-06,
      "loss": 0.1687,
      "step": 4855
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.219668888924731,
      "learning_rate": 3.3082025696117723e-06,
      "loss": 0.1868,
      "step": 4856
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4259146369178248,
      "learning_rate": 3.3062518823918664e-06,
      "loss": 0.169,
      "step": 4857
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3230502918526674,
      "learning_rate": 3.3043014863468985e-06,
      "loss": 0.1477,
      "step": 4858
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4561973467871694,
      "learning_rate": 3.3023513818121645e-06,
      "loss": 0.206,
      "step": 4859
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5642780010992288,
      "learning_rate": 3.3004015691229086e-06,
      "loss": 0.234,
      "step": 4860
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.685222151297987,
      "learning_rate": 3.298452048614326e-06,
      "loss": 0.7027,
      "step": 4861
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5570667320243394,
      "learning_rate": 3.2965028206215627e-06,
      "loss": 0.188,
      "step": 4862
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.442847940576609,
      "learning_rate": 3.2945538854797134e-06,
      "loss": 0.1447,
      "step": 4863
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2312948900126146,
      "learning_rate": 3.292605243523821e-06,
      "loss": 0.1203,
      "step": 4864
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3750809788202938,
      "learning_rate": 3.2906568950888827e-06,
      "loss": 0.1733,
      "step": 4865
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4931539833234682,
      "learning_rate": 3.288708840509838e-06,
      "loss": 0.2091,
      "step": 4866
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.4405938636784255,
      "learning_rate": 3.286761080121585e-06,
      "loss": 0.6482,
      "step": 4867
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2571094740441753,
      "learning_rate": 3.2848136142589637e-06,
      "loss": 0.1812,
      "step": 4868
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.587821408826734,
      "learning_rate": 3.282866443256765e-06,
      "loss": 0.6259,
      "step": 4869
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3238654873590892,
      "learning_rate": 3.280919567449733e-06,
      "loss": 0.2118,
      "step": 4870
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4117434265045379,
      "learning_rate": 3.278972987172556e-06,
      "loss": 0.2117,
      "step": 4871
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.503907416982156,
      "learning_rate": 3.277026702759874e-06,
      "loss": 0.2015,
      "step": 4872
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4048193152802135,
      "learning_rate": 3.275080714546277e-06,
      "loss": 0.1876,
      "step": 4873
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.428121318393855,
      "learning_rate": 3.2731350228663024e-06,
      "loss": 0.1832,
      "step": 4874
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4135611528321803,
      "learning_rate": 3.2711896280544343e-06,
      "loss": 0.2183,
      "step": 4875
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5344358246194536,
      "learning_rate": 3.2692445304451128e-06,
      "loss": 0.1912,
      "step": 4876
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5460506887982708,
      "learning_rate": 3.2672997303727217e-06,
      "loss": 0.231,
      "step": 4877
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.27860492772584,
      "learning_rate": 3.265355228171594e-06,
      "loss": 0.1488,
      "step": 4878
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3129367201439452,
      "learning_rate": 3.263411024176012e-06,
      "loss": 0.1587,
      "step": 4879
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3317410829057663,
      "learning_rate": 3.2614671187202075e-06,
      "loss": 0.1916,
      "step": 4880
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3620972275039978,
      "learning_rate": 3.2595235121383608e-06,
      "loss": 0.1835,
      "step": 4881
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4567214083557853,
      "learning_rate": 3.2575802047645977e-06,
      "loss": 0.2273,
      "step": 4882
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2912603609421198,
      "learning_rate": 3.2556371969329992e-06,
      "loss": 0.1534,
      "step": 4883
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4842863708660128,
      "learning_rate": 3.2536944889775868e-06,
      "loss": 0.1623,
      "step": 4884
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4102574982842278,
      "learning_rate": 3.2517520812323382e-06,
      "loss": 0.1959,
      "step": 4885
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.539624246519049,
      "learning_rate": 3.2498099740311752e-06,
      "loss": 0.4345,
      "step": 4886
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.373197305802169,
      "learning_rate": 3.2478681677079655e-06,
      "loss": 0.211,
      "step": 4887
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5461346511025298,
      "learning_rate": 3.2459266625965315e-06,
      "loss": 0.1943,
      "step": 4888
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6181252107250037,
      "learning_rate": 3.24398545903064e-06,
      "loss": 0.1975,
      "step": 4889
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3612982439228853,
      "learning_rate": 3.2420445573440027e-06,
      "loss": 0.1934,
      "step": 4890
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.8068047953477935,
      "learning_rate": 3.2401039578702886e-06,
      "loss": 0.5916,
      "step": 4891
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4565443982803117,
      "learning_rate": 3.2381636609431065e-06,
      "loss": 0.1659,
      "step": 4892
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3312419396217745,
      "learning_rate": 3.2362236668960144e-06,
      "loss": 0.1797,
      "step": 4893
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4161178802196672,
      "learning_rate": 3.234283976062521e-06,
      "loss": 0.1916,
      "step": 4894
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.545199703660511,
      "learning_rate": 3.2323445887760837e-06,
      "loss": 0.4169,
      "step": 4895
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5008623571832065,
      "learning_rate": 3.230405505370104e-06,
      "loss": 0.1872,
      "step": 4896
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4744933026994607,
      "learning_rate": 3.228466726177932e-06,
      "loss": 0.2106,
      "step": 4897
    },
    {
      "epoch": 0.63,
      "grad_norm": 11.312407511030113,
      "learning_rate": 3.2265282515328676e-06,
      "loss": 0.6071,
      "step": 4898
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.047332519561194,
      "learning_rate": 3.2245900817681576e-06,
      "loss": 0.4166,
      "step": 4899
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.452074252104239,
      "learning_rate": 3.2226522172169928e-06,
      "loss": 0.2357,
      "step": 4900
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.40251856220212,
      "learning_rate": 3.220714658212518e-06,
      "loss": 0.1712,
      "step": 4901
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.239823729190593,
      "learning_rate": 3.2187774050878213e-06,
      "loss": 0.8101,
      "step": 4902
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4604637608408506,
      "learning_rate": 3.2168404581759362e-06,
      "loss": 0.1803,
      "step": 4903
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5140770205145764,
      "learning_rate": 3.21490381780985e-06,
      "loss": 0.2236,
      "step": 4904
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.713558356870783,
      "learning_rate": 3.212967484322491e-06,
      "loss": 0.5642,
      "step": 4905
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.433466284870531,
      "learning_rate": 3.2110314580467404e-06,
      "loss": 0.2286,
      "step": 4906
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2877353777304235,
      "learning_rate": 3.209095739315421e-06,
      "loss": 0.195,
      "step": 4907
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4027032640358676,
      "learning_rate": 3.2071603284613052e-06,
      "loss": 0.231,
      "step": 4908
    },
    {
      "epoch": 0.63,
      "grad_norm": 8.097145218071805,
      "learning_rate": 3.2052252258171142e-06,
      "loss": 0.7016,
      "step": 4909
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.403287677467636,
      "learning_rate": 3.2032904317155133e-06,
      "loss": 0.1421,
      "step": 4910
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3665326117459233,
      "learning_rate": 3.2013559464891162e-06,
      "loss": 0.1998,
      "step": 4911
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3454757628236267,
      "learning_rate": 3.199421770470482e-06,
      "loss": 0.1688,
      "step": 4912
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5275359041921628,
      "learning_rate": 3.1974879039921214e-06,
      "loss": 0.2364,
      "step": 4913
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4241226659499575,
      "learning_rate": 3.1955543473864868e-06,
      "loss": 0.179,
      "step": 4914
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3451012392931037,
      "learning_rate": 3.1936211009859786e-06,
      "loss": 0.204,
      "step": 4915
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4395519941771533,
      "learning_rate": 3.1916881651229447e-06,
      "loss": 0.1601,
      "step": 4916
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.207103839035296,
      "learning_rate": 3.189755540129679e-06,
      "loss": 0.1247,
      "step": 4917
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.299892504385191,
      "learning_rate": 3.1878232263384214e-06,
      "loss": 0.1862,
      "step": 4918
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.324305573305628,
      "learning_rate": 3.1858912240813607e-06,
      "loss": 0.1924,
      "step": 4919
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4806026761928204,
      "learning_rate": 3.183959533690629e-06,
      "loss": 0.2067,
      "step": 4920
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.696621870330636,
      "learning_rate": 3.1820281554983056e-06,
      "loss": 0.5525,
      "step": 4921
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3671552526944621,
      "learning_rate": 3.1800970898364194e-06,
      "loss": 0.1762,
      "step": 4922
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.272168892462716,
      "learning_rate": 3.178166337036941e-06,
      "loss": 0.1788,
      "step": 4923
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3404428556565517,
      "learning_rate": 3.176235897431791e-06,
      "loss": 0.1937,
      "step": 4924
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2700997785505925,
      "learning_rate": 3.1743057713528325e-06,
      "loss": 0.1631,
      "step": 4925
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.063059924010094,
      "learning_rate": 3.172375959131876e-06,
      "loss": 0.7248,
      "step": 4926
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6623076501158969,
      "learning_rate": 3.1704464611006813e-06,
      "loss": 0.2469,
      "step": 4927
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4987203442761203,
      "learning_rate": 3.16851727759095e-06,
      "loss": 0.1751,
      "step": 4928
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.863583117089164,
      "learning_rate": 3.1665884089343296e-06,
      "loss": 0.5409,
      "step": 4929
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3902192958286885,
      "learning_rate": 3.1646598554624174e-06,
      "loss": 0.2208,
      "step": 4930
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2984311251024243,
      "learning_rate": 3.1627316175067515e-06,
      "loss": 0.2013,
      "step": 4931
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1865541599144616,
      "learning_rate": 3.160803695398822e-06,
      "loss": 0.1393,
      "step": 4932
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.436035501254407,
      "learning_rate": 3.1588760894700575e-06,
      "loss": 0.6013,
      "step": 4933
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3561968136317177,
      "learning_rate": 3.156948800051839e-06,
      "loss": 0.1864,
      "step": 4934
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6566480054497594,
      "learning_rate": 3.1550218274754887e-06,
      "loss": 0.1986,
      "step": 4935
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.275827751526579,
      "learning_rate": 3.153095172072273e-06,
      "loss": 0.5399,
      "step": 4936
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3173537610469932,
      "learning_rate": 3.15116883417341e-06,
      "loss": 0.1653,
      "step": 4937
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2868691706362707,
      "learning_rate": 3.1492428141100575e-06,
      "loss": 0.1654,
      "step": 4938
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.8376978388441403,
      "learning_rate": 3.1473171122133207e-06,
      "loss": 0.5235,
      "step": 4939
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.128020714943564,
      "learning_rate": 3.14539172881425e-06,
      "loss": 0.7494,
      "step": 4940
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.325955999517964,
      "learning_rate": 3.1434666642438416e-06,
      "loss": 0.1601,
      "step": 4941
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4871294458972224,
      "learning_rate": 3.141541918833037e-06,
      "loss": 0.1973,
      "step": 4942
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3675120785268717,
      "learning_rate": 3.139617492912722e-06,
      "loss": 0.1699,
      "step": 4943
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.570934871999149,
      "learning_rate": 3.1376933868137267e-06,
      "loss": 0.2167,
      "step": 4944
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4607077242812554,
      "learning_rate": 3.1357696008668285e-06,
      "loss": 0.2137,
      "step": 4945
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.428289609287828,
      "learning_rate": 3.133846135402748e-06,
      "loss": 0.185,
      "step": 4946
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3917835876692761,
      "learning_rate": 3.1319229907521502e-06,
      "loss": 0.1797,
      "step": 4947
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5234316012203415,
      "learning_rate": 3.130000167245648e-06,
      "loss": 0.2237,
      "step": 4948
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4710225239111618,
      "learning_rate": 3.128077665213794e-06,
      "loss": 0.1971,
      "step": 4949
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3825122694021257,
      "learning_rate": 3.1261554849870925e-06,
      "loss": 0.2022,
      "step": 4950
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4262446429517996,
      "learning_rate": 3.124233626895985e-06,
      "loss": 0.2031,
      "step": 4951
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2400891035159496,
      "learning_rate": 3.1223120912708647e-06,
      "loss": 0.1285,
      "step": 4952
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3179863879771123,
      "learning_rate": 3.1203908784420635e-06,
      "loss": 0.2159,
      "step": 4953
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4251663585815408,
      "learning_rate": 3.1184699887398594e-06,
      "loss": 0.2073,
      "step": 4954
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.877114221344478,
      "learning_rate": 3.116549422494478e-06,
      "loss": 0.5977,
      "step": 4955
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4629473813167344,
      "learning_rate": 3.1146291800360863e-06,
      "loss": 0.1872,
      "step": 4956
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4191940890546122,
      "learning_rate": 3.1127092616947935e-06,
      "loss": 0.1887,
      "step": 4957
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3512624943770148,
      "learning_rate": 3.1107896678006577e-06,
      "loss": 0.154,
      "step": 4958
    },
    {
      "epoch": 0.63,
      "grad_norm": 8.57958207276849,
      "learning_rate": 3.108870398683681e-06,
      "loss": 0.4756,
      "step": 4959
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3567900972836395,
      "learning_rate": 3.106951454673808e-06,
      "loss": 0.1959,
      "step": 4960
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.318522056874331,
      "learning_rate": 3.105032836100925e-06,
      "loss": 0.2115,
      "step": 4961
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.531323086250577,
      "learning_rate": 3.1031145432948644e-06,
      "loss": 0.5959,
      "step": 4962
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.379392989554044,
      "learning_rate": 3.1011965765854056e-06,
      "loss": 0.1784,
      "step": 4963
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.48872845454248,
      "learning_rate": 3.099278936302269e-06,
      "loss": 0.1859,
      "step": 4964
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.581452171951147,
      "learning_rate": 3.097361622775117e-06,
      "loss": 0.2207,
      "step": 4965
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.154430844345693,
      "learning_rate": 3.095444636333559e-06,
      "loss": 0.4929,
      "step": 4966
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3324361829494697,
      "learning_rate": 3.0935279773071492e-06,
      "loss": 0.1682,
      "step": 4967
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4025857925858303,
      "learning_rate": 3.0916116460253787e-06,
      "loss": 0.1787,
      "step": 4968
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4637416707492092,
      "learning_rate": 3.0896956428176916e-06,
      "loss": 0.2132,
      "step": 4969
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5733919902503535,
      "learning_rate": 3.0877799680134713e-06,
      "loss": 0.2006,
      "step": 4970
    },
    {
      "epoch": 0.64,
      "grad_norm": 8.444895934665492,
      "learning_rate": 3.085864621942042e-06,
      "loss": 0.5591,
      "step": 4971
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.36104602973759,
      "learning_rate": 3.0839496049326745e-06,
      "loss": 0.2342,
      "step": 4972
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4681124470836902,
      "learning_rate": 3.0820349173145823e-06,
      "loss": 0.1998,
      "step": 4973
    },
    {
      "epoch": 0.64,
      "grad_norm": 20.9081319486792,
      "learning_rate": 3.080120559416924e-06,
      "loss": 0.6035,
      "step": 4974
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5372620208530272,
      "learning_rate": 3.0782065315687975e-06,
      "loss": 0.2381,
      "step": 4975
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.289401990701361,
      "learning_rate": 3.0762928340992483e-06,
      "loss": 0.1872,
      "step": 4976
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4739187980453017,
      "learning_rate": 3.0743794673372605e-06,
      "loss": 0.2102,
      "step": 4977
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4998235618919742,
      "learning_rate": 3.072466431611768e-06,
      "loss": 0.2056,
      "step": 4978
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.619011415164856,
      "learning_rate": 3.0705537272516406e-06,
      "loss": 0.5471,
      "step": 4979
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.329001275268879,
      "learning_rate": 3.0686413545856975e-06,
      "loss": 0.2025,
      "step": 4980
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3279893914637735,
      "learning_rate": 3.0667293139426944e-06,
      "loss": 0.1335,
      "step": 4981
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.340002374150066,
      "learning_rate": 3.064817605651336e-06,
      "loss": 0.1673,
      "step": 4982
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.415370063206269,
      "learning_rate": 3.0629062300402636e-06,
      "loss": 0.1839,
      "step": 4983
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.35674140921504,
      "learning_rate": 3.0609951874380685e-06,
      "loss": 0.2206,
      "step": 4984
    },
    {
      "epoch": 0.64,
      "grad_norm": 11.620830169124915,
      "learning_rate": 3.05908447817328e-06,
      "loss": 0.6326,
      "step": 4985
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.668697237142749,
      "learning_rate": 3.057174102574367e-06,
      "loss": 0.1925,
      "step": 4986
    },
    {
      "epoch": 0.64,
      "grad_norm": 57.61297430236201,
      "learning_rate": 3.0552640609697514e-06,
      "loss": 0.5014,
      "step": 4987
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3055837534298396,
      "learning_rate": 3.0533543536877897e-06,
      "loss": 0.1505,
      "step": 4988
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5324120360267623,
      "learning_rate": 3.0514449810567827e-06,
      "loss": 0.2379,
      "step": 4989
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2135861505956818,
      "learning_rate": 3.049535943404971e-06,
      "loss": 0.175,
      "step": 4990
    },
    {
      "epoch": 0.64,
      "grad_norm": 9.843255655706187,
      "learning_rate": 3.047627241060544e-06,
      "loss": 0.5275,
      "step": 4991
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5023150497849236,
      "learning_rate": 3.0457188743516275e-06,
      "loss": 0.2182,
      "step": 4992
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2435745834973861,
      "learning_rate": 3.043810843606292e-06,
      "loss": 0.1574,
      "step": 4993
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4776723406785188,
      "learning_rate": 3.041903149152552e-06,
      "loss": 0.1802,
      "step": 4994
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2534986372942751,
      "learning_rate": 3.039995791318358e-06,
      "loss": 0.2051,
      "step": 4995
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2875945953561367,
      "learning_rate": 3.038088770431612e-06,
      "loss": 0.2001,
      "step": 4996
    },
    {
      "epoch": 0.64,
      "grad_norm": 10.18993251593194,
      "learning_rate": 3.0361820868201497e-06,
      "loss": 0.7081,
      "step": 4997
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3485463450835542,
      "learning_rate": 3.0342757408117544e-06,
      "loss": 0.1653,
      "step": 4998
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3361753873203188,
      "learning_rate": 3.0323697327341483e-06,
      "loss": 0.2076,
      "step": 4999
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5732852150058063,
      "learning_rate": 3.030464062914995e-06,
      "loss": 0.2269,
      "step": 5000
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.442920057199149,
      "learning_rate": 3.028558731681902e-06,
      "loss": 0.2181,
      "step": 5001
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4063396692994772,
      "learning_rate": 3.0266537393624185e-06,
      "loss": 0.1963,
      "step": 5002
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4079249797972184,
      "learning_rate": 3.024749086284034e-06,
      "loss": 0.2098,
      "step": 5003
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.375231642810665,
      "learning_rate": 3.02284477277418e-06,
      "loss": 0.4828,
      "step": 5004
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.726374710641954,
      "learning_rate": 3.020940799160229e-06,
      "loss": 0.5558,
      "step": 5005
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4055830018082236,
      "learning_rate": 3.0190371657695005e-06,
      "loss": 0.1728,
      "step": 5006
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.511937545084834,
      "learning_rate": 3.017133872929249e-06,
      "loss": 0.1752,
      "step": 5007
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4984367302560848,
      "learning_rate": 3.0152309209666703e-06,
      "loss": 0.1713,
      "step": 5008
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3429653470012348,
      "learning_rate": 3.0133283102089067e-06,
      "loss": 0.1788,
      "step": 5009
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6186532953999575,
      "learning_rate": 3.0114260409830386e-06,
      "loss": 0.2641,
      "step": 5010
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4759789423170797,
      "learning_rate": 3.009524113616086e-06,
      "loss": 0.2308,
      "step": 5011
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4166499727088129,
      "learning_rate": 3.0076225284350147e-06,
      "loss": 0.1616,
      "step": 5012
    },
    {
      "epoch": 0.64,
      "grad_norm": 10.14494135093183,
      "learning_rate": 3.005721285766729e-06,
      "loss": 0.6831,
      "step": 5013
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.207994892961058,
      "learning_rate": 3.0038203859380717e-06,
      "loss": 0.7337,
      "step": 5014
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.335534407833,
      "learning_rate": 3.0019198292758327e-06,
      "loss": 0.1918,
      "step": 5015
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.364618715458839,
      "learning_rate": 3.0000196161067398e-06,
      "loss": 0.1778,
      "step": 5016
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.52613250967692,
      "learning_rate": 2.998119746757461e-06,
      "loss": 0.2336,
      "step": 5017
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4571953848437942,
      "learning_rate": 2.9962202215546043e-06,
      "loss": 0.2192,
      "step": 5018
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.264867692557448,
      "learning_rate": 2.994321040824722e-06,
      "loss": 0.5485,
      "step": 5019
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.436019416387297,
      "learning_rate": 2.9924222048943046e-06,
      "loss": 0.2017,
      "step": 5020
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.966765572491479,
      "learning_rate": 2.990523714089785e-06,
      "loss": 0.4712,
      "step": 5021
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5130780269902684,
      "learning_rate": 2.988625568737532e-06,
      "loss": 0.2467,
      "step": 5022
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.409937331648496,
      "learning_rate": 2.986727769163862e-06,
      "loss": 0.4768,
      "step": 5023
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2944717478382726,
      "learning_rate": 2.9848303156950297e-06,
      "loss": 0.1381,
      "step": 5024
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5443128300924451,
      "learning_rate": 2.982933208657228e-06,
      "loss": 0.1893,
      "step": 5025
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3849654560322258,
      "learning_rate": 2.9810364483765913e-06,
      "loss": 0.1725,
      "step": 5026
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2496183413893427,
      "learning_rate": 2.9791400351791954e-06,
      "loss": 0.1533,
      "step": 5027
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6164483407746597,
      "learning_rate": 2.9772439693910554e-06,
      "loss": 0.2045,
      "step": 5028
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.485373164385519,
      "learning_rate": 2.975348251338126e-06,
      "loss": 0.1869,
      "step": 5029
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.347741391749622,
      "learning_rate": 2.973452881346305e-06,
      "loss": 0.1398,
      "step": 5030
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.563355427689588,
      "learning_rate": 2.9715578597414268e-06,
      "loss": 0.6064,
      "step": 5031
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.502092240211848,
      "learning_rate": 2.9696631868492664e-06,
      "loss": 0.1826,
      "step": 5032
    },
    {
      "epoch": 0.64,
      "grad_norm": 14.896212669046985,
      "learning_rate": 2.9677688629955438e-06,
      "loss": 0.5327,
      "step": 5033
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.45330267024653,
      "learning_rate": 2.965874888505913e-06,
      "loss": 0.1624,
      "step": 5034
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5891186529987804,
      "learning_rate": 2.9639812637059717e-06,
      "loss": 0.193,
      "step": 5035
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3118413300336031,
      "learning_rate": 2.962087988921253e-06,
      "loss": 0.1725,
      "step": 5036
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2900050579558922,
      "learning_rate": 2.9601950644772352e-06,
      "loss": 0.1492,
      "step": 5037
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4302805870165956,
      "learning_rate": 2.958302490699334e-06,
      "loss": 0.1354,
      "step": 5038
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4639121834071098,
      "learning_rate": 2.9564102679129027e-06,
      "loss": 0.1811,
      "step": 5039
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3156435188747502,
      "learning_rate": 2.954518396443239e-06,
      "loss": 0.1692,
      "step": 5040
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.092959767714257,
      "learning_rate": 2.9526268766155753e-06,
      "loss": 0.4946,
      "step": 5041
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3285724868877982,
      "learning_rate": 2.9507357087550857e-06,
      "loss": 0.1955,
      "step": 5042
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.486193316350365,
      "learning_rate": 2.948844893186886e-06,
      "loss": 0.1859,
      "step": 5043
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.276755047563612,
      "learning_rate": 2.9469544302360283e-06,
      "loss": 0.2094,
      "step": 5044
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3646821194079577,
      "learning_rate": 2.945064320227505e-06,
      "loss": 0.1787,
      "step": 5045
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2448287494537953,
      "learning_rate": 2.9431745634862484e-06,
      "loss": 0.1589,
      "step": 5046
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4748461175636842,
      "learning_rate": 2.9412851603371277e-06,
      "loss": 0.2058,
      "step": 5047
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.625060115369323,
      "learning_rate": 2.9393961111049564e-06,
      "loss": 0.1775,
      "step": 5048
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.469529791977539,
      "learning_rate": 2.937507416114481e-06,
      "loss": 0.1795,
      "step": 5049
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5128662262945025,
      "learning_rate": 2.9356190756903913e-06,
      "loss": 0.1883,
      "step": 5050
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3754708268901072,
      "learning_rate": 2.9337310901573134e-06,
      "loss": 0.2167,
      "step": 5051
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.23649940733848,
      "learning_rate": 2.9318434598398173e-06,
      "loss": 0.1729,
      "step": 5052
    },
    {
      "epoch": 0.65,
      "grad_norm": 14.702920773621205,
      "learning_rate": 2.929956185062407e-06,
      "loss": 0.5919,
      "step": 5053
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3599079894460915,
      "learning_rate": 2.9280692661495247e-06,
      "loss": 0.1662,
      "step": 5054
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3181582612749612,
      "learning_rate": 2.9261827034255575e-06,
      "loss": 0.1639,
      "step": 5055
    },
    {
      "epoch": 0.65,
      "grad_norm": 51.25914152263893,
      "learning_rate": 2.9242964972148245e-06,
      "loss": 0.6843,
      "step": 5056
    },
    {
      "epoch": 0.65,
      "grad_norm": 7.16119636642733,
      "learning_rate": 2.922410647841586e-06,
      "loss": 0.5682,
      "step": 5057
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4292867076694942,
      "learning_rate": 2.920525155630044e-06,
      "loss": 0.2012,
      "step": 5058
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.389299237220389,
      "learning_rate": 2.918640020904334e-06,
      "loss": 0.1975,
      "step": 5059
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.477089311096776,
      "learning_rate": 2.9167552439885333e-06,
      "loss": 0.4907,
      "step": 5060
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4069000612286557,
      "learning_rate": 2.914870825206659e-06,
      "loss": 0.1783,
      "step": 5061
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5262679698830124,
      "learning_rate": 2.9129867648826623e-06,
      "loss": 0.1897,
      "step": 5062
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.412993976860651,
      "learning_rate": 2.9111030633404354e-06,
      "loss": 0.193,
      "step": 5063
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.55758322752055,
      "learning_rate": 2.9092197209038087e-06,
      "loss": 0.7185,
      "step": 5064
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4849118601917306,
      "learning_rate": 2.90733673789655e-06,
      "loss": 0.2238,
      "step": 5065
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3824589342018823,
      "learning_rate": 2.905454114642365e-06,
      "loss": 0.164,
      "step": 5066
    },
    {
      "epoch": 0.65,
      "grad_norm": 13.095602328315444,
      "learning_rate": 2.9035718514649013e-06,
      "loss": 0.5777,
      "step": 5067
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4412228617798537,
      "learning_rate": 2.9016899486877404e-06,
      "loss": 0.188,
      "step": 5068
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.29165688298871,
      "learning_rate": 2.8998084066344005e-06,
      "loss": 0.1783,
      "step": 5069
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5481286003143773,
      "learning_rate": 2.8979272256283453e-06,
      "loss": 0.1617,
      "step": 5070
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4761169092927509,
      "learning_rate": 2.89604640599297e-06,
      "loss": 0.1998,
      "step": 5071
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4589978010108204,
      "learning_rate": 2.8941659480516083e-06,
      "loss": 0.1946,
      "step": 5072
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.528650519688426,
      "learning_rate": 2.892285852127532e-06,
      "loss": 0.1876,
      "step": 5073
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.360810434474507,
      "learning_rate": 2.8904061185439513e-06,
      "loss": 0.609,
      "step": 5074
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.938382977273774,
      "learning_rate": 2.8885267476240173e-06,
      "loss": 0.4291,
      "step": 5075
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5069810667114207,
      "learning_rate": 2.8866477396908143e-06,
      "loss": 0.1995,
      "step": 5076
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4419070475703875,
      "learning_rate": 2.8847690950673642e-06,
      "loss": 0.1873,
      "step": 5077
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4657261334802143,
      "learning_rate": 2.882890814076629e-06,
      "loss": 0.2412,
      "step": 5078
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4343269905450138,
      "learning_rate": 2.8810128970415035e-06,
      "loss": 0.2151,
      "step": 5079
    },
    {
      "epoch": 0.65,
      "grad_norm": 7.109726795801517,
      "learning_rate": 2.879135344284829e-06,
      "loss": 0.5695,
      "step": 5080
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.523259012075086,
      "learning_rate": 2.8772581561293743e-06,
      "loss": 0.208,
      "step": 5081
    },
    {
      "epoch": 0.65,
      "grad_norm": 10.044465418888441,
      "learning_rate": 2.875381332897853e-06,
      "loss": 0.5315,
      "step": 5082
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2153965483999303,
      "learning_rate": 2.873504874912911e-06,
      "loss": 0.1454,
      "step": 5083
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.820284210112551,
      "learning_rate": 2.8716287824971344e-06,
      "loss": 0.513,
      "step": 5084
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5913237281062549,
      "learning_rate": 2.869753055973044e-06,
      "loss": 0.219,
      "step": 5085
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.990986766571134,
      "learning_rate": 2.8678776956630995e-06,
      "loss": 0.4377,
      "step": 5086
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3519630280114527,
      "learning_rate": 2.866002701889694e-06,
      "loss": 0.1758,
      "step": 5087
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3595385536718376,
      "learning_rate": 2.8641280749751655e-06,
      "loss": 0.1838,
      "step": 5088
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4189900270602518,
      "learning_rate": 2.8622538152417804e-06,
      "loss": 0.1935,
      "step": 5089
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2771354030837618,
      "learning_rate": 2.8603799230117495e-06,
      "loss": 0.1669,
      "step": 5090
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.665319769093634,
      "learning_rate": 2.858506398607214e-06,
      "loss": 0.4676,
      "step": 5091
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.449542537721442,
      "learning_rate": 2.8566332423502553e-06,
      "loss": 0.2037,
      "step": 5092
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.420905718387713,
      "learning_rate": 2.8547604545628903e-06,
      "loss": 0.1921,
      "step": 5093
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.3720982546728995,
      "learning_rate": 2.852888035567073e-06,
      "loss": 0.3753,
      "step": 5094
    },
    {
      "epoch": 0.65,
      "grad_norm": 10.816391521811243,
      "learning_rate": 2.851015985684691e-06,
      "loss": 0.5533,
      "step": 5095
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.104731975007369,
      "learning_rate": 2.8491443052375767e-06,
      "loss": 0.4287,
      "step": 5096
    },
    {
      "epoch": 0.65,
      "grad_norm": 10.28466523632451,
      "learning_rate": 2.84727299454749e-06,
      "loss": 0.523,
      "step": 5097
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3585332489635817,
      "learning_rate": 2.845402053936133e-06,
      "loss": 0.1522,
      "step": 5098
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.125055216821156,
      "learning_rate": 2.8435314837251414e-06,
      "loss": 0.4462,
      "step": 5099
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.216898759166165,
      "learning_rate": 2.841661284236088e-06,
      "loss": 0.1654,
      "step": 5100
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3896216985096137,
      "learning_rate": 2.8397914557904816e-06,
      "loss": 0.1875,
      "step": 5101
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4797071364447194,
      "learning_rate": 2.837921998709765e-06,
      "loss": 0.2234,
      "step": 5102
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3628553150881635,
      "learning_rate": 2.8360529133153237e-06,
      "loss": 0.1703,
      "step": 5103
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4095604515751896,
      "learning_rate": 2.8341841999284725e-06,
      "loss": 0.1766,
      "step": 5104
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.624395810909887,
      "learning_rate": 2.832315858870467e-06,
      "loss": 0.2736,
      "step": 5105
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.278078868547567,
      "learning_rate": 2.830447890462492e-06,
      "loss": 0.1673,
      "step": 5106
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4738750282034756,
      "learning_rate": 2.828580295025679e-06,
      "loss": 0.1954,
      "step": 5107
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.444588651255552,
      "learning_rate": 2.8267130728810866e-06,
      "loss": 0.2419,
      "step": 5108
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3853011200479584,
      "learning_rate": 2.824846224349712e-06,
      "loss": 0.2094,
      "step": 5109
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4876029326195113,
      "learning_rate": 2.822979749752486e-06,
      "loss": 0.2014,
      "step": 5110
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.243183327289994,
      "learning_rate": 2.821113649410281e-06,
      "loss": 0.5254,
      "step": 5111
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3491194294904623,
      "learning_rate": 2.8192479236438987e-06,
      "loss": 0.1819,
      "step": 5112
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.436969699462176,
      "learning_rate": 2.8173825727740808e-06,
      "loss": 0.1894,
      "step": 5113
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.7376379681918945,
      "learning_rate": 2.815517597121501e-06,
      "loss": 0.4994,
      "step": 5114
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3062123824350114,
      "learning_rate": 2.8136529970067715e-06,
      "loss": 0.2056,
      "step": 5115
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.828624889787796,
      "learning_rate": 2.8117887727504355e-06,
      "loss": 0.5375,
      "step": 5116
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2806683924850935,
      "learning_rate": 2.809924924672978e-06,
      "loss": 0.1844,
      "step": 5117
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2934793821291821,
      "learning_rate": 2.8080614530948174e-06,
      "loss": 0.1842,
      "step": 5118
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3222507431267705,
      "learning_rate": 2.806198358336304e-06,
      "loss": 0.1808,
      "step": 5119
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.7643963952089114,
      "learning_rate": 2.804335640717726e-06,
      "loss": 0.3927,
      "step": 5120
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2789292473512925,
      "learning_rate": 2.802473300559305e-06,
      "loss": 0.144,
      "step": 5121
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.38912117880381,
      "learning_rate": 2.8006113381812e-06,
      "loss": 0.1588,
      "step": 5122
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5783197821492485,
      "learning_rate": 2.798749753903501e-06,
      "loss": 0.1682,
      "step": 5123
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.820550591816738,
      "learning_rate": 2.7968885480462406e-06,
      "loss": 0.7252,
      "step": 5124
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.548645799509419,
      "learning_rate": 2.7950277209293774e-06,
      "loss": 0.2233,
      "step": 5125
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2767422013307312,
      "learning_rate": 2.7931672728728137e-06,
      "loss": 0.163,
      "step": 5126
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.811434172818997,
      "learning_rate": 2.7913072041963797e-06,
      "loss": 0.5244,
      "step": 5127
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4305200293747884,
      "learning_rate": 2.789447515219842e-06,
      "loss": 0.1572,
      "step": 5128
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4043739259321446,
      "learning_rate": 2.7875882062629033e-06,
      "loss": 0.1836,
      "step": 5129
    },
    {
      "epoch": 0.66,
      "grad_norm": 15.264330633660046,
      "learning_rate": 2.7857292776452003e-06,
      "loss": 0.4864,
      "step": 5130
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.056272535944773,
      "learning_rate": 2.783870729686303e-06,
      "loss": 0.6018,
      "step": 5131
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.315824082959716,
      "learning_rate": 2.782012562705721e-06,
      "loss": 0.1594,
      "step": 5132
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.131266491514421,
      "learning_rate": 2.780154777022892e-06,
      "loss": 0.6646,
      "step": 5133
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4776151448700103,
      "learning_rate": 2.77829737295719e-06,
      "loss": 0.2171,
      "step": 5134
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5046147513697086,
      "learning_rate": 2.7764403508279275e-06,
      "loss": 0.2027,
      "step": 5135
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3220207908902952,
      "learning_rate": 2.774583710954346e-06,
      "loss": 0.1603,
      "step": 5136
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.804079216140304,
      "learning_rate": 2.7727274536556248e-06,
      "loss": 0.7058,
      "step": 5137
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4831112031298932,
      "learning_rate": 2.770871579250872e-06,
      "loss": 0.1857,
      "step": 5138
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.005007266325222,
      "learning_rate": 2.76901608805914e-06,
      "loss": 0.4921,
      "step": 5139
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2572385186097546,
      "learning_rate": 2.767160980399406e-06,
      "loss": 0.158,
      "step": 5140
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.351678424476889,
      "learning_rate": 2.7653062565905842e-06,
      "loss": 0.1718,
      "step": 5141
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3749815535840335,
      "learning_rate": 2.7634519169515237e-06,
      "loss": 0.1672,
      "step": 5142
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5166909673680815,
      "learning_rate": 2.7615979618010057e-06,
      "loss": 0.2171,
      "step": 5143
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.616914173718923,
      "learning_rate": 2.759744391457749e-06,
      "loss": 0.2098,
      "step": 5144
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4497000274894343,
      "learning_rate": 2.757891206240404e-06,
      "loss": 0.1892,
      "step": 5145
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5361946285924855,
      "learning_rate": 2.7560384064675505e-06,
      "loss": 0.2503,
      "step": 5146
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5450617468125951,
      "learning_rate": 2.7541859924577106e-06,
      "loss": 0.1872,
      "step": 5147
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3616175431660773,
      "learning_rate": 2.7523339645293357e-06,
      "loss": 0.181,
      "step": 5148
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3508472654018273,
      "learning_rate": 2.750482323000808e-06,
      "loss": 0.1854,
      "step": 5149
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3728901829356521,
      "learning_rate": 2.748631068190449e-06,
      "loss": 0.1854,
      "step": 5150
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3638344433615837,
      "learning_rate": 2.746780200416509e-06,
      "loss": 0.1577,
      "step": 5151
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3076257092954833,
      "learning_rate": 2.7449297199971713e-06,
      "loss": 0.1626,
      "step": 5152
    },
    {
      "epoch": 0.66,
      "grad_norm": 20.331259955824912,
      "learning_rate": 2.743079627250559e-06,
      "loss": 0.4808,
      "step": 5153
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.436645870804933,
      "learning_rate": 2.741229922494724e-06,
      "loss": 0.4759,
      "step": 5154
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1318150540449081,
      "learning_rate": 2.7393806060476525e-06,
      "loss": 0.1553,
      "step": 5155
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2859506338999838,
      "learning_rate": 2.737531678227261e-06,
      "loss": 0.1713,
      "step": 5156
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.187985286089472,
      "learning_rate": 2.7356831393514033e-06,
      "loss": 0.1708,
      "step": 5157
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3890274270186287,
      "learning_rate": 2.733834989737864e-06,
      "loss": 0.1554,
      "step": 5158
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.416174658523014,
      "learning_rate": 2.7319872297043604e-06,
      "loss": 0.2467,
      "step": 5159
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4130310791038492,
      "learning_rate": 2.7301398595685456e-06,
      "loss": 0.2161,
      "step": 5160
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4595498916895093,
      "learning_rate": 2.7282928796480047e-06,
      "loss": 0.1925,
      "step": 5161
    },
    {
      "epoch": 0.66,
      "grad_norm": 8.756208296974346,
      "learning_rate": 2.726446290260251e-06,
      "loss": 0.4828,
      "step": 5162
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3718102159754222,
      "learning_rate": 2.72460009172274e-06,
      "loss": 0.1867,
      "step": 5163
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.138053196956338,
      "learning_rate": 2.722754284352851e-06,
      "loss": 0.1401,
      "step": 5164
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.05015769668348,
      "learning_rate": 2.7209088684679015e-06,
      "loss": 0.6486,
      "step": 5165
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4518974340924973,
      "learning_rate": 2.719063844385138e-06,
      "loss": 0.1728,
      "step": 5166
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.466570938083344,
      "learning_rate": 2.7172192124217418e-06,
      "loss": 0.2102,
      "step": 5167
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4149825900350117,
      "learning_rate": 2.7153749728948286e-06,
      "loss": 0.1814,
      "step": 5168
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2518809587894115,
      "learning_rate": 2.7135311261214425e-06,
      "loss": 0.1971,
      "step": 5169
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.494361984335687,
      "learning_rate": 2.711687672418564e-06,
      "loss": 0.4557,
      "step": 5170
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3309463261906223,
      "learning_rate": 2.7098446121031008e-06,
      "loss": 0.1715,
      "step": 5171
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.416601647860765,
      "learning_rate": 2.7080019454919e-06,
      "loss": 0.2066,
      "step": 5172
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6141514351743325,
      "learning_rate": 2.706159672901736e-06,
      "loss": 0.204,
      "step": 5173
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2863621908137397,
      "learning_rate": 2.7043177946493148e-06,
      "loss": 0.1807,
      "step": 5174
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4802197287578374,
      "learning_rate": 2.7024763110512808e-06,
      "loss": 0.1928,
      "step": 5175
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.29113475417484,
      "learning_rate": 2.700635222424204e-06,
      "loss": 0.6324,
      "step": 5176
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6355960951962198,
      "learning_rate": 2.6987945290845884e-06,
      "loss": 0.2544,
      "step": 5177
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3363274788016215,
      "learning_rate": 2.6969542313488716e-06,
      "loss": 0.1556,
      "step": 5178
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.341758954438275,
      "learning_rate": 2.695114329533422e-06,
      "loss": 0.1917,
      "step": 5179
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3471936573560526,
      "learning_rate": 2.693274823954537e-06,
      "loss": 0.191,
      "step": 5180
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3183084915190886,
      "learning_rate": 2.6914357149284522e-06,
      "loss": 0.1614,
      "step": 5181
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.424446813532815,
      "learning_rate": 2.689597002771333e-06,
      "loss": 0.1516,
      "step": 5182
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4291692099088453,
      "learning_rate": 2.6877586877992735e-06,
      "loss": 0.1804,
      "step": 5183
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.470833470431681,
      "learning_rate": 2.6859207703283012e-06,
      "loss": 0.1762,
      "step": 5184
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4222928915304451,
      "learning_rate": 2.6840832506743757e-06,
      "loss": 0.1821,
      "step": 5185
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4421074315723963,
      "learning_rate": 2.6822461291533874e-06,
      "loss": 0.1362,
      "step": 5186
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.44533475596481,
      "learning_rate": 2.6804094060811604e-06,
      "loss": 0.195,
      "step": 5187
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7409761004132849,
      "learning_rate": 2.6785730817734446e-06,
      "loss": 0.2564,
      "step": 5188
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.496396609013809,
      "learning_rate": 2.67673715654593e-06,
      "loss": 0.2103,
      "step": 5189
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.438126906058636,
      "learning_rate": 2.6749016307142307e-06,
      "loss": 0.2235,
      "step": 5190
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.360227391841782,
      "learning_rate": 2.6730665045938975e-06,
      "loss": 0.155,
      "step": 5191
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4455594308986413,
      "learning_rate": 2.671231778500408e-06,
      "loss": 0.1862,
      "step": 5192
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3434152630616087,
      "learning_rate": 2.669397452749173e-06,
      "loss": 0.1373,
      "step": 5193
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4258639259317738,
      "learning_rate": 2.667563527655535e-06,
      "loss": 0.179,
      "step": 5194
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.366220297847372,
      "learning_rate": 2.6657300035347643e-06,
      "loss": 0.4097,
      "step": 5195
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4291420278632294,
      "learning_rate": 2.6638968807020693e-06,
      "loss": 0.1913,
      "step": 5196
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.273788633170711,
      "learning_rate": 2.6620641594725834e-06,
      "loss": 0.5861,
      "step": 5197
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3342592708495322,
      "learning_rate": 2.6602318401613715e-06,
      "loss": 0.2202,
      "step": 5198
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.319210774442695,
      "learning_rate": 2.6583999230834294e-06,
      "loss": 0.1616,
      "step": 5199
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.319393469123398,
      "learning_rate": 2.656568408553689e-06,
      "loss": 0.1583,
      "step": 5200
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4282252733665337,
      "learning_rate": 2.654737296887006e-06,
      "loss": 0.1893,
      "step": 5201
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2562800858988354,
      "learning_rate": 2.652906588398171e-06,
      "loss": 0.2209,
      "step": 5202
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4981383254855776,
      "learning_rate": 2.6510762834019006e-06,
      "loss": 0.2223,
      "step": 5203
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.874336789209758,
      "learning_rate": 2.64924638221285e-06,
      "loss": 0.7036,
      "step": 5204
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3271008394035793,
      "learning_rate": 2.6474168851455995e-06,
      "loss": 0.1686,
      "step": 5205
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.432380441633629,
      "learning_rate": 2.645587792514659e-06,
      "loss": 0.577,
      "step": 5206
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3688968054287889,
      "learning_rate": 2.6437591046344724e-06,
      "loss": 0.1677,
      "step": 5207
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.48763889250952,
      "learning_rate": 2.641930821819409e-06,
      "loss": 0.2257,
      "step": 5208
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2825989376275493,
      "learning_rate": 2.6401029443837767e-06,
      "loss": 0.1473,
      "step": 5209
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2777090609630843,
      "learning_rate": 2.6382754726418047e-06,
      "loss": 0.2022,
      "step": 5210
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4431801275455367,
      "learning_rate": 2.63644840690766e-06,
      "loss": 0.203,
      "step": 5211
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.046508761364367,
      "learning_rate": 2.634621747495435e-06,
      "loss": 0.4253,
      "step": 5212
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.493813665164751,
      "learning_rate": 2.632795494719154e-06,
      "loss": 0.4425,
      "step": 5213
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4128085352137254,
      "learning_rate": 2.6309696488927692e-06,
      "loss": 0.2052,
      "step": 5214
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3696824190705998,
      "learning_rate": 2.6291442103301667e-06,
      "loss": 0.2136,
      "step": 5215
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4542839208702307,
      "learning_rate": 2.6273191793451575e-06,
      "loss": 0.1736,
      "step": 5216
    },
    {
      "epoch": 0.67,
      "grad_norm": 8.471039895367817,
      "learning_rate": 2.6254945562514877e-06,
      "loss": 0.7211,
      "step": 5217
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3289347049033315,
      "learning_rate": 2.623670341362834e-06,
      "loss": 0.1806,
      "step": 5218
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3711718161598396,
      "learning_rate": 2.621846534992797e-06,
      "loss": 0.2166,
      "step": 5219
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4472882858062799,
      "learning_rate": 2.62002313745491e-06,
      "loss": 0.1951,
      "step": 5220
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5736169132425302,
      "learning_rate": 2.6182001490626374e-06,
      "loss": 0.2056,
      "step": 5221
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.656194724695291,
      "learning_rate": 2.6163775701293714e-06,
      "loss": 0.4359,
      "step": 5222
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2664863520339236,
      "learning_rate": 2.6145554009684343e-06,
      "loss": 0.1599,
      "step": 5223
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2750757269215343,
      "learning_rate": 2.612733641893076e-06,
      "loss": 0.1645,
      "step": 5224
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.413035083464526,
      "learning_rate": 2.610912293216483e-06,
      "loss": 0.2215,
      "step": 5225
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4602541514385603,
      "learning_rate": 2.6090913552517626e-06,
      "loss": 0.1802,
      "step": 5226
    },
    {
      "epoch": 0.67,
      "grad_norm": 10.811433416816083,
      "learning_rate": 2.6072708283119543e-06,
      "loss": 0.6153,
      "step": 5227
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3813601858747642,
      "learning_rate": 2.6054507127100315e-06,
      "loss": 0.2187,
      "step": 5228
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.341276984691384,
      "learning_rate": 2.6036310087588913e-06,
      "loss": 0.2084,
      "step": 5229
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3783609293320087,
      "learning_rate": 2.601811716771362e-06,
      "loss": 0.1607,
      "step": 5230
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5695129651658026,
      "learning_rate": 2.599992837060198e-06,
      "loss": 0.1557,
      "step": 5231
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4285198345472838,
      "learning_rate": 2.598174369938091e-06,
      "loss": 0.1621,
      "step": 5232
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5034665205710178,
      "learning_rate": 2.596356315717654e-06,
      "loss": 0.2369,
      "step": 5233
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.395482228629099,
      "learning_rate": 2.594538674711432e-06,
      "loss": 0.6192,
      "step": 5234
    },
    {
      "epoch": 0.67,
      "grad_norm": 9.08742334039031,
      "learning_rate": 2.5927214472318983e-06,
      "loss": 0.4018,
      "step": 5235
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3585832437507688,
      "learning_rate": 2.5909046335914533e-06,
      "loss": 0.1762,
      "step": 5236
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4563598539032379,
      "learning_rate": 2.5890882341024328e-06,
      "loss": 0.1744,
      "step": 5237
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.401050605059021,
      "learning_rate": 2.5872722490770917e-06,
      "loss": 0.1928,
      "step": 5238
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.152411746647394,
      "learning_rate": 2.5854566788276247e-06,
      "loss": 0.1429,
      "step": 5239
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4015658803659221,
      "learning_rate": 2.583641523666146e-06,
      "loss": 0.1731,
      "step": 5240
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3704452333666717,
      "learning_rate": 2.5818267839047018e-06,
      "loss": 0.1883,
      "step": 5241
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.363497053216782,
      "learning_rate": 2.5800124598552676e-06,
      "loss": 0.3557,
      "step": 5242
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.503031275737606,
      "learning_rate": 2.5781985518297468e-06,
      "loss": 0.185,
      "step": 5243
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6309731766558262,
      "learning_rate": 2.576385060139971e-06,
      "loss": 0.251,
      "step": 5244
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.807192378787799,
      "learning_rate": 2.574571985097698e-06,
      "loss": 0.4941,
      "step": 5245
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.706380459126579,
      "learning_rate": 2.572759327014618e-06,
      "loss": 0.7309,
      "step": 5246
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4201699331259867,
      "learning_rate": 2.5709470862023517e-06,
      "loss": 0.2323,
      "step": 5247
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.458010096912857,
      "learning_rate": 2.5691352629724407e-06,
      "loss": 0.1866,
      "step": 5248
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4603700535992183,
      "learning_rate": 2.567323857636359e-06,
      "loss": 0.2237,
      "step": 5249
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.877772686705728,
      "learning_rate": 2.565512870505509e-06,
      "loss": 0.4367,
      "step": 5250
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4765639664601784,
      "learning_rate": 2.5637023018912187e-06,
      "loss": 0.1608,
      "step": 5251
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.49349640951492,
      "learning_rate": 2.561892152104745e-06,
      "loss": 0.5303,
      "step": 5252
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5203183893193362,
      "learning_rate": 2.5600824214572774e-06,
      "loss": 0.2111,
      "step": 5253
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4430272055960915,
      "learning_rate": 2.558273110259925e-06,
      "loss": 0.1847,
      "step": 5254
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.576991497148407,
      "learning_rate": 2.5564642188237345e-06,
      "loss": 0.5297,
      "step": 5255
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2422072167809208,
      "learning_rate": 2.5546557474596725e-06,
      "loss": 0.2137,
      "step": 5256
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4945240227386392,
      "learning_rate": 2.5528476964786365e-06,
      "loss": 0.1867,
      "step": 5257
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2476026761900325,
      "learning_rate": 2.5510400661914503e-06,
      "loss": 0.2003,
      "step": 5258
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.415538091855744,
      "learning_rate": 2.5492328569088666e-06,
      "loss": 0.1594,
      "step": 5259
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.959920690087776,
      "learning_rate": 2.5474260689415687e-06,
      "loss": 0.4179,
      "step": 5260
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.393724734513312,
      "learning_rate": 2.5456197026001613e-06,
      "loss": 0.207,
      "step": 5261
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.474745007857217,
      "learning_rate": 2.543813758195181e-06,
      "loss": 0.1983,
      "step": 5262
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4751570120963147,
      "learning_rate": 2.54200823603709e-06,
      "loss": 0.1853,
      "step": 5263
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.53327353244692,
      "learning_rate": 2.5402031364362772e-06,
      "loss": 0.2342,
      "step": 5264
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.930312581032029,
      "learning_rate": 2.5383984597030638e-06,
      "loss": 0.5757,
      "step": 5265
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.435110061167129,
      "learning_rate": 2.5365942061476924e-06,
      "loss": 0.1975,
      "step": 5266
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.381270612717725,
      "learning_rate": 2.5347903760803327e-06,
      "loss": 0.1852,
      "step": 5267
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2989763613019238,
      "learning_rate": 2.532986969811089e-06,
      "loss": 0.1666,
      "step": 5268
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.456213693616179,
      "learning_rate": 2.5311839876499843e-06,
      "loss": 0.1653,
      "step": 5269
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8114893705532393,
      "learning_rate": 2.5293814299069742e-06,
      "loss": 0.1738,
      "step": 5270
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3661684509447585,
      "learning_rate": 2.5275792968919376e-06,
      "loss": 0.1817,
      "step": 5271
    },
    {
      "epoch": 0.67,
      "grad_norm": 18.028612621208328,
      "learning_rate": 2.525777588914683e-06,
      "loss": 0.5882,
      "step": 5272
    },
    {
      "epoch": 0.67,
      "grad_norm": 63.872060285424745,
      "learning_rate": 2.5239763062849424e-06,
      "loss": 0.6265,
      "step": 5273
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.081078955362501,
      "learning_rate": 2.5221754493123798e-06,
      "loss": 0.6438,
      "step": 5274
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.115171609422927,
      "learning_rate": 2.520375018306584e-06,
      "loss": 0.6349,
      "step": 5275
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5032247818278457,
      "learning_rate": 2.5185750135770694e-06,
      "loss": 0.2158,
      "step": 5276
    },
    {
      "epoch": 0.68,
      "grad_norm": 10.61851980153102,
      "learning_rate": 2.516775435433276e-06,
      "loss": 0.4777,
      "step": 5277
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2579790554437664,
      "learning_rate": 2.514976284184575e-06,
      "loss": 0.1646,
      "step": 5278
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3014645198271437,
      "learning_rate": 2.5131775601402588e-06,
      "loss": 0.1843,
      "step": 5279
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.963462222397379,
      "learning_rate": 2.511379263609547e-06,
      "loss": 0.7074,
      "step": 5280
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5049787043800584,
      "learning_rate": 2.5095813949015934e-06,
      "loss": 0.1888,
      "step": 5281
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.384583243258133,
      "learning_rate": 2.507783954325467e-06,
      "loss": 0.1857,
      "step": 5282
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3524099839694455,
      "learning_rate": 2.505986942190173e-06,
      "loss": 0.1497,
      "step": 5283
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5933972500746358,
      "learning_rate": 2.504190358804637e-06,
      "loss": 0.2139,
      "step": 5284
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4715452297627107,
      "learning_rate": 2.5023942044777114e-06,
      "loss": 0.237,
      "step": 5285
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4136207336487752,
      "learning_rate": 2.5005984795181783e-06,
      "loss": 0.1835,
      "step": 5286
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.784547836747873,
      "learning_rate": 2.498803184234741e-06,
      "loss": 0.438,
      "step": 5287
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4057345743913578,
      "learning_rate": 2.497008318936031e-06,
      "loss": 0.1499,
      "step": 5288
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.51939165066354,
      "learning_rate": 2.49521388393061e-06,
      "loss": 0.1858,
      "step": 5289
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3224560816079685,
      "learning_rate": 2.4934198795269603e-06,
      "loss": 0.1513,
      "step": 5290
    },
    {
      "epoch": 0.68,
      "grad_norm": 8.341500138482667,
      "learning_rate": 2.4916263060334903e-06,
      "loss": 0.5333,
      "step": 5291
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.310700698647966,
      "learning_rate": 2.48983316375854e-06,
      "loss": 0.6558,
      "step": 5292
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5977628038003957,
      "learning_rate": 2.488040453010369e-06,
      "loss": 0.2542,
      "step": 5293
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2839235107606273,
      "learning_rate": 2.486248174097165e-06,
      "loss": 0.2121,
      "step": 5294
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4495660851694936,
      "learning_rate": 2.4844563273270404e-06,
      "loss": 0.1917,
      "step": 5295
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2858919986792279,
      "learning_rate": 2.482664913008037e-06,
      "loss": 0.1847,
      "step": 5296
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3707788468278712,
      "learning_rate": 2.4808739314481196e-06,
      "loss": 0.1729,
      "step": 5297
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.465363233450658,
      "learning_rate": 2.4790833829551764e-06,
      "loss": 0.6479,
      "step": 5298
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3784377252353808,
      "learning_rate": 2.4772932678370255e-06,
      "loss": 0.1736,
      "step": 5299
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.331753230187265,
      "learning_rate": 2.475503586401407e-06,
      "loss": 0.2041,
      "step": 5300
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5146794427877184,
      "learning_rate": 2.473714338955987e-06,
      "loss": 0.2073,
      "step": 5301
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2946927642884203,
      "learning_rate": 2.4719255258083614e-06,
      "loss": 0.1713,
      "step": 5302
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2085808686689654,
      "learning_rate": 2.4701371472660433e-06,
      "loss": 0.1551,
      "step": 5303
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.419383243836705,
      "learning_rate": 2.4683492036364805e-06,
      "loss": 0.2164,
      "step": 5304
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.227976986157876,
      "learning_rate": 2.466561695227039e-06,
      "loss": 0.5131,
      "step": 5305
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4175259427028641,
      "learning_rate": 2.4647746223450123e-06,
      "loss": 0.1841,
      "step": 5306
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.541568530898554,
      "learning_rate": 2.462987985297619e-06,
      "loss": 0.19,
      "step": 5307
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4177823537012886,
      "learning_rate": 2.461201784392002e-06,
      "loss": 0.1702,
      "step": 5308
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4470928090724537,
      "learning_rate": 2.459416019935229e-06,
      "loss": 0.1908,
      "step": 5309
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3361034565950598,
      "learning_rate": 2.4576306922342956e-06,
      "loss": 0.1754,
      "step": 5310
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3384833007019614,
      "learning_rate": 2.4558458015961213e-06,
      "loss": 0.154,
      "step": 5311
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3782182667931577,
      "learning_rate": 2.4540613483275484e-06,
      "loss": 0.1772,
      "step": 5312
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.361563408693365,
      "learning_rate": 2.4522773327353446e-06,
      "loss": 0.1791,
      "step": 5313
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.599091138215553,
      "learning_rate": 2.4504937551262024e-06,
      "loss": 0.2019,
      "step": 5314
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5428301295830016,
      "learning_rate": 2.448710615806741e-06,
      "loss": 0.1861,
      "step": 5315
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3840111450247778,
      "learning_rate": 2.4469279150834996e-06,
      "loss": 0.216,
      "step": 5316
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.203749969005446,
      "learning_rate": 2.445145653262949e-06,
      "loss": 0.1639,
      "step": 5317
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4047130311251945,
      "learning_rate": 2.4433638306514794e-06,
      "loss": 0.1634,
      "step": 5318
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.541461394308983,
      "learning_rate": 2.4415824475554046e-06,
      "loss": 0.5094,
      "step": 5319
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.27974828402653,
      "learning_rate": 2.439801504280968e-06,
      "loss": 0.573,
      "step": 5320
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.987847874939284,
      "learning_rate": 2.438021001134334e-06,
      "loss": 0.5583,
      "step": 5321
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3495821997739614,
      "learning_rate": 2.436240938421591e-06,
      "loss": 0.2004,
      "step": 5322
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4454674550139177,
      "learning_rate": 2.4344613164487513e-06,
      "loss": 0.203,
      "step": 5323
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.541065017164165,
      "learning_rate": 2.432682135521752e-06,
      "loss": 0.2173,
      "step": 5324
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.053335672564174,
      "learning_rate": 2.430903395946459e-06,
      "loss": 0.5835,
      "step": 5325
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3564989749102674,
      "learning_rate": 2.429125098028655e-06,
      "loss": 0.1927,
      "step": 5326
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.324745842337667,
      "learning_rate": 2.427347242074051e-06,
      "loss": 0.6536,
      "step": 5327
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5747115326379686,
      "learning_rate": 2.4255698283882794e-06,
      "loss": 0.2357,
      "step": 5328
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.10227352752096,
      "learning_rate": 2.4237928572769016e-06,
      "loss": 0.1647,
      "step": 5329
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4941439784265593,
      "learning_rate": 2.4220163290453973e-06,
      "loss": 0.2027,
      "step": 5330
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.11773939273154,
      "learning_rate": 2.4202402439991708e-06,
      "loss": 0.1619,
      "step": 5331
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5264302097663547,
      "learning_rate": 2.4184646024435547e-06,
      "loss": 0.1986,
      "step": 5332
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4784881422134353,
      "learning_rate": 2.416689404683802e-06,
      "loss": 0.2091,
      "step": 5333
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.225272930763302,
      "learning_rate": 2.4149146510250885e-06,
      "loss": 0.4611,
      "step": 5334
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3658955403746809,
      "learning_rate": 2.4131403417725156e-06,
      "loss": 0.2,
      "step": 5335
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.529544597375931,
      "learning_rate": 2.4113664772311076e-06,
      "loss": 0.1873,
      "step": 5336
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4402113925733209,
      "learning_rate": 2.40959305770581e-06,
      "loss": 0.1826,
      "step": 5337
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6462204149632658,
      "learning_rate": 2.4078200835014983e-06,
      "loss": 0.2113,
      "step": 5338
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.645239565819005,
      "learning_rate": 2.4060475549229634e-06,
      "loss": 0.6956,
      "step": 5339
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5732693162775355,
      "learning_rate": 2.4042754722749274e-06,
      "loss": 0.2371,
      "step": 5340
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.4246421758985655,
      "learning_rate": 2.40250383586203e-06,
      "loss": 0.6139,
      "step": 5341
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2464616885433646,
      "learning_rate": 2.400732645988836e-06,
      "loss": 0.1692,
      "step": 5342
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3088541895819912,
      "learning_rate": 2.398961902959833e-06,
      "loss": 0.1542,
      "step": 5343
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3805356524017314,
      "learning_rate": 2.3971916070794337e-06,
      "loss": 0.2022,
      "step": 5344
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5839725994924458,
      "learning_rate": 2.3954217586519692e-06,
      "loss": 0.2076,
      "step": 5345
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2802951243391198,
      "learning_rate": 2.3936523579817005e-06,
      "loss": 0.1679,
      "step": 5346
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4133033952161447,
      "learning_rate": 2.391883405372805e-06,
      "loss": 0.229,
      "step": 5347
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3234903904705697,
      "learning_rate": 2.3901149011293906e-06,
      "loss": 0.1687,
      "step": 5348
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5901447926767596,
      "learning_rate": 2.388346845555481e-06,
      "loss": 0.1522,
      "step": 5349
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.729249231465275,
      "learning_rate": 2.386579238955025e-06,
      "loss": 0.6744,
      "step": 5350
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4305536986973966,
      "learning_rate": 2.384812081631895e-06,
      "loss": 0.1747,
      "step": 5351
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.26780429470465,
      "learning_rate": 2.3830453738898852e-06,
      "loss": 0.2127,
      "step": 5352
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4452910659315799,
      "learning_rate": 2.3812791160327154e-06,
      "loss": 0.1871,
      "step": 5353
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1921031239794075,
      "learning_rate": 2.379513308364024e-06,
      "loss": 0.12,
      "step": 5354
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3125580073066054,
      "learning_rate": 2.377747951187375e-06,
      "loss": 0.1614,
      "step": 5355
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3337510104913997,
      "learning_rate": 2.3759830448062497e-06,
      "loss": 0.1887,
      "step": 5356
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4163719655620546,
      "learning_rate": 2.3742185895240623e-06,
      "loss": 0.1409,
      "step": 5357
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.190004536960456,
      "learning_rate": 2.3724545856441394e-06,
      "loss": 0.6013,
      "step": 5358
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4225171995809676,
      "learning_rate": 2.370691033469735e-06,
      "loss": 0.193,
      "step": 5359
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.489080773652712,
      "learning_rate": 2.368927933304021e-06,
      "loss": 0.2043,
      "step": 5360
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.731402968054587,
      "learning_rate": 2.3671652854500995e-06,
      "loss": 0.2375,
      "step": 5361
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.465954312024473,
      "learning_rate": 2.3654030902109877e-06,
      "loss": 0.1921,
      "step": 5362
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5256723352476846,
      "learning_rate": 2.3636413478896276e-06,
      "loss": 0.1857,
      "step": 5363
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2774822802167465,
      "learning_rate": 2.361880058788883e-06,
      "loss": 0.1617,
      "step": 5364
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.289616096453146,
      "learning_rate": 2.3601192232115376e-06,
      "loss": 0.1517,
      "step": 5365
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.3504709359241565,
      "learning_rate": 2.3583588414603046e-06,
      "loss": 0.5538,
      "step": 5366
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.327367017128275,
      "learning_rate": 2.3565989138378094e-06,
      "loss": 0.1856,
      "step": 5367
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.136402335016454,
      "learning_rate": 2.3548394406466074e-06,
      "loss": 0.4779,
      "step": 5368
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3916748923045383,
      "learning_rate": 2.3530804221891707e-06,
      "loss": 0.1821,
      "step": 5369
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3148408970548093,
      "learning_rate": 2.3513218587678952e-06,
      "loss": 0.1792,
      "step": 5370
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4390712450073762,
      "learning_rate": 2.349563750685099e-06,
      "loss": 0.2124,
      "step": 5371
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3120172022323229,
      "learning_rate": 2.3478060982430193e-06,
      "loss": 0.1449,
      "step": 5372
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3178499179116743,
      "learning_rate": 2.3460489017438176e-06,
      "loss": 0.1522,
      "step": 5373
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3749725447657244,
      "learning_rate": 2.3442921614895783e-06,
      "loss": 0.1655,
      "step": 5374
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6321681978447928,
      "learning_rate": 2.342535877782302e-06,
      "loss": 0.2269,
      "step": 5375
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.39557321135757,
      "learning_rate": 2.340780050923918e-06,
      "loss": 0.185,
      "step": 5376
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.522335184401836,
      "learning_rate": 2.339024681216272e-06,
      "loss": 0.1956,
      "step": 5377
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3575109746268366,
      "learning_rate": 2.3372697689611317e-06,
      "loss": 0.1953,
      "step": 5378
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3266428226732725,
      "learning_rate": 2.3355153144601873e-06,
      "loss": 0.1701,
      "step": 5379
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4399985771523496,
      "learning_rate": 2.3337613180150497e-06,
      "loss": 0.2053,
      "step": 5380
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.425187867297747,
      "learning_rate": 2.3320077799272493e-06,
      "loss": 0.2025,
      "step": 5381
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2917692964754048,
      "learning_rate": 2.3302547004982434e-06,
      "loss": 0.1514,
      "step": 5382
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2979040714438053,
      "learning_rate": 2.328502080029405e-06,
      "loss": 0.1881,
      "step": 5383
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.293905657381074,
      "learning_rate": 2.326749918822028e-06,
      "loss": 0.1757,
      "step": 5384
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5893330829101497,
      "learning_rate": 2.3249982171773322e-06,
      "loss": 0.1896,
      "step": 5385
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.566427518991508,
      "learning_rate": 2.323246975396454e-06,
      "loss": 0.5616,
      "step": 5386
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3242345831660642,
      "learning_rate": 2.321496193780453e-06,
      "loss": 0.1856,
      "step": 5387
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6207924017243704,
      "learning_rate": 2.3197458726303064e-06,
      "loss": 0.2179,
      "step": 5388
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4082641490297538,
      "learning_rate": 2.3179960122469173e-06,
      "loss": 0.1874,
      "step": 5389
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2589290651556828,
      "learning_rate": 2.3162466129311074e-06,
      "loss": 0.169,
      "step": 5390
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3645610275689604,
      "learning_rate": 2.314497674983617e-06,
      "loss": 0.178,
      "step": 5391
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4217253727690837,
      "learning_rate": 2.3127491987051086e-06,
      "loss": 0.2036,
      "step": 5392
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5376832537857463,
      "learning_rate": 2.311001184396165e-06,
      "loss": 0.1781,
      "step": 5393
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4691351226480294,
      "learning_rate": 2.3092536323572933e-06,
      "loss": 0.2076,
      "step": 5394
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.252481319856805,
      "learning_rate": 2.307506542888916e-06,
      "loss": 0.1429,
      "step": 5395
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.322347875495963,
      "learning_rate": 2.305759916291376e-06,
      "loss": 0.1826,
      "step": 5396
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6090461694102869,
      "learning_rate": 2.3040137528649424e-06,
      "loss": 0.1839,
      "step": 5397
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.386825272738509,
      "learning_rate": 2.3022680529097995e-06,
      "loss": 0.18,
      "step": 5398
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.1651817293574425,
      "learning_rate": 2.300522816726053e-06,
      "loss": 0.6478,
      "step": 5399
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3787380491883126,
      "learning_rate": 2.298778044613729e-06,
      "loss": 0.1905,
      "step": 5400
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.293315659525096,
      "learning_rate": 2.2970337368727753e-06,
      "loss": 0.1578,
      "step": 5401
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4801843889920492,
      "learning_rate": 2.2952898938030554e-06,
      "loss": 0.2006,
      "step": 5402
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.313989589274908,
      "learning_rate": 2.293546515704358e-06,
      "loss": 0.1913,
      "step": 5403
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.226904777278391,
      "learning_rate": 2.2918036028763928e-06,
      "loss": 0.1279,
      "step": 5404
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5222619099522106,
      "learning_rate": 2.290061155618784e-06,
      "loss": 0.2259,
      "step": 5405
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.883154383727351,
      "learning_rate": 2.2883191742310795e-06,
      "loss": 0.45,
      "step": 5406
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.454515639466296,
      "learning_rate": 2.2865776590127447e-06,
      "loss": 0.194,
      "step": 5407
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.387182559000789,
      "learning_rate": 2.2848366102631674e-06,
      "loss": 0.1833,
      "step": 5408
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4773354517709407,
      "learning_rate": 2.283096028281652e-06,
      "loss": 0.1802,
      "step": 5409
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5017213931703262,
      "learning_rate": 2.281355913367428e-06,
      "loss": 0.2055,
      "step": 5410
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.753289919223414,
      "learning_rate": 2.2796162658196397e-06,
      "loss": 0.6322,
      "step": 5411
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.348878795903292,
      "learning_rate": 2.2778770859373504e-06,
      "loss": 0.1864,
      "step": 5412
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.387619173519608,
      "learning_rate": 2.27613837401955e-06,
      "loss": 0.1767,
      "step": 5413
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4680712170764396,
      "learning_rate": 2.2744001303651407e-06,
      "loss": 0.1791,
      "step": 5414
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.369861931226433,
      "learning_rate": 2.2726623552729473e-06,
      "loss": 0.1827,
      "step": 5415
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6339889194870636,
      "learning_rate": 2.2709250490417124e-06,
      "loss": 0.2373,
      "step": 5416
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6969211281963767,
      "learning_rate": 2.2691882119700983e-06,
      "loss": 0.2264,
      "step": 5417
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3108513020263763,
      "learning_rate": 2.2674518443566908e-06,
      "loss": 0.1657,
      "step": 5418
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4614087332824146,
      "learning_rate": 2.265715946499989e-06,
      "loss": 0.1734,
      "step": 5419
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5518567701430177,
      "learning_rate": 2.2639805186984148e-06,
      "loss": 0.1803,
      "step": 5420
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2862669853035078,
      "learning_rate": 2.2622455612503064e-06,
      "loss": 0.1902,
      "step": 5421
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5195864431052235,
      "learning_rate": 2.2605110744539266e-06,
      "loss": 0.1989,
      "step": 5422
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.239039659211695,
      "learning_rate": 2.258777058607452e-06,
      "loss": 0.1741,
      "step": 5423
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5043006140470483,
      "learning_rate": 2.257043514008978e-06,
      "loss": 0.2031,
      "step": 5424
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4587548036664766,
      "learning_rate": 2.255310440956525e-06,
      "loss": 0.1887,
      "step": 5425
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5205091903677748,
      "learning_rate": 2.2535778397480263e-06,
      "loss": 0.2177,
      "step": 5426
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.448137617276376,
      "learning_rate": 2.251845710681336e-06,
      "loss": 0.1953,
      "step": 5427
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.923499296029548,
      "learning_rate": 2.250114054054228e-06,
      "loss": 0.6598,
      "step": 5428
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3815292984929315,
      "learning_rate": 2.2483828701643933e-06,
      "loss": 0.1837,
      "step": 5429
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4777980996288151,
      "learning_rate": 2.2466521593094404e-06,
      "loss": 0.1781,
      "step": 5430
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.385144448988506,
      "learning_rate": 2.2449219217869013e-06,
      "loss": 0.1809,
      "step": 5431
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4227031993637664,
      "learning_rate": 2.243192157894225e-06,
      "loss": 0.1942,
      "step": 5432
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.466377421345802,
      "learning_rate": 2.241462867928777e-06,
      "loss": 0.1841,
      "step": 5433
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.597346885262702,
      "learning_rate": 2.239734052187841e-06,
      "loss": 0.1892,
      "step": 5434
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2641157166178618,
      "learning_rate": 2.2380057109686213e-06,
      "loss": 0.1516,
      "step": 5435
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.342322843155115,
      "learning_rate": 2.2362778445682403e-06,
      "loss": 0.1949,
      "step": 5436
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1929234687838361,
      "learning_rate": 2.234550453283737e-06,
      "loss": 0.147,
      "step": 5437
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4365093259087027,
      "learning_rate": 2.2328235374120694e-06,
      "loss": 0.1759,
      "step": 5438
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.825611300595502,
      "learning_rate": 2.231097097250115e-06,
      "loss": 0.6137,
      "step": 5439
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5266653686650753,
      "learning_rate": 2.229371133094671e-06,
      "loss": 0.2389,
      "step": 5440
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2757831714094772,
      "learning_rate": 2.22764564524245e-06,
      "loss": 0.1862,
      "step": 5441
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4271101474550305,
      "learning_rate": 2.225920633990082e-06,
      "loss": 0.1833,
      "step": 5442
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3910199434705315,
      "learning_rate": 2.2241960996341166e-06,
      "loss": 0.2322,
      "step": 5443
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.681887559822497,
      "learning_rate": 2.2224720424710222e-06,
      "loss": 0.5383,
      "step": 5444
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.409936143349533,
      "learning_rate": 2.2207484627971817e-06,
      "loss": 0.4611,
      "step": 5445
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2934121909678589,
      "learning_rate": 2.2190253609089014e-06,
      "loss": 0.1417,
      "step": 5446
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4055831716802598,
      "learning_rate": 2.217302737102402e-06,
      "loss": 0.1958,
      "step": 5447
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3922213405655104,
      "learning_rate": 2.2155805916738215e-06,
      "loss": 0.1801,
      "step": 5448
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.405507700438091,
      "learning_rate": 2.2138589249192156e-06,
      "loss": 0.1802,
      "step": 5449
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.488008329440083,
      "learning_rate": 2.212137737134562e-06,
      "loss": 0.2036,
      "step": 5450
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4338550215774795,
      "learning_rate": 2.2104170286157506e-06,
      "loss": 0.1828,
      "step": 5451
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3508611306724336,
      "learning_rate": 2.20869679965859e-06,
      "loss": 0.1577,
      "step": 5452
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1345548612001803,
      "learning_rate": 2.206977050558811e-06,
      "loss": 0.1693,
      "step": 5453
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.26986844734452,
      "learning_rate": 2.205257781612057e-06,
      "loss": 0.167,
      "step": 5454
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3674926169564723,
      "learning_rate": 2.2035389931138896e-06,
      "loss": 0.1952,
      "step": 5455
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5321015688342412,
      "learning_rate": 2.201820685359788e-06,
      "loss": 0.2065,
      "step": 5456
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5362775776892703,
      "learning_rate": 2.20010285864515e-06,
      "loss": 0.1655,
      "step": 5457
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3915643511760947,
      "learning_rate": 2.198385513265289e-06,
      "loss": 0.1601,
      "step": 5458
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.116031118825757,
      "learning_rate": 2.1966686495154375e-06,
      "loss": 0.5299,
      "step": 5459
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4609012829951267,
      "learning_rate": 2.194952267690744e-06,
      "loss": 0.2029,
      "step": 5460
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4274111702080068,
      "learning_rate": 2.193236368086275e-06,
      "loss": 0.155,
      "step": 5461
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.8312277790268565,
      "learning_rate": 2.191520950997014e-06,
      "loss": 0.3334,
      "step": 5462
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7406102659563545,
      "learning_rate": 2.1898060167178604e-06,
      "loss": 0.198,
      "step": 5463
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4838245139095119,
      "learning_rate": 2.18809156554363e-06,
      "loss": 0.2126,
      "step": 5464
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2423063656765283,
      "learning_rate": 2.1863775977690588e-06,
      "loss": 0.1479,
      "step": 5465
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5088312206973562,
      "learning_rate": 2.1846641136887947e-06,
      "loss": 0.1782,
      "step": 5466
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.315579148158016,
      "learning_rate": 2.182951113597408e-06,
      "loss": 0.1829,
      "step": 5467
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3255515705177376,
      "learning_rate": 2.1812385977893844e-06,
      "loss": 0.1546,
      "step": 5468
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.2396236271178145,
      "learning_rate": 2.1795265665591236e-06,
      "loss": 0.5282,
      "step": 5469
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.429521982194587,
      "learning_rate": 2.177815020200944e-06,
      "loss": 0.6609,
      "step": 5470
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.585582048262242,
      "learning_rate": 2.17610395900908e-06,
      "loss": 0.2369,
      "step": 5471
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.366541682973003,
      "learning_rate": 2.174393383277683e-06,
      "loss": 0.6173,
      "step": 5472
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3936156051261976,
      "learning_rate": 2.1726832933008185e-06,
      "loss": 0.2105,
      "step": 5473
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.199439635764082,
      "learning_rate": 2.170973689372475e-06,
      "loss": 0.1831,
      "step": 5474
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3260523111327263,
      "learning_rate": 2.1692645717865515e-06,
      "loss": 0.1499,
      "step": 5475
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7202172894735415,
      "learning_rate": 2.1675559408368623e-06,
      "loss": 0.1942,
      "step": 5476
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3567545841852828,
      "learning_rate": 2.165847796817145e-06,
      "loss": 0.1606,
      "step": 5477
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.432136294096259,
      "learning_rate": 2.164140140021049e-06,
      "loss": 0.1847,
      "step": 5478
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3795234096126523,
      "learning_rate": 2.1624329707421374e-06,
      "loss": 0.1776,
      "step": 5479
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3353848767803613,
      "learning_rate": 2.1607262892738956e-06,
      "loss": 0.2189,
      "step": 5480
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3758818887325688,
      "learning_rate": 2.1590200959097173e-06,
      "loss": 0.2151,
      "step": 5481
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5761506705011632,
      "learning_rate": 2.157314390942923e-06,
      "loss": 0.2402,
      "step": 5482
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3285422021967048,
      "learning_rate": 2.1556091746667392e-06,
      "loss": 0.1755,
      "step": 5483
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.120741520575493,
      "learning_rate": 2.1539044473743136e-06,
      "loss": 0.5631,
      "step": 5484
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4448959456467982,
      "learning_rate": 2.152200209358708e-06,
      "loss": 0.1835,
      "step": 5485
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.339068010979368,
      "learning_rate": 2.1504964609128994e-06,
      "loss": 0.2149,
      "step": 5486
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.406380981309929,
      "learning_rate": 2.1487932023297853e-06,
      "loss": 0.2055,
      "step": 5487
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1400127939608924,
      "learning_rate": 2.147090433902172e-06,
      "loss": 0.1431,
      "step": 5488
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.411931439874426,
      "learning_rate": 2.1453881559227883e-06,
      "loss": 0.1748,
      "step": 5489
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.441213029730731,
      "learning_rate": 2.143686368684274e-06,
      "loss": 0.1766,
      "step": 5490
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2737957486576543,
      "learning_rate": 2.141985072479187e-06,
      "loss": 0.202,
      "step": 5491
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6180475731688824,
      "learning_rate": 2.1402842675999978e-06,
      "loss": 0.1936,
      "step": 5492
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.397667665158094,
      "learning_rate": 2.1385839543390967e-06,
      "loss": 0.1658,
      "step": 5493
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.467682095716239,
      "learning_rate": 2.136884132988784e-06,
      "loss": 0.5589,
      "step": 5494
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.533636025660675,
      "learning_rate": 2.1351848038412832e-06,
      "loss": 0.1629,
      "step": 5495
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3850694120182823,
      "learning_rate": 2.1334859671887236e-06,
      "loss": 0.2402,
      "step": 5496
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4143702689404543,
      "learning_rate": 2.1317876233231606e-06,
      "loss": 0.1874,
      "step": 5497
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.844682320367659,
      "learning_rate": 2.1300897725365555e-06,
      "loss": 0.4681,
      "step": 5498
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2804629783160486,
      "learning_rate": 2.12839241512079e-06,
      "loss": 0.146,
      "step": 5499
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1577644153831672,
      "learning_rate": 2.1266955513676584e-06,
      "loss": 0.1494,
      "step": 5500
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3783554944396261,
      "learning_rate": 2.124999181568872e-06,
      "loss": 0.1836,
      "step": 5501
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2851039454093809,
      "learning_rate": 2.1233033060160542e-06,
      "loss": 0.1842,
      "step": 5502
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.415511382867731,
      "learning_rate": 2.121607925000749e-06,
      "loss": 0.1766,
      "step": 5503
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4149897934192361,
      "learning_rate": 2.1199130388144098e-06,
      "loss": 0.1626,
      "step": 5504
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.472144544173594,
      "learning_rate": 2.1182186477484094e-06,
      "loss": 0.1981,
      "step": 5505
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3312525156453632,
      "learning_rate": 2.1165247520940317e-06,
      "loss": 0.1444,
      "step": 5506
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4008438168488095,
      "learning_rate": 2.114831352142478e-06,
      "loss": 0.1959,
      "step": 5507
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.569428686106568,
      "learning_rate": 2.1131384481848614e-06,
      "loss": 0.6655,
      "step": 5508
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2814341554127096,
      "learning_rate": 2.111446040512212e-06,
      "loss": 0.1688,
      "step": 5509
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.528858741450897,
      "learning_rate": 2.1097541294154773e-06,
      "loss": 0.4478,
      "step": 5510
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.73276830583751,
      "learning_rate": 2.108062715185514e-06,
      "loss": 0.5653,
      "step": 5511
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.229382259960209,
      "learning_rate": 2.1063717981130952e-06,
      "loss": 0.7126,
      "step": 5512
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.212914345080245,
      "learning_rate": 2.10468137848891e-06,
      "loss": 0.6339,
      "step": 5513
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3602062861050765,
      "learning_rate": 2.102991456603562e-06,
      "loss": 0.202,
      "step": 5514
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.413544638911263,
      "learning_rate": 2.1013020327475683e-06,
      "loss": 0.195,
      "step": 5515
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3968991123089414,
      "learning_rate": 2.099613107211359e-06,
      "loss": 0.1523,
      "step": 5516
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3744901152873332,
      "learning_rate": 2.0979246802852794e-06,
      "loss": 0.1954,
      "step": 5517
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3422575069912117,
      "learning_rate": 2.096236752259592e-06,
      "loss": 0.1661,
      "step": 5518
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2862793725791561,
      "learning_rate": 2.0945493234244714e-06,
      "loss": 0.1433,
      "step": 5519
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5276530215624327,
      "learning_rate": 2.0928623940700044e-06,
      "loss": 0.1567,
      "step": 5520
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7388997796561687,
      "learning_rate": 2.091175964486194e-06,
      "loss": 0.2499,
      "step": 5521
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.257265654501349,
      "learning_rate": 2.0894900349629576e-06,
      "loss": 0.527,
      "step": 5522
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4972056104322833,
      "learning_rate": 2.087804605790124e-06,
      "loss": 0.1467,
      "step": 5523
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3301138902885676,
      "learning_rate": 2.08611967725744e-06,
      "loss": 0.1811,
      "step": 5524
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.292352259096499,
      "learning_rate": 2.0844352496545652e-06,
      "loss": 0.1482,
      "step": 5525
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4779981359919239,
      "learning_rate": 2.0827513232710716e-06,
      "loss": 0.1905,
      "step": 5526
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.326218363499624,
      "learning_rate": 2.081067898396445e-06,
      "loss": 0.1772,
      "step": 5527
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4311526502283873,
      "learning_rate": 2.0793849753200855e-06,
      "loss": 0.1988,
      "step": 5528
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.760071579858007,
      "learning_rate": 2.077702554331308e-06,
      "loss": 0.6025,
      "step": 5529
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.458306409811951,
      "learning_rate": 2.0760206357193373e-06,
      "loss": 0.2108,
      "step": 5530
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4463082912046912,
      "learning_rate": 2.0743392197733193e-06,
      "loss": 0.1577,
      "step": 5531
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.325891247422393,
      "learning_rate": 2.0726583067823046e-06,
      "loss": 0.5625,
      "step": 5532
    },
    {
      "epoch": 0.71,
      "grad_norm": 8.03988560714396,
      "learning_rate": 2.0709778970352657e-06,
      "loss": 0.496,
      "step": 5533
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.437620374742736,
      "learning_rate": 2.069297990821082e-06,
      "loss": 0.2014,
      "step": 5534
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2914405035100167,
      "learning_rate": 2.0676185884285495e-06,
      "loss": 0.1605,
      "step": 5535
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5900664847818156,
      "learning_rate": 2.0659396901463764e-06,
      "loss": 0.2349,
      "step": 5536
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5142938723994919,
      "learning_rate": 2.064261296263185e-06,
      "loss": 0.2075,
      "step": 5537
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.352825957588408,
      "learning_rate": 2.0625834070675094e-06,
      "loss": 0.1744,
      "step": 5538
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.896863736519962,
      "learning_rate": 2.0609060228478017e-06,
      "loss": 0.4959,
      "step": 5539
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2911060258640257,
      "learning_rate": 2.0592291438924213e-06,
      "loss": 0.1731,
      "step": 5540
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.684813973632871,
      "learning_rate": 2.0575527704896414e-06,
      "loss": 0.585,
      "step": 5541
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5662626822581882,
      "learning_rate": 2.055876902927654e-06,
      "loss": 0.2281,
      "step": 5542
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4911675001620204,
      "learning_rate": 2.0542015414945577e-06,
      "loss": 0.1744,
      "step": 5543
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2224886367230319,
      "learning_rate": 2.0525266864783676e-06,
      "loss": 0.153,
      "step": 5544
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3074163182099652,
      "learning_rate": 2.050852338167008e-06,
      "loss": 0.1681,
      "step": 5545
    },
    {
      "epoch": 0.71,
      "grad_norm": 23.756183191343645,
      "learning_rate": 2.049178496848323e-06,
      "loss": 0.4849,
      "step": 5546
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.989225783873516,
      "learning_rate": 2.0475051628100635e-06,
      "loss": 0.5869,
      "step": 5547
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4284256112295985,
      "learning_rate": 2.045832336339894e-06,
      "loss": 0.1665,
      "step": 5548
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2962913115002601,
      "learning_rate": 2.044160017725394e-06,
      "loss": 0.2042,
      "step": 5549
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.600869071171168,
      "learning_rate": 2.042488207254054e-06,
      "loss": 0.2217,
      "step": 5550
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3344231265262847,
      "learning_rate": 2.040816905213276e-06,
      "loss": 0.2097,
      "step": 5551
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5513910267753563,
      "learning_rate": 2.0391461118903788e-06,
      "loss": 0.1971,
      "step": 5552
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.431983022958495,
      "learning_rate": 2.0374758275725893e-06,
      "loss": 0.4761,
      "step": 5553
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.26395162967894,
      "learning_rate": 2.0358060525470507e-06,
      "loss": 0.1703,
      "step": 5554
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3244564411327824,
      "learning_rate": 2.0341367871008154e-06,
      "loss": 0.1985,
      "step": 5555
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.445650947638917,
      "learning_rate": 2.0324680315208505e-06,
      "loss": 0.2151,
      "step": 5556
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3965436639507807,
      "learning_rate": 2.0307997860940333e-06,
      "loss": 0.1815,
      "step": 5557
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.397577227980064,
      "learning_rate": 2.0291320511071544e-06,
      "loss": 0.1826,
      "step": 5558
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3406371732594118,
      "learning_rate": 2.0274648268469154e-06,
      "loss": 0.1984,
      "step": 5559
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5518081357670404,
      "learning_rate": 2.025798113599933e-06,
      "loss": 0.1733,
      "step": 5560
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4718604351689097,
      "learning_rate": 2.0241319116527376e-06,
      "loss": 0.2017,
      "step": 5561
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.301563195617497,
      "learning_rate": 2.022466221291765e-06,
      "loss": 0.1571,
      "step": 5562
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3299537419533733,
      "learning_rate": 2.0208010428033675e-06,
      "loss": 0.1545,
      "step": 5563
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3445809233640735,
      "learning_rate": 2.0191363764738087e-06,
      "loss": 0.1636,
      "step": 5564
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3528799889746674,
      "learning_rate": 2.017472222589264e-06,
      "loss": 0.1912,
      "step": 5565
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4920552588378406,
      "learning_rate": 2.0158085814358187e-06,
      "loss": 0.1884,
      "step": 5566
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.483017420946941,
      "learning_rate": 2.014145453299476e-06,
      "loss": 0.1835,
      "step": 5567
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3910950318754554,
      "learning_rate": 2.012482838466145e-06,
      "loss": 0.1858,
      "step": 5568
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.3084104725388395,
      "learning_rate": 2.010820737221646e-06,
      "loss": 0.6132,
      "step": 5569
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.976292300781279,
      "learning_rate": 2.0091591498517184e-06,
      "loss": 0.4779,
      "step": 5570
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4492127764875777,
      "learning_rate": 2.007498076642005e-06,
      "loss": 0.185,
      "step": 5571
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.476245211787532,
      "learning_rate": 2.0058375178780644e-06,
      "loss": 0.2109,
      "step": 5572
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4854051037161684,
      "learning_rate": 2.004177473845366e-06,
      "loss": 0.1708,
      "step": 5573
    },
    {
      "epoch": 0.71,
      "grad_norm": 13.500310693213674,
      "learning_rate": 2.0025179448292886e-06,
      "loss": 0.6986,
      "step": 5574
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4083782104006706,
      "learning_rate": 2.000858931115128e-06,
      "loss": 0.1713,
      "step": 5575
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.448910116291861,
      "learning_rate": 1.999200432988086e-06,
      "loss": 0.2195,
      "step": 5576
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.47624778184355,
      "learning_rate": 1.997542450733278e-06,
      "loss": 0.5891,
      "step": 5577
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.498252008931897,
      "learning_rate": 1.9958849846357287e-06,
      "loss": 0.2167,
      "step": 5578
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3455319959924588,
      "learning_rate": 1.994228034980378e-06,
      "loss": 0.1531,
      "step": 5579
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3336816700588645,
      "learning_rate": 1.992571602052075e-06,
      "loss": 0.2148,
      "step": 5580
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.460621447954724,
      "learning_rate": 1.9909156861355767e-06,
      "loss": 0.1647,
      "step": 5581
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.389474649960917,
      "learning_rate": 1.9892602875155582e-06,
      "loss": 0.5139,
      "step": 5582
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2160785255202862,
      "learning_rate": 1.9876054064765993e-06,
      "loss": 0.1477,
      "step": 5583
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.535511615152349,
      "learning_rate": 1.9859510433031943e-06,
      "loss": 0.6654,
      "step": 5584
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.11452174434848,
      "learning_rate": 1.984297198279746e-06,
      "loss": 0.4252,
      "step": 5585
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4237814126913662,
      "learning_rate": 1.982643871690571e-06,
      "loss": 0.2074,
      "step": 5586
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5189736605184927,
      "learning_rate": 1.980991063819893e-06,
      "loss": 0.2027,
      "step": 5587
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5197093590065784,
      "learning_rate": 1.9793387749518517e-06,
      "loss": 0.1507,
      "step": 5588
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.368648627743058,
      "learning_rate": 1.9776870053704917e-06,
      "loss": 0.4397,
      "step": 5589
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4114086183228813,
      "learning_rate": 1.976035755359775e-06,
      "loss": 0.1562,
      "step": 5590
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3692525527006958,
      "learning_rate": 1.974385025203569e-06,
      "loss": 0.202,
      "step": 5591
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3465232139079382,
      "learning_rate": 1.9727348151856535e-06,
      "loss": 0.1891,
      "step": 5592
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3383487552048867,
      "learning_rate": 1.9710851255897173e-06,
      "loss": 0.1697,
      "step": 5593
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4835043983655725,
      "learning_rate": 1.969435956699363e-06,
      "loss": 0.2,
      "step": 5594
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.447325364359195,
      "learning_rate": 1.967787308798099e-06,
      "loss": 0.1822,
      "step": 5595
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2290061961842413,
      "learning_rate": 1.966139182169351e-06,
      "loss": 0.1758,
      "step": 5596
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.185643530911175,
      "learning_rate": 1.9644915770964472e-06,
      "loss": 0.1815,
      "step": 5597
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3080711446999422,
      "learning_rate": 1.9628444938626336e-06,
      "loss": 0.1433,
      "step": 5598
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5278609751380454,
      "learning_rate": 1.9611979327510617e-06,
      "loss": 0.1991,
      "step": 5599
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.592518249474307,
      "learning_rate": 1.9595518940447933e-06,
      "loss": 0.1633,
      "step": 5600
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2659358098869051,
      "learning_rate": 1.9579063780268026e-06,
      "loss": 0.1728,
      "step": 5601
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6186335385176605,
      "learning_rate": 1.9562613849799704e-06,
      "loss": 0.1772,
      "step": 5602
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3734843145207565,
      "learning_rate": 1.9546169151870943e-06,
      "loss": 0.1861,
      "step": 5603
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2673587260020267,
      "learning_rate": 1.9529729689308756e-06,
      "loss": 0.1832,
      "step": 5604
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5127027832619726,
      "learning_rate": 1.9513295464939274e-06,
      "loss": 0.1847,
      "step": 5605
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.217592277070974,
      "learning_rate": 1.9496866481587717e-06,
      "loss": 0.1319,
      "step": 5606
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.17368689860246,
      "learning_rate": 1.9480442742078455e-06,
      "loss": 0.1655,
      "step": 5607
    },
    {
      "epoch": 0.72,
      "grad_norm": 8.31311513132549,
      "learning_rate": 1.9464024249234895e-06,
      "loss": 0.6843,
      "step": 5608
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2062712601864496,
      "learning_rate": 1.9447611005879573e-06,
      "loss": 0.1488,
      "step": 5609
    },
    {
      "epoch": 0.72,
      "grad_norm": 10.408915747068265,
      "learning_rate": 1.9431203014834093e-06,
      "loss": 0.7516,
      "step": 5610
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7691445838024027,
      "learning_rate": 1.9414800278919223e-06,
      "loss": 0.1984,
      "step": 5611
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1877233410129542,
      "learning_rate": 1.9398402800954746e-06,
      "loss": 0.1495,
      "step": 5612
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3222239676265612,
      "learning_rate": 1.9382010583759604e-06,
      "loss": 0.1192,
      "step": 5613
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5301772280647628,
      "learning_rate": 1.936562363015179e-06,
      "loss": 0.2079,
      "step": 5614
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3462102622979955,
      "learning_rate": 1.9349241942948405e-06,
      "loss": 0.1782,
      "step": 5615
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2898202485127042,
      "learning_rate": 1.9332865524965677e-06,
      "loss": 0.1832,
      "step": 5616
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6315348664426554,
      "learning_rate": 1.9316494379018876e-06,
      "loss": 0.1782,
      "step": 5617
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1678253836726433,
      "learning_rate": 1.9300128507922417e-06,
      "loss": 0.1976,
      "step": 5618
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5632682748655369,
      "learning_rate": 1.9283767914489777e-06,
      "loss": 0.2033,
      "step": 5619
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.2344018983813765,
      "learning_rate": 1.926741260153352e-06,
      "loss": 0.3766,
      "step": 5620
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3525996815378518,
      "learning_rate": 1.925106257186532e-06,
      "loss": 0.2032,
      "step": 5621
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4889084967591402,
      "learning_rate": 1.923471782829594e-06,
      "loss": 0.2279,
      "step": 5622
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.308461450875394,
      "learning_rate": 1.921837837363521e-06,
      "loss": 0.1708,
      "step": 5623
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2731168794514685,
      "learning_rate": 1.92020442106921e-06,
      "loss": 0.1451,
      "step": 5624
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3459786557361144,
      "learning_rate": 1.918571534227462e-06,
      "loss": 0.169,
      "step": 5625
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.057260649924866,
      "learning_rate": 1.9169391771189915e-06,
      "loss": 0.4606,
      "step": 5626
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4613884143365685,
      "learning_rate": 1.915307350024419e-06,
      "loss": 0.2065,
      "step": 5627
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.426156013321384,
      "learning_rate": 1.913676053224273e-06,
      "loss": 0.164,
      "step": 5628
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3699915971383378,
      "learning_rate": 1.9120452869989943e-06,
      "loss": 0.1948,
      "step": 5629
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.366374752869595,
      "learning_rate": 1.9104150516289283e-06,
      "loss": 0.208,
      "step": 5630
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4622956630636499,
      "learning_rate": 1.9087853473943313e-06,
      "loss": 0.1947,
      "step": 5631
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.189167724665759,
      "learning_rate": 1.9071561745753715e-06,
      "loss": 0.703,
      "step": 5632
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3496609625866873,
      "learning_rate": 1.90552753345212e-06,
      "loss": 0.1992,
      "step": 5633
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4322776668828738,
      "learning_rate": 1.9038994243045582e-06,
      "loss": 0.1822,
      "step": 5634
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3481733106042295,
      "learning_rate": 1.90227184741258e-06,
      "loss": 0.1778,
      "step": 5635
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.642006552031785,
      "learning_rate": 1.9006448030559832e-06,
      "loss": 0.5441,
      "step": 5636
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3735228736148812,
      "learning_rate": 1.899018291514476e-06,
      "loss": 0.1904,
      "step": 5637
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4558450255549304,
      "learning_rate": 1.897392313067672e-06,
      "loss": 0.2207,
      "step": 5638
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.393790115600832,
      "learning_rate": 1.8957668679950997e-06,
      "loss": 0.1768,
      "step": 5639
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4580331108936178,
      "learning_rate": 1.8941419565761903e-06,
      "loss": 0.2133,
      "step": 5640
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4029259867355324,
      "learning_rate": 1.892517579090285e-06,
      "loss": 0.1512,
      "step": 5641
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5205078992738408,
      "learning_rate": 1.8908937358166323e-06,
      "loss": 0.197,
      "step": 5642
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4422298930459299,
      "learning_rate": 1.8892704270343887e-06,
      "loss": 0.2004,
      "step": 5643
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.274275258064602,
      "learning_rate": 1.8876476530226235e-06,
      "loss": 0.178,
      "step": 5644
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.176759818107953,
      "learning_rate": 1.8860254140603063e-06,
      "loss": 0.138,
      "step": 5645
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3203764703708496,
      "learning_rate": 1.8844037104263225e-06,
      "loss": 0.175,
      "step": 5646
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3832028750715308,
      "learning_rate": 1.8827825423994595e-06,
      "loss": 0.1889,
      "step": 5647
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.446204095939191,
      "learning_rate": 1.8811619102584155e-06,
      "loss": 0.2001,
      "step": 5648
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.448392226218094,
      "learning_rate": 1.8795418142817962e-06,
      "loss": 0.6225,
      "step": 5649
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4302065594210018,
      "learning_rate": 1.877922254748114e-06,
      "loss": 0.1891,
      "step": 5650
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4699444170801168,
      "learning_rate": 1.876303231935791e-06,
      "loss": 0.168,
      "step": 5651
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3985114276424588,
      "learning_rate": 1.8746847461231533e-06,
      "loss": 0.1594,
      "step": 5652
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5599911866392835,
      "learning_rate": 1.8730667975884398e-06,
      "loss": 0.1773,
      "step": 5653
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5379939384840176,
      "learning_rate": 1.8714493866097955e-06,
      "loss": 0.1962,
      "step": 5654
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.562144614984445,
      "learning_rate": 1.8698325134652711e-06,
      "loss": 0.3333,
      "step": 5655
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4895944586085927,
      "learning_rate": 1.8682161784328262e-06,
      "loss": 0.2035,
      "step": 5656
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.545075542344156,
      "learning_rate": 1.8666003817903267e-06,
      "loss": 0.6014,
      "step": 5657
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4776905627891244,
      "learning_rate": 1.8649851238155465e-06,
      "loss": 0.2272,
      "step": 5658
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.480092802583579,
      "learning_rate": 1.8633704047861667e-06,
      "loss": 0.5667,
      "step": 5659
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2365564051621003,
      "learning_rate": 1.8617562249797788e-06,
      "loss": 0.1884,
      "step": 5660
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4476366964550038,
      "learning_rate": 1.8601425846738775e-06,
      "loss": 0.1888,
      "step": 5661
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1903182766609854,
      "learning_rate": 1.858529484145864e-06,
      "loss": 0.1512,
      "step": 5662
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.450574129137202,
      "learning_rate": 1.8569169236730533e-06,
      "loss": 0.1838,
      "step": 5663
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5183194161301634,
      "learning_rate": 1.8553049035326615e-06,
      "loss": 0.1704,
      "step": 5664
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4418655690716096,
      "learning_rate": 1.8536934240018129e-06,
      "loss": 0.168,
      "step": 5665
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.291923674420243,
      "learning_rate": 1.852082485357538e-06,
      "loss": 0.1319,
      "step": 5666
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4529219441654155,
      "learning_rate": 1.8504720878767797e-06,
      "loss": 0.1843,
      "step": 5667
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3585625984208487,
      "learning_rate": 1.8488622318363814e-06,
      "loss": 0.2172,
      "step": 5668
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6288692131594031,
      "learning_rate": 1.847252917513097e-06,
      "loss": 0.2016,
      "step": 5669
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.353580851485481,
      "learning_rate": 1.845644145183586e-06,
      "loss": 0.2114,
      "step": 5670
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3948473357254998,
      "learning_rate": 1.844035915124413e-06,
      "loss": 0.1827,
      "step": 5671
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2051787521071529,
      "learning_rate": 1.8424282276120547e-06,
      "loss": 0.1283,
      "step": 5672
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.610216483536434,
      "learning_rate": 1.84082108292289e-06,
      "loss": 0.7399,
      "step": 5673
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3613826458731306,
      "learning_rate": 1.8392144813332041e-06,
      "loss": 0.1737,
      "step": 5674
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4358629692868343,
      "learning_rate": 1.8376084231191932e-06,
      "loss": 0.1854,
      "step": 5675
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4653287702736741,
      "learning_rate": 1.8360029085569558e-06,
      "loss": 0.2174,
      "step": 5676
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3883158286505306,
      "learning_rate": 1.8343979379224991e-06,
      "loss": 0.2037,
      "step": 5677
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.1809609024933065,
      "learning_rate": 1.832793511491735e-06,
      "loss": 0.4242,
      "step": 5678
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2108656167984122,
      "learning_rate": 1.831189629540484e-06,
      "loss": 0.1558,
      "step": 5679
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.642642979758562,
      "learning_rate": 1.8295862923444702e-06,
      "loss": 0.2342,
      "step": 5680
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.347472019231469,
      "learning_rate": 1.8279835001793272e-06,
      "loss": 0.438,
      "step": 5681
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3312519528754096,
      "learning_rate": 1.8263812533205955e-06,
      "loss": 0.1987,
      "step": 5682
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3254256360648922,
      "learning_rate": 1.8247795520437177e-06,
      "loss": 0.1678,
      "step": 5683
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.365864678596024,
      "learning_rate": 1.8231783966240458e-06,
      "loss": 0.1735,
      "step": 5684
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.55142972449211,
      "learning_rate": 1.8215777873368363e-06,
      "loss": 0.2056,
      "step": 5685
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6504928823036218,
      "learning_rate": 1.8199777244572525e-06,
      "loss": 0.2179,
      "step": 5686
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4361248019940025,
      "learning_rate": 1.8183782082603618e-06,
      "loss": 0.2127,
      "step": 5687
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3498422744165495,
      "learning_rate": 1.8167792390211435e-06,
      "loss": 0.1697,
      "step": 5688
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.029293704713685,
      "learning_rate": 1.8151808170144751e-06,
      "loss": 0.5983,
      "step": 5689
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.379591650417307,
      "learning_rate": 1.813582942515148e-06,
      "loss": 0.1892,
      "step": 5690
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3794444265381751,
      "learning_rate": 1.8119856157978534e-06,
      "loss": 0.2178,
      "step": 5691
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.095497781384439,
      "learning_rate": 1.8103888371371898e-06,
      "loss": 0.5262,
      "step": 5692
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.302543202187366,
      "learning_rate": 1.8087926068076622e-06,
      "loss": 0.1772,
      "step": 5693
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.426722521191241,
      "learning_rate": 1.8071969250836813e-06,
      "loss": 0.1905,
      "step": 5694
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3504949312835182,
      "learning_rate": 1.8056017922395607e-06,
      "loss": 0.2129,
      "step": 5695
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4273896728052173,
      "learning_rate": 1.8040072085495276e-06,
      "loss": 0.1818,
      "step": 5696
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.002122551178587,
      "learning_rate": 1.8024131742877054e-06,
      "loss": 0.4323,
      "step": 5697
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3075504416266623,
      "learning_rate": 1.8008196897281287e-06,
      "loss": 0.1742,
      "step": 5698
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.459838023910851,
      "learning_rate": 1.7992267551447334e-06,
      "loss": 0.194,
      "step": 5699
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.884272284782092,
      "learning_rate": 1.7976343708113675e-06,
      "loss": 0.687,
      "step": 5700
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2704025969174966,
      "learning_rate": 1.7960425370017782e-06,
      "loss": 0.136,
      "step": 5701
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.377923350412178,
      "learning_rate": 1.794451253989618e-06,
      "loss": 0.1634,
      "step": 5702
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3949324546063737,
      "learning_rate": 1.7928605220484513e-06,
      "loss": 0.1995,
      "step": 5703
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5326456450440256,
      "learning_rate": 1.7912703414517413e-06,
      "loss": 0.1868,
      "step": 5704
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7235945293262218,
      "learning_rate": 1.7896807124728582e-06,
      "loss": 0.1718,
      "step": 5705
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4079484259221764,
      "learning_rate": 1.788091635385078e-06,
      "loss": 0.168,
      "step": 5706
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.372884149878535,
      "learning_rate": 1.7865031104615809e-06,
      "loss": 0.2001,
      "step": 5707
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.886444157494076,
      "learning_rate": 1.784915137975452e-06,
      "loss": 0.2497,
      "step": 5708
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.366382324448913,
      "learning_rate": 1.783327718199685e-06,
      "loss": 0.1687,
      "step": 5709
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3561357266287477,
      "learning_rate": 1.7817408514071722e-06,
      "loss": 0.1882,
      "step": 5710
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3886299224960577,
      "learning_rate": 1.780154537870718e-06,
      "loss": 0.1739,
      "step": 5711
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3784807712214437,
      "learning_rate": 1.7785687778630268e-06,
      "loss": 0.1731,
      "step": 5712
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3619155204971924,
      "learning_rate": 1.776983571656708e-06,
      "loss": 0.1821,
      "step": 5713
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5207511517488794,
      "learning_rate": 1.7753989195242782e-06,
      "loss": 0.1706,
      "step": 5714
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.394384955668569,
      "learning_rate": 1.7738148217381568e-06,
      "loss": 0.1756,
      "step": 5715
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3602839332803045,
      "learning_rate": 1.772231278570667e-06,
      "loss": 0.1975,
      "step": 5716
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.683592426828661,
      "learning_rate": 1.7706482902940397e-06,
      "loss": 0.496,
      "step": 5717
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.432270082482894,
      "learning_rate": 1.7690658571804109e-06,
      "loss": 0.5838,
      "step": 5718
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4463642148170452,
      "learning_rate": 1.7674839795018173e-06,
      "loss": 0.2155,
      "step": 5719
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3643835234355075,
      "learning_rate": 1.7659026575302025e-06,
      "loss": 0.163,
      "step": 5720
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.518155095475931,
      "learning_rate": 1.764321891537414e-06,
      "loss": 0.1775,
      "step": 5721
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5855331180563097,
      "learning_rate": 1.7627416817952032e-06,
      "loss": 0.2026,
      "step": 5722
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2862461924593938,
      "learning_rate": 1.7611620285752246e-06,
      "loss": 0.158,
      "step": 5723
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.113056833001286,
      "learning_rate": 1.7595829321490437e-06,
      "loss": 0.1521,
      "step": 5724
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4060171838140612,
      "learning_rate": 1.7580043927881224e-06,
      "loss": 0.1677,
      "step": 5725
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6646586863022683,
      "learning_rate": 1.756426410763829e-06,
      "loss": 0.2021,
      "step": 5726
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.761114451478809,
      "learning_rate": 1.7548489863474393e-06,
      "loss": 0.5455,
      "step": 5727
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2523907026306713,
      "learning_rate": 1.753272119810131e-06,
      "loss": 0.151,
      "step": 5728
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3655079456267265,
      "learning_rate": 1.7516958114229837e-06,
      "loss": 0.1846,
      "step": 5729
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.999056863198303,
      "learning_rate": 1.7501200614569847e-06,
      "loss": 0.564,
      "step": 5730
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.314122330377528,
      "learning_rate": 1.7485448701830205e-06,
      "loss": 0.6555,
      "step": 5731
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3348307755688364,
      "learning_rate": 1.7469702378718894e-06,
      "loss": 0.1599,
      "step": 5732
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2599715568521983,
      "learning_rate": 1.7453961647942868e-06,
      "loss": 0.1614,
      "step": 5733
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.78588797092808,
      "learning_rate": 1.743822651220814e-06,
      "loss": 0.5439,
      "step": 5734
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.605935066736502,
      "learning_rate": 1.7422496974219761e-06,
      "loss": 0.5397,
      "step": 5735
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4816260724762973,
      "learning_rate": 1.7406773036681807e-06,
      "loss": 0.198,
      "step": 5736
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.354590315157479,
      "learning_rate": 1.7391054702297439e-06,
      "loss": 0.1954,
      "step": 5737
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5440924174334305,
      "learning_rate": 1.737534197376879e-06,
      "loss": 0.1974,
      "step": 5738
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2952707405027328,
      "learning_rate": 1.7359634853797081e-06,
      "loss": 0.1594,
      "step": 5739
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.509070748117032,
      "learning_rate": 1.7343933345082547e-06,
      "loss": 0.226,
      "step": 5740
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5679149325369406,
      "learning_rate": 1.7328237450324454e-06,
      "loss": 0.1836,
      "step": 5741
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3850256970950254,
      "learning_rate": 1.73125471722211e-06,
      "loss": 0.1886,
      "step": 5742
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4713199892656286,
      "learning_rate": 1.7296862513469836e-06,
      "loss": 0.1657,
      "step": 5743
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4614926130320758,
      "learning_rate": 1.7281183476767016e-06,
      "loss": 0.1848,
      "step": 5744
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.000850560449002,
      "learning_rate": 1.7265510064808084e-06,
      "loss": 0.5387,
      "step": 5745
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3444716890147728,
      "learning_rate": 1.7249842280287442e-06,
      "loss": 0.1832,
      "step": 5746
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3463425611022555,
      "learning_rate": 1.7234180125898608e-06,
      "loss": 0.1798,
      "step": 5747
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.336889779647927,
      "learning_rate": 1.721852360433406e-06,
      "loss": 0.1648,
      "step": 5748
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4510291809380547,
      "learning_rate": 1.7202872718285341e-06,
      "loss": 0.2061,
      "step": 5749
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.177627295919857,
      "learning_rate": 1.7187227470443013e-06,
      "loss": 0.5863,
      "step": 5750
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7282074312845679,
      "learning_rate": 1.7171587863496686e-06,
      "loss": 0.1849,
      "step": 5751
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5547721839288111,
      "learning_rate": 1.715595390013497e-06,
      "loss": 0.2033,
      "step": 5752
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.334058315723912,
      "learning_rate": 1.7140325583045553e-06,
      "loss": 0.6142,
      "step": 5753
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6181227095380564,
      "learning_rate": 1.7124702914915097e-06,
      "loss": 0.2269,
      "step": 5754
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.035937684821825,
      "learning_rate": 1.7109085898429345e-06,
      "loss": 0.5393,
      "step": 5755
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6540284307202249,
      "learning_rate": 1.7093474536273037e-06,
      "loss": 0.1838,
      "step": 5756
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.396431985967137,
      "learning_rate": 1.7077868831129935e-06,
      "loss": 0.1715,
      "step": 5757
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5536272198648147,
      "learning_rate": 1.7062268785682852e-06,
      "loss": 0.2233,
      "step": 5758
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4329897190863239,
      "learning_rate": 1.7046674402613594e-06,
      "loss": 0.1763,
      "step": 5759
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2171865594661944,
      "learning_rate": 1.703108568460305e-06,
      "loss": 0.1432,
      "step": 5760
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4195940213606677,
      "learning_rate": 1.7015502634331083e-06,
      "loss": 0.1754,
      "step": 5761
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.306683376856773,
      "learning_rate": 1.6999925254476606e-06,
      "loss": 0.1894,
      "step": 5762
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.758488109752004,
      "learning_rate": 1.698435354771753e-06,
      "loss": 0.3564,
      "step": 5763
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5501178760371974,
      "learning_rate": 1.6968787516730845e-06,
      "loss": 0.208,
      "step": 5764
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2485363189862118,
      "learning_rate": 1.6953227164192516e-06,
      "loss": 0.1648,
      "step": 5765
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.393962010195419,
      "learning_rate": 1.6937672492777547e-06,
      "loss": 0.1787,
      "step": 5766
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3082407350091376,
      "learning_rate": 1.6922123505159955e-06,
      "loss": 0.2123,
      "step": 5767
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.138687761578784,
      "learning_rate": 1.6906580204012818e-06,
      "loss": 0.5748,
      "step": 5768
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4544620353139552,
      "learning_rate": 1.68910425920082e-06,
      "loss": 0.1828,
      "step": 5769
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.580271878562585,
      "learning_rate": 1.6875510671817186e-06,
      "loss": 0.202,
      "step": 5770
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3817820573816646,
      "learning_rate": 1.6859984446109906e-06,
      "loss": 0.2056,
      "step": 5771
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.460218632751898,
      "learning_rate": 1.6844463917555487e-06,
      "loss": 0.226,
      "step": 5772
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2495429414646502,
      "learning_rate": 1.6828949088822077e-06,
      "loss": 0.1504,
      "step": 5773
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7237373612192837,
      "learning_rate": 1.6813439962576872e-06,
      "loss": 0.2174,
      "step": 5774
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.4054480764365325,
      "learning_rate": 1.6797936541486082e-06,
      "loss": 0.5358,
      "step": 5775
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.751417153299252,
      "learning_rate": 1.6782438828214913e-06,
      "loss": 0.5959,
      "step": 5776
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5449034014633296,
      "learning_rate": 1.6766946825427605e-06,
      "loss": 0.1957,
      "step": 5777
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3309122850522002,
      "learning_rate": 1.6751460535787407e-06,
      "loss": 0.1433,
      "step": 5778
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.426592126033138,
      "learning_rate": 1.6735979961956588e-06,
      "loss": 0.1648,
      "step": 5779
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.231642795420568,
      "learning_rate": 1.6720505106596429e-06,
      "loss": 0.1697,
      "step": 5780
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4618019285900168,
      "learning_rate": 1.6705035972367272e-06,
      "loss": 0.1686,
      "step": 5781
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.418996274416929,
      "learning_rate": 1.6689572561928397e-06,
      "loss": 0.1732,
      "step": 5782
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5318656872950729,
      "learning_rate": 1.6674114877938185e-06,
      "loss": 0.1938,
      "step": 5783
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4426304815063529,
      "learning_rate": 1.6658662923053974e-06,
      "loss": 0.209,
      "step": 5784
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5068386554001503,
      "learning_rate": 1.664321669993213e-06,
      "loss": 0.1943,
      "step": 5785
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3138142819183,
      "learning_rate": 1.6627776211228041e-06,
      "loss": 0.1342,
      "step": 5786
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.414762518412117,
      "learning_rate": 1.661234145959611e-06,
      "loss": 0.1765,
      "step": 5787
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4088468706014807,
      "learning_rate": 1.6596912447689723e-06,
      "loss": 0.2326,
      "step": 5788
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5033917810978108,
      "learning_rate": 1.6581489178161348e-06,
      "loss": 0.1952,
      "step": 5789
    },
    {
      "epoch": 0.74,
      "grad_norm": 26.87007674848195,
      "learning_rate": 1.6566071653662403e-06,
      "loss": 0.6396,
      "step": 5790
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3202246564551716,
      "learning_rate": 1.6550659876843317e-06,
      "loss": 0.1543,
      "step": 5791
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.403113535293451,
      "learning_rate": 1.65352538503536e-06,
      "loss": 0.5276,
      "step": 5792
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5326880798398081,
      "learning_rate": 1.6519853576841698e-06,
      "loss": 0.181,
      "step": 5793
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4114146048210208,
      "learning_rate": 1.6504459058955108e-06,
      "loss": 0.1479,
      "step": 5794
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4776453033513879,
      "learning_rate": 1.6489070299340298e-06,
      "loss": 0.1818,
      "step": 5795
    },
    {
      "epoch": 0.74,
      "grad_norm": 8.108783051008784,
      "learning_rate": 1.6473687300642815e-06,
      "loss": 0.7201,
      "step": 5796
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3519438381783326,
      "learning_rate": 1.6458310065507154e-06,
      "loss": 0.1875,
      "step": 5797
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.04123618573534,
      "learning_rate": 1.6442938596576842e-06,
      "loss": 0.436,
      "step": 5798
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.384607725910033,
      "learning_rate": 1.6427572896494408e-06,
      "loss": 0.1985,
      "step": 5799
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3062513252607026,
      "learning_rate": 1.6412212967901386e-06,
      "loss": 0.154,
      "step": 5800
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.827606576436322,
      "learning_rate": 1.6396858813438355e-06,
      "loss": 0.5463,
      "step": 5801
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5919551812399964,
      "learning_rate": 1.638151043574485e-06,
      "loss": 0.1726,
      "step": 5802
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7937048455518583,
      "learning_rate": 1.6366167837459429e-06,
      "loss": 0.2351,
      "step": 5803
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.403227763797239,
      "learning_rate": 1.6350831021219686e-06,
      "loss": 0.2088,
      "step": 5804
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4256231198569151,
      "learning_rate": 1.6335499989662185e-06,
      "loss": 0.1881,
      "step": 5805
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.8663706975399466,
      "learning_rate": 1.6320174745422513e-06,
      "loss": 0.7078,
      "step": 5806
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3794632085629874,
      "learning_rate": 1.6304855291135247e-06,
      "loss": 0.1819,
      "step": 5807
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4886688900382663,
      "learning_rate": 1.6289541629433986e-06,
      "loss": 0.1781,
      "step": 5808
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4183886821680765,
      "learning_rate": 1.6274233762951308e-06,
      "loss": 0.218,
      "step": 5809
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4296372965838722,
      "learning_rate": 1.6258931694318831e-06,
      "loss": 0.1813,
      "step": 5810
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5308397403935174,
      "learning_rate": 1.624363542616717e-06,
      "loss": 0.1891,
      "step": 5811
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3080458390081144,
      "learning_rate": 1.6228344961125914e-06,
      "loss": 0.1658,
      "step": 5812
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6002165539054167,
      "learning_rate": 1.6213060301823674e-06,
      "loss": 0.2224,
      "step": 5813
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.397803505646373,
      "learning_rate": 1.6197781450888067e-06,
      "loss": 0.1523,
      "step": 5814
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4798539818818406,
      "learning_rate": 1.618250841094569e-06,
      "loss": 0.2272,
      "step": 5815
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.212871426101267,
      "learning_rate": 1.616724118462215e-06,
      "loss": 0.624,
      "step": 5816
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5197899475898233,
      "learning_rate": 1.6151979774542087e-06,
      "loss": 0.1813,
      "step": 5817
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4571028994501056,
      "learning_rate": 1.6136724183329106e-06,
      "loss": 0.1821,
      "step": 5818
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5703719434047174,
      "learning_rate": 1.6121474413605792e-06,
      "loss": 0.1912,
      "step": 5819
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3521337915879998,
      "learning_rate": 1.61062304679938e-06,
      "loss": 0.2031,
      "step": 5820
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.675941472001607,
      "learning_rate": 1.609099234911372e-06,
      "loss": 0.4267,
      "step": 5821
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.162084316317529,
      "learning_rate": 1.6075760059585166e-06,
      "loss": 0.626,
      "step": 5822
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.135586040309193,
      "learning_rate": 1.6060533602026734e-06,
      "loss": 0.7037,
      "step": 5823
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4993300538012004,
      "learning_rate": 1.6045312979056027e-06,
      "loss": 0.1624,
      "step": 5824
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3718292635754108,
      "learning_rate": 1.6030098193289667e-06,
      "loss": 0.1977,
      "step": 5825
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3461052840103418,
      "learning_rate": 1.601488924734324e-06,
      "loss": 0.1503,
      "step": 5826
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.493207310158811,
      "learning_rate": 1.5999686143831344e-06,
      "loss": 0.1847,
      "step": 5827
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.291108830906837,
      "learning_rate": 1.5984488885367543e-06,
      "loss": 0.5888,
      "step": 5828
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3409653953891687,
      "learning_rate": 1.5969297474564465e-06,
      "loss": 0.1927,
      "step": 5829
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4564556189413507,
      "learning_rate": 1.595411191403367e-06,
      "loss": 0.1849,
      "step": 5830
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.369682823202802,
      "learning_rate": 1.5938932206385716e-06,
      "loss": 0.1698,
      "step": 5831
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.749098213265127,
      "learning_rate": 1.5923758354230196e-06,
      "loss": 0.5232,
      "step": 5832
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6035280261887845,
      "learning_rate": 1.5908590360175663e-06,
      "loss": 0.2063,
      "step": 5833
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3700369845715004,
      "learning_rate": 1.5893428226829672e-06,
      "loss": 0.1786,
      "step": 5834
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4472768020788491,
      "learning_rate": 1.5878271956798762e-06,
      "loss": 0.2174,
      "step": 5835
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.426260715868786,
      "learning_rate": 1.5863121552688477e-06,
      "loss": 0.1857,
      "step": 5836
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2620043871061681,
      "learning_rate": 1.5847977017103327e-06,
      "loss": 0.1688,
      "step": 5837
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6005491050726288,
      "learning_rate": 1.5832838352646856e-06,
      "loss": 0.2028,
      "step": 5838
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4185157375386552,
      "learning_rate": 1.5817705561921587e-06,
      "loss": 0.153,
      "step": 5839
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4384710729824355,
      "learning_rate": 1.5802578647529005e-06,
      "loss": 0.2046,
      "step": 5840
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3402665250164365,
      "learning_rate": 1.5787457612069607e-06,
      "loss": 0.1708,
      "step": 5841
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.62666674798467,
      "learning_rate": 1.577234245814287e-06,
      "loss": 0.222,
      "step": 5842
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3699108709097845,
      "learning_rate": 1.5757233188347265e-06,
      "loss": 0.1587,
      "step": 5843
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2568441602284808,
      "learning_rate": 1.5742129805280249e-06,
      "loss": 0.152,
      "step": 5844
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.366063166036423,
      "learning_rate": 1.572703231153826e-06,
      "loss": 0.166,
      "step": 5845
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3623415638260923,
      "learning_rate": 1.5711940709716755e-06,
      "loss": 0.2124,
      "step": 5846
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2373607119469539,
      "learning_rate": 1.5696855002410127e-06,
      "loss": 0.1578,
      "step": 5847
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2534940673610242,
      "learning_rate": 1.5681775192211819e-06,
      "loss": 0.162,
      "step": 5848
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.876692382939296,
      "learning_rate": 1.5666701281714202e-06,
      "loss": 0.5582,
      "step": 5849
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3686540572787234,
      "learning_rate": 1.5651633273508666e-06,
      "loss": 0.1558,
      "step": 5850
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4744264047800162,
      "learning_rate": 1.5636571170185565e-06,
      "loss": 0.1904,
      "step": 5851
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2688174176245006,
      "learning_rate": 1.5621514974334246e-06,
      "loss": 0.1847,
      "step": 5852
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3332800624776073,
      "learning_rate": 1.5606464688543066e-06,
      "loss": 0.1527,
      "step": 5853
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3787910874983171,
      "learning_rate": 1.5591420315399324e-06,
      "loss": 0.2232,
      "step": 5854
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4040094739678157,
      "learning_rate": 1.5576381857489337e-06,
      "loss": 0.1915,
      "step": 5855
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.399158679056446,
      "learning_rate": 1.556134931739836e-06,
      "loss": 0.5228,
      "step": 5856
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3445132172550678,
      "learning_rate": 1.5546322697710697e-06,
      "loss": 0.1834,
      "step": 5857
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3501650941379997,
      "learning_rate": 1.553130200100959e-06,
      "loss": 0.1661,
      "step": 5858
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.901637775483595,
      "learning_rate": 1.5516287229877242e-06,
      "loss": 0.6324,
      "step": 5859
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.530025912115187,
      "learning_rate": 1.5501278386894907e-06,
      "loss": 0.1468,
      "step": 5860
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5146558605571927,
      "learning_rate": 1.5486275474642765e-06,
      "loss": 0.2407,
      "step": 5861
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5831592267122585,
      "learning_rate": 1.5471278495699982e-06,
      "loss": 0.2044,
      "step": 5862
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.289679808807945,
      "learning_rate": 1.545628745264472e-06,
      "loss": 0.2014,
      "step": 5863
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4994241944577114,
      "learning_rate": 1.5441302348054105e-06,
      "loss": 0.2304,
      "step": 5864
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.265881186966355,
      "learning_rate": 1.5426323184504244e-06,
      "loss": 0.483,
      "step": 5865
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6769867490186139,
      "learning_rate": 1.5411349964570254e-06,
      "loss": 0.231,
      "step": 5866
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.584310034641871,
      "learning_rate": 1.5396382690826173e-06,
      "loss": 0.2145,
      "step": 5867
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3148229979031054,
      "learning_rate": 1.5381421365845083e-06,
      "loss": 0.1888,
      "step": 5868
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4614485247170088,
      "learning_rate": 1.5366465992198997e-06,
      "loss": 0.1676,
      "step": 5869
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.531764634621869,
      "learning_rate": 1.5351516572458913e-06,
      "loss": 0.1913,
      "step": 5870
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2775286936369015,
      "learning_rate": 1.5336573109194807e-06,
      "loss": 0.1627,
      "step": 5871
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.661085886963072,
      "learning_rate": 1.5321635604975637e-06,
      "loss": 0.4301,
      "step": 5872
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.524874668015195,
      "learning_rate": 1.530670406236932e-06,
      "loss": 0.2104,
      "step": 5873
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.275120460578636,
      "learning_rate": 1.5291778483942771e-06,
      "loss": 0.1256,
      "step": 5874
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.03400658928566,
      "learning_rate": 1.527685887226189e-06,
      "loss": 0.3936,
      "step": 5875
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.353120182255288,
      "learning_rate": 1.5261945229891512e-06,
      "loss": 0.4816,
      "step": 5876
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3374209571584104,
      "learning_rate": 1.5247037559395467e-06,
      "loss": 0.1255,
      "step": 5877
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3535105253936026,
      "learning_rate": 1.5232135863336556e-06,
      "loss": 0.1844,
      "step": 5878
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4870202504200982,
      "learning_rate": 1.5217240144276558e-06,
      "loss": 0.2197,
      "step": 5879
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3628363834623263,
      "learning_rate": 1.5202350404776196e-06,
      "loss": 0.151,
      "step": 5880
    },
    {
      "epoch": 0.75,
      "grad_norm": 10.30905537668229,
      "learning_rate": 1.5187466647395227e-06,
      "loss": 0.5684,
      "step": 5881
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3462059258881516,
      "learning_rate": 1.5172588874692318e-06,
      "loss": 0.1667,
      "step": 5882
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.028999374852722,
      "learning_rate": 1.5157717089225144e-06,
      "loss": 0.6258,
      "step": 5883
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5754232362912226,
      "learning_rate": 1.5142851293550303e-06,
      "loss": 0.1993,
      "step": 5884
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3576101037585966,
      "learning_rate": 1.5127991490223449e-06,
      "loss": 0.1678,
      "step": 5885
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4796538234713974,
      "learning_rate": 1.5113137681799123e-06,
      "loss": 0.1745,
      "step": 5886
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3787283769217826,
      "learning_rate": 1.5098289870830869e-06,
      "loss": 0.1854,
      "step": 5887
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.206284239763585,
      "learning_rate": 1.508344805987119e-06,
      "loss": 0.1507,
      "step": 5888
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2234174509299973,
      "learning_rate": 1.5068612251471592e-06,
      "loss": 0.167,
      "step": 5889
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5026452592844386,
      "learning_rate": 1.5053782448182509e-06,
      "loss": 0.2067,
      "step": 5890
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3365376910319884,
      "learning_rate": 1.5038958652553354e-06,
      "loss": 0.1668,
      "step": 5891
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.273478120802703,
      "learning_rate": 1.502414086713251e-06,
      "loss": 0.1457,
      "step": 5892
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5540362304269175,
      "learning_rate": 1.5009329094467313e-06,
      "loss": 0.2291,
      "step": 5893
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5396759976664485,
      "learning_rate": 1.49945233371041e-06,
      "loss": 0.1755,
      "step": 5894
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.325709184906715,
      "learning_rate": 1.497972359758813e-06,
      "loss": 0.5822,
      "step": 5895
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.549350054749919,
      "learning_rate": 1.4964929878463685e-06,
      "loss": 0.2097,
      "step": 5896
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4250634991934963,
      "learning_rate": 1.4950142182273947e-06,
      "loss": 0.1588,
      "step": 5897
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.369823431233023,
      "learning_rate": 1.493536051156111e-06,
      "loss": 0.6618,
      "step": 5898
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2439787072757822,
      "learning_rate": 1.4920584868866295e-06,
      "loss": 0.1924,
      "step": 5899
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4712264559511132,
      "learning_rate": 1.4905815256729621e-06,
      "loss": 0.1545,
      "step": 5900
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.346128800937466,
      "learning_rate": 1.4891051677690156e-06,
      "loss": 0.1741,
      "step": 5901
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5782615452645827,
      "learning_rate": 1.4876294134285902e-06,
      "loss": 0.1781,
      "step": 5902
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8660151573474417,
      "learning_rate": 1.4861542629053882e-06,
      "loss": 0.2651,
      "step": 5903
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3950134769638318,
      "learning_rate": 1.4846797164530051e-06,
      "loss": 0.1915,
      "step": 5904
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.268064984926393,
      "learning_rate": 1.4832057743249329e-06,
      "loss": 0.1717,
      "step": 5905
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.24051430991757,
      "learning_rate": 1.4817324367745573e-06,
      "loss": 0.4462,
      "step": 5906
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3717756040379168,
      "learning_rate": 1.4802597040551636e-06,
      "loss": 0.1836,
      "step": 5907
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.328665145407037,
      "learning_rate": 1.4787875764199312e-06,
      "loss": 0.455,
      "step": 5908
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4123106856903298,
      "learning_rate": 1.4773160541219338e-06,
      "loss": 0.187,
      "step": 5909
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.369964393317383,
      "learning_rate": 1.4758451374141469e-06,
      "loss": 0.2016,
      "step": 5910
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.548311234876524,
      "learning_rate": 1.474374826549435e-06,
      "loss": 0.524,
      "step": 5911
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3637110017631553,
      "learning_rate": 1.4729051217805645e-06,
      "loss": 0.193,
      "step": 5912
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3552929293241724,
      "learning_rate": 1.4714360233601933e-06,
      "loss": 0.1905,
      "step": 5913
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4195760854289459,
      "learning_rate": 1.4699675315408756e-06,
      "loss": 0.2092,
      "step": 5914
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.399764422299543,
      "learning_rate": 1.468499646575064e-06,
      "loss": 0.2052,
      "step": 5915
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.598745881786379,
      "learning_rate": 1.4670323687151012e-06,
      "loss": 0.1943,
      "step": 5916
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3900206591317479,
      "learning_rate": 1.4655656982132338e-06,
      "loss": 0.1618,
      "step": 5917
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4273053065551589,
      "learning_rate": 1.464099635321598e-06,
      "loss": 0.1968,
      "step": 5918
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3471116447169735,
      "learning_rate": 1.4626341802922262e-06,
      "loss": 0.1823,
      "step": 5919
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3940689972920786,
      "learning_rate": 1.461169333377047e-06,
      "loss": 0.1817,
      "step": 5920
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4466169591117102,
      "learning_rate": 1.459705094827884e-06,
      "loss": 0.2164,
      "step": 5921
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5568029760339486,
      "learning_rate": 1.4582414648964594e-06,
      "loss": 0.199,
      "step": 5922
    },
    {
      "epoch": 0.76,
      "grad_norm": 10.391995605530784,
      "learning_rate": 1.4567784438343868e-06,
      "loss": 0.6362,
      "step": 5923
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.347344219997779,
      "learning_rate": 1.455316031893174e-06,
      "loss": 0.4611,
      "step": 5924
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2671633740139037,
      "learning_rate": 1.4538542293242307e-06,
      "loss": 0.1698,
      "step": 5925
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.477400899594276,
      "learning_rate": 1.4523930363788562e-06,
      "loss": 0.1981,
      "step": 5926
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2516347151346323,
      "learning_rate": 1.450932453308246e-06,
      "loss": 0.1692,
      "step": 5927
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.438897972131763,
      "learning_rate": 1.4494724803634912e-06,
      "loss": 0.1903,
      "step": 5928
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2764142683600186,
      "learning_rate": 1.448013117795578e-06,
      "loss": 0.1942,
      "step": 5929
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3683594623494941,
      "learning_rate": 1.446554365855387e-06,
      "loss": 0.1977,
      "step": 5930
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4040542122607242,
      "learning_rate": 1.445096224793695e-06,
      "loss": 0.2016,
      "step": 5931
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5672920158207049,
      "learning_rate": 1.4436386948611763e-06,
      "loss": 0.1735,
      "step": 5932
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2150230247346205,
      "learning_rate": 1.442181776308394e-06,
      "loss": 0.1335,
      "step": 5933
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3324809155054775,
      "learning_rate": 1.4407254693858108e-06,
      "loss": 0.1624,
      "step": 5934
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4437701557986071,
      "learning_rate": 1.4392697743437816e-06,
      "loss": 0.1862,
      "step": 5935
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3238539887775724,
      "learning_rate": 1.437814691432558e-06,
      "loss": 0.1772,
      "step": 5936
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3090271103428381,
      "learning_rate": 1.4363602209022837e-06,
      "loss": 0.1723,
      "step": 5937
    },
    {
      "epoch": 0.76,
      "grad_norm": 8.330142524610425,
      "learning_rate": 1.4349063630030018e-06,
      "loss": 0.6072,
      "step": 5938
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.091130560310398,
      "learning_rate": 1.4334531179846455e-06,
      "loss": 0.5805,
      "step": 5939
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5854859459803543,
      "learning_rate": 1.432000486097046e-06,
      "loss": 0.2322,
      "step": 5940
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3125701227670339,
      "learning_rate": 1.4305484675899272e-06,
      "loss": 0.1619,
      "step": 5941
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.348651289410746,
      "learning_rate": 1.4290970627129075e-06,
      "loss": 0.1325,
      "step": 5942
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.7196392648708345,
      "learning_rate": 1.4276462717154999e-06,
      "loss": 0.6584,
      "step": 5943
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.3129917089910785,
      "learning_rate": 1.4261960948471122e-06,
      "loss": 0.536,
      "step": 5944
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3177688997734471,
      "learning_rate": 1.424746532357046e-06,
      "loss": 0.1579,
      "step": 5945
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3861736099889437,
      "learning_rate": 1.4232975844944997e-06,
      "loss": 0.1482,
      "step": 5946
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4402010396553617,
      "learning_rate": 1.4218492515085636e-06,
      "loss": 0.1477,
      "step": 5947
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4570985083330315,
      "learning_rate": 1.4204015336482213e-06,
      "loss": 0.1635,
      "step": 5948
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2577966852626312,
      "learning_rate": 1.418954431162355e-06,
      "loss": 0.1726,
      "step": 5949
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.44787312623616,
      "learning_rate": 1.417507944299737e-06,
      "loss": 0.2031,
      "step": 5950
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4474705626275903,
      "learning_rate": 1.4160620733090351e-06,
      "loss": 0.1949,
      "step": 5951
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.970057118932273,
      "learning_rate": 1.4146168184388099e-06,
      "loss": 0.4525,
      "step": 5952
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4863073432701497,
      "learning_rate": 1.413172179937521e-06,
      "loss": 0.1734,
      "step": 5953
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4720899128302176,
      "learning_rate": 1.4117281580535158e-06,
      "loss": 0.1796,
      "step": 5954
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4265016817156606,
      "learning_rate": 1.410284753035039e-06,
      "loss": 0.196,
      "step": 5955
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5286905589614712,
      "learning_rate": 1.4088419651302288e-06,
      "loss": 0.1641,
      "step": 5956
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.48607850517793,
      "learning_rate": 1.407399794587117e-06,
      "loss": 0.1858,
      "step": 5957
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5285732951432331,
      "learning_rate": 1.4059582416536282e-06,
      "loss": 0.1796,
      "step": 5958
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.336705563665438,
      "learning_rate": 1.4045173065775852e-06,
      "loss": 0.4719,
      "step": 5959
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3234774718672773,
      "learning_rate": 1.4030769896066975e-06,
      "loss": 0.1693,
      "step": 5960
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3296519176966608,
      "learning_rate": 1.4016372909885762e-06,
      "loss": 0.1663,
      "step": 5961
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5464223214259143,
      "learning_rate": 1.4001982109707201e-06,
      "loss": 0.184,
      "step": 5962
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.600852697530997,
      "learning_rate": 1.3987597498005245e-06,
      "loss": 0.3823,
      "step": 5963
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3348337990019137,
      "learning_rate": 1.397321907725277e-06,
      "loss": 0.1872,
      "step": 5964
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.458434745879235,
      "learning_rate": 1.3958846849921593e-06,
      "loss": 0.1731,
      "step": 5965
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5538672753233267,
      "learning_rate": 1.3944480818482448e-06,
      "loss": 0.2021,
      "step": 5966
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.416901891411164,
      "learning_rate": 1.393012098540505e-06,
      "loss": 0.1931,
      "step": 5967
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5088194750075847,
      "learning_rate": 1.3915767353158022e-06,
      "loss": 0.1681,
      "step": 5968
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3238753894244126,
      "learning_rate": 1.3901419924208908e-06,
      "loss": 0.1806,
      "step": 5969
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4527677604331286,
      "learning_rate": 1.3887078701024204e-06,
      "loss": 0.2088,
      "step": 5970
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.277147519358723,
      "learning_rate": 1.3872743686069328e-06,
      "loss": 0.1307,
      "step": 5971
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2648324949150536,
      "learning_rate": 1.3858414881808634e-06,
      "loss": 0.1414,
      "step": 5972
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.348246105257814,
      "learning_rate": 1.3844092290705396e-06,
      "loss": 0.6242,
      "step": 5973
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3351016665777184,
      "learning_rate": 1.382977591522186e-06,
      "loss": 0.1593,
      "step": 5974
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2953993530729755,
      "learning_rate": 1.3815465757819174e-06,
      "loss": 0.197,
      "step": 5975
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4914289026153211,
      "learning_rate": 1.3801161820957386e-06,
      "loss": 0.2012,
      "step": 5976
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5249412571555185,
      "learning_rate": 1.378686410709556e-06,
      "loss": 0.1764,
      "step": 5977
    },
    {
      "epoch": 0.77,
      "grad_norm": 8.19458758196158,
      "learning_rate": 1.3772572618691604e-06,
      "loss": 0.5486,
      "step": 5978
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3297506443568554,
      "learning_rate": 1.3758287358202404e-06,
      "loss": 0.1806,
      "step": 5979
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3375193018242175,
      "learning_rate": 1.3744008328083758e-06,
      "loss": 0.168,
      "step": 5980
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3866356080454267,
      "learning_rate": 1.3729735530790378e-06,
      "loss": 0.1809,
      "step": 5981
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.262394192440566,
      "learning_rate": 1.3715468968775952e-06,
      "loss": 0.1437,
      "step": 5982
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.354681196919386,
      "learning_rate": 1.3701208644493064e-06,
      "loss": 0.1698,
      "step": 5983
    },
    {
      "epoch": 0.77,
      "grad_norm": 8.100068784839833,
      "learning_rate": 1.3686954560393218e-06,
      "loss": 0.6421,
      "step": 5984
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5066789986230442,
      "learning_rate": 1.3672706718926849e-06,
      "loss": 0.2126,
      "step": 5985
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.35195403735445,
      "learning_rate": 1.3658465122543346e-06,
      "loss": 0.1692,
      "step": 5986
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.487907305564798,
      "learning_rate": 1.3644229773690997e-06,
      "loss": 0.1886,
      "step": 5987
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2022768774816661,
      "learning_rate": 1.3630000674817011e-06,
      "loss": 0.1511,
      "step": 5988
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5137503828928818,
      "learning_rate": 1.361577782836756e-06,
      "loss": 0.1671,
      "step": 5989
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4766047103053368,
      "learning_rate": 1.3601561236787702e-06,
      "loss": 0.1816,
      "step": 5990
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.270167601899381,
      "learning_rate": 1.3587350902521435e-06,
      "loss": 0.6094,
      "step": 5991
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3630448878371761,
      "learning_rate": 1.357314682801168e-06,
      "loss": 0.1908,
      "step": 5992
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.391480226630229,
      "learning_rate": 1.3558949015700278e-06,
      "loss": 0.1906,
      "step": 5993
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4397195455302298,
      "learning_rate": 1.3544757468027986e-06,
      "loss": 0.1584,
      "step": 5994
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.383458522741505,
      "learning_rate": 1.3530572187434531e-06,
      "loss": 0.1756,
      "step": 5995
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.390013371208714,
      "learning_rate": 1.351639317635849e-06,
      "loss": 0.1884,
      "step": 5996
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.412732808620388,
      "learning_rate": 1.3502220437237429e-06,
      "loss": 0.1756,
      "step": 5997
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4465752292084204,
      "learning_rate": 1.3488053972507792e-06,
      "loss": 0.2148,
      "step": 5998
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3129887491328964,
      "learning_rate": 1.3473893784604963e-06,
      "loss": 0.1366,
      "step": 5999
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4891341944159262,
      "learning_rate": 1.345973987596324e-06,
      "loss": 0.172,
      "step": 6000
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3290061924259795,
      "learning_rate": 1.3445592249015843e-06,
      "loss": 0.1539,
      "step": 6001
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4412131362260299,
      "learning_rate": 1.3431450906194892e-06,
      "loss": 0.1602,
      "step": 6002
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7493931501255167,
      "learning_rate": 1.3417315849931495e-06,
      "loss": 0.2469,
      "step": 6003
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3683224040911781,
      "learning_rate": 1.3403187082655584e-06,
      "loss": 0.1583,
      "step": 6004
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.60120224068066,
      "learning_rate": 1.3389064606796098e-06,
      "loss": 0.2236,
      "step": 6005
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4373529531247011,
      "learning_rate": 1.3374948424780836e-06,
      "loss": 0.2172,
      "step": 6006
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4349437244596206,
      "learning_rate": 1.336083853903653e-06,
      "loss": 0.1699,
      "step": 6007
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1538014222397077,
      "learning_rate": 1.3346734951988844e-06,
      "loss": 0.1357,
      "step": 6008
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.2366396572343685,
      "learning_rate": 1.333263766606232e-06,
      "loss": 0.5825,
      "step": 6009
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.026801695952292,
      "learning_rate": 1.3318546683680483e-06,
      "loss": 0.4949,
      "step": 6010
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.132189584972634,
      "learning_rate": 1.3304462007265716e-06,
      "loss": 0.1475,
      "step": 6011
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3273867963813626,
      "learning_rate": 1.3290383639239347e-06,
      "loss": 0.1489,
      "step": 6012
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5030657047409484,
      "learning_rate": 1.3276311582021583e-06,
      "loss": 0.1783,
      "step": 6013
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.310279744314773,
      "learning_rate": 1.3262245838031618e-06,
      "loss": 0.5811,
      "step": 6014
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.235741883382616,
      "learning_rate": 1.3248186409687491e-06,
      "loss": 0.1492,
      "step": 6015
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.499596709558661,
      "learning_rate": 1.3234133299406183e-06,
      "loss": 0.1798,
      "step": 6016
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.383346386119651,
      "learning_rate": 1.3220086509603569e-06,
      "loss": 0.1941,
      "step": 6017
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5565780491118695,
      "learning_rate": 1.3206046042694493e-06,
      "loss": 0.1921,
      "step": 6018
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3273356925126758,
      "learning_rate": 1.3192011901092654e-06,
      "loss": 0.1638,
      "step": 6019
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3240092910532169,
      "learning_rate": 1.3177984087210682e-06,
      "loss": 0.1938,
      "step": 6020
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.79718882055623,
      "learning_rate": 1.3163962603460123e-06,
      "loss": 0.4551,
      "step": 6021
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5501401787328182,
      "learning_rate": 1.3149947452251422e-06,
      "loss": 0.2071,
      "step": 6022
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2648764181524812,
      "learning_rate": 1.3135938635993966e-06,
      "loss": 0.1583,
      "step": 6023
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3391726437562934,
      "learning_rate": 1.312193615709601e-06,
      "loss": 0.1756,
      "step": 6024
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5808548889524991,
      "learning_rate": 1.310794001796477e-06,
      "loss": 0.219,
      "step": 6025
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3626050435146213,
      "learning_rate": 1.3093950221006329e-06,
      "loss": 0.1859,
      "step": 6026
    },
    {
      "epoch": 0.77,
      "grad_norm": 11.808739934463746,
      "learning_rate": 1.30799667686257e-06,
      "loss": 0.5963,
      "step": 6027
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3770219237454073,
      "learning_rate": 1.3065989663226797e-06,
      "loss": 0.1837,
      "step": 6028
    },
    {
      "epoch": 0.77,
      "grad_norm": 8.063701911917738,
      "learning_rate": 1.3052018907212448e-06,
      "loss": 0.4453,
      "step": 6029
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.557362174201385,
      "learning_rate": 1.303805450298437e-06,
      "loss": 0.5924,
      "step": 6030
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5009528700864436,
      "learning_rate": 1.3024096452943236e-06,
      "loss": 0.1972,
      "step": 6031
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4004060233863673,
      "learning_rate": 1.301014475948857e-06,
      "loss": 0.1508,
      "step": 6032
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.292435686593866,
      "learning_rate": 1.2996199425018858e-06,
      "loss": 0.4547,
      "step": 6033
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2584898687938453,
      "learning_rate": 1.2982260451931445e-06,
      "loss": 0.1756,
      "step": 6034
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3970188106011563,
      "learning_rate": 1.2968327842622612e-06,
      "loss": 0.1741,
      "step": 6035
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5330613904890353,
      "learning_rate": 1.2954401599487531e-06,
      "loss": 0.2223,
      "step": 6036
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.249249204862837,
      "learning_rate": 1.2940481724920284e-06,
      "loss": 0.6749,
      "step": 6037
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4038831385257766,
      "learning_rate": 1.292656822131384e-06,
      "loss": 0.2096,
      "step": 6038
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3379532540250676,
      "learning_rate": 1.291266109106013e-06,
      "loss": 0.177,
      "step": 6039
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3631244930636455,
      "learning_rate": 1.2898760336549931e-06,
      "loss": 0.178,
      "step": 6040
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4283386269634022,
      "learning_rate": 1.2884865960172931e-06,
      "loss": 0.1883,
      "step": 6041
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.0802391043709925,
      "learning_rate": 1.287097796431776e-06,
      "loss": 0.5577,
      "step": 6042
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4383726419619092,
      "learning_rate": 1.2857096351371917e-06,
      "loss": 0.1649,
      "step": 6043
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4818715735751855,
      "learning_rate": 1.2843221123721804e-06,
      "loss": 0.1913,
      "step": 6044
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4222010794279467,
      "learning_rate": 1.2829352283752728e-06,
      "loss": 0.1992,
      "step": 6045
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3693843658438427,
      "learning_rate": 1.2815489833848927e-06,
      "loss": 0.1747,
      "step": 6046
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3359777886394428,
      "learning_rate": 1.28016337763935e-06,
      "loss": 0.1652,
      "step": 6047
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.212649396861182,
      "learning_rate": 1.278778411376847e-06,
      "loss": 0.189,
      "step": 6048
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2731868358270513,
      "learning_rate": 1.2773940848354754e-06,
      "loss": 0.1434,
      "step": 6049
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3897935064558067,
      "learning_rate": 1.2760103982532152e-06,
      "loss": 0.1762,
      "step": 6050
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.638481345993579,
      "learning_rate": 1.274627351867941e-06,
      "loss": 0.234,
      "step": 6051
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4915035914471424,
      "learning_rate": 1.2732449459174134e-06,
      "loss": 0.1495,
      "step": 6052
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.92976813202983,
      "learning_rate": 1.2718631806392823e-06,
      "loss": 0.5346,
      "step": 6053
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.983559404461299,
      "learning_rate": 1.2704820562710923e-06,
      "loss": 0.3699,
      "step": 6054
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2727724430513538,
      "learning_rate": 1.2691015730502732e-06,
      "loss": 0.1587,
      "step": 6055
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3720773573818688,
      "learning_rate": 1.2677217312141455e-06,
      "loss": 0.1683,
      "step": 6056
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2041379630490656,
      "learning_rate": 1.2663425309999205e-06,
      "loss": 0.1295,
      "step": 6057
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.863066694622708,
      "learning_rate": 1.2649639726446994e-06,
      "loss": 0.2707,
      "step": 6058
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.547451358215098,
      "learning_rate": 1.2635860563854695e-06,
      "loss": 0.1606,
      "step": 6059
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.188570099579676,
      "learning_rate": 1.2622087824591129e-06,
      "loss": 0.4883,
      "step": 6060
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4475128905893033,
      "learning_rate": 1.2608321511024007e-06,
      "loss": 0.1438,
      "step": 6061
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.384225783969732,
      "learning_rate": 1.2594561625519891e-06,
      "loss": 0.1795,
      "step": 6062
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5413456672987205,
      "learning_rate": 1.258080817044428e-06,
      "loss": 0.2036,
      "step": 6063
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.272162737411395,
      "learning_rate": 1.256706114816154e-06,
      "loss": 0.1774,
      "step": 6064
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2523015477316803,
      "learning_rate": 1.2553320561034955e-06,
      "loss": 0.1592,
      "step": 6065
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.7438920429463334,
      "learning_rate": 1.2539586411426664e-06,
      "loss": 0.446,
      "step": 6066
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.259417128941448,
      "learning_rate": 1.2525858701697762e-06,
      "loss": 0.5619,
      "step": 6067
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5369420015770012,
      "learning_rate": 1.2512137434208188e-06,
      "loss": 0.1932,
      "step": 6068
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6793715933495552,
      "learning_rate": 1.2498422611316767e-06,
      "loss": 0.1818,
      "step": 6069
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2988497405028756,
      "learning_rate": 1.2484714235381278e-06,
      "loss": 0.1699,
      "step": 6070
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6649815102876053,
      "learning_rate": 1.2471012308758324e-06,
      "loss": 0.203,
      "step": 6071
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3778160222509497,
      "learning_rate": 1.2457316833803424e-06,
      "loss": 0.191,
      "step": 6072
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.492574201471644,
      "learning_rate": 1.244362781287099e-06,
      "loss": 0.1841,
      "step": 6073
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4470503440993079,
      "learning_rate": 1.2429945248314317e-06,
      "loss": 0.1896,
      "step": 6074
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6105948303556468,
      "learning_rate": 1.2416269142485615e-06,
      "loss": 0.1684,
      "step": 6075
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3893625813885326,
      "learning_rate": 1.2402599497735961e-06,
      "loss": 0.1655,
      "step": 6076
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4123879487568425,
      "learning_rate": 1.2388936316415317e-06,
      "loss": 0.1786,
      "step": 6077
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.36372573576362,
      "learning_rate": 1.237527960087253e-06,
      "loss": 0.1861,
      "step": 6078
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.534976663135379,
      "learning_rate": 1.2361629353455378e-06,
      "loss": 0.1786,
      "step": 6079
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.346066588977614,
      "learning_rate": 1.234798557651048e-06,
      "loss": 0.4075,
      "step": 6080
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2684929890317724,
      "learning_rate": 1.233434827238335e-06,
      "loss": 0.1508,
      "step": 6081
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3994973313608223,
      "learning_rate": 1.2320717443418422e-06,
      "loss": 0.168,
      "step": 6082
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4064938579878818,
      "learning_rate": 1.2307093091958983e-06,
      "loss": 0.1695,
      "step": 6083
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6242977206797522,
      "learning_rate": 1.2293475220347212e-06,
      "loss": 0.2208,
      "step": 6084
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.465366433868263,
      "learning_rate": 1.2279863830924183e-06,
      "loss": 0.1824,
      "step": 6085
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4398860357877883,
      "learning_rate": 1.2266258926029851e-06,
      "loss": 0.2066,
      "step": 6086
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.0698477197004745,
      "learning_rate": 1.2252660508003045e-06,
      "loss": 0.5483,
      "step": 6087
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3937827292974068,
      "learning_rate": 1.2239068579181497e-06,
      "loss": 0.1936,
      "step": 6088
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.314431444376913,
      "learning_rate": 1.222548314190184e-06,
      "loss": 0.1791,
      "step": 6089
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5790241621932368,
      "learning_rate": 1.2211904198499551e-06,
      "loss": 0.214,
      "step": 6090
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4196011674809688,
      "learning_rate": 1.2198331751309006e-06,
      "loss": 0.1744,
      "step": 6091
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3613101842620983,
      "learning_rate": 1.2184765802663468e-06,
      "loss": 0.1532,
      "step": 6092
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.445616792310713,
      "learning_rate": 1.2171206354895081e-06,
      "loss": 0.1905,
      "step": 6093
    },
    {
      "epoch": 0.78,
      "grad_norm": 12.146798160345408,
      "learning_rate": 1.2157653410334875e-06,
      "loss": 0.734,
      "step": 6094
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4622747668952238,
      "learning_rate": 1.214410697131273e-06,
      "loss": 0.1859,
      "step": 6095
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5126107999617953,
      "learning_rate": 1.2130567040157465e-06,
      "loss": 0.1934,
      "step": 6096
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.536909909323699,
      "learning_rate": 1.2117033619196762e-06,
      "loss": 0.1782,
      "step": 6097
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.387022183739942,
      "learning_rate": 1.2103506710757156e-06,
      "loss": 0.1982,
      "step": 6098
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2517718784757839,
      "learning_rate": 1.2089986317164075e-06,
      "loss": 0.1974,
      "step": 6099
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.25450686755148,
      "learning_rate": 1.2076472440741844e-06,
      "loss": 0.585,
      "step": 6100
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.324526230641254,
      "learning_rate": 1.2062965083813643e-06,
      "loss": 0.1359,
      "step": 6101
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3771999964104167,
      "learning_rate": 1.2049464248701537e-06,
      "loss": 0.156,
      "step": 6102
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4396739679326735,
      "learning_rate": 1.20359699377265e-06,
      "loss": 0.1739,
      "step": 6103
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4723454193414054,
      "learning_rate": 1.202248215320835e-06,
      "loss": 0.1817,
      "step": 6104
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5505269770703023,
      "learning_rate": 1.2009000897465782e-06,
      "loss": 0.15,
      "step": 6105
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6773517142710634,
      "learning_rate": 1.1995526172816385e-06,
      "loss": 0.1584,
      "step": 6106
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4217858208060492,
      "learning_rate": 1.1982057981576634e-06,
      "loss": 0.1954,
      "step": 6107
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.777196603459419,
      "learning_rate": 1.196859632606186e-06,
      "loss": 0.6577,
      "step": 6108
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4184045174198374,
      "learning_rate": 1.1955141208586257e-06,
      "loss": 0.1904,
      "step": 6109
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5297906201039275,
      "learning_rate": 1.1941692631462954e-06,
      "loss": 0.1986,
      "step": 6110
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6089425856611874,
      "learning_rate": 1.1928250597003893e-06,
      "loss": 0.1901,
      "step": 6111
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.161272640507757,
      "learning_rate": 1.1914815107519922e-06,
      "loss": 0.5416,
      "step": 6112
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.1309597147892365,
      "learning_rate": 1.1901386165320755e-06,
      "loss": 0.7567,
      "step": 6113
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2850885304271997,
      "learning_rate": 1.1887963772714982e-06,
      "loss": 0.1714,
      "step": 6114
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.359546033066446,
      "learning_rate": 1.1874547932010054e-06,
      "loss": 0.1975,
      "step": 6115
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.469028353011656,
      "learning_rate": 1.1861138645512343e-06,
      "loss": 0.6088,
      "step": 6116
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2745562211331294,
      "learning_rate": 1.1847735915527026e-06,
      "loss": 0.135,
      "step": 6117
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.131741168832009,
      "learning_rate": 1.183433974435822e-06,
      "loss": 0.4738,
      "step": 6118
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3943306098841513,
      "learning_rate": 1.1820950134308862e-06,
      "loss": 0.1817,
      "step": 6119
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.272129522546109,
      "learning_rate": 1.1807567087680787e-06,
      "loss": 0.1636,
      "step": 6120
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.348737881423158,
      "learning_rate": 1.1794190606774696e-06,
      "loss": 0.1737,
      "step": 6121
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.511265140971372,
      "learning_rate": 1.178082069389016e-06,
      "loss": 0.1879,
      "step": 6122
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5538587418728225,
      "learning_rate": 1.1767457351325605e-06,
      "loss": 0.175,
      "step": 6123
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3101795820378515,
      "learning_rate": 1.1754100581378365e-06,
      "loss": 0.1974,
      "step": 6124
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.477865764880873,
      "learning_rate": 1.174075038634463e-06,
      "loss": 0.1717,
      "step": 6125
    },
    {
      "epoch": 0.78,
      "grad_norm": 11.972712111220137,
      "learning_rate": 1.1727406768519444e-06,
      "loss": 0.6128,
      "step": 6126
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.170405120335117,
      "learning_rate": 1.171406973019673e-06,
      "loss": 0.1138,
      "step": 6127
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5265895111965164,
      "learning_rate": 1.1700739273669277e-06,
      "loss": 0.2029,
      "step": 6128
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3405496152121,
      "learning_rate": 1.168741540122874e-06,
      "loss": 0.1626,
      "step": 6129
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.710806217285736,
      "learning_rate": 1.1674098115165645e-06,
      "loss": 0.5866,
      "step": 6130
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6147770191093591,
      "learning_rate": 1.166078741776941e-06,
      "loss": 0.1919,
      "step": 6131
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4722481284485183,
      "learning_rate": 1.1647483311328285e-06,
      "loss": 0.1868,
      "step": 6132
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3958996691677683,
      "learning_rate": 1.1634185798129383e-06,
      "loss": 0.2147,
      "step": 6133
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3048080899573133,
      "learning_rate": 1.1620894880458732e-06,
      "loss": 0.146,
      "step": 6134
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5788624073886268,
      "learning_rate": 1.1607610560601179e-06,
      "loss": 0.1893,
      "step": 6135
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5662048577580099,
      "learning_rate": 1.1594332840840455e-06,
      "loss": 0.2353,
      "step": 6136
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.300149335670335,
      "learning_rate": 1.1581061723459153e-06,
      "loss": 0.1555,
      "step": 6137
    },
    {
      "epoch": 0.79,
      "grad_norm": 13.402749070926124,
      "learning_rate": 1.1567797210738713e-06,
      "loss": 0.549,
      "step": 6138
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.522291598080312,
      "learning_rate": 1.1554539304959494e-06,
      "loss": 0.6862,
      "step": 6139
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.373982382895216,
      "learning_rate": 1.1541288008400665e-06,
      "loss": 0.1971,
      "step": 6140
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2797702822987111,
      "learning_rate": 1.1528043323340281e-06,
      "loss": 0.1753,
      "step": 6141
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3479680463986619,
      "learning_rate": 1.151480525205525e-06,
      "loss": 0.1409,
      "step": 6142
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.959161890290408,
      "learning_rate": 1.1501573796821348e-06,
      "loss": 0.5647,
      "step": 6143
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6929343153626855,
      "learning_rate": 1.148834895991323e-06,
      "loss": 0.2146,
      "step": 6144
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.219449510319721,
      "learning_rate": 1.147513074360438e-06,
      "loss": 0.6841,
      "step": 6145
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5210702377676926,
      "learning_rate": 1.1461919150167189e-06,
      "loss": 0.1597,
      "step": 6146
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2191026390447184,
      "learning_rate": 1.1448714181872867e-06,
      "loss": 0.169,
      "step": 6147
    },
    {
      "epoch": 0.79,
      "grad_norm": 15.489272180489275,
      "learning_rate": 1.1435515840991502e-06,
      "loss": 0.6322,
      "step": 6148
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.610299821488926,
      "learning_rate": 1.1422324129792039e-06,
      "loss": 0.4748,
      "step": 6149
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.4177508239041865,
      "learning_rate": 1.1409139050542295e-06,
      "loss": 0.5122,
      "step": 6150
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2464906866792227,
      "learning_rate": 1.1395960605508916e-06,
      "loss": 0.1863,
      "step": 6151
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3354580799052544,
      "learning_rate": 1.1382788796957456e-06,
      "loss": 0.1241,
      "step": 6152
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5334396565342039,
      "learning_rate": 1.1369623627152276e-06,
      "loss": 0.2105,
      "step": 6153
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.476495405059449,
      "learning_rate": 1.1356465098356656e-06,
      "loss": 0.1945,
      "step": 6154
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3682181640517874,
      "learning_rate": 1.1343313212832674e-06,
      "loss": 0.1762,
      "step": 6155
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6399454500992796,
      "learning_rate": 1.13301679728413e-06,
      "loss": 0.1726,
      "step": 6156
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3960632655384508,
      "learning_rate": 1.1317029380642353e-06,
      "loss": 0.1567,
      "step": 6157
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.637818027127679,
      "learning_rate": 1.1303897438494503e-06,
      "loss": 0.4093,
      "step": 6158
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4264488117058016,
      "learning_rate": 1.1290772148655278e-06,
      "loss": 0.2262,
      "step": 6159
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.482312792133076,
      "learning_rate": 1.1277653513381083e-06,
      "loss": 0.6039,
      "step": 6160
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6506242142171328,
      "learning_rate": 1.1264541534927148e-06,
      "loss": 0.1934,
      "step": 6161
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4701472293898152,
      "learning_rate": 1.1251436215547596e-06,
      "loss": 0.1979,
      "step": 6162
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.894267665296833,
      "learning_rate": 1.1238337557495372e-06,
      "loss": 0.5056,
      "step": 6163
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5450822046989452,
      "learning_rate": 1.1225245563022285e-06,
      "loss": 0.201,
      "step": 6164
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2756595535804032,
      "learning_rate": 1.1212160234378999e-06,
      "loss": 0.1481,
      "step": 6165
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4120910385317398,
      "learning_rate": 1.1199081573815023e-06,
      "loss": 0.1987,
      "step": 6166
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4602986961660818,
      "learning_rate": 1.1186009583578761e-06,
      "loss": 0.1945,
      "step": 6167
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.154486110456084,
      "learning_rate": 1.1172944265917419e-06,
      "loss": 0.6162,
      "step": 6168
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.221076345544045,
      "learning_rate": 1.1159885623077076e-06,
      "loss": 0.5461,
      "step": 6169
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3595216217507826,
      "learning_rate": 1.1146833657302659e-06,
      "loss": 0.1483,
      "step": 6170
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.456902661570526,
      "learning_rate": 1.1133788370837972e-06,
      "loss": 0.182,
      "step": 6171
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.220676126731934,
      "learning_rate": 1.1120749765925643e-06,
      "loss": 0.5176,
      "step": 6172
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.483474137046647,
      "learning_rate": 1.1107717844807153e-06,
      "loss": 0.1869,
      "step": 6173
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4998874515118708,
      "learning_rate": 1.1094692609722829e-06,
      "loss": 0.1662,
      "step": 6174
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5403349762459133,
      "learning_rate": 1.108167406291189e-06,
      "loss": 0.184,
      "step": 6175
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.367044781045921,
      "learning_rate": 1.1068662206612363e-06,
      "loss": 0.186,
      "step": 6176
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.462909487167777,
      "learning_rate": 1.1055657043061124e-06,
      "loss": 0.5176,
      "step": 6177
    },
    {
      "epoch": 0.79,
      "grad_norm": 11.035181743094409,
      "learning_rate": 1.104265857449393e-06,
      "loss": 0.3489,
      "step": 6178
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5977115920066283,
      "learning_rate": 1.1029666803145356e-06,
      "loss": 0.1557,
      "step": 6179
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3860539524102347,
      "learning_rate": 1.101668173124883e-06,
      "loss": 0.1616,
      "step": 6180
    },
    {
      "epoch": 0.79,
      "grad_norm": 10.585798386977071,
      "learning_rate": 1.1003703361036644e-06,
      "loss": 0.552,
      "step": 6181
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5983351786006017,
      "learning_rate": 1.0990731694739947e-06,
      "loss": 0.2094,
      "step": 6182
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.397407697848756,
      "learning_rate": 1.0977766734588707e-06,
      "loss": 0.2005,
      "step": 6183
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.453931668948545,
      "learning_rate": 1.0964808482811751e-06,
      "loss": 0.1645,
      "step": 6184
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.0378385343387,
      "learning_rate": 1.0951856941636752e-06,
      "loss": 0.4974,
      "step": 6185
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.326533026586988,
      "learning_rate": 1.093891211329023e-06,
      "loss": 0.1752,
      "step": 6186
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3518229998564288,
      "learning_rate": 1.0925973999997535e-06,
      "loss": 0.1591,
      "step": 6187
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2975226028159588,
      "learning_rate": 1.091304260398291e-06,
      "loss": 0.1725,
      "step": 6188
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4558959733701,
      "learning_rate": 1.0900117927469384e-06,
      "loss": 0.1686,
      "step": 6189
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.161437158966273,
      "learning_rate": 1.088719997267888e-06,
      "loss": 0.5915,
      "step": 6190
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.448479661379486,
      "learning_rate": 1.087428874183214e-06,
      "loss": 0.2205,
      "step": 6191
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3129697943944552,
      "learning_rate": 1.0861384237148749e-06,
      "loss": 0.1792,
      "step": 6192
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5341270501351907,
      "learning_rate": 1.0848486460847135e-06,
      "loss": 0.1962,
      "step": 6193
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4652668098623303,
      "learning_rate": 1.083559541514459e-06,
      "loss": 0.185,
      "step": 6194
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.419020651841159,
      "learning_rate": 1.0822711102257205e-06,
      "loss": 0.6131,
      "step": 6195
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4377824780877322,
      "learning_rate": 1.0809833524399981e-06,
      "loss": 0.168,
      "step": 6196
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3706027805486691,
      "learning_rate": 1.079696268378671e-06,
      "loss": 0.2274,
      "step": 6197
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.543674756592445,
      "learning_rate": 1.0784098582630015e-06,
      "loss": 0.1392,
      "step": 6198
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.356246792480859,
      "learning_rate": 1.0771241223141415e-06,
      "loss": 0.1469,
      "step": 6199
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4354390477991437,
      "learning_rate": 1.0758390607531232e-06,
      "loss": 0.1949,
      "step": 6200
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3869000733893628,
      "learning_rate": 1.0745546738008627e-06,
      "loss": 0.1874,
      "step": 6201
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3666509348864644,
      "learning_rate": 1.0732709616781606e-06,
      "loss": 0.1912,
      "step": 6202
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4530780972998738,
      "learning_rate": 1.071987924605704e-06,
      "loss": 0.1943,
      "step": 6203
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.622810773484806,
      "learning_rate": 1.0707055628040602e-06,
      "loss": 0.5748,
      "step": 6204
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4362797113612071,
      "learning_rate": 1.0694238764936827e-06,
      "loss": 0.197,
      "step": 6205
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5219418672555918,
      "learning_rate": 1.0681428658949083e-06,
      "loss": 0.1723,
      "step": 6206
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2079401061428057,
      "learning_rate": 1.0668625312279552e-06,
      "loss": 0.1504,
      "step": 6207
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.600598388400776,
      "learning_rate": 1.0655828727129319e-06,
      "loss": 0.2109,
      "step": 6208
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5836780073673116,
      "learning_rate": 1.0643038905698239e-06,
      "loss": 0.1908,
      "step": 6209
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2287220848495237,
      "learning_rate": 1.0630255850185024e-06,
      "loss": 0.1496,
      "step": 6210
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.2270088388021625,
      "learning_rate": 1.0617479562787258e-06,
      "loss": 0.3945,
      "step": 6211
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5520931284971502,
      "learning_rate": 1.060471004570131e-06,
      "loss": 0.1843,
      "step": 6212
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5298853085890352,
      "learning_rate": 1.0591947301122424e-06,
      "loss": 0.1722,
      "step": 6213
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4930368826902423,
      "learning_rate": 1.0579191331244648e-06,
      "loss": 0.1776,
      "step": 6214
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3864910964668928,
      "learning_rate": 1.0566442138260885e-06,
      "loss": 0.2088,
      "step": 6215
    },
    {
      "epoch": 0.8,
      "grad_norm": 9.168655038782738,
      "learning_rate": 1.0553699724362866e-06,
      "loss": 0.6225,
      "step": 6216
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2654123460307158,
      "learning_rate": 1.0540964091741157e-06,
      "loss": 0.1469,
      "step": 6217
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.417417502571745,
      "learning_rate": 1.0528235242585188e-06,
      "loss": 0.5723,
      "step": 6218
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.400078763975922,
      "learning_rate": 1.0515513179083176e-06,
      "loss": 0.5047,
      "step": 6219
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.249353481646167,
      "learning_rate": 1.050279790342219e-06,
      "loss": 0.1571,
      "step": 6220
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2388940125360473,
      "learning_rate": 1.0490089417788135e-06,
      "loss": 0.1261,
      "step": 6221
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5049999310355864,
      "learning_rate": 1.0477387724365752e-06,
      "loss": 0.1708,
      "step": 6222
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2710208331829536,
      "learning_rate": 1.0464692825338584e-06,
      "loss": 0.1755,
      "step": 6223
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4598525410237446,
      "learning_rate": 1.045200472288907e-06,
      "loss": 0.2398,
      "step": 6224
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5663642415767562,
      "learning_rate": 1.0439323419198422e-06,
      "loss": 0.2002,
      "step": 6225
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4591700348892287,
      "learning_rate": 1.0426648916446692e-06,
      "loss": 0.1863,
      "step": 6226
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.535095345505745,
      "learning_rate": 1.0413981216812802e-06,
      "loss": 0.1805,
      "step": 6227
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2832311938299765,
      "learning_rate": 1.0401320322474456e-06,
      "loss": 0.1596,
      "step": 6228
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.333448655171871,
      "learning_rate": 1.0388666235608219e-06,
      "loss": 0.6087,
      "step": 6229
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3269274662294126,
      "learning_rate": 1.0376018958389472e-06,
      "loss": 0.1743,
      "step": 6230
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.475222035050357,
      "learning_rate": 1.036337849299241e-06,
      "loss": 0.177,
      "step": 6231
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5552103905850734,
      "learning_rate": 1.0350744841590106e-06,
      "loss": 0.2186,
      "step": 6232
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4882262171127179,
      "learning_rate": 1.0338118006354413e-06,
      "loss": 0.1757,
      "step": 6233
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0858422125034204,
      "learning_rate": 1.032549798945604e-06,
      "loss": 0.1216,
      "step": 6234
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.516299409600919,
      "learning_rate": 1.0312884793064493e-06,
      "loss": 0.1776,
      "step": 6235
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.837731371901007,
      "learning_rate": 1.0300278419348158e-06,
      "loss": 0.6637,
      "step": 6236
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4880306025652148,
      "learning_rate": 1.02876788704742e-06,
      "loss": 0.1898,
      "step": 6237
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.485834638931664,
      "learning_rate": 1.0275086148608614e-06,
      "loss": 0.1871,
      "step": 6238
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5235971266827033,
      "learning_rate": 1.026250025591627e-06,
      "loss": 0.1991,
      "step": 6239
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3650847254745027,
      "learning_rate": 1.0249921194560803e-06,
      "loss": 0.1618,
      "step": 6240
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2879356236327637,
      "learning_rate": 1.0237348966704708e-06,
      "loss": 0.146,
      "step": 6241
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3469832252937248,
      "learning_rate": 1.022478357450929e-06,
      "loss": 0.1731,
      "step": 6242
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2752813852102318,
      "learning_rate": 1.0212225020134693e-06,
      "loss": 0.1432,
      "step": 6243
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3155062475433792,
      "learning_rate": 1.0199673305739854e-06,
      "loss": 0.1366,
      "step": 6244
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3505627512468654,
      "learning_rate": 1.0187128433482601e-06,
      "loss": 0.1785,
      "step": 6245
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.366020295772551,
      "learning_rate": 1.0174590405519502e-06,
      "loss": 0.1758,
      "step": 6246
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5415159595611456,
      "learning_rate": 1.0162059224006027e-06,
      "loss": 0.2053,
      "step": 6247
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4525519860593021,
      "learning_rate": 1.0149534891096408e-06,
      "loss": 0.1691,
      "step": 6248
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4668286191353048,
      "learning_rate": 1.0137017408943729e-06,
      "loss": 0.1619,
      "step": 6249
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3749073678506463,
      "learning_rate": 1.0124506779699882e-06,
      "loss": 0.1884,
      "step": 6250
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.437769999757957,
      "learning_rate": 1.0112003005515603e-06,
      "loss": 0.1536,
      "step": 6251
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4925777266102966,
      "learning_rate": 1.0099506088540418e-06,
      "loss": 0.2029,
      "step": 6252
    },
    {
      "epoch": 0.8,
      "grad_norm": 8.912867888660912,
      "learning_rate": 1.0087016030922709e-06,
      "loss": 0.4869,
      "step": 6253
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4433529851815328,
      "learning_rate": 1.0074532834809647e-06,
      "loss": 0.1481,
      "step": 6254
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4883041450427856,
      "learning_rate": 1.0062056502347257e-06,
      "loss": 0.173,
      "step": 6255
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5106618689892408,
      "learning_rate": 1.0049587035680353e-06,
      "loss": 0.1834,
      "step": 6256
    },
    {
      "epoch": 0.8,
      "grad_norm": 19.181173534879942,
      "learning_rate": 1.003712443695259e-06,
      "loss": 0.6015,
      "step": 6257
    },
    {
      "epoch": 0.8,
      "grad_norm": 8.463219406316718,
      "learning_rate": 1.0024668708306418e-06,
      "loss": 0.606,
      "step": 6258
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3075608261675749,
      "learning_rate": 1.001221985188312e-06,
      "loss": 0.1712,
      "step": 6259
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7517996221751377,
      "learning_rate": 9.99977786982282e-07,
      "loss": 0.1747,
      "step": 6260
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6140896652620333,
      "learning_rate": 9.987342764264424e-07,
      "loss": 0.211,
      "step": 6261
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4553720013188438,
      "learning_rate": 9.974914537345675e-07,
      "loss": 0.1621,
      "step": 6262
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.584367291441707,
      "learning_rate": 9.96249319120311e-07,
      "loss": 0.5611,
      "step": 6263
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4963903704380066,
      "learning_rate": 9.95007872797214e-07,
      "loss": 0.1954,
      "step": 6264
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.242516438978962,
      "learning_rate": 9.937671149786933e-07,
      "loss": 0.4993,
      "step": 6265
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3247757802189075,
      "learning_rate": 9.925270458780496e-07,
      "loss": 0.1654,
      "step": 6266
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5394064218350088,
      "learning_rate": 9.91287665708464e-07,
      "loss": 0.1704,
      "step": 6267
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3667853371498864,
      "learning_rate": 9.900489746830034e-07,
      "loss": 0.1672,
      "step": 6268
    },
    {
      "epoch": 0.8,
      "grad_norm": 9.51444309979402,
      "learning_rate": 9.888109730146112e-07,
      "loss": 0.544,
      "step": 6269
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.322114269404511,
      "learning_rate": 9.87573660916114e-07,
      "loss": 0.1782,
      "step": 6270
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2257493064001976,
      "learning_rate": 9.863370386002214e-07,
      "loss": 0.1288,
      "step": 6271
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.159912660488381,
      "learning_rate": 9.851011062795201e-07,
      "loss": 0.4374,
      "step": 6272
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4136161432311516,
      "learning_rate": 9.83865864166485e-07,
      "loss": 0.1627,
      "step": 6273
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.338456092012816,
      "learning_rate": 9.826313124734654e-07,
      "loss": 0.128,
      "step": 6274
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5032210524014114,
      "learning_rate": 9.813974514126977e-07,
      "loss": 0.187,
      "step": 6275
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.021647383676687,
      "learning_rate": 9.801642811962964e-07,
      "loss": 0.6561,
      "step": 6276
    },
    {
      "epoch": 0.8,
      "grad_norm": 8.684621815952466,
      "learning_rate": 9.789318020362564e-07,
      "loss": 0.6574,
      "step": 6277
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.501403833784775,
      "learning_rate": 9.777000141444564e-07,
      "loss": 0.194,
      "step": 6278
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5417597399160317,
      "learning_rate": 9.764689177326542e-07,
      "loss": 0.2078,
      "step": 6279
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3158484699766209,
      "learning_rate": 9.75238513012488e-07,
      "loss": 0.1668,
      "step": 6280
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.448592327373514,
      "learning_rate": 9.74008800195481e-07,
      "loss": 0.1718,
      "step": 6281
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.929824578171731,
      "learning_rate": 9.727797794930361e-07,
      "loss": 0.475,
      "step": 6282
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4101975889045106,
      "learning_rate": 9.715514511164343e-07,
      "loss": 0.1495,
      "step": 6283
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.975211062961501,
      "learning_rate": 9.703238152768402e-07,
      "loss": 0.6299,
      "step": 6284
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4413314175444565,
      "learning_rate": 9.690968721852978e-07,
      "loss": 0.1547,
      "step": 6285
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.4619895361785,
      "learning_rate": 9.678706220527333e-07,
      "loss": 0.7616,
      "step": 6286
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5039879200008865,
      "learning_rate": 9.666450650899533e-07,
      "loss": 0.1705,
      "step": 6287
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3046969581887566,
      "learning_rate": 9.654202015076442e-07,
      "loss": 0.1586,
      "step": 6288
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5045185813308746,
      "learning_rate": 9.641960315163768e-07,
      "loss": 0.1321,
      "step": 6289
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.6391729719420205,
      "learning_rate": 9.629725553265983e-07,
      "loss": 0.3999,
      "step": 6290
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.580624072813515,
      "learning_rate": 9.617497731486374e-07,
      "loss": 0.5967,
      "step": 6291
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2918551301697105,
      "learning_rate": 9.605276851927075e-07,
      "loss": 0.1586,
      "step": 6292
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3594436864260833,
      "learning_rate": 9.593062916688982e-07,
      "loss": 0.1873,
      "step": 6293
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.575590312911236,
      "learning_rate": 9.580855927871808e-07,
      "loss": 0.1758,
      "step": 6294
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5693529010440916,
      "learning_rate": 9.56865588757407e-07,
      "loss": 0.1863,
      "step": 6295
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3148483399696718,
      "learning_rate": 9.556462797893113e-07,
      "loss": 0.1786,
      "step": 6296
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5033133038281996,
      "learning_rate": 9.544276660925067e-07,
      "loss": 0.1864,
      "step": 6297
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.237361646637826,
      "learning_rate": 9.532097478764862e-07,
      "loss": 0.1441,
      "step": 6298
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4036999696186658,
      "learning_rate": 9.519925253506246e-07,
      "loss": 0.1672,
      "step": 6299
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4865845101256596,
      "learning_rate": 9.507759987241755e-07,
      "loss": 0.1787,
      "step": 6300
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4838056753300553,
      "learning_rate": 9.495601682062755e-07,
      "loss": 0.1698,
      "step": 6301
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.644963497549756,
      "learning_rate": 9.483450340059386e-07,
      "loss": 0.6175,
      "step": 6302
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5315110048791531,
      "learning_rate": 9.47130596332062e-07,
      "loss": 0.1826,
      "step": 6303
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5071706833132112,
      "learning_rate": 9.459168553934211e-07,
      "loss": 0.1467,
      "step": 6304
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.392711420997582,
      "learning_rate": 9.447038113986717e-07,
      "loss": 0.528,
      "step": 6305
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4552372603276378,
      "learning_rate": 9.434914645563498e-07,
      "loss": 0.1694,
      "step": 6306
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.302107210893766,
      "learning_rate": 9.422798150748724e-07,
      "loss": 0.1507,
      "step": 6307
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5346154678164192,
      "learning_rate": 9.410688631625364e-07,
      "loss": 0.1926,
      "step": 6308
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2872659946727176,
      "learning_rate": 9.398586090275164e-07,
      "loss": 0.1481,
      "step": 6309
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2258093156180776,
      "learning_rate": 9.386490528778702e-07,
      "loss": 0.1706,
      "step": 6310
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5383274079021878,
      "learning_rate": 9.374401949215367e-07,
      "loss": 0.1784,
      "step": 6311
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8250775871641267,
      "learning_rate": 9.362320353663313e-07,
      "loss": 0.2096,
      "step": 6312
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5545535141962628,
      "learning_rate": 9.350245744199499e-07,
      "loss": 0.2073,
      "step": 6313
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.916862118937182,
      "learning_rate": 9.338178122899693e-07,
      "loss": 0.4762,
      "step": 6314
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5226567422551314,
      "learning_rate": 9.32611749183846e-07,
      "loss": 0.2004,
      "step": 6315
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2574383899937627,
      "learning_rate": 9.31406385308915e-07,
      "loss": 0.1472,
      "step": 6316
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.527801493074588,
      "learning_rate": 9.302017208723951e-07,
      "loss": 0.6464,
      "step": 6317
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.285327471813423,
      "learning_rate": 9.289977560813789e-07,
      "loss": 0.1545,
      "step": 6318
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3350510846716093,
      "learning_rate": 9.27794491142845e-07,
      "loss": 0.2416,
      "step": 6319
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4448005439698215,
      "learning_rate": 9.265919262636469e-07,
      "loss": 0.1835,
      "step": 6320
    },
    {
      "epoch": 0.81,
      "grad_norm": 10.770231744014964,
      "learning_rate": 9.253900616505202e-07,
      "loss": 0.4572,
      "step": 6321
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3645127787885503,
      "learning_rate": 9.241888975100782e-07,
      "loss": 0.1802,
      "step": 6322
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.390823610557108,
      "learning_rate": 9.229884340488149e-07,
      "loss": 0.1629,
      "step": 6323
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.857852399573938,
      "learning_rate": 9.217886714731056e-07,
      "loss": 0.5628,
      "step": 6324
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5997519088792407,
      "learning_rate": 9.205896099892019e-07,
      "loss": 0.188,
      "step": 6325
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5325072111169356,
      "learning_rate": 9.193912498032376e-07,
      "loss": 0.1751,
      "step": 6326
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.821677460597623,
      "learning_rate": 9.181935911212231e-07,
      "loss": 0.5368,
      "step": 6327
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4377880383059538,
      "learning_rate": 9.169966341490499e-07,
      "loss": 0.2017,
      "step": 6328
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.398839529151077,
      "learning_rate": 9.158003790924908e-07,
      "loss": 0.1713,
      "step": 6329
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5249781163368068,
      "learning_rate": 9.146048261571944e-07,
      "loss": 0.1805,
      "step": 6330
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6499399788608566,
      "learning_rate": 9.134099755486892e-07,
      "loss": 0.2068,
      "step": 6331
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3272778990245022,
      "learning_rate": 9.122158274723863e-07,
      "loss": 0.1641,
      "step": 6332
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.027893488039328,
      "learning_rate": 9.110223821335723e-07,
      "loss": 0.6848,
      "step": 6333
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6323393200703045,
      "learning_rate": 9.098296397374146e-07,
      "loss": 0.1691,
      "step": 6334
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.279153860847879,
      "learning_rate": 9.086376004889591e-07,
      "loss": 0.1775,
      "step": 6335
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3277338200871238,
      "learning_rate": 9.074462645931309e-07,
      "loss": 0.173,
      "step": 6336
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4646032446233832,
      "learning_rate": 9.062556322547333e-07,
      "loss": 0.1965,
      "step": 6337
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4923440634377312,
      "learning_rate": 9.050657036784516e-07,
      "loss": 0.2385,
      "step": 6338
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.378747002257013,
      "learning_rate": 9.038764790688492e-07,
      "loss": 0.1861,
      "step": 6339
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3758740047200404,
      "learning_rate": 9.026879586303666e-07,
      "loss": 0.1338,
      "step": 6340
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.338562822679211,
      "learning_rate": 9.015001425673242e-07,
      "loss": 0.1878,
      "step": 6341
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4292096189166044,
      "learning_rate": 9.003130310839203e-07,
      "loss": 0.1461,
      "step": 6342
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4638851613695605,
      "learning_rate": 8.99126624384235e-07,
      "loss": 0.1724,
      "step": 6343
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.535973815786625,
      "learning_rate": 8.979409226722224e-07,
      "loss": 0.6,
      "step": 6344
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4553565556594423,
      "learning_rate": 8.967559261517217e-07,
      "loss": 0.1711,
      "step": 6345
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3721567508516845,
      "learning_rate": 8.955716350264454e-07,
      "loss": 0.1879,
      "step": 6346
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.223600036745847,
      "learning_rate": 8.943880494999884e-07,
      "loss": 0.147,
      "step": 6347
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.404309323957642,
      "learning_rate": 8.932051697758227e-07,
      "loss": 0.4588,
      "step": 6348
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3870588985753567,
      "learning_rate": 8.920229960572973e-07,
      "loss": 0.1805,
      "step": 6349
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.52555057728179,
      "learning_rate": 8.908415285476435e-07,
      "loss": 0.1653,
      "step": 6350
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5044116525922122,
      "learning_rate": 8.89660767449968e-07,
      "loss": 0.2041,
      "step": 6351
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.22419890474807,
      "learning_rate": 8.884807129672568e-07,
      "loss": 0.5148,
      "step": 6352
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.447031182564299,
      "learning_rate": 8.873013653023765e-07,
      "loss": 0.1813,
      "step": 6353
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.278813863517362,
      "learning_rate": 8.861227246580706e-07,
      "loss": 0.1702,
      "step": 6354
    },
    {
      "epoch": 0.81,
      "grad_norm": 10.369570960792426,
      "learning_rate": 8.849447912369591e-07,
      "loss": 0.3888,
      "step": 6355
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.46063481798872,
      "learning_rate": 8.837675652415451e-07,
      "loss": 0.1654,
      "step": 6356
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.405847138719362,
      "learning_rate": 8.82591046874206e-07,
      "loss": 0.1932,
      "step": 6357
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3121625817151517,
      "learning_rate": 8.814152363371992e-07,
      "loss": 0.1831,
      "step": 6358
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4146220496411457,
      "learning_rate": 8.802401338326582e-07,
      "loss": 0.2119,
      "step": 6359
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3984354254398192,
      "learning_rate": 8.790657395626001e-07,
      "loss": 0.1607,
      "step": 6360
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.473013104105561,
      "learning_rate": 8.778920537289154e-07,
      "loss": 0.2023,
      "step": 6361
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.538929680256893,
      "learning_rate": 8.767190765333744e-07,
      "loss": 0.5838,
      "step": 6362
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4906221970381237,
      "learning_rate": 8.755468081776252e-07,
      "loss": 0.1731,
      "step": 6363
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3721408648857996,
      "learning_rate": 8.743752488631946e-07,
      "loss": 0.1713,
      "step": 6364
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.282017358786757,
      "learning_rate": 8.732043987914856e-07,
      "loss": 0.1945,
      "step": 6365
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4888369414436378,
      "learning_rate": 8.720342581637836e-07,
      "loss": 0.1899,
      "step": 6366
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3156398011952455,
      "learning_rate": 8.708648271812469e-07,
      "loss": 0.1549,
      "step": 6367
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4586798353499781,
      "learning_rate": 8.696961060449166e-07,
      "loss": 0.1691,
      "step": 6368
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.311509022097438,
      "learning_rate": 8.685280949557084e-07,
      "loss": 0.1827,
      "step": 6369
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2738050423369873,
      "learning_rate": 8.673607941144169e-07,
      "loss": 0.1579,
      "step": 6370
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1209075162987494,
      "learning_rate": 8.661942037217141e-07,
      "loss": 0.1219,
      "step": 6371
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.457503395975815,
      "learning_rate": 8.650283239781515e-07,
      "loss": 0.1917,
      "step": 6372
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4886621602410988,
      "learning_rate": 8.638631550841553e-07,
      "loss": 0.1897,
      "step": 6373
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7136422662499564,
      "learning_rate": 8.626986972400326e-07,
      "loss": 0.2177,
      "step": 6374
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4505775936038336,
      "learning_rate": 8.615349506459691e-07,
      "loss": 0.2095,
      "step": 6375
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.424944833297931,
      "learning_rate": 8.603719155020246e-07,
      "loss": 0.1679,
      "step": 6376
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.42693329496506,
      "learning_rate": 8.592095920081383e-07,
      "loss": 0.1681,
      "step": 6377
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3032897902830074,
      "learning_rate": 8.580479803641279e-07,
      "loss": 0.2113,
      "step": 6378
    },
    {
      "epoch": 0.82,
      "grad_norm": 8.147950992704423,
      "learning_rate": 8.568870807696872e-07,
      "loss": 0.5985,
      "step": 6379
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3410313508614289,
      "learning_rate": 8.557268934243868e-07,
      "loss": 0.1788,
      "step": 6380
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3991410089384828,
      "learning_rate": 8.545674185276792e-07,
      "loss": 0.1927,
      "step": 6381
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5882571057618435,
      "learning_rate": 8.534086562788907e-07,
      "loss": 0.1699,
      "step": 6382
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3528252239179837,
      "learning_rate": 8.52250606877224e-07,
      "loss": 0.1837,
      "step": 6383
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.40289616183659,
      "learning_rate": 8.510932705217645e-07,
      "loss": 0.6486,
      "step": 6384
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.954770092032652,
      "learning_rate": 8.499366474114695e-07,
      "loss": 0.6152,
      "step": 6385
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.482546109937985,
      "learning_rate": 8.487807377451767e-07,
      "loss": 0.6509,
      "step": 6386
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3696279427646216,
      "learning_rate": 8.476255417216007e-07,
      "loss": 0.1901,
      "step": 6387
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3166110925172978,
      "learning_rate": 8.464710595393306e-07,
      "loss": 0.192,
      "step": 6388
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3708631110911824,
      "learning_rate": 8.453172913968382e-07,
      "loss": 0.1569,
      "step": 6389
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.050185870684018,
      "learning_rate": 8.441642374924692e-07,
      "loss": 0.4553,
      "step": 6390
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4051372866127025,
      "learning_rate": 8.430118980244462e-07,
      "loss": 0.1789,
      "step": 6391
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.431335482326751,
      "learning_rate": 8.418602731908687e-07,
      "loss": 0.1599,
      "step": 6392
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.9383759920499335,
      "learning_rate": 8.407093631897168e-07,
      "loss": 0.7165,
      "step": 6393
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3992474698743722,
      "learning_rate": 8.395591682188442e-07,
      "loss": 0.1858,
      "step": 6394
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.247955976224084,
      "learning_rate": 8.384096884759807e-07,
      "loss": 0.142,
      "step": 6395
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.495400418976924,
      "learning_rate": 8.372609241587387e-07,
      "loss": 0.1886,
      "step": 6396
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3743089234846546,
      "learning_rate": 8.361128754646025e-07,
      "loss": 0.1648,
      "step": 6397
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5419250554353936,
      "learning_rate": 8.349655425909348e-07,
      "loss": 0.1826,
      "step": 6398
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2756732145280685,
      "learning_rate": 8.338189257349755e-07,
      "loss": 0.1415,
      "step": 6399
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.702524245084277,
      "learning_rate": 8.326730250938414e-07,
      "loss": 0.4789,
      "step": 6400
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2608509742255924,
      "learning_rate": 8.31527840864525e-07,
      "loss": 0.1314,
      "step": 6401
    },
    {
      "epoch": 0.82,
      "grad_norm": 25.114358380478407,
      "learning_rate": 8.303833732438988e-07,
      "loss": 0.5784,
      "step": 6402
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3544702255228884,
      "learning_rate": 8.29239622428708e-07,
      "loss": 0.1604,
      "step": 6403
    },
    {
      "epoch": 0.82,
      "grad_norm": 12.000634991117362,
      "learning_rate": 8.280965886155789e-07,
      "loss": 0.5475,
      "step": 6404
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4793647489421988,
      "learning_rate": 8.26954272001011e-07,
      "loss": 0.2008,
      "step": 6405
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3762056391784334,
      "learning_rate": 8.25812672781382e-07,
      "loss": 0.1905,
      "step": 6406
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3641852310522116,
      "learning_rate": 8.246717911529456e-07,
      "loss": 0.1655,
      "step": 6407
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4783036221970312,
      "learning_rate": 8.235316273118333e-07,
      "loss": 0.1868,
      "step": 6408
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2603486850254917,
      "learning_rate": 8.223921814540503e-07,
      "loss": 0.1534,
      "step": 6409
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4688593707744038,
      "learning_rate": 8.212534537754841e-07,
      "loss": 0.1865,
      "step": 6410
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3191025567214942,
      "learning_rate": 8.201154444718917e-07,
      "loss": 0.1841,
      "step": 6411
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.591851363552277,
      "learning_rate": 8.189781537389135e-07,
      "loss": 0.2048,
      "step": 6412
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6437635066729306,
      "learning_rate": 8.178415817720609e-07,
      "loss": 0.2022,
      "step": 6413
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3069557847677549,
      "learning_rate": 8.16705728766724e-07,
      "loss": 0.1503,
      "step": 6414
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3484252444221836,
      "learning_rate": 8.155705949181691e-07,
      "loss": 0.1726,
      "step": 6415
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5806950510909406,
      "learning_rate": 8.144361804215384e-07,
      "loss": 0.1993,
      "step": 6416
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4556481400659402,
      "learning_rate": 8.133024854718524e-07,
      "loss": 0.1766,
      "step": 6417
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3576204181168616,
      "learning_rate": 8.121695102640053e-07,
      "loss": 0.1447,
      "step": 6418
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3923731525549286,
      "learning_rate": 8.110372549927692e-07,
      "loss": 0.178,
      "step": 6419
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6092776518621732,
      "learning_rate": 8.099057198527899e-07,
      "loss": 0.2141,
      "step": 6420
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4810145741745786,
      "learning_rate": 8.087749050385952e-07,
      "loss": 0.1677,
      "step": 6421
    },
    {
      "epoch": 0.82,
      "grad_norm": 8.611818241945812,
      "learning_rate": 8.076448107445822e-07,
      "loss": 0.5399,
      "step": 6422
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.619708916399224,
      "learning_rate": 8.065154371650286e-07,
      "loss": 0.2133,
      "step": 6423
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.557020169233651,
      "learning_rate": 8.053867844940855e-07,
      "loss": 0.5406,
      "step": 6424
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.423390223029999,
      "learning_rate": 8.042588529257828e-07,
      "loss": 0.5672,
      "step": 6425
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1947074344088668,
      "learning_rate": 8.031316426540254e-07,
      "loss": 0.1133,
      "step": 6426
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.390677485444262,
      "learning_rate": 8.02005153872592e-07,
      "loss": 0.1884,
      "step": 6427
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3636192665974773,
      "learning_rate": 8.008793867751402e-07,
      "loss": 0.1861,
      "step": 6428
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.545963573410636,
      "learning_rate": 7.997543415552011e-07,
      "loss": 0.2321,
      "step": 6429
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.213706212259862,
      "learning_rate": 7.986300184061857e-07,
      "loss": 0.6222,
      "step": 6430
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4875782296550357,
      "learning_rate": 7.975064175213748e-07,
      "loss": 0.1865,
      "step": 6431
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.751606440364249,
      "learning_rate": 7.963835390939317e-07,
      "loss": 0.5365,
      "step": 6432
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3792492274244095,
      "learning_rate": 7.952613833168909e-07,
      "loss": 0.1796,
      "step": 6433
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.348317480476053,
      "learning_rate": 7.941399503831637e-07,
      "loss": 0.168,
      "step": 6434
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.257838933141048,
      "learning_rate": 7.930192404855375e-07,
      "loss": 0.1572,
      "step": 6435
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4007725384530794,
      "learning_rate": 7.918992538166753e-07,
      "loss": 0.1695,
      "step": 6436
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3244703922705137,
      "learning_rate": 7.907799905691144e-07,
      "loss": 0.1439,
      "step": 6437
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.525392660453871,
      "learning_rate": 7.896614509352724e-07,
      "loss": 0.2179,
      "step": 6438
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.602502970161896,
      "learning_rate": 7.885436351074355e-07,
      "loss": 0.1645,
      "step": 6439
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5262711604454502,
      "learning_rate": 7.874265432777728e-07,
      "loss": 0.1866,
      "step": 6440
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.30018466373041,
      "learning_rate": 7.863101756383235e-07,
      "loss": 0.1885,
      "step": 6441
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3451764515265532,
      "learning_rate": 7.851945323810045e-07,
      "loss": 0.1767,
      "step": 6442
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.540653860115125,
      "learning_rate": 7.840796136976075e-07,
      "loss": 0.4756,
      "step": 6443
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4686654391843268,
      "learning_rate": 7.829654197797998e-07,
      "loss": 0.1989,
      "step": 6444
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4719879772005404,
      "learning_rate": 7.818519508191236e-07,
      "loss": 0.1755,
      "step": 6445
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.242166209538563,
      "learning_rate": 7.807392070069992e-07,
      "loss": 0.5101,
      "step": 6446
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4866996080994153,
      "learning_rate": 7.796271885347189e-07,
      "loss": 0.1671,
      "step": 6447
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.454325299725053,
      "learning_rate": 7.785158955934508e-07,
      "loss": 0.162,
      "step": 6448
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4749841528629533,
      "learning_rate": 7.774053283742406e-07,
      "loss": 0.1732,
      "step": 6449
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3507045818785037,
      "learning_rate": 7.762954870680067e-07,
      "loss": 0.1536,
      "step": 6450
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4692912750426108,
      "learning_rate": 7.751863718655444e-07,
      "loss": 0.1572,
      "step": 6451
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.4977791935240665,
      "learning_rate": 7.740779829575218e-07,
      "loss": 0.4407,
      "step": 6452
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4500069166536358,
      "learning_rate": 7.729703205344863e-07,
      "loss": 0.2255,
      "step": 6453
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5867231487125566,
      "learning_rate": 7.718633847868568e-07,
      "loss": 0.2194,
      "step": 6454
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2125547820695262,
      "learning_rate": 7.707571759049281e-07,
      "loss": 0.1735,
      "step": 6455
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4513567252843098,
      "learning_rate": 7.696516940788701e-07,
      "loss": 0.1844,
      "step": 6456
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5433450398008264,
      "learning_rate": 7.685469394987271e-07,
      "loss": 0.1887,
      "step": 6457
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3459887336356289,
      "learning_rate": 7.67442912354422e-07,
      "loss": 0.1609,
      "step": 6458
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5598398520944878,
      "learning_rate": 7.663396128357481e-07,
      "loss": 0.1901,
      "step": 6459
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5629748869096733,
      "learning_rate": 7.652370411323745e-07,
      "loss": 0.1682,
      "step": 6460
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.652312339518471,
      "learning_rate": 7.641351974338478e-07,
      "loss": 0.5711,
      "step": 6461
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2440392588649016,
      "learning_rate": 7.630340819295879e-07,
      "loss": 0.1521,
      "step": 6462
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.155700594091241,
      "learning_rate": 7.619336948088879e-07,
      "loss": 0.5026,
      "step": 6463
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.8275372184281915,
      "learning_rate": 7.608340362609174e-07,
      "loss": 0.5773,
      "step": 6464
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4113998892021369,
      "learning_rate": 7.597351064747211e-07,
      "loss": 0.1878,
      "step": 6465
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4274320973127674,
      "learning_rate": 7.586369056392162e-07,
      "loss": 0.1831,
      "step": 6466
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6597893881546208,
      "learning_rate": 7.575394339431969e-07,
      "loss": 0.2125,
      "step": 6467
    },
    {
      "epoch": 0.83,
      "grad_norm": 8.450858919658712,
      "learning_rate": 7.564426915753331e-07,
      "loss": 0.6656,
      "step": 6468
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.244905997335831,
      "learning_rate": 7.553466787241665e-07,
      "loss": 0.1352,
      "step": 6469
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.85132925034517,
      "learning_rate": 7.542513955781139e-07,
      "loss": 0.4518,
      "step": 6470
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6251276717763985,
      "learning_rate": 7.53156842325467e-07,
      "loss": 0.1873,
      "step": 6471
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2631210397994852,
      "learning_rate": 7.520630191543932e-07,
      "loss": 0.1678,
      "step": 6472
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6928085035221379,
      "learning_rate": 7.509699262529308e-07,
      "loss": 0.2406,
      "step": 6473
    },
    {
      "epoch": 0.83,
      "grad_norm": 11.278789800987797,
      "learning_rate": 7.498775638089989e-07,
      "loss": 0.6755,
      "step": 6474
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3865123474625105,
      "learning_rate": 7.487859320103847e-07,
      "loss": 0.1569,
      "step": 6475
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5128785100239148,
      "learning_rate": 7.476950310447523e-07,
      "loss": 0.168,
      "step": 6476
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5026456301314202,
      "learning_rate": 7.466048610996423e-07,
      "loss": 0.1774,
      "step": 6477
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.050782389374096,
      "learning_rate": 7.455154223624661e-07,
      "loss": 0.4563,
      "step": 6478
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3072941346237965,
      "learning_rate": 7.444267150205108e-07,
      "loss": 0.1797,
      "step": 6479
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.373486277572935,
      "learning_rate": 7.433387392609387e-07,
      "loss": 0.4859,
      "step": 6480
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.400358478101089,
      "learning_rate": 7.422514952707832e-07,
      "loss": 0.1937,
      "step": 6481
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4127343481856054,
      "learning_rate": 7.411649832369566e-07,
      "loss": 0.1856,
      "step": 6482
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5354443245849787,
      "learning_rate": 7.400792033462428e-07,
      "loss": 0.1715,
      "step": 6483
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2698130618358325,
      "learning_rate": 7.389941557852987e-07,
      "loss": 0.1718,
      "step": 6484
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3039542094297696,
      "learning_rate": 7.379098407406554e-07,
      "loss": 0.1717,
      "step": 6485
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4004479946054647,
      "learning_rate": 7.368262583987229e-07,
      "loss": 0.1682,
      "step": 6486
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4460587527542714,
      "learning_rate": 7.357434089457788e-07,
      "loss": 0.1905,
      "step": 6487
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5087145166089353,
      "learning_rate": 7.346612925679774e-07,
      "loss": 0.1464,
      "step": 6488
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.71716505175345,
      "learning_rate": 7.33579909451349e-07,
      "loss": 0.5816,
      "step": 6489
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2963248091551944,
      "learning_rate": 7.324992597817948e-07,
      "loss": 0.1683,
      "step": 6490
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2916486884946918,
      "learning_rate": 7.314193437450911e-07,
      "loss": 0.1673,
      "step": 6491
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3447529682080375,
      "learning_rate": 7.30340161526888e-07,
      "loss": 0.164,
      "step": 6492
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6888904373304432,
      "learning_rate": 7.292617133127083e-07,
      "loss": 0.191,
      "step": 6493
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2844379357392242,
      "learning_rate": 7.281839992879503e-07,
      "loss": 0.1501,
      "step": 6494
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.465976901230881,
      "learning_rate": 7.271070196378859e-07,
      "loss": 0.1573,
      "step": 6495
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.295158836929715,
      "learning_rate": 7.260307745476619e-07,
      "loss": 0.1701,
      "step": 6496
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5141709146879756,
      "learning_rate": 7.249552642022956e-07,
      "loss": 0.1787,
      "step": 6497
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5100057919574483,
      "learning_rate": 7.238804887866796e-07,
      "loss": 0.1622,
      "step": 6498
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4386860013481402,
      "learning_rate": 7.228064484855807e-07,
      "loss": 0.1554,
      "step": 6499
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.652253929043126,
      "learning_rate": 7.217331434836395e-07,
      "loss": 0.5181,
      "step": 6500
    },
    {
      "epoch": 0.83,
      "grad_norm": 10.326213719357634,
      "learning_rate": 7.206605739653683e-07,
      "loss": 0.6605,
      "step": 6501
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.049544636285201,
      "learning_rate": 7.195887401151536e-07,
      "loss": 0.6079,
      "step": 6502
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4734086878025672,
      "learning_rate": 7.185176421172573e-07,
      "loss": 0.1866,
      "step": 6503
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3836850411218382,
      "learning_rate": 7.174472801558147e-07,
      "loss": 0.1903,
      "step": 6504
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.397585244734763,
      "learning_rate": 7.163776544148321e-07,
      "loss": 0.2037,
      "step": 6505
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4182783714668998,
      "learning_rate": 7.153087650781909e-07,
      "loss": 0.1694,
      "step": 6506
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3943055739667205,
      "learning_rate": 7.142406123296452e-07,
      "loss": 0.1835,
      "step": 6507
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.147950963223466,
      "learning_rate": 7.131731963528232e-07,
      "loss": 0.7113,
      "step": 6508
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3885301766593945,
      "learning_rate": 7.121065173312253e-07,
      "loss": 0.1566,
      "step": 6509
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4070092892330217,
      "learning_rate": 7.110405754482269e-07,
      "loss": 0.1806,
      "step": 6510
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3984038869353033,
      "learning_rate": 7.09975370887076e-07,
      "loss": 0.1402,
      "step": 6511
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.639896401467398,
      "learning_rate": 7.089109038308928e-07,
      "loss": 0.5201,
      "step": 6512
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4382329075441995,
      "learning_rate": 7.078471744626708e-07,
      "loss": 0.1709,
      "step": 6513
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3527515078550665,
      "learning_rate": 7.067841829652794e-07,
      "loss": 0.1664,
      "step": 6514
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.621080465126018,
      "learning_rate": 7.057219295214579e-07,
      "loss": 0.2196,
      "step": 6515
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4436750879540554,
      "learning_rate": 7.046604143138198e-07,
      "loss": 0.183,
      "step": 6516
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4604391348094719,
      "learning_rate": 7.035996375248527e-07,
      "loss": 0.2146,
      "step": 6517
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.718754020211077,
      "learning_rate": 7.025395993369166e-07,
      "loss": 0.6879,
      "step": 6518
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2536603208406802,
      "learning_rate": 7.01480299932244e-07,
      "loss": 0.1348,
      "step": 6519
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3090788002900895,
      "learning_rate": 7.004217394929402e-07,
      "loss": 0.1521,
      "step": 6520
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.560174479065981,
      "learning_rate": 6.993639182009843e-07,
      "loss": 0.1737,
      "step": 6521
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5044561271621362,
      "learning_rate": 6.983068362382272e-07,
      "loss": 0.2019,
      "step": 6522
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5626989066759647,
      "learning_rate": 6.972504937863955e-07,
      "loss": 0.2223,
      "step": 6523
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5409431273688126,
      "learning_rate": 6.961948910270844e-07,
      "loss": 0.2014,
      "step": 6524
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3323821212108884,
      "learning_rate": 6.951400281417669e-07,
      "loss": 0.1789,
      "step": 6525
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4543987451207963,
      "learning_rate": 6.940859053117843e-07,
      "loss": 0.1808,
      "step": 6526
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.483695907657301,
      "learning_rate": 6.930325227183537e-07,
      "loss": 0.2094,
      "step": 6527
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.647908836948302,
      "learning_rate": 6.919798805425626e-07,
      "loss": 0.2024,
      "step": 6528
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4831744364028938,
      "learning_rate": 6.909279789653734e-07,
      "loss": 0.1637,
      "step": 6529
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.281455919499814,
      "learning_rate": 6.89876818167618e-07,
      "loss": 0.1458,
      "step": 6530
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3423224000630802,
      "learning_rate": 6.888263983300048e-07,
      "loss": 0.1696,
      "step": 6531
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4993680401786789,
      "learning_rate": 6.877767196331147e-07,
      "loss": 0.164,
      "step": 6532
    },
    {
      "epoch": 0.84,
      "grad_norm": 7.41531522794484,
      "learning_rate": 6.867277822573975e-07,
      "loss": 0.513,
      "step": 6533
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5264961375193282,
      "learning_rate": 6.856795863831789e-07,
      "loss": 0.2152,
      "step": 6534
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5487186921304494,
      "learning_rate": 6.846321321906551e-07,
      "loss": 0.1855,
      "step": 6535
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4209983367950556,
      "learning_rate": 6.835854198598957e-07,
      "loss": 0.1805,
      "step": 6536
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.454768483481851,
      "learning_rate": 6.825394495708415e-07,
      "loss": 0.1839,
      "step": 6537
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.487991526582857,
      "learning_rate": 6.814942215033099e-07,
      "loss": 0.6765,
      "step": 6538
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.924297231099055,
      "learning_rate": 6.804497358369855e-07,
      "loss": 0.5848,
      "step": 6539
    },
    {
      "epoch": 0.84,
      "grad_norm": 8.425556273922558,
      "learning_rate": 6.794059927514268e-07,
      "loss": 0.5167,
      "step": 6540
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4291862507296287,
      "learning_rate": 6.783629924260682e-07,
      "loss": 0.2018,
      "step": 6541
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2710132423410727,
      "learning_rate": 6.773207350402117e-07,
      "loss": 0.1949,
      "step": 6542
    },
    {
      "epoch": 0.84,
      "grad_norm": 7.46217544543931,
      "learning_rate": 6.762792207730334e-07,
      "loss": 0.5982,
      "step": 6543
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2668310055887657,
      "learning_rate": 6.752384498035824e-07,
      "loss": 0.1762,
      "step": 6544
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5311508788640844,
      "learning_rate": 6.741984223107773e-07,
      "loss": 0.1893,
      "step": 6545
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.418203231935403,
      "learning_rate": 6.731591384734138e-07,
      "loss": 0.2129,
      "step": 6546
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3855713689254963,
      "learning_rate": 6.721205984701551e-07,
      "loss": 0.1537,
      "step": 6547
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.417649356746784,
      "learning_rate": 6.71082802479539e-07,
      "loss": 0.1886,
      "step": 6548
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.413188636295602,
      "learning_rate": 6.70045750679974e-07,
      "loss": 0.2109,
      "step": 6549
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4460922149433666,
      "learning_rate": 6.690094432497407e-07,
      "loss": 0.1941,
      "step": 6550
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5430924615831128,
      "learning_rate": 6.679738803669944e-07,
      "loss": 0.19,
      "step": 6551
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.635008871813503,
      "learning_rate": 6.669390622097577e-07,
      "loss": 0.1663,
      "step": 6552
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3977201333020237,
      "learning_rate": 6.65904988955931e-07,
      "loss": 0.1731,
      "step": 6553
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.395798515392245,
      "learning_rate": 6.648716607832811e-07,
      "loss": 0.5989,
      "step": 6554
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.290733752235696,
      "learning_rate": 6.638390778694504e-07,
      "loss": 0.1547,
      "step": 6555
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.8233683231056075,
      "learning_rate": 6.628072403919511e-07,
      "loss": 0.4831,
      "step": 6556
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4567351134129571,
      "learning_rate": 6.617761485281687e-07,
      "loss": 0.1594,
      "step": 6557
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.462436245976243,
      "learning_rate": 6.607458024553576e-07,
      "loss": 0.2021,
      "step": 6558
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3694961521799218,
      "learning_rate": 6.597162023506492e-07,
      "loss": 0.1696,
      "step": 6559
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4687328681777647,
      "learning_rate": 6.586873483910416e-07,
      "loss": 0.1842,
      "step": 6560
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3871136008325378,
      "learning_rate": 6.576592407534088e-07,
      "loss": 0.1741,
      "step": 6561
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.715815313621055,
      "learning_rate": 6.566318796144933e-07,
      "loss": 0.4847,
      "step": 6562
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2637062084952984,
      "learning_rate": 6.556052651509104e-07,
      "loss": 0.1487,
      "step": 6563
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2696733635326038,
      "learning_rate": 6.545793975391468e-07,
      "loss": 0.1517,
      "step": 6564
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.613022627248835,
      "learning_rate": 6.535542769555609e-07,
      "loss": 0.1975,
      "step": 6565
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.635889932141604,
      "learning_rate": 6.525299035763827e-07,
      "loss": 0.2183,
      "step": 6566
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.695718050174527,
      "learning_rate": 6.515062775777148e-07,
      "loss": 0.611,
      "step": 6567
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7882103490703307,
      "learning_rate": 6.504833991355292e-07,
      "loss": 0.228,
      "step": 6568
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4630928788047934,
      "learning_rate": 6.494612684256718e-07,
      "loss": 0.1578,
      "step": 6569
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4649127235470831,
      "learning_rate": 6.484398856238582e-07,
      "loss": 0.1915,
      "step": 6570
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5975013308862716,
      "learning_rate": 6.474192509056759e-07,
      "loss": 0.1957,
      "step": 6571
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4930075762140864,
      "learning_rate": 6.463993644465843e-07,
      "loss": 0.152,
      "step": 6572
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3565119092589815,
      "learning_rate": 6.453802264219117e-07,
      "loss": 0.1762,
      "step": 6573
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2282107519727388,
      "learning_rate": 6.443618370068622e-07,
      "loss": 0.1325,
      "step": 6574
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3921591904774118,
      "learning_rate": 6.43344196376508e-07,
      "loss": 0.1941,
      "step": 6575
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4837849687209035,
      "learning_rate": 6.423273047057932e-07,
      "loss": 0.1716,
      "step": 6576
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.165468858155727,
      "learning_rate": 6.413111621695322e-07,
      "loss": 0.4215,
      "step": 6577
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3981886841195406,
      "learning_rate": 6.402957689424139e-07,
      "loss": 0.1608,
      "step": 6578
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6057114216617596,
      "learning_rate": 6.392811251989944e-07,
      "loss": 0.1991,
      "step": 6579
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3829267280430835,
      "learning_rate": 6.382672311137039e-07,
      "loss": 0.162,
      "step": 6580
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3472105964107441,
      "learning_rate": 6.372540868608401e-07,
      "loss": 0.2175,
      "step": 6581
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.871124249846487,
      "learning_rate": 6.362416926145775e-07,
      "loss": 0.813,
      "step": 6582
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.407878439381304,
      "learning_rate": 6.352300485489571e-07,
      "loss": 0.5684,
      "step": 6583
    },
    {
      "epoch": 0.84,
      "grad_norm": 7.087513617564081,
      "learning_rate": 6.342191548378923e-07,
      "loss": 0.7069,
      "step": 6584
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.306033564160291,
      "learning_rate": 6.33209011655167e-07,
      "loss": 0.1557,
      "step": 6585
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3604031514407808,
      "learning_rate": 6.321996191744368e-07,
      "loss": 0.1593,
      "step": 6586
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3964783389243185,
      "learning_rate": 6.311909775692265e-07,
      "loss": 0.1569,
      "step": 6587
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6030875292256532,
      "learning_rate": 6.301830870129349e-07,
      "loss": 0.2066,
      "step": 6588
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3260592828102171,
      "learning_rate": 6.291759476788312e-07,
      "loss": 0.1527,
      "step": 6589
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.40243177507469,
      "learning_rate": 6.281695597400533e-07,
      "loss": 0.5882,
      "step": 6590
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.212117563604929,
      "learning_rate": 6.271639233696103e-07,
      "loss": 0.5545,
      "step": 6591
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.547964307251132,
      "learning_rate": 6.261590387403832e-07,
      "loss": 0.6013,
      "step": 6592
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4024535405042107,
      "learning_rate": 6.251549060251233e-07,
      "loss": 0.1865,
      "step": 6593
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.221529350800002,
      "learning_rate": 6.241515253964515e-07,
      "loss": 0.5612,
      "step": 6594
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.454359481317415,
      "learning_rate": 6.231488970268628e-07,
      "loss": 0.1918,
      "step": 6595
    },
    {
      "epoch": 0.84,
      "grad_norm": 8.575478134760164,
      "learning_rate": 6.221470210887182e-07,
      "loss": 0.7787,
      "step": 6596
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4755927502598352,
      "learning_rate": 6.211458977542545e-07,
      "loss": 0.2205,
      "step": 6597
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.02769100354036,
      "learning_rate": 6.201455271955747e-07,
      "loss": 0.4555,
      "step": 6598
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.516402273880873,
      "learning_rate": 6.191459095846547e-07,
      "loss": 0.1877,
      "step": 6599
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.596553443440491,
      "learning_rate": 6.181470450933397e-07,
      "loss": 0.5212,
      "step": 6600
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5150085793297725,
      "learning_rate": 6.171489338933467e-07,
      "loss": 0.1892,
      "step": 6601
    },
    {
      "epoch": 0.85,
      "grad_norm": 9.966064901934143,
      "learning_rate": 6.161515761562614e-07,
      "loss": 0.4783,
      "step": 6602
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.379465392578485,
      "learning_rate": 6.151549720535433e-07,
      "loss": 0.1747,
      "step": 6603
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4065064396420726,
      "learning_rate": 6.141591217565185e-07,
      "loss": 0.1678,
      "step": 6604
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4654208287426707,
      "learning_rate": 6.131640254363847e-07,
      "loss": 0.1864,
      "step": 6605
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.9989544520306035,
      "learning_rate": 6.121696832642126e-07,
      "loss": 0.7112,
      "step": 6606
    },
    {
      "epoch": 0.85,
      "grad_norm": 8.058403578106768,
      "learning_rate": 6.111760954109402e-07,
      "loss": 0.5511,
      "step": 6607
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.38106886525874,
      "learning_rate": 6.101832620473763e-07,
      "loss": 0.1619,
      "step": 6608
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.89649331323385,
      "learning_rate": 6.091911833441999e-07,
      "loss": 0.5848,
      "step": 6609
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4540139155829883,
      "learning_rate": 6.081998594719629e-07,
      "loss": 0.197,
      "step": 6610
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.434457923780956,
      "learning_rate": 6.07209290601084e-07,
      "loss": 0.169,
      "step": 6611
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4744456450663408,
      "learning_rate": 6.062194769018542e-07,
      "loss": 0.1933,
      "step": 6612
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.856441305099984,
      "learning_rate": 6.05230418544433e-07,
      "loss": 0.7659,
      "step": 6613
    },
    {
      "epoch": 0.85,
      "grad_norm": 9.957887499082013,
      "learning_rate": 6.042421156988498e-07,
      "loss": 0.5508,
      "step": 6614
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.0376379252446375,
      "learning_rate": 6.032545685350088e-07,
      "loss": 0.5319,
      "step": 6615
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.234488914237844,
      "learning_rate": 6.022677772226781e-07,
      "loss": 0.5253,
      "step": 6616
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4915964728795663,
      "learning_rate": 6.012817419314992e-07,
      "loss": 0.1863,
      "step": 6617
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3510307610170316,
      "learning_rate": 6.002964628309838e-07,
      "loss": 0.155,
      "step": 6618
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.323053305397317,
      "learning_rate": 5.993119400905123e-07,
      "loss": 0.1742,
      "step": 6619
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3794278227240433,
      "learning_rate": 5.983281738793351e-07,
      "loss": 0.1766,
      "step": 6620
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5675555462053952,
      "learning_rate": 5.973451643665734e-07,
      "loss": 0.203,
      "step": 6621
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4281245334236425,
      "learning_rate": 5.963629117212183e-07,
      "loss": 0.1899,
      "step": 6622
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.2791198924754905,
      "learning_rate": 5.95381416112129e-07,
      "loss": 0.5149,
      "step": 6623
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.396528249406539,
      "learning_rate": 5.944006777080363e-07,
      "loss": 0.1872,
      "step": 6624
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5188945288568507,
      "learning_rate": 5.934206966775429e-07,
      "loss": 0.2006,
      "step": 6625
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3964138617594866,
      "learning_rate": 5.924414731891171e-07,
      "loss": 0.1797,
      "step": 6626
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3613397195774621,
      "learning_rate": 5.914630074110989e-07,
      "loss": 0.1609,
      "step": 6627
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4553447244113287,
      "learning_rate": 5.904852995116977e-07,
      "loss": 0.1492,
      "step": 6628
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4304177252580137,
      "learning_rate": 5.895083496589932e-07,
      "loss": 0.174,
      "step": 6629
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4097491511116353,
      "learning_rate": 5.885321580209324e-07,
      "loss": 0.1735,
      "step": 6630
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3763532780626582,
      "learning_rate": 5.875567247653374e-07,
      "loss": 0.1185,
      "step": 6631
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.231625387324125,
      "learning_rate": 5.865820500598951e-07,
      "loss": 0.6494,
      "step": 6632
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3619186912023613,
      "learning_rate": 5.85608134072162e-07,
      "loss": 0.163,
      "step": 6633
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.958829265076497,
      "learning_rate": 5.846349769695675e-07,
      "loss": 0.6459,
      "step": 6634
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.406024580909653,
      "learning_rate": 5.836625789194078e-07,
      "loss": 0.2007,
      "step": 6635
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3910921661433278,
      "learning_rate": 5.826909400888492e-07,
      "loss": 0.1511,
      "step": 6636
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4048670599476936,
      "learning_rate": 5.81720060644928e-07,
      "loss": 0.2086,
      "step": 6637
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.734565467464551,
      "learning_rate": 5.807499407545486e-07,
      "loss": 0.6157,
      "step": 6638
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.364902190844915,
      "learning_rate": 5.797805805844875e-07,
      "loss": 0.1747,
      "step": 6639
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7081212164799553,
      "learning_rate": 5.788119803013881e-07,
      "loss": 0.1983,
      "step": 6640
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4146683061656644,
      "learning_rate": 5.778441400717644e-07,
      "loss": 0.1701,
      "step": 6641
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5371368862144732,
      "learning_rate": 5.768770600619978e-07,
      "loss": 0.1876,
      "step": 6642
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2803480895105268,
      "learning_rate": 5.759107404383429e-07,
      "loss": 0.1389,
      "step": 6643
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.541381642361484,
      "learning_rate": 5.749451813669205e-07,
      "loss": 0.4814,
      "step": 6644
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.478347204475633,
      "learning_rate": 5.739803830137192e-07,
      "loss": 0.1941,
      "step": 6645
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4504525834186148,
      "learning_rate": 5.730163455446025e-07,
      "loss": 0.1876,
      "step": 6646
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.207451810446637,
      "learning_rate": 5.720530691252979e-07,
      "loss": 0.1449,
      "step": 6647
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3913527901509568,
      "learning_rate": 5.71090553921404e-07,
      "loss": 0.1221,
      "step": 6648
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.215727918457952,
      "learning_rate": 5.701288000983884e-07,
      "loss": 0.5588,
      "step": 6649
    },
    {
      "epoch": 0.85,
      "grad_norm": 11.775911817583989,
      "learning_rate": 5.691678078215873e-07,
      "loss": 0.4009,
      "step": 6650
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1360317753616762,
      "learning_rate": 5.682075772562051e-07,
      "loss": 0.1262,
      "step": 6651
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.074594786130584,
      "learning_rate": 5.672481085673199e-07,
      "loss": 0.5203,
      "step": 6652
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.574739734138673,
      "learning_rate": 5.662894019198722e-07,
      "loss": 0.2211,
      "step": 6653
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4457303857688006,
      "learning_rate": 5.653314574786778e-07,
      "loss": 0.1634,
      "step": 6654
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3902949864375893,
      "learning_rate": 5.643742754084164e-07,
      "loss": 0.1857,
      "step": 6655
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3001976799230668,
      "learning_rate": 5.634178558736397e-07,
      "loss": 0.1425,
      "step": 6656
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6082254667422393,
      "learning_rate": 5.624621990387669e-07,
      "loss": 0.1897,
      "step": 6657
    },
    {
      "epoch": 0.85,
      "grad_norm": 9.738229650068511,
      "learning_rate": 5.615073050680859e-07,
      "loss": 0.753,
      "step": 6658
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4259050218261446,
      "learning_rate": 5.605531741257536e-07,
      "loss": 0.1954,
      "step": 6659
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4778741913380038,
      "learning_rate": 5.595998063757985e-07,
      "loss": 0.2107,
      "step": 6660
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2528146419100943,
      "learning_rate": 5.586472019821132e-07,
      "loss": 0.164,
      "step": 6661
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3800690696626299,
      "learning_rate": 5.576953611084635e-07,
      "loss": 0.1774,
      "step": 6662
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.501261802367207,
      "learning_rate": 5.567442839184805e-07,
      "loss": 0.5474,
      "step": 6663
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4718311903163939,
      "learning_rate": 5.557939705756665e-07,
      "loss": 0.1937,
      "step": 6664
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7499133676438203,
      "learning_rate": 5.548444212433901e-07,
      "loss": 0.2198,
      "step": 6665
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.128796502199356,
      "learning_rate": 5.538956360848891e-07,
      "loss": 0.2457,
      "step": 6666
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4681824549469011,
      "learning_rate": 5.529476152632735e-07,
      "loss": 0.1997,
      "step": 6667
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4314806898539818,
      "learning_rate": 5.520003589415168e-07,
      "loss": 0.1619,
      "step": 6668
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4173035781949537,
      "learning_rate": 5.510538672824645e-07,
      "loss": 0.1563,
      "step": 6669
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.361328633912398,
      "learning_rate": 5.501081404488279e-07,
      "loss": 0.1589,
      "step": 6670
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4219159316864358,
      "learning_rate": 5.491631786031904e-07,
      "loss": 0.1644,
      "step": 6671
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2907241151958615,
      "learning_rate": 5.482189819080014e-07,
      "loss": 0.1362,
      "step": 6672
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.397911014806611,
      "learning_rate": 5.472755505255783e-07,
      "loss": 0.6032,
      "step": 6673
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4782003911063921,
      "learning_rate": 5.463328846181081e-07,
      "loss": 0.1718,
      "step": 6674
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.592588703366502,
      "learning_rate": 5.453909843476468e-07,
      "loss": 0.5739,
      "step": 6675
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3627346935709497,
      "learning_rate": 5.444498498761181e-07,
      "loss": 0.17,
      "step": 6676
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.742957523570659,
      "learning_rate": 5.43509481365313e-07,
      "loss": 0.7834,
      "step": 6677
    },
    {
      "epoch": 0.85,
      "grad_norm": 16.45424559494015,
      "learning_rate": 5.425698789768924e-07,
      "loss": 0.5535,
      "step": 6678
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2759924530158537,
      "learning_rate": 5.416310428723837e-07,
      "loss": 0.1573,
      "step": 6679
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4455800107091175,
      "learning_rate": 5.40692973213185e-07,
      "loss": 0.2273,
      "step": 6680
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.015717481245905,
      "learning_rate": 5.397556701605605e-07,
      "loss": 0.4652,
      "step": 6681
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3751597217088583,
      "learning_rate": 5.388191338756443e-07,
      "loss": 0.1827,
      "step": 6682
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3532613554060389,
      "learning_rate": 5.378833645194375e-07,
      "loss": 0.1762,
      "step": 6683
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4113212096855996,
      "learning_rate": 5.369483622528104e-07,
      "loss": 0.1561,
      "step": 6684
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.586279817775568,
      "learning_rate": 5.360141272364994e-07,
      "loss": 0.1837,
      "step": 6685
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4043318635972935,
      "learning_rate": 5.350806596311109e-07,
      "loss": 0.1839,
      "step": 6686
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5568864170647267,
      "learning_rate": 5.341479595971177e-07,
      "loss": 0.226,
      "step": 6687
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2395118283567228,
      "learning_rate": 5.332160272948628e-07,
      "loss": 0.1598,
      "step": 6688
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2914774216295473,
      "learning_rate": 5.322848628845578e-07,
      "loss": 0.1501,
      "step": 6689
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4665621550024817,
      "learning_rate": 5.313544665262782e-07,
      "loss": 0.2,
      "step": 6690
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6640267898239887,
      "learning_rate": 5.304248383799715e-07,
      "loss": 0.1606,
      "step": 6691
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4218691568735184,
      "learning_rate": 5.294959786054505e-07,
      "loss": 0.1835,
      "step": 6692
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.539474411728536,
      "learning_rate": 5.285678873623973e-07,
      "loss": 0.5224,
      "step": 6693
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1801622149202757,
      "learning_rate": 5.276405648103616e-07,
      "loss": 0.111,
      "step": 6694
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4583311995159942,
      "learning_rate": 5.267140111087604e-07,
      "loss": 0.1711,
      "step": 6695
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5028687060402328,
      "learning_rate": 5.257882264168795e-07,
      "loss": 0.1928,
      "step": 6696
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3770405003778206,
      "learning_rate": 5.248632108938728e-07,
      "loss": 0.2121,
      "step": 6697
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5174462982582504,
      "learning_rate": 5.239389646987592e-07,
      "loss": 0.1793,
      "step": 6698
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4757121670631745,
      "learning_rate": 5.230154879904292e-07,
      "loss": 0.146,
      "step": 6699
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.365399663082726,
      "learning_rate": 5.220927809276383e-07,
      "loss": 0.149,
      "step": 6700
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.569526777105663,
      "learning_rate": 5.211708436690111e-07,
      "loss": 0.1739,
      "step": 6701
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4434241862736958,
      "learning_rate": 5.20249676373038e-07,
      "loss": 0.1815,
      "step": 6702
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.403049336188495,
      "learning_rate": 5.193292791980803e-07,
      "loss": 0.1681,
      "step": 6703
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2740341363313878,
      "learning_rate": 5.184096523023635e-07,
      "loss": 0.1714,
      "step": 6704
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.416103493416008,
      "learning_rate": 5.174907958439829e-07,
      "loss": 0.1737,
      "step": 6705
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3472572062098291,
      "learning_rate": 5.165727099808998e-07,
      "loss": 0.1344,
      "step": 6706
    },
    {
      "epoch": 0.86,
      "grad_norm": 8.394862741055158,
      "learning_rate": 5.156553948709436e-07,
      "loss": 0.7114,
      "step": 6707
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3285296520459433,
      "learning_rate": 5.147388506718127e-07,
      "loss": 0.1681,
      "step": 6708
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.392122006623985,
      "learning_rate": 5.138230775410713e-07,
      "loss": 0.1594,
      "step": 6709
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2738012658425035,
      "learning_rate": 5.129080756361499e-07,
      "loss": 0.1652,
      "step": 6710
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3639523606512123,
      "learning_rate": 5.119938451143502e-07,
      "loss": 0.1919,
      "step": 6711
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.414686112297834,
      "learning_rate": 5.110803861328378e-07,
      "loss": 0.1965,
      "step": 6712
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1892011475066466,
      "learning_rate": 5.10167698848647e-07,
      "loss": 0.1558,
      "step": 6713
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3345280584435633,
      "learning_rate": 5.092557834186795e-07,
      "loss": 0.1369,
      "step": 6714
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6520561397105586,
      "learning_rate": 5.083446399997044e-07,
      "loss": 0.1918,
      "step": 6715
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4976472799892966,
      "learning_rate": 5.074342687483557e-07,
      "loss": 0.1504,
      "step": 6716
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4421982789695005,
      "learning_rate": 5.06524669821139e-07,
      "loss": 0.1743,
      "step": 6717
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3633747237551865,
      "learning_rate": 5.056158433744251e-07,
      "loss": 0.1664,
      "step": 6718
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3177687755204797,
      "learning_rate": 5.047077895644514e-07,
      "loss": 0.1435,
      "step": 6719
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1905820579519322,
      "learning_rate": 5.038005085473224e-07,
      "loss": 0.1173,
      "step": 6720
    },
    {
      "epoch": 0.86,
      "grad_norm": 13.533478813377132,
      "learning_rate": 5.028940004790106e-07,
      "loss": 0.5466,
      "step": 6721
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3689992573337841,
      "learning_rate": 5.019882655153547e-07,
      "loss": 0.1618,
      "step": 6722
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.379890098581834,
      "learning_rate": 5.010833038120605e-07,
      "loss": 0.1816,
      "step": 6723
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3621454144597782,
      "learning_rate": 5.001791155247032e-07,
      "loss": 0.1713,
      "step": 6724
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4397837864376812,
      "learning_rate": 4.992757008087218e-07,
      "loss": 0.2004,
      "step": 6725
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2902893442441585,
      "learning_rate": 4.983730598194247e-07,
      "loss": 0.1377,
      "step": 6726
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5465315757883444,
      "learning_rate": 4.974711927119863e-07,
      "loss": 0.1696,
      "step": 6727
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2535106374095837,
      "learning_rate": 4.96570099641448e-07,
      "loss": 0.1721,
      "step": 6728
    },
    {
      "epoch": 0.86,
      "grad_norm": 9.848794899527826,
      "learning_rate": 4.95669780762717e-07,
      "loss": 0.5468,
      "step": 6729
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.479621307348149,
      "learning_rate": 4.9477023623057e-07,
      "loss": 0.1825,
      "step": 6730
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.942649545370746,
      "learning_rate": 4.938714661996469e-07,
      "loss": 0.5828,
      "step": 6731
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4284724286438968,
      "learning_rate": 4.929734708244599e-07,
      "loss": 0.1702,
      "step": 6732
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5680718073134619,
      "learning_rate": 4.920762502593829e-07,
      "loss": 0.1838,
      "step": 6733
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5416376917954837,
      "learning_rate": 4.911798046586591e-07,
      "loss": 0.2383,
      "step": 6734
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4371811408128818,
      "learning_rate": 4.902841341763964e-07,
      "loss": 0.1546,
      "step": 6735
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.455249561254473,
      "learning_rate": 4.893892389665728e-07,
      "loss": 0.1865,
      "step": 6736
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2967914819219457,
      "learning_rate": 4.884951191830312e-07,
      "loss": 0.1847,
      "step": 6737
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2467287050581652,
      "learning_rate": 4.876017749794787e-07,
      "loss": 0.1441,
      "step": 6738
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.477287911243924,
      "learning_rate": 4.867092065094947e-07,
      "loss": 0.4988,
      "step": 6739
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4634482675073845,
      "learning_rate": 4.858174139265209e-07,
      "loss": 0.1756,
      "step": 6740
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4543106647371438,
      "learning_rate": 4.849263973838664e-07,
      "loss": 0.1724,
      "step": 6741
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3701021188080034,
      "learning_rate": 4.840361570347069e-07,
      "loss": 0.1634,
      "step": 6742
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.959163073108736,
      "learning_rate": 4.831466930320861e-07,
      "loss": 0.4138,
      "step": 6743
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.534967480089364,
      "learning_rate": 4.822580055289117e-07,
      "loss": 0.2082,
      "step": 6744
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4872118738894047,
      "learning_rate": 4.813700946779598e-07,
      "loss": 0.1627,
      "step": 6745
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.558076932675987,
      "learning_rate": 4.804829606318745e-07,
      "loss": 0.5928,
      "step": 6746
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.519910915031489,
      "learning_rate": 4.795966035431626e-07,
      "loss": 0.5965,
      "step": 6747
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.067084470264168,
      "learning_rate": 4.787110235642001e-07,
      "loss": 0.6944,
      "step": 6748
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2019549736337192,
      "learning_rate": 4.778262208472273e-07,
      "loss": 0.1566,
      "step": 6749
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3455840693954473,
      "learning_rate": 4.769421955443531e-07,
      "loss": 0.1503,
      "step": 6750
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.45846785666949,
      "learning_rate": 4.7605894780755113e-07,
      "loss": 0.1851,
      "step": 6751
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.447186927547401,
      "learning_rate": 4.751764777886614e-07,
      "loss": 0.1772,
      "step": 6752
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.538217022216494,
      "learning_rate": 4.742947856393909e-07,
      "loss": 0.1831,
      "step": 6753
    },
    {
      "epoch": 0.86,
      "grad_norm": 8.057202754727667,
      "learning_rate": 4.734138715113146e-07,
      "loss": 0.5859,
      "step": 6754
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5691484523349155,
      "learning_rate": 4.7253373555587014e-07,
      "loss": 0.204,
      "step": 6755
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3929578914140441,
      "learning_rate": 4.7165437792436373e-07,
      "loss": 0.1526,
      "step": 6756
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6045639255323392,
      "learning_rate": 4.7077579876796675e-07,
      "loss": 0.1686,
      "step": 6757
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1370559286975135,
      "learning_rate": 4.698979982377172e-07,
      "loss": 0.1225,
      "step": 6758
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.363851093188849,
      "learning_rate": 4.690209764845183e-07,
      "loss": 0.1348,
      "step": 6759
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.755801255225337,
      "learning_rate": 4.6814473365914217e-07,
      "loss": 0.4957,
      "step": 6760
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.386279531692675,
      "learning_rate": 4.6726926991222386e-07,
      "loss": 0.183,
      "step": 6761
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2832423569428293,
      "learning_rate": 4.663945853942653e-07,
      "loss": 0.176,
      "step": 6762
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5038225017921476,
      "learning_rate": 4.655206802556361e-07,
      "loss": 0.2241,
      "step": 6763
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6882388284875476,
      "learning_rate": 4.646475546465706e-07,
      "loss": 0.2262,
      "step": 6764
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5373540224452957,
      "learning_rate": 4.6377520871716874e-07,
      "loss": 0.1862,
      "step": 6765
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3438870268583363,
      "learning_rate": 4.629036426173955e-07,
      "loss": 0.21,
      "step": 6766
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5980647166327548,
      "learning_rate": 4.6203285649708605e-07,
      "loss": 0.1982,
      "step": 6767
    },
    {
      "epoch": 0.87,
      "grad_norm": 9.9654697418933,
      "learning_rate": 4.6116285050593733e-07,
      "loss": 0.4089,
      "step": 6768
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4797742387570558,
      "learning_rate": 4.6029362479351303e-07,
      "loss": 0.2135,
      "step": 6769
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.475315520232875,
      "learning_rate": 4.5942517950924305e-07,
      "loss": 0.1814,
      "step": 6770
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2929740602078805,
      "learning_rate": 4.5855751480242404e-07,
      "loss": 0.1421,
      "step": 6771
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6104999625086498,
      "learning_rate": 4.5769063082221623e-07,
      "loss": 0.1644,
      "step": 6772
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.527656473443429,
      "learning_rate": 4.568245277176486e-07,
      "loss": 0.1477,
      "step": 6773
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3632011436659088,
      "learning_rate": 4.559592056376133e-07,
      "loss": 0.1941,
      "step": 6774
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5293720181241823,
      "learning_rate": 4.5509466473087017e-07,
      "loss": 0.1681,
      "step": 6775
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3247046299532215,
      "learning_rate": 4.5423090514604305e-07,
      "loss": 0.1871,
      "step": 6776
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4371463177367,
      "learning_rate": 4.533679270316227e-07,
      "loss": 0.1946,
      "step": 6777
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.349743904204749,
      "learning_rate": 4.525057305359648e-07,
      "loss": 0.5798,
      "step": 6778
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.615245122741383,
      "learning_rate": 4.5164431580729087e-07,
      "loss": 0.1978,
      "step": 6779
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.410397400625344,
      "learning_rate": 4.507836829936868e-07,
      "loss": 0.2048,
      "step": 6780
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2506961217293173,
      "learning_rate": 4.499238322431071e-07,
      "loss": 0.1873,
      "step": 6781
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2140660110038244,
      "learning_rate": 4.4906476370337017e-07,
      "loss": 0.1537,
      "step": 6782
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4065852050980492,
      "learning_rate": 4.482064775221595e-07,
      "loss": 0.1928,
      "step": 6783
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.440584373749016,
      "learning_rate": 4.4734897384702434e-07,
      "loss": 0.2136,
      "step": 6784
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3361265275969,
      "learning_rate": 4.4649225282537954e-07,
      "loss": 0.1872,
      "step": 6785
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3797995872372733,
      "learning_rate": 4.4563631460450506e-07,
      "loss": 0.1514,
      "step": 6786
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2937711135769399,
      "learning_rate": 4.4478115933154597e-07,
      "loss": 0.1482,
      "step": 6787
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2245075785544446,
      "learning_rate": 4.439267871535152e-07,
      "loss": 0.1477,
      "step": 6788
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.485609208999942,
      "learning_rate": 4.430731982172887e-07,
      "loss": 0.1738,
      "step": 6789
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.860002235830165,
      "learning_rate": 4.4222039266960616e-07,
      "loss": 0.5597,
      "step": 6790
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.493571545007999,
      "learning_rate": 4.413683706570776e-07,
      "loss": 0.2101,
      "step": 6791
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3135642693975764,
      "learning_rate": 4.4051713232617423e-07,
      "loss": 0.1763,
      "step": 6792
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.222610872521175,
      "learning_rate": 4.396666778232345e-07,
      "loss": 0.1493,
      "step": 6793
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5087645665933385,
      "learning_rate": 4.388170072944603e-07,
      "loss": 0.1645,
      "step": 6794
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.595494326599528,
      "learning_rate": 4.3796812088591934e-07,
      "loss": 0.2307,
      "step": 6795
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3667068578257804,
      "learning_rate": 4.37120018743547e-07,
      "loss": 0.1754,
      "step": 6796
    },
    {
      "epoch": 0.87,
      "grad_norm": 11.67722974751224,
      "learning_rate": 4.362727010131407e-07,
      "loss": 0.6404,
      "step": 6797
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2347301202032959,
      "learning_rate": 4.354261678403643e-07,
      "loss": 0.1474,
      "step": 6798
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.344553132901298,
      "learning_rate": 4.345804193707459e-07,
      "loss": 0.1804,
      "step": 6799
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.462475205231292,
      "learning_rate": 4.3373545574968143e-07,
      "loss": 0.1954,
      "step": 6800
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2734729952067283,
      "learning_rate": 4.3289127712242864e-07,
      "loss": 0.1475,
      "step": 6801
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.2258483756222605,
      "learning_rate": 4.320478836341113e-07,
      "loss": 0.5196,
      "step": 6802
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5804911405161808,
      "learning_rate": 4.312052754297197e-07,
      "loss": 0.2329,
      "step": 6803
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3413573489615358,
      "learning_rate": 4.303634526541067e-07,
      "loss": 0.1748,
      "step": 6804
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.895902001101423,
      "learning_rate": 4.2952241545199226e-07,
      "loss": 0.4724,
      "step": 6805
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3083775875325745,
      "learning_rate": 4.2868216396796057e-07,
      "loss": 0.1721,
      "step": 6806
    },
    {
      "epoch": 0.87,
      "grad_norm": 8.55722589906439,
      "learning_rate": 4.2784269834645955e-07,
      "loss": 0.4782,
      "step": 6807
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3951769049140512,
      "learning_rate": 4.270040187318031e-07,
      "loss": 0.1532,
      "step": 6808
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.528148143133713,
      "learning_rate": 4.2616612526817146e-07,
      "loss": 0.4793,
      "step": 6809
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5274292073935007,
      "learning_rate": 4.253290180996067e-07,
      "loss": 0.1652,
      "step": 6810
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6358226847119894,
      "learning_rate": 4.244926973700181e-07,
      "loss": 0.1846,
      "step": 6811
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.192538094538986,
      "learning_rate": 4.2365716322317894e-07,
      "loss": 0.1662,
      "step": 6812
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.366465815443758,
      "learning_rate": 4.228224158027272e-07,
      "loss": 0.1843,
      "step": 6813
    },
    {
      "epoch": 0.87,
      "grad_norm": 31.450186175458033,
      "learning_rate": 4.2198845525216524e-07,
      "loss": 0.4787,
      "step": 6814
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.358239129405376,
      "learning_rate": 4.2115528171486063e-07,
      "loss": 0.1629,
      "step": 6815
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.2937357180844655,
      "learning_rate": 4.2032289533404434e-07,
      "loss": 0.4876,
      "step": 6816
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4589079775419929,
      "learning_rate": 4.194912962528158e-07,
      "loss": 0.1971,
      "step": 6817
    },
    {
      "epoch": 0.87,
      "grad_norm": 9.78710741234337,
      "learning_rate": 4.1866048461413454e-07,
      "loss": 0.4534,
      "step": 6818
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4678891294040946,
      "learning_rate": 4.17830460560828e-07,
      "loss": 0.2091,
      "step": 6819
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4624995742903402,
      "learning_rate": 4.170012242355864e-07,
      "loss": 0.1921,
      "step": 6820
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5760535629793522,
      "learning_rate": 4.161727757809653e-07,
      "loss": 0.188,
      "step": 6821
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.39942458417362,
      "learning_rate": 4.153451153393839e-07,
      "loss": 0.1649,
      "step": 6822
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.337946266971665,
      "learning_rate": 4.145182430531264e-07,
      "loss": 0.7821,
      "step": 6823
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1834887988536742,
      "learning_rate": 4.1369215906434337e-07,
      "loss": 0.1286,
      "step": 6824
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4919289712177928,
      "learning_rate": 4.1286686351504735e-07,
      "loss": 0.1973,
      "step": 6825
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.492523838080868,
      "learning_rate": 4.120423565471154e-07,
      "loss": 0.1972,
      "step": 6826
    },
    {
      "epoch": 0.87,
      "grad_norm": 11.490662958474879,
      "learning_rate": 4.1121863830229037e-07,
      "loss": 0.4355,
      "step": 6827
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3901437233126834,
      "learning_rate": 4.1039570892217993e-07,
      "loss": 0.1486,
      "step": 6828
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4549052195059176,
      "learning_rate": 4.0957356854825436e-07,
      "loss": 0.2002,
      "step": 6829
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.379498410596255,
      "learning_rate": 4.087522173218489e-07,
      "loss": 0.467,
      "step": 6830
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4322592789035833,
      "learning_rate": 4.079316553841628e-07,
      "loss": 0.2034,
      "step": 6831
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4723930269307637,
      "learning_rate": 4.0711188287626156e-07,
      "loss": 0.1995,
      "step": 6832
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.464209404983984,
      "learning_rate": 4.062928999390736e-07,
      "loss": 0.1987,
      "step": 6833
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2522826088664407,
      "learning_rate": 4.0547470671339075e-07,
      "loss": 0.1922,
      "step": 6834
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.884756429772454,
      "learning_rate": 4.0465730333986983e-07,
      "loss": 0.6044,
      "step": 6835
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3626181529959616,
      "learning_rate": 4.0384068995903136e-07,
      "loss": 0.1578,
      "step": 6836
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.814608938118087,
      "learning_rate": 4.03024866711263e-07,
      "loss": 0.5352,
      "step": 6837
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3802328251997167,
      "learning_rate": 4.0220983373681133e-07,
      "loss": 0.1784,
      "step": 6838
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.99772734884005,
      "learning_rate": 4.013955911757922e-07,
      "loss": 0.1869,
      "step": 6839
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4156549562459968,
      "learning_rate": 4.00582139168183e-07,
      "loss": 0.1455,
      "step": 6840
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5907029587019845,
      "learning_rate": 3.997694778538247e-07,
      "loss": 0.1479,
      "step": 6841
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.478105908879924,
      "learning_rate": 3.9895760737242384e-07,
      "loss": 0.567,
      "step": 6842
    },
    {
      "epoch": 0.88,
      "grad_norm": 12.10515887372666,
      "learning_rate": 3.9814652786355046e-07,
      "loss": 0.5305,
      "step": 6843
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3968072512884722,
      "learning_rate": 3.973362394666369e-07,
      "loss": 0.2032,
      "step": 6844
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.541708047845802,
      "learning_rate": 3.965267423209834e-07,
      "loss": 0.5998,
      "step": 6845
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3805575219108286,
      "learning_rate": 3.957180365657498e-07,
      "loss": 0.1707,
      "step": 6846
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4296677646629072,
      "learning_rate": 3.949101223399643e-07,
      "loss": 0.1647,
      "step": 6847
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4134058686874624,
      "learning_rate": 3.9410299978251523e-07,
      "loss": 0.1991,
      "step": 6848
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5293794810438337,
      "learning_rate": 3.932966690321571e-07,
      "loss": 0.2332,
      "step": 6849
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2596490211371023,
      "learning_rate": 3.92491130227507e-07,
      "loss": 0.1753,
      "step": 6850
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3369186288264272,
      "learning_rate": 3.9168638350704614e-07,
      "loss": 0.2192,
      "step": 6851
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3721016756960613,
      "learning_rate": 3.9088242900911957e-07,
      "loss": 0.1806,
      "step": 6852
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2851843274631376,
      "learning_rate": 3.900792668719372e-07,
      "loss": 0.1869,
      "step": 6853
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4436323377557725,
      "learning_rate": 3.8927689723357253e-07,
      "loss": 0.1906,
      "step": 6854
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5265203518413188,
      "learning_rate": 3.8847532023195967e-07,
      "loss": 0.2205,
      "step": 6855
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4326807106113306,
      "learning_rate": 3.876745360049017e-07,
      "loss": 0.1621,
      "step": 6856
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4776328067224753,
      "learning_rate": 3.868745446900618e-07,
      "loss": 0.2016,
      "step": 6857
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3488583865921449,
      "learning_rate": 3.860753464249678e-07,
      "loss": 0.1674,
      "step": 6858
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3865371825888602,
      "learning_rate": 3.8527694134700975e-07,
      "loss": 0.1631,
      "step": 6859
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3481373401858607,
      "learning_rate": 3.844793295934451e-07,
      "loss": 0.183,
      "step": 6860
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1360976380827994,
      "learning_rate": 3.836825113013909e-07,
      "loss": 0.1506,
      "step": 6861
    },
    {
      "epoch": 0.88,
      "grad_norm": 15.713271150188632,
      "learning_rate": 3.8288648660783023e-07,
      "loss": 0.5716,
      "step": 6862
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3793346177669876,
      "learning_rate": 3.820912556496087e-07,
      "loss": 0.1972,
      "step": 6863
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4306646549885034,
      "learning_rate": 3.8129681856343526e-07,
      "loss": 0.1799,
      "step": 6864
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4404900465111006,
      "learning_rate": 3.8050317548588343e-07,
      "loss": 0.1632,
      "step": 6865
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3510943536404227,
      "learning_rate": 3.7971032655339026e-07,
      "loss": 0.172,
      "step": 6866
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4906796122489119,
      "learning_rate": 3.789182719022538e-07,
      "loss": 0.2161,
      "step": 6867
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2501296353231262,
      "learning_rate": 3.781270116686392e-07,
      "loss": 0.1652,
      "step": 6868
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6002819332305993,
      "learning_rate": 3.7733654598857303e-07,
      "loss": 0.194,
      "step": 6869
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2758828247773046,
      "learning_rate": 3.7654687499794453e-07,
      "loss": 0.1641,
      "step": 6870
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.425597733842452,
      "learning_rate": 3.757579988325083e-07,
      "loss": 0.2109,
      "step": 6871
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.804446548129026,
      "learning_rate": 3.7496991762788095e-07,
      "loss": 0.52,
      "step": 6872
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4624152603205725,
      "learning_rate": 3.7418263151954184e-07,
      "loss": 0.1473,
      "step": 6873
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4002628115122362,
      "learning_rate": 3.7339614064283545e-07,
      "loss": 0.1547,
      "step": 6874
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3913951769048136,
      "learning_rate": 3.7261044513296927e-07,
      "loss": 0.1832,
      "step": 6875
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5506530010957824,
      "learning_rate": 3.71825545125013e-07,
      "loss": 0.1773,
      "step": 6876
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.46404467114836,
      "learning_rate": 3.710414407538998e-07,
      "loss": 0.1601,
      "step": 6877
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3972049664505288,
      "learning_rate": 3.7025813215442685e-07,
      "loss": 0.1723,
      "step": 6878
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.236364094586105,
      "learning_rate": 3.694756194612531e-07,
      "loss": 0.159,
      "step": 6879
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.619508208407572,
      "learning_rate": 3.686939028089015e-07,
      "loss": 0.6685,
      "step": 6880
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3716805870532685,
      "learning_rate": 3.6791298233175955e-07,
      "loss": 0.191,
      "step": 6881
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2946150686610183,
      "learning_rate": 3.671328581640754e-07,
      "loss": 0.1822,
      "step": 6882
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3503806326965084,
      "learning_rate": 3.663535304399607e-07,
      "loss": 0.1638,
      "step": 6883
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5178460078463694,
      "learning_rate": 3.6557499929339325e-07,
      "loss": 0.173,
      "step": 6884
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.399503818275666,
      "learning_rate": 3.647972648582104e-07,
      "loss": 0.2213,
      "step": 6885
    },
    {
      "epoch": 0.88,
      "grad_norm": 24.292939077449994,
      "learning_rate": 3.640203272681131e-07,
      "loss": 0.5268,
      "step": 6886
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5878420258599257,
      "learning_rate": 3.6324418665666717e-07,
      "loss": 0.1797,
      "step": 6887
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5301874971931544,
      "learning_rate": 3.624688431572981e-07,
      "loss": 0.2155,
      "step": 6888
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.307403632619412,
      "learning_rate": 3.6169429690329825e-07,
      "loss": 0.1428,
      "step": 6889
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5121692628266095,
      "learning_rate": 3.6092054802782107e-07,
      "loss": 0.2233,
      "step": 6890
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.449546590849423,
      "learning_rate": 3.601475966638829e-07,
      "loss": 0.2137,
      "step": 6891
    },
    {
      "epoch": 0.88,
      "grad_norm": 13.64762998218543,
      "learning_rate": 3.5937544294436134e-07,
      "loss": 0.6196,
      "step": 6892
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3963769809296065,
      "learning_rate": 3.586040870020008e-07,
      "loss": 0.167,
      "step": 6893
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4482463777285501,
      "learning_rate": 3.5783352896940513e-07,
      "loss": 0.1697,
      "step": 6894
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2093514504919602,
      "learning_rate": 3.570637689790418e-07,
      "loss": 0.132,
      "step": 6895
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4766025670563112,
      "learning_rate": 3.562948071632427e-07,
      "loss": 0.1565,
      "step": 6896
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2544236203571895,
      "learning_rate": 3.55526643654201e-07,
      "loss": 0.1437,
      "step": 6897
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2830671341646005,
      "learning_rate": 3.5475927858397263e-07,
      "loss": 0.1754,
      "step": 6898
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4720068473306658,
      "learning_rate": 3.539927120844766e-07,
      "loss": 0.1613,
      "step": 6899
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3502598463937552,
      "learning_rate": 3.532269442874942e-07,
      "loss": 0.1809,
      "step": 6900
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5045671020139844,
      "learning_rate": 3.524619753246694e-07,
      "loss": 0.1685,
      "step": 6901
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.357825670560253,
      "learning_rate": 3.516978053275111e-07,
      "loss": 0.1684,
      "step": 6902
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4882989701207823,
      "learning_rate": 3.5093443442738627e-07,
      "loss": 0.2192,
      "step": 6903
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3976539598195055,
      "learning_rate": 3.5017186275553004e-07,
      "loss": 0.1904,
      "step": 6904
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4513136481982614,
      "learning_rate": 3.494100904430359e-07,
      "loss": 0.1652,
      "step": 6905
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5603565395129348,
      "learning_rate": 3.486491176208617e-07,
      "loss": 0.1937,
      "step": 6906
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2609123774165087,
      "learning_rate": 3.478889444198269e-07,
      "loss": 0.1472,
      "step": 6907
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5014440568120426,
      "learning_rate": 3.471295709706146e-07,
      "loss": 0.1408,
      "step": 6908
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.303745581825776,
      "learning_rate": 3.4637099740376934e-07,
      "loss": 0.5234,
      "step": 6909
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.397695745485007,
      "learning_rate": 3.4561322384969843e-07,
      "loss": 0.1706,
      "step": 6910
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4409653391002075,
      "learning_rate": 3.448562504386738e-07,
      "loss": 0.1907,
      "step": 6911
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.457667676365661,
      "learning_rate": 3.4410007730082685e-07,
      "loss": 0.1843,
      "step": 6912
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.768677204596899,
      "learning_rate": 3.4334470456615255e-07,
      "loss": 0.4976,
      "step": 6913
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3930106947469416,
      "learning_rate": 3.4259013236450755e-07,
      "loss": 0.1498,
      "step": 6914
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2277093905315033,
      "learning_rate": 3.4183636082561257e-07,
      "loss": 0.1627,
      "step": 6915
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.41250016593572,
      "learning_rate": 3.4108339007904834e-07,
      "loss": 0.3607,
      "step": 6916
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4353570615614737,
      "learning_rate": 3.4033122025426077e-07,
      "loss": 0.1583,
      "step": 6917
    },
    {
      "epoch": 0.89,
      "grad_norm": 8.289712088962615,
      "learning_rate": 3.395798514805565e-07,
      "loss": 0.7083,
      "step": 6918
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3363005108908137,
      "learning_rate": 3.3882928388710376e-07,
      "loss": 0.1468,
      "step": 6919
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2177385566472236,
      "learning_rate": 3.380795176029328e-07,
      "loss": 0.1654,
      "step": 6920
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4386328884770943,
      "learning_rate": 3.3733055275693983e-07,
      "loss": 0.1835,
      "step": 6921
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1949559051986722,
      "learning_rate": 3.3658238947787857e-07,
      "loss": 0.1336,
      "step": 6922
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.157742908803322,
      "learning_rate": 3.3583502789436783e-07,
      "loss": 0.5477,
      "step": 6923
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5243630858024524,
      "learning_rate": 3.3508846813488647e-07,
      "loss": 0.1846,
      "step": 6924
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6679671610177327,
      "learning_rate": 3.3434271032777856e-07,
      "loss": 0.1997,
      "step": 6925
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4673247904688973,
      "learning_rate": 3.335977546012481e-07,
      "loss": 0.1887,
      "step": 6926
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.386439738216116,
      "learning_rate": 3.328536010833605e-07,
      "loss": 0.4969,
      "step": 6927
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.170583238989705,
      "learning_rate": 3.3211024990204565e-07,
      "loss": 0.5766,
      "step": 6928
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4413555550020989,
      "learning_rate": 3.3136770118509243e-07,
      "loss": 0.1984,
      "step": 6929
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2354990952469498,
      "learning_rate": 3.3062595506015594e-07,
      "loss": 0.1938,
      "step": 6930
    },
    {
      "epoch": 0.89,
      "grad_norm": 33.9340650599836,
      "learning_rate": 3.2988501165474864e-07,
      "loss": 0.6903,
      "step": 6931
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3172879021659027,
      "learning_rate": 3.291448710962497e-07,
      "loss": 0.1465,
      "step": 6932
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3212366195677676,
      "learning_rate": 3.284055335118963e-07,
      "loss": 0.1412,
      "step": 6933
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.520461413539026,
      "learning_rate": 3.276669990287895e-07,
      "loss": 0.1663,
      "step": 6934
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3402800199243132,
      "learning_rate": 3.269292677738922e-07,
      "loss": 0.1366,
      "step": 6935
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5200461824349318,
      "learning_rate": 3.261923398740285e-07,
      "loss": 0.1701,
      "step": 6936
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.514039854235311,
      "learning_rate": 3.2545621545588434e-07,
      "loss": 0.4697,
      "step": 6937
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.40166455934208,
      "learning_rate": 3.2472089464600844e-07,
      "loss": 0.1327,
      "step": 6938
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4312834920290596,
      "learning_rate": 3.2398637757081187e-07,
      "loss": 0.15,
      "step": 6939
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.390721924335015,
      "learning_rate": 3.232526643565664e-07,
      "loss": 0.1481,
      "step": 6940
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.443019554651668,
      "learning_rate": 3.2251975512940516e-07,
      "loss": 0.1692,
      "step": 6941
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3826216558597693,
      "learning_rate": 3.2178765001532386e-07,
      "loss": 0.208,
      "step": 6942
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4472919328071925,
      "learning_rate": 3.210563491401808e-07,
      "loss": 0.1852,
      "step": 6943
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4268085967163882,
      "learning_rate": 3.203258526296937e-07,
      "loss": 0.1828,
      "step": 6944
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.372408455412643,
      "learning_rate": 3.195961606094433e-07,
      "loss": 0.4586,
      "step": 6945
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3216854606056654,
      "learning_rate": 3.188672732048731e-07,
      "loss": 0.1564,
      "step": 6946
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5512572043603445,
      "learning_rate": 3.181391905412867e-07,
      "loss": 0.196,
      "step": 6947
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2846899044150546,
      "learning_rate": 3.1741191274385076e-07,
      "loss": 0.151,
      "step": 6948
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.329070472272496,
      "learning_rate": 3.1668543993759293e-07,
      "loss": 0.1913,
      "step": 6949
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3702320065386002,
      "learning_rate": 3.159597722474006e-07,
      "loss": 0.1906,
      "step": 6950
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.411347013077508,
      "learning_rate": 3.152349097980262e-07,
      "loss": 0.1818,
      "step": 6951
    },
    {
      "epoch": 0.89,
      "grad_norm": 8.46133421810505,
      "learning_rate": 3.1451085271408053e-07,
      "loss": 0.5169,
      "step": 6952
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2945186854798958,
      "learning_rate": 3.137876011200386e-07,
      "loss": 0.1734,
      "step": 6953
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.392678550640383,
      "learning_rate": 3.130651551402353e-07,
      "loss": 0.159,
      "step": 6954
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.170227302971849,
      "learning_rate": 3.123435148988674e-07,
      "loss": 0.6656,
      "step": 6955
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.362961993000117,
      "learning_rate": 3.116226805199929e-07,
      "loss": 0.1673,
      "step": 6956
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4530558501434778,
      "learning_rate": 3.1090265212753214e-07,
      "loss": 0.2095,
      "step": 6957
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2925933183525595,
      "learning_rate": 3.101834298452661e-07,
      "loss": 0.1817,
      "step": 6958
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5101122416522266,
      "learning_rate": 3.094650137968369e-07,
      "loss": 0.2165,
      "step": 6959
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5007655326151512,
      "learning_rate": 3.0874740410575033e-07,
      "loss": 0.2109,
      "step": 6960
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3222861978278446,
      "learning_rate": 3.080306008953704e-07,
      "loss": 0.1842,
      "step": 6961
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.421561177350518,
      "learning_rate": 3.0731460428892414e-07,
      "loss": 0.1752,
      "step": 6962
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5884635541087229,
      "learning_rate": 3.065994144094997e-07,
      "loss": 0.1942,
      "step": 6963
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.45715406184932,
      "learning_rate": 3.0588503138004597e-07,
      "loss": 0.1899,
      "step": 6964
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3162619941313243,
      "learning_rate": 3.051714553233748e-07,
      "loss": 0.1927,
      "step": 6965
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4547574779342105,
      "learning_rate": 3.044586863621568e-07,
      "loss": 0.1704,
      "step": 6966
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5367318829375658,
      "learning_rate": 3.0374672461892574e-07,
      "loss": 0.193,
      "step": 6967
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3985312873238644,
      "learning_rate": 3.0303557021607754e-07,
      "loss": 0.1845,
      "step": 6968
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.104236876257716,
      "learning_rate": 3.023252232758667e-07,
      "loss": 0.5083,
      "step": 6969
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.647436869203171,
      "learning_rate": 3.0161568392040986e-07,
      "loss": 0.1585,
      "step": 6970
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.547914590317278,
      "learning_rate": 3.009069522716851e-07,
      "loss": 0.1626,
      "step": 6971
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4430008998262356,
      "learning_rate": 3.0019902845153216e-07,
      "loss": 0.1839,
      "step": 6972
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.327766606158593,
      "learning_rate": 2.994919125816498e-07,
      "loss": 0.1557,
      "step": 6973
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5240336374149352,
      "learning_rate": 2.9878560478360186e-07,
      "loss": 0.2017,
      "step": 6974
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4891021860314095,
      "learning_rate": 2.980801051788085e-07,
      "loss": 0.1859,
      "step": 6975
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4816220603072614,
      "learning_rate": 2.973754138885554e-07,
      "loss": 0.2237,
      "step": 6976
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4062577356912709,
      "learning_rate": 2.9667153103398573e-07,
      "loss": 0.2388,
      "step": 6977
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6405482453831997,
      "learning_rate": 2.9596845673610597e-07,
      "loss": 0.1894,
      "step": 6978
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5314599385264176,
      "learning_rate": 2.9526619111578223e-07,
      "loss": 0.2077,
      "step": 6979
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4678909152405515,
      "learning_rate": 2.945647342937413e-07,
      "loss": 0.193,
      "step": 6980
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.592671824019978,
      "learning_rate": 2.9386408639057394e-07,
      "loss": 0.2021,
      "step": 6981
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1962659544615077,
      "learning_rate": 2.9316424752672766e-07,
      "loss": 0.1362,
      "step": 6982
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3933561499714793,
      "learning_rate": 2.9246521782251403e-07,
      "loss": 0.1577,
      "step": 6983
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5640779283295303,
      "learning_rate": 2.91766997398103e-07,
      "loss": 0.1667,
      "step": 6984
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.057996439837103,
      "learning_rate": 2.91069586373528e-07,
      "loss": 0.4744,
      "step": 6985
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4445237462457592,
      "learning_rate": 2.9037298486868205e-07,
      "loss": 0.1771,
      "step": 6986
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2437975189126163,
      "learning_rate": 2.8967719300331875e-07,
      "loss": 0.1408,
      "step": 6987
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.173970071192578,
      "learning_rate": 2.8898221089705194e-07,
      "loss": 0.6001,
      "step": 6988
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.229885236393641,
      "learning_rate": 2.882880386693582e-07,
      "loss": 0.1481,
      "step": 6989
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.41296198084457,
      "learning_rate": 2.8759467643957375e-07,
      "loss": 0.5958,
      "step": 6990
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.82072356339115,
      "learning_rate": 2.8690212432689546e-07,
      "loss": 0.6564,
      "step": 6991
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4454602352146082,
      "learning_rate": 2.862103824503809e-07,
      "loss": 0.1948,
      "step": 6992
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5902367051520567,
      "learning_rate": 2.855194509289483e-07,
      "loss": 0.2171,
      "step": 6993
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.455312476044108,
      "learning_rate": 2.8482932988137647e-07,
      "loss": 0.5543,
      "step": 6994
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.529360178580649,
      "learning_rate": 2.8414001942630556e-07,
      "loss": 0.2002,
      "step": 6995
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4773431450609003,
      "learning_rate": 2.834515196822374e-07,
      "loss": 0.1631,
      "step": 6996
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.245756446153727,
      "learning_rate": 2.8276383076753175e-07,
      "loss": 0.5918,
      "step": 6997
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.466034249485542,
      "learning_rate": 2.8207695280041025e-07,
      "loss": 0.2094,
      "step": 6998
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2950113237943164,
      "learning_rate": 2.813908858989556e-07,
      "loss": 0.1693,
      "step": 6999
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.0238716878425,
      "learning_rate": 2.8070563018111063e-07,
      "loss": 0.5775,
      "step": 7000
    },
    {
      "epoch": 0.9,
      "grad_norm": 9.029931295343705,
      "learning_rate": 2.8002118576467784e-07,
      "loss": 0.5142,
      "step": 7001
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.447234611969665,
      "learning_rate": 2.7933755276732257e-07,
      "loss": 0.5583,
      "step": 7002
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.746835577393832,
      "learning_rate": 2.7865473130656794e-07,
      "loss": 0.2082,
      "step": 7003
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5598004239142276,
      "learning_rate": 2.779727214997996e-07,
      "loss": 0.1879,
      "step": 7004
    },
    {
      "epoch": 0.9,
      "grad_norm": 11.458274581902621,
      "learning_rate": 2.7729152346426366e-07,
      "loss": 0.4137,
      "step": 7005
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4577559692501116,
      "learning_rate": 2.766111373170649e-07,
      "loss": 0.1714,
      "step": 7006
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4653793796019423,
      "learning_rate": 2.7593156317516966e-07,
      "loss": 0.1721,
      "step": 7007
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5078072121060153,
      "learning_rate": 2.752528011554051e-07,
      "loss": 0.1924,
      "step": 7008
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.40508853397322,
      "learning_rate": 2.7457485137445725e-07,
      "loss": 0.1759,
      "step": 7009
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3465479971777565,
      "learning_rate": 2.738977139488752e-07,
      "loss": 0.143,
      "step": 7010
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4312782585351829,
      "learning_rate": 2.732213889950652e-07,
      "loss": 0.2099,
      "step": 7011
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.377156040826906,
      "learning_rate": 2.725458766292954e-07,
      "loss": 0.1919,
      "step": 7012
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4029729725722833,
      "learning_rate": 2.718711769676957e-07,
      "loss": 0.2068,
      "step": 7013
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3672569359390032,
      "learning_rate": 2.711972901262538e-07,
      "loss": 0.1724,
      "step": 7014
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2718551385461139,
      "learning_rate": 2.705242162208188e-07,
      "loss": 0.1983,
      "step": 7015
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4889179755105793,
      "learning_rate": 2.698519553670992e-07,
      "loss": 0.1789,
      "step": 7016
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.535497305803631,
      "learning_rate": 2.6918050768066527e-07,
      "loss": 0.1522,
      "step": 7017
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5806012416742834,
      "learning_rate": 2.68509873276947e-07,
      "loss": 0.1974,
      "step": 7018
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.506283628811392,
      "learning_rate": 2.678400522712332e-07,
      "loss": 0.1922,
      "step": 7019
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6108473422242802,
      "learning_rate": 2.6717104477867464e-07,
      "loss": 0.1904,
      "step": 7020
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.405141758088476,
      "learning_rate": 2.665028509142803e-07,
      "loss": 0.179,
      "step": 7021
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.773426270781811,
      "learning_rate": 2.6583547079292224e-07,
      "loss": 0.4277,
      "step": 7022
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.336821741645328,
      "learning_rate": 2.651689045293293e-07,
      "loss": 0.5627,
      "step": 7023
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4485807042617858,
      "learning_rate": 2.645031522380925e-07,
      "loss": 0.175,
      "step": 7024
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3729576356057207,
      "learning_rate": 2.638382140336626e-07,
      "loss": 0.1921,
      "step": 7025
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.878609432556394,
      "learning_rate": 2.631740900303503e-07,
      "loss": 0.572,
      "step": 7026
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.044123802170206,
      "learning_rate": 2.6251078034232605e-07,
      "loss": 0.6825,
      "step": 7027
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2966284734929399,
      "learning_rate": 2.6184828508362016e-07,
      "loss": 0.1736,
      "step": 7028
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.364470380931779,
      "learning_rate": 2.6118660436812326e-07,
      "loss": 0.1824,
      "step": 7029
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.330875658922846,
      "learning_rate": 2.605257383095855e-07,
      "loss": 0.1744,
      "step": 7030
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5344867473220805,
      "learning_rate": 2.5986568702161817e-07,
      "loss": 0.1916,
      "step": 7031
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6905979757840441,
      "learning_rate": 2.5920645061769225e-07,
      "loss": 0.207,
      "step": 7032
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.544967645149101,
      "learning_rate": 2.585480292111375e-07,
      "loss": 0.5673,
      "step": 7033
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3414776627748477,
      "learning_rate": 2.578904229151441e-07,
      "loss": 0.1815,
      "step": 7034
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5456627873117816,
      "learning_rate": 2.5723363184276207e-07,
      "loss": 0.2013,
      "step": 7035
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3550498212226851,
      "learning_rate": 2.5657765610690223e-07,
      "loss": 0.1649,
      "step": 7036
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.587345977175478,
      "learning_rate": 2.559224958203321e-07,
      "loss": 0.2119,
      "step": 7037
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3990066520649884,
      "learning_rate": 2.552681510956845e-07,
      "loss": 0.2029,
      "step": 7038
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4552527654855894,
      "learning_rate": 2.546146220454471e-07,
      "loss": 0.1912,
      "step": 7039
    },
    {
      "epoch": 0.9,
      "grad_norm": 9.383656010569133,
      "learning_rate": 2.539619087819689e-07,
      "loss": 0.5856,
      "step": 7040
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.426127692384848,
      "learning_rate": 2.533100114174597e-07,
      "loss": 0.1836,
      "step": 7041
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3615499798067434,
      "learning_rate": 2.526589300639881e-07,
      "loss": 0.1544,
      "step": 7042
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.411828132830264,
      "learning_rate": 2.520086648334824e-07,
      "loss": 0.1999,
      "step": 7043
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3270351893749868,
      "learning_rate": 2.5135921583773036e-07,
      "loss": 0.1374,
      "step": 7044
    },
    {
      "epoch": 0.9,
      "grad_norm": 8.31205014705302,
      "learning_rate": 2.507105831883794e-07,
      "loss": 0.4903,
      "step": 7045
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.767683341545522,
      "learning_rate": 2.5006276699693854e-07,
      "loss": 0.7286,
      "step": 7046
    },
    {
      "epoch": 0.9,
      "grad_norm": 11.434700155862142,
      "learning_rate": 2.4941576737477435e-07,
      "loss": 0.5668,
      "step": 7047
    },
    {
      "epoch": 0.9,
      "grad_norm": 8.45831369969746,
      "learning_rate": 2.487695844331128e-07,
      "loss": 0.7126,
      "step": 7048
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.976063900232673,
      "learning_rate": 2.481242182830401e-07,
      "loss": 0.6599,
      "step": 7049
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4312737665340522,
      "learning_rate": 2.4747966903550355e-07,
      "loss": 0.1725,
      "step": 7050
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4632816695544704,
      "learning_rate": 2.4683593680130734e-07,
      "loss": 0.1619,
      "step": 7051
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5044988402240864,
      "learning_rate": 2.461930216911168e-07,
      "loss": 0.1334,
      "step": 7052
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.74739092598359,
      "learning_rate": 2.455509238154574e-07,
      "loss": 0.7147,
      "step": 7053
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3881023490364681,
      "learning_rate": 2.4490964328471257e-07,
      "loss": 0.1655,
      "step": 7054
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4129818292776672,
      "learning_rate": 2.442691802091257e-07,
      "loss": 0.175,
      "step": 7055
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.960608565904123,
      "learning_rate": 2.4362953469879934e-07,
      "loss": 0.489,
      "step": 7056
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.155276196557764,
      "learning_rate": 2.429907068636972e-07,
      "loss": 0.55,
      "step": 7057
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4261941228741186,
      "learning_rate": 2.423526968136397e-07,
      "loss": 0.1686,
      "step": 7058
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.651084696854624,
      "learning_rate": 2.4171550465830974e-07,
      "loss": 0.1918,
      "step": 7059
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.398104050667251,
      "learning_rate": 2.4107913050724627e-07,
      "loss": 0.5164,
      "step": 7060
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.999371704622976,
      "learning_rate": 2.4044357446985134e-07,
      "loss": 0.504,
      "step": 7061
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4675693602938864,
      "learning_rate": 2.398088366553836e-07,
      "loss": 0.1722,
      "step": 7062
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.521871426667267,
      "learning_rate": 2.3917491717296184e-07,
      "loss": 0.1667,
      "step": 7063
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3802861233996229,
      "learning_rate": 2.385418161315639e-07,
      "loss": 0.1779,
      "step": 7064
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3349099424141957,
      "learning_rate": 2.3790953364002722e-07,
      "loss": 0.1823,
      "step": 7065
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.108379560973643,
      "learning_rate": 2.372780698070476e-07,
      "loss": 0.5775,
      "step": 7066
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.496034249146795,
      "learning_rate": 2.3664742474118317e-07,
      "loss": 0.1876,
      "step": 7067
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7119658513079599,
      "learning_rate": 2.3601759855084672e-07,
      "loss": 0.1992,
      "step": 7068
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.18947806337989,
      "learning_rate": 2.3538859134431547e-07,
      "loss": 0.1404,
      "step": 7069
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3380807910482526,
      "learning_rate": 2.347604032297207e-07,
      "loss": 0.1727,
      "step": 7070
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.270171572657038,
      "learning_rate": 2.3413303431505606e-07,
      "loss": 0.1701,
      "step": 7071
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.124597349142281,
      "learning_rate": 2.3350648470817416e-07,
      "loss": 0.4009,
      "step": 7072
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4418898757349194,
      "learning_rate": 2.3288075451678381e-07,
      "loss": 0.1581,
      "step": 7073
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5437768044705733,
      "learning_rate": 2.3225584384845845e-07,
      "loss": 0.1772,
      "step": 7074
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.448015528187404,
      "learning_rate": 2.3163175281062545e-07,
      "loss": 0.1776,
      "step": 7075
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.389401906712368,
      "learning_rate": 2.31008481510574e-07,
      "loss": 0.152,
      "step": 7076
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5032726634005433,
      "learning_rate": 2.3038603005545113e-07,
      "loss": 0.1907,
      "step": 7077
    },
    {
      "epoch": 0.91,
      "grad_norm": 11.333070161470395,
      "learning_rate": 2.2976439855226406e-07,
      "loss": 0.4245,
      "step": 7078
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.602929061129649,
      "learning_rate": 2.2914358710787842e-07,
      "loss": 0.2371,
      "step": 7079
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4117213480380904,
      "learning_rate": 2.2852359582901828e-07,
      "loss": 0.1841,
      "step": 7080
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.478336442154045,
      "learning_rate": 2.2790442482226727e-07,
      "loss": 0.4395,
      "step": 7081
    },
    {
      "epoch": 0.91,
      "grad_norm": 14.700327607294177,
      "learning_rate": 2.2728607419406967e-07,
      "loss": 0.5698,
      "step": 7082
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2601490665159365,
      "learning_rate": 2.2666854405072546e-07,
      "loss": 0.1396,
      "step": 7083
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3808902541955865,
      "learning_rate": 2.2605183449839585e-07,
      "loss": 0.1842,
      "step": 7084
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6802796238034272,
      "learning_rate": 2.2543594564309989e-07,
      "loss": 0.2221,
      "step": 7085
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5312497331628596,
      "learning_rate": 2.2482087759071625e-07,
      "loss": 0.1939,
      "step": 7086
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.767269479358581,
      "learning_rate": 2.2420663044698254e-07,
      "loss": 0.5543,
      "step": 7087
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.2475664699341165,
      "learning_rate": 2.2359320431749432e-07,
      "loss": 0.5824,
      "step": 7088
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5897562302129002,
      "learning_rate": 2.2298059930770833e-07,
      "loss": 0.1813,
      "step": 7089
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.355242462826402,
      "learning_rate": 2.2236881552293642e-07,
      "loss": 0.1783,
      "step": 7090
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.723792563539857,
      "learning_rate": 2.2175785306835285e-07,
      "loss": 0.5526,
      "step": 7091
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.225198948516342,
      "learning_rate": 2.21147712048988e-07,
      "loss": 0.5602,
      "step": 7092
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.442122188023854,
      "learning_rate": 2.2053839256973297e-07,
      "loss": 0.1604,
      "step": 7093
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6190201560315993,
      "learning_rate": 2.1992989473533566e-07,
      "loss": 0.2098,
      "step": 7094
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5083255564959284,
      "learning_rate": 2.1932221865040572e-07,
      "loss": 0.1909,
      "step": 7095
    },
    {
      "epoch": 0.91,
      "grad_norm": 9.973295850209851,
      "learning_rate": 2.187153644194079e-07,
      "loss": 0.5217,
      "step": 7096
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2780703967907983,
      "learning_rate": 2.1810933214666928e-07,
      "loss": 0.1604,
      "step": 7097
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4036293158391824,
      "learning_rate": 2.1750412193637216e-07,
      "loss": 0.1704,
      "step": 7098
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3425026220595333,
      "learning_rate": 2.1689973389256047e-07,
      "loss": 0.1445,
      "step": 7099
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2056796250875919,
      "learning_rate": 2.1629616811913502e-07,
      "loss": 0.1631,
      "step": 7100
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.169950617407075,
      "learning_rate": 2.1569342471985556e-07,
      "loss": 0.3941,
      "step": 7101
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4238406686843867,
      "learning_rate": 2.150915037983403e-07,
      "loss": 0.1932,
      "step": 7102
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.156609034526518,
      "learning_rate": 2.1449040545806766e-07,
      "loss": 0.4995,
      "step": 7103
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.464151049150612,
      "learning_rate": 2.1389012980237267e-07,
      "loss": 0.2001,
      "step": 7104
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3144400243471983,
      "learning_rate": 2.1329067693444893e-07,
      "loss": 0.1658,
      "step": 7105
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3546424096507046,
      "learning_rate": 2.126920469573507e-07,
      "loss": 0.1921,
      "step": 7106
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4060040116157246,
      "learning_rate": 2.1209423997398893e-07,
      "loss": 0.1683,
      "step": 7107
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4608002004686191,
      "learning_rate": 2.1149725608713368e-07,
      "loss": 0.1747,
      "step": 7108
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3180644193084092,
      "learning_rate": 2.109010953994123e-07,
      "loss": 0.136,
      "step": 7109
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4081688205965388,
      "learning_rate": 2.1030575801331332e-07,
      "loss": 0.1594,
      "step": 7110
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4009704699262875,
      "learning_rate": 2.097112440311816e-07,
      "loss": 0.1936,
      "step": 7111
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.528087558656676,
      "learning_rate": 2.0911755355522089e-07,
      "loss": 0.1541,
      "step": 7112
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4300154813814487,
      "learning_rate": 2.0852468668749294e-07,
      "loss": 0.1789,
      "step": 7113
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.312166037621835,
      "learning_rate": 2.0793264352991894e-07,
      "loss": 0.1607,
      "step": 7114
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3882277283671325,
      "learning_rate": 2.0734142418427806e-07,
      "loss": 0.1672,
      "step": 7115
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.53239831361709,
      "learning_rate": 2.067510287522073e-07,
      "loss": 0.1693,
      "step": 7116
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6709450061370277,
      "learning_rate": 2.0616145733520276e-07,
      "loss": 0.2405,
      "step": 7117
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.277988201641677,
      "learning_rate": 2.0557271003461942e-07,
      "loss": 0.1583,
      "step": 7118
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3940303164785002,
      "learning_rate": 2.0498478695166857e-07,
      "loss": 0.1539,
      "step": 7119
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.164997007917478,
      "learning_rate": 2.0439768818742156e-07,
      "loss": 0.4664,
      "step": 7120
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4680867556198658,
      "learning_rate": 2.0381141384280711e-07,
      "loss": 0.1825,
      "step": 7121
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2999863036149153,
      "learning_rate": 2.0322596401861294e-07,
      "loss": 0.1396,
      "step": 7122
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4742207492413313,
      "learning_rate": 2.026413388154841e-07,
      "loss": 0.1621,
      "step": 7123
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4788385361407068,
      "learning_rate": 2.020575383339246e-07,
      "loss": 0.218,
      "step": 7124
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.335334599940033,
      "learning_rate": 2.0147456267429754e-07,
      "loss": 0.1659,
      "step": 7125
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.379594958541275,
      "learning_rate": 2.0089241193682273e-07,
      "loss": 0.1394,
      "step": 7126
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5551625257823412,
      "learning_rate": 2.003110862215779e-07,
      "loss": 0.1858,
      "step": 7127
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6704648491778622,
      "learning_rate": 1.9973058562850033e-07,
      "loss": 0.1893,
      "step": 7128
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.408357972114624,
      "learning_rate": 1.9915091025738464e-07,
      "loss": 0.1397,
      "step": 7129
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.94261980293549,
      "learning_rate": 1.985720602078828e-07,
      "loss": 0.6757,
      "step": 7130
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5698819309128422,
      "learning_rate": 1.9799403557950793e-07,
      "loss": 0.1915,
      "step": 7131
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3372192614672993,
      "learning_rate": 1.9741683647162724e-07,
      "loss": 0.1506,
      "step": 7132
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.697403404598429,
      "learning_rate": 1.9684046298346858e-07,
      "loss": 0.4601,
      "step": 7133
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4394217254750252,
      "learning_rate": 1.9626491521411773e-07,
      "loss": 0.1459,
      "step": 7134
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3378035466886593,
      "learning_rate": 1.956901932625177e-07,
      "loss": 0.1434,
      "step": 7135
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.265312413296288,
      "learning_rate": 1.9511629722747004e-07,
      "loss": 0.1652,
      "step": 7136
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5131658823956897,
      "learning_rate": 1.9454322720763364e-07,
      "loss": 0.1997,
      "step": 7137
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3960028772517057,
      "learning_rate": 1.939709833015252e-07,
      "loss": 0.174,
      "step": 7138
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5233123541585716,
      "learning_rate": 1.9339956560752216e-07,
      "loss": 0.1968,
      "step": 7139
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.823449419952525,
      "learning_rate": 1.9282897422385593e-07,
      "loss": 0.5759,
      "step": 7140
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.52951738297842,
      "learning_rate": 1.9225920924861917e-07,
      "loss": 0.2178,
      "step": 7141
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4244413381318064,
      "learning_rate": 1.9169027077975965e-07,
      "loss": 0.2081,
      "step": 7142
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.625429652138192,
      "learning_rate": 1.9112215891508635e-07,
      "loss": 0.1958,
      "step": 7143
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2815522204056704,
      "learning_rate": 1.905548737522628e-07,
      "loss": 0.1537,
      "step": 7144
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.82717888289317,
      "learning_rate": 1.899884153888115e-07,
      "loss": 0.4813,
      "step": 7145
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3603422461599448,
      "learning_rate": 1.8942278392211466e-07,
      "loss": 0.1607,
      "step": 7146
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.452972324046191,
      "learning_rate": 1.8885797944941052e-07,
      "loss": 0.1819,
      "step": 7147
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.283024609183638,
      "learning_rate": 1.8829400206779536e-07,
      "loss": 0.135,
      "step": 7148
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.403389221170893,
      "learning_rate": 1.8773085187422325e-07,
      "loss": 0.1836,
      "step": 7149
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2357954616028628,
      "learning_rate": 1.8716852896550618e-07,
      "loss": 0.1685,
      "step": 7150
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3421043577454883,
      "learning_rate": 1.8660703343831354e-07,
      "loss": 0.1672,
      "step": 7151
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3069071172180893,
      "learning_rate": 1.8604636538917365e-07,
      "loss": 0.1911,
      "step": 7152
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1599358914069686,
      "learning_rate": 1.8548652491447217e-07,
      "loss": 0.1355,
      "step": 7153
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.512668406391628,
      "learning_rate": 1.8492751211045156e-07,
      "loss": 0.2073,
      "step": 7154
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.859525346474024,
      "learning_rate": 1.8436932707321276e-07,
      "loss": 0.4874,
      "step": 7155
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.450824979562763,
      "learning_rate": 1.8381196989871453e-07,
      "loss": 0.1955,
      "step": 7156
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.279452864757562,
      "learning_rate": 1.8325544068277244e-07,
      "loss": 0.1578,
      "step": 7157
    },
    {
      "epoch": 0.92,
      "grad_norm": 29.45503180615539,
      "learning_rate": 1.8269973952106057e-07,
      "loss": 0.6148,
      "step": 7158
    },
    {
      "epoch": 0.92,
      "grad_norm": 7.035722745099297,
      "learning_rate": 1.8214486650911022e-07,
      "loss": 0.5672,
      "step": 7159
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3688742455605114,
      "learning_rate": 1.8159082174231012e-07,
      "loss": 0.1995,
      "step": 7160
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.37145963644259,
      "learning_rate": 1.8103760531590851e-07,
      "loss": 0.1862,
      "step": 7161
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4607102873952442,
      "learning_rate": 1.8048521732500878e-07,
      "loss": 0.2266,
      "step": 7162
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3728660146489944,
      "learning_rate": 1.7993365786457217e-07,
      "loss": 0.1519,
      "step": 7163
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5337199191037985,
      "learning_rate": 1.793829270294195e-07,
      "loss": 0.1888,
      "step": 7164
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3903050319489396,
      "learning_rate": 1.7883302491422673e-07,
      "loss": 0.1552,
      "step": 7165
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.461830084372644,
      "learning_rate": 1.782839516135282e-07,
      "loss": 0.1943,
      "step": 7166
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.9383573288146785,
      "learning_rate": 1.777357072217173e-07,
      "loss": 0.6204,
      "step": 7167
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4858933609306315,
      "learning_rate": 1.7718829183304254e-07,
      "loss": 0.1761,
      "step": 7168
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2717970416838715,
      "learning_rate": 1.766417055416114e-07,
      "loss": 0.1586,
      "step": 7169
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4288897078144887,
      "learning_rate": 1.7609594844138767e-07,
      "loss": 0.2173,
      "step": 7170
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.421132530844567,
      "learning_rate": 1.7555102062619454e-07,
      "loss": 0.1914,
      "step": 7171
    },
    {
      "epoch": 0.92,
      "grad_norm": 7.783487663883801,
      "learning_rate": 1.7500692218971048e-07,
      "loss": 0.6022,
      "step": 7172
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4368578784777528,
      "learning_rate": 1.7446365322547231e-07,
      "loss": 0.17,
      "step": 7173
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4827176057750828,
      "learning_rate": 1.7392121382687533e-07,
      "loss": 0.2137,
      "step": 7174
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.359164014078232,
      "learning_rate": 1.733796040871699e-07,
      "loss": 0.1642,
      "step": 7175
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3945640299745288,
      "learning_rate": 1.7283882409946552e-07,
      "loss": 0.1686,
      "step": 7176
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4099896590901242,
      "learning_rate": 1.7229887395672884e-07,
      "loss": 0.1587,
      "step": 7177
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3957923050350216,
      "learning_rate": 1.7175975375178343e-07,
      "loss": 0.1923,
      "step": 7178
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.634142928047715,
      "learning_rate": 1.7122146357730908e-07,
      "loss": 0.1883,
      "step": 7179
    },
    {
      "epoch": 0.92,
      "grad_norm": 15.05832227138776,
      "learning_rate": 1.7068400352584613e-07,
      "loss": 0.5303,
      "step": 7180
    },
    {
      "epoch": 0.92,
      "grad_norm": 7.547144166446662,
      "learning_rate": 1.7014737368978795e-07,
      "loss": 0.469,
      "step": 7181
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.477317874458856,
      "learning_rate": 1.6961157416139018e-07,
      "loss": 0.1631,
      "step": 7182
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2719958219026772,
      "learning_rate": 1.690766050327608e-07,
      "loss": 0.1211,
      "step": 7183
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.2640659802659355,
      "learning_rate": 1.685424663958679e-07,
      "loss": 0.4868,
      "step": 7184
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4838756690891914,
      "learning_rate": 1.680091583425364e-07,
      "loss": 0.1971,
      "step": 7185
    },
    {
      "epoch": 0.92,
      "grad_norm": 11.122165293107617,
      "learning_rate": 1.674766809644479e-07,
      "loss": 0.5998,
      "step": 7186
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3952499552760294,
      "learning_rate": 1.6694503435314035e-07,
      "loss": 0.1906,
      "step": 7187
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.371389059075517,
      "learning_rate": 1.6641421860001172e-07,
      "loss": 0.1797,
      "step": 7188
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4602167698121564,
      "learning_rate": 1.6588423379631458e-07,
      "loss": 0.1676,
      "step": 7189
    },
    {
      "epoch": 0.92,
      "grad_norm": 8.285963385184276,
      "learning_rate": 1.6535508003315937e-07,
      "loss": 0.6354,
      "step": 7190
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.692523605246867,
      "learning_rate": 1.6482675740151444e-07,
      "loss": 0.4999,
      "step": 7191
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1981516030847363,
      "learning_rate": 1.642992659922038e-07,
      "loss": 0.1294,
      "step": 7192
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4363849456470237,
      "learning_rate": 1.6377260589590939e-07,
      "loss": 0.1647,
      "step": 7193
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4058390100154026,
      "learning_rate": 1.632467772031704e-07,
      "loss": 0.1589,
      "step": 7194
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.465764783422582,
      "learning_rate": 1.6272178000438288e-07,
      "loss": 0.1704,
      "step": 7195
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.709981523763056,
      "learning_rate": 1.621976143898002e-07,
      "loss": 0.6074,
      "step": 7196
    },
    {
      "epoch": 0.92,
      "grad_norm": 8.67914119335301,
      "learning_rate": 1.6167428044953138e-07,
      "loss": 0.5606,
      "step": 7197
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3157509716318696,
      "learning_rate": 1.6115177827354556e-07,
      "loss": 0.1445,
      "step": 7198
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2037126643696545,
      "learning_rate": 1.6063010795166533e-07,
      "loss": 0.1638,
      "step": 7199
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3122645981068743,
      "learning_rate": 1.6010926957357232e-07,
      "loss": 0.1321,
      "step": 7200
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3321407075599156,
      "learning_rate": 1.5958926322880487e-07,
      "loss": 0.1773,
      "step": 7201
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9219610306442825,
      "learning_rate": 1.590700890067576e-07,
      "loss": 0.1998,
      "step": 7202
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.843050625952519,
      "learning_rate": 1.5855174699668298e-07,
      "loss": 0.6094,
      "step": 7203
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4145773929914538,
      "learning_rate": 1.580342372876903e-07,
      "loss": 0.1646,
      "step": 7204
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4768619211605336,
      "learning_rate": 1.5751755996874452e-07,
      "loss": 0.1779,
      "step": 7205
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.029909206745744,
      "learning_rate": 1.5700171512866956e-07,
      "loss": 0.5091,
      "step": 7206
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4164576425214326,
      "learning_rate": 1.5648670285614397e-07,
      "loss": 0.228,
      "step": 7207
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4601840348423698,
      "learning_rate": 1.559725232397058e-07,
      "loss": 0.2175,
      "step": 7208
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4850539006212469,
      "learning_rate": 1.5545917636774655e-07,
      "loss": 0.2077,
      "step": 7209
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5471511466620531,
      "learning_rate": 1.5494666232851896e-07,
      "loss": 0.1789,
      "step": 7210
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5046007986254784,
      "learning_rate": 1.5443498121012813e-07,
      "loss": 0.1751,
      "step": 7211
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.098279471883155,
      "learning_rate": 1.5392413310053866e-07,
      "loss": 0.6422,
      "step": 7212
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3356362964537132,
      "learning_rate": 1.5341411808757146e-07,
      "loss": 0.1665,
      "step": 7213
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5883365912379874,
      "learning_rate": 1.5290493625890413e-07,
      "loss": 0.2039,
      "step": 7214
    },
    {
      "epoch": 0.92,
      "grad_norm": 7.35286778943639,
      "learning_rate": 1.5239658770206945e-07,
      "loss": 0.5181,
      "step": 7215
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5747130173504835,
      "learning_rate": 1.5188907250446028e-07,
      "loss": 0.1934,
      "step": 7216
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3838419933363315,
      "learning_rate": 1.513823907533235e-07,
      "loss": 0.1743,
      "step": 7217
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.391891399705947,
      "learning_rate": 1.508765425357639e-07,
      "loss": 0.1498,
      "step": 7218
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3411785197310666,
      "learning_rate": 1.5037152793874244e-07,
      "loss": 0.1598,
      "step": 7219
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4098826794746284,
      "learning_rate": 1.4986734704907745e-07,
      "loss": 0.1798,
      "step": 7220
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.023026679191991,
      "learning_rate": 1.4936399995344298e-07,
      "loss": 0.4749,
      "step": 7221
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3474208397035898,
      "learning_rate": 1.4886148673836975e-07,
      "loss": 0.1591,
      "step": 7222
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.069841468978265,
      "learning_rate": 1.4835980749024592e-07,
      "loss": 0.5401,
      "step": 7223
    },
    {
      "epoch": 0.92,
      "grad_norm": 15.236861242063323,
      "learning_rate": 1.4785896229531692e-07,
      "loss": 0.4648,
      "step": 7224
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4885901100197436,
      "learning_rate": 1.473589512396828e-07,
      "loss": 0.2058,
      "step": 7225
    },
    {
      "epoch": 0.92,
      "grad_norm": 10.071853012572538,
      "learning_rate": 1.46859774409302e-07,
      "loss": 0.5759,
      "step": 7226
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6057203370776216,
      "learning_rate": 1.4636143188998808e-07,
      "loss": 0.2081,
      "step": 7227
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2939280272932496,
      "learning_rate": 1.4586392376741254e-07,
      "loss": 0.1371,
      "step": 7228
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.44450405589997,
      "learning_rate": 1.4536725012710252e-07,
      "loss": 0.1817,
      "step": 7229
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3167800472582478,
      "learning_rate": 1.4487141105444136e-07,
      "loss": 0.1621,
      "step": 7230
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.482912717299135,
      "learning_rate": 1.4437640663467034e-07,
      "loss": 0.1976,
      "step": 7231
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5329313463072807,
      "learning_rate": 1.4388223695288695e-07,
      "loss": 0.1967,
      "step": 7232
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3299075045738111,
      "learning_rate": 1.433889020940432e-07,
      "loss": 0.1588,
      "step": 7233
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2239676454686081,
      "learning_rate": 1.4289640214294963e-07,
      "loss": 0.1766,
      "step": 7234
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.774098375568992,
      "learning_rate": 1.424047371842735e-07,
      "loss": 0.5528,
      "step": 7235
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4015532454788444,
      "learning_rate": 1.4191390730253718e-07,
      "loss": 0.1335,
      "step": 7236
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2317564418570501,
      "learning_rate": 1.4142391258211985e-07,
      "loss": 0.1398,
      "step": 7237
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.436122230361409,
      "learning_rate": 1.409347531072569e-07,
      "loss": 0.1425,
      "step": 7238
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6937894154593316,
      "learning_rate": 1.4044642896204107e-07,
      "loss": 0.1674,
      "step": 7239
    },
    {
      "epoch": 0.93,
      "grad_norm": 10.803354631532986,
      "learning_rate": 1.3995894023042135e-07,
      "loss": 0.6101,
      "step": 7240
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.3269013165921395,
      "learning_rate": 1.394722869962023e-07,
      "loss": 0.5061,
      "step": 7241
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.219921624343162,
      "learning_rate": 1.3898646934304538e-07,
      "loss": 0.1582,
      "step": 7242
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5667859592171387,
      "learning_rate": 1.3850148735446767e-07,
      "loss": 0.1625,
      "step": 7243
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3726326027473652,
      "learning_rate": 1.38017341113843e-07,
      "loss": 0.1671,
      "step": 7244
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3962585712023172,
      "learning_rate": 1.3753403070440263e-07,
      "loss": 0.1297,
      "step": 7245
    },
    {
      "epoch": 0.93,
      "grad_norm": 14.462897908616782,
      "learning_rate": 1.3705155620923337e-07,
      "loss": 0.7602,
      "step": 7246
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.063554249897834,
      "learning_rate": 1.3656991771127781e-07,
      "loss": 0.4938,
      "step": 7247
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3654573348990202,
      "learning_rate": 1.3608911529333467e-07,
      "loss": 0.1608,
      "step": 7248
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4680324748052795,
      "learning_rate": 1.3560914903806065e-07,
      "loss": 0.1956,
      "step": 7249
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5338112774327608,
      "learning_rate": 1.3513001902796642e-07,
      "loss": 0.2227,
      "step": 7250
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.361193411577292,
      "learning_rate": 1.3465172534541936e-07,
      "loss": 0.1678,
      "step": 7251
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4728355918492892,
      "learning_rate": 1.34174268072646e-07,
      "loss": 0.1742,
      "step": 7252
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2275952951010456,
      "learning_rate": 1.3369764729172453e-07,
      "loss": 0.1714,
      "step": 7253
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.373762143076465,
      "learning_rate": 1.3322186308459274e-07,
      "loss": 0.1864,
      "step": 7254
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4585883061839298,
      "learning_rate": 1.3274691553304352e-07,
      "loss": 0.1755,
      "step": 7255
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.530605559383586,
      "learning_rate": 1.322728047187255e-07,
      "loss": 0.6136,
      "step": 7256
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5687011475769108,
      "learning_rate": 1.31799530723144e-07,
      "loss": 0.1862,
      "step": 7257
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.324408551088491,
      "learning_rate": 1.3132709362766006e-07,
      "loss": 0.1355,
      "step": 7258
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4041283774422952,
      "learning_rate": 1.308554935134909e-07,
      "loss": 0.1709,
      "step": 7259
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6427535183394701,
      "learning_rate": 1.3038473046171063e-07,
      "loss": 0.2144,
      "step": 7260
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3129589935021615,
      "learning_rate": 1.299148045532489e-07,
      "loss": 0.1487,
      "step": 7261
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.791299255167242,
      "learning_rate": 1.2944571586888998e-07,
      "loss": 0.4093,
      "step": 7262
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4921848299432237,
      "learning_rate": 1.2897746448927828e-07,
      "loss": 0.201,
      "step": 7263
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3735267100886837,
      "learning_rate": 1.2851005049490939e-07,
      "loss": 0.1673,
      "step": 7264
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4104455869031813,
      "learning_rate": 1.2804347396613848e-07,
      "loss": 0.1766,
      "step": 7265
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4187811531844776,
      "learning_rate": 1.2757773498317416e-07,
      "loss": 0.197,
      "step": 7266
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5727855904277033,
      "learning_rate": 1.2711283362608351e-07,
      "loss": 0.2137,
      "step": 7267
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.256830162885882,
      "learning_rate": 1.266487699747887e-07,
      "loss": 0.1542,
      "step": 7268
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4361515160494396,
      "learning_rate": 1.2618554410906648e-07,
      "loss": 0.1969,
      "step": 7269
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.753349115682825,
      "learning_rate": 1.2572315610855201e-07,
      "loss": 0.6696,
      "step": 7270
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.405878519827229,
      "learning_rate": 1.252616060527334e-07,
      "loss": 0.5085,
      "step": 7271
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2421461871505322,
      "learning_rate": 1.248008940209583e-07,
      "loss": 0.1363,
      "step": 7272
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4575388693126703,
      "learning_rate": 1.243410200924272e-07,
      "loss": 0.2067,
      "step": 7273
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2849166183470446,
      "learning_rate": 1.2388198434619803e-07,
      "loss": 0.146,
      "step": 7274
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.299467959280238,
      "learning_rate": 1.2342378686118538e-07,
      "loss": 0.1799,
      "step": 7275
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.521191814211609,
      "learning_rate": 1.2296642771615741e-07,
      "loss": 0.2198,
      "step": 7276
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.261018896331814,
      "learning_rate": 1.2250990698974009e-07,
      "loss": 0.1306,
      "step": 7277
    },
    {
      "epoch": 0.93,
      "grad_norm": 8.094538686011935,
      "learning_rate": 1.2205422476041452e-07,
      "loss": 0.3784,
      "step": 7278
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6026372447783528,
      "learning_rate": 1.2159938110651803e-07,
      "loss": 0.1894,
      "step": 7279
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4693442270239638,
      "learning_rate": 1.2114537610624255e-07,
      "loss": 0.2038,
      "step": 7280
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.315154660886358,
      "learning_rate": 1.206922098376373e-07,
      "loss": 0.1352,
      "step": 7281
    },
    {
      "epoch": 0.93,
      "grad_norm": 8.020097750224952,
      "learning_rate": 1.2023988237860718e-07,
      "loss": 0.6806,
      "step": 7282
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5461742615654506,
      "learning_rate": 1.1978839380691277e-07,
      "loss": 0.1805,
      "step": 7283
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3913491261236242,
      "learning_rate": 1.1933774420016974e-07,
      "loss": 0.1513,
      "step": 7284
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5428550764534585,
      "learning_rate": 1.188879336358495e-07,
      "loss": 0.1863,
      "step": 7285
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4241068404099755,
      "learning_rate": 1.184389621912807e-07,
      "loss": 0.136,
      "step": 7286
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6148403373616085,
      "learning_rate": 1.1799082994364553e-07,
      "loss": 0.1603,
      "step": 7287
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3774693690800874,
      "learning_rate": 1.17543536969984e-07,
      "loss": 0.1723,
      "step": 7288
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.559204019837223,
      "learning_rate": 1.1709708334719128e-07,
      "loss": 0.4545,
      "step": 7289
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.348496770482337,
      "learning_rate": 1.1665146915201652e-07,
      "loss": 0.1787,
      "step": 7290
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.376136826190946,
      "learning_rate": 1.1620669446106735e-07,
      "loss": 0.1584,
      "step": 7291
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3441779245341914,
      "learning_rate": 1.157627593508054e-07,
      "loss": 0.1484,
      "step": 7292
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3330106205451178,
      "learning_rate": 1.153196638975479e-07,
      "loss": 0.137,
      "step": 7293
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3640537470206517,
      "learning_rate": 1.148774081774684e-07,
      "loss": 0.1895,
      "step": 7294
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5236692772607277,
      "learning_rate": 1.1443599226659497e-07,
      "loss": 0.2025,
      "step": 7295
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4125315141453993,
      "learning_rate": 1.1399541624081357e-07,
      "loss": 0.177,
      "step": 7296
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.272498904961072,
      "learning_rate": 1.1355568017586305e-07,
      "loss": 0.1631,
      "step": 7297
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.240857405015692,
      "learning_rate": 1.1311678414734018e-07,
      "loss": 0.437,
      "step": 7298
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4235015086750316,
      "learning_rate": 1.1267872823069459e-07,
      "loss": 0.2039,
      "step": 7299
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5334663203729448,
      "learning_rate": 1.1224151250123549e-07,
      "loss": 0.187,
      "step": 7300
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5820199556248613,
      "learning_rate": 1.1180513703412388e-07,
      "loss": 0.1917,
      "step": 7301
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4372584122271714,
      "learning_rate": 1.1136960190437751e-07,
      "loss": 0.1351,
      "step": 7302
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5850327419413208,
      "learning_rate": 1.1093490718687094e-07,
      "loss": 0.2012,
      "step": 7303
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3492284100819754,
      "learning_rate": 1.1050105295633274e-07,
      "loss": 0.1685,
      "step": 7304
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2132529870555941,
      "learning_rate": 1.1006803928734711e-07,
      "loss": 0.1508,
      "step": 7305
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.446303151054239,
      "learning_rate": 1.0963586625435507e-07,
      "loss": 0.1581,
      "step": 7306
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6074218653518841,
      "learning_rate": 1.0920453393165109e-07,
      "loss": 0.2172,
      "step": 7307
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4760407916676264,
      "learning_rate": 1.087740423933864e-07,
      "loss": 0.2101,
      "step": 7308
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4392248182537786,
      "learning_rate": 1.0834439171356848e-07,
      "loss": 0.1964,
      "step": 7309
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4401846431801213,
      "learning_rate": 1.0791558196605823e-07,
      "loss": 0.1652,
      "step": 7310
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3468351288947322,
      "learning_rate": 1.0748761322457334e-07,
      "loss": 0.1653,
      "step": 7311
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4358329207184246,
      "learning_rate": 1.0706048556268667e-07,
      "loss": 0.1875,
      "step": 7312
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3260190690722076,
      "learning_rate": 1.0663419905382666e-07,
      "loss": 0.1649,
      "step": 7313
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2170160461508888,
      "learning_rate": 1.0620875377127637e-07,
      "loss": 0.152,
      "step": 7314
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3708604518246297,
      "learning_rate": 1.0578414978817508e-07,
      "loss": 0.1777,
      "step": 7315
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4624398699582728,
      "learning_rate": 1.0536038717751607e-07,
      "loss": 0.1924,
      "step": 7316
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3927382826066625,
      "learning_rate": 1.0493746601215105e-07,
      "loss": 0.1835,
      "step": 7317
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.383901021738157,
      "learning_rate": 1.0451538636478353e-07,
      "loss": 0.5256,
      "step": 7318
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7400734205366823,
      "learning_rate": 1.0409414830797493e-07,
      "loss": 0.2015,
      "step": 7319
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4412342694864455,
      "learning_rate": 1.0367375191414064e-07,
      "loss": 0.1635,
      "step": 7320
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7280093896993094,
      "learning_rate": 1.032541972555512e-07,
      "loss": 0.1874,
      "step": 7321
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3970415267023037,
      "learning_rate": 1.0283548440433332e-07,
      "loss": 0.1309,
      "step": 7322
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3376175728784874,
      "learning_rate": 1.0241761343246781e-07,
      "loss": 0.1873,
      "step": 7323
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5189853205759791,
      "learning_rate": 1.0200058441179272e-07,
      "loss": 0.1848,
      "step": 7324
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4903796264068594,
      "learning_rate": 1.0158439741399961e-07,
      "loss": 0.1881,
      "step": 7325
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.716396793278513,
      "learning_rate": 1.0116905251063625e-07,
      "loss": 0.5788,
      "step": 7326
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.532369225178833,
      "learning_rate": 1.0075454977310384e-07,
      "loss": 0.1954,
      "step": 7327
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.446494026223758,
      "learning_rate": 1.0034088927266206e-07,
      "loss": 0.1974,
      "step": 7328
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4947233522276417,
      "learning_rate": 9.99280710804229e-08,
      "loss": 0.1882,
      "step": 7329
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4173387296585291,
      "learning_rate": 9.951609526735517e-08,
      "loss": 0.1558,
      "step": 7330
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3624651410515591,
      "learning_rate": 9.910496190428164e-08,
      "loss": 0.1732,
      "step": 7331
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.258813392301987,
      "learning_rate": 9.869467106188135e-08,
      "loss": 0.098,
      "step": 7332
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5562807672939263,
      "learning_rate": 9.828522281068787e-08,
      "loss": 0.1989,
      "step": 7333
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4539752728568762,
      "learning_rate": 9.787661722108988e-08,
      "loss": 0.1768,
      "step": 7334
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4820072053027105,
      "learning_rate": 9.74688543633323e-08,
      "loss": 0.2154,
      "step": 7335
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.437456743461236,
      "learning_rate": 9.706193430751298e-08,
      "loss": 0.186,
      "step": 7336
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7345016542650629,
      "learning_rate": 9.665585712358704e-08,
      "loss": 0.1814,
      "step": 7337
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.470507811786504,
      "learning_rate": 9.625062288136367e-08,
      "loss": 0.1859,
      "step": 7338
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.069912248147817,
      "learning_rate": 9.584623165050766e-08,
      "loss": 0.5678,
      "step": 7339
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.604450854661282,
      "learning_rate": 9.544268350053843e-08,
      "loss": 0.2105,
      "step": 7340
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4903503394289572,
      "learning_rate": 9.50399785008299e-08,
      "loss": 0.2128,
      "step": 7341
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3778121956628353,
      "learning_rate": 9.463811672061284e-08,
      "loss": 0.1721,
      "step": 7342
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3914581403599136,
      "learning_rate": 9.423709822897087e-08,
      "loss": 0.1597,
      "step": 7343
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2309284591514702,
      "learning_rate": 9.383692309484382e-08,
      "loss": 0.1715,
      "step": 7344
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.324429775945907,
      "learning_rate": 9.343759138702724e-08,
      "loss": 0.1553,
      "step": 7345
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4159498687577035,
      "learning_rate": 9.303910317417064e-08,
      "loss": 0.2038,
      "step": 7346
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4283403168045696,
      "learning_rate": 9.264145852477868e-08,
      "loss": 0.1623,
      "step": 7347
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.354353838237953,
      "learning_rate": 9.224465750721057e-08,
      "loss": 0.4647,
      "step": 7348
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4981792799838975,
      "learning_rate": 9.184870018968173e-08,
      "loss": 0.2114,
      "step": 7349
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5268666931544566,
      "learning_rate": 9.14535866402616e-08,
      "loss": 0.1746,
      "step": 7350
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5796130035308953,
      "learning_rate": 9.10593169268742e-08,
      "loss": 0.1872,
      "step": 7351
    },
    {
      "epoch": 0.94,
      "grad_norm": 10.030360255073376,
      "learning_rate": 9.066589111729973e-08,
      "loss": 0.4785,
      "step": 7352
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2614725199153258,
      "learning_rate": 9.027330927917244e-08,
      "loss": 0.1432,
      "step": 7353
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3719100357772422,
      "learning_rate": 8.988157147998222e-08,
      "loss": 0.1596,
      "step": 7354
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.577360834651534,
      "learning_rate": 8.949067778707188e-08,
      "loss": 0.1909,
      "step": 7355
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4680838658035718,
      "learning_rate": 8.91006282676421e-08,
      "loss": 0.1942,
      "step": 7356
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.300025547866343,
      "learning_rate": 8.871142298874647e-08,
      "loss": 0.1575,
      "step": 7357
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7135803650250436,
      "learning_rate": 8.832306201729368e-08,
      "loss": 0.175,
      "step": 7358
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4956979678619813,
      "learning_rate": 8.793554542004756e-08,
      "loss": 0.1921,
      "step": 7359
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3881397482905102,
      "learning_rate": 8.754887326362649e-08,
      "loss": 0.1641,
      "step": 7360
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.968008415456132,
      "learning_rate": 8.71630456145045e-08,
      "loss": 0.5318,
      "step": 7361
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5957111155553685,
      "learning_rate": 8.677806253900967e-08,
      "loss": 0.2373,
      "step": 7362
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7919606919591757,
      "learning_rate": 8.639392410332403e-08,
      "loss": 0.1832,
      "step": 7363
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4588059008329737,
      "learning_rate": 8.601063037348644e-08,
      "loss": 0.1909,
      "step": 7364
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3026692417372714,
      "learning_rate": 8.562818141538976e-08,
      "loss": 0.1375,
      "step": 7365
    },
    {
      "epoch": 0.94,
      "grad_norm": 8.519739315574942,
      "learning_rate": 8.52465772947808e-08,
      "loss": 0.6749,
      "step": 7366
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3602999314006925,
      "learning_rate": 8.486581807726157e-08,
      "loss": 0.1721,
      "step": 7367
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.259257607602777,
      "learning_rate": 8.448590382829025e-08,
      "loss": 0.1467,
      "step": 7368
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4916854616173465,
      "learning_rate": 8.410683461317682e-08,
      "loss": 0.2204,
      "step": 7369
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6532644534670438,
      "learning_rate": 8.372861049708859e-08,
      "loss": 0.2408,
      "step": 7370
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3455553869388315,
      "learning_rate": 8.335123154504688e-08,
      "loss": 0.1551,
      "step": 7371
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3699366016082637,
      "learning_rate": 8.297469782192702e-08,
      "loss": 0.1688,
      "step": 7372
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.078787311825206,
      "learning_rate": 8.25990093924589e-08,
      "loss": 0.6583,
      "step": 7373
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3339908701685808,
      "learning_rate": 8.222416632122864e-08,
      "loss": 0.1328,
      "step": 7374
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.596732549094482,
      "learning_rate": 8.185016867267693e-08,
      "loss": 0.1599,
      "step": 7375
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4326741467653632,
      "learning_rate": 8.147701651109674e-08,
      "loss": 0.1736,
      "step": 7376
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.501964813233697,
      "learning_rate": 8.11047099006379e-08,
      "loss": 0.1943,
      "step": 7377
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.465875405566382,
      "learning_rate": 8.073324890530421e-08,
      "loss": 0.1687,
      "step": 7378
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.551971103095541,
      "learning_rate": 8.036263358895402e-08,
      "loss": 0.2046,
      "step": 7379
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5446262071146868,
      "learning_rate": 7.999286401529971e-08,
      "loss": 0.1496,
      "step": 7380
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6249746763164132,
      "learning_rate": 7.962394024791043e-08,
      "loss": 0.1977,
      "step": 7381
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.27263848759829,
      "learning_rate": 7.925586235020766e-08,
      "loss": 0.1425,
      "step": 7382
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4201124607303426,
      "learning_rate": 7.888863038546801e-08,
      "loss": 0.1927,
      "step": 7383
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.820958335507368,
      "learning_rate": 7.852224441682377e-08,
      "loss": 0.1844,
      "step": 7384
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.692689538850068,
      "learning_rate": 7.81567045072601e-08,
      "loss": 0.5906,
      "step": 7385
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5017948463412174,
      "learning_rate": 7.779201071961784e-08,
      "loss": 0.1755,
      "step": 7386
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.625069812009352,
      "learning_rate": 7.742816311659185e-08,
      "loss": 0.1862,
      "step": 7387
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.892467553640659,
      "learning_rate": 7.706516176073209e-08,
      "loss": 0.603,
      "step": 7388
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4820919481348893,
      "learning_rate": 7.670300671444309e-08,
      "loss": 0.1592,
      "step": 7389
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.324061475047632,
      "learning_rate": 7.634169803998226e-08,
      "loss": 0.1668,
      "step": 7390
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.13059125260358,
      "learning_rate": 7.598123579946382e-08,
      "loss": 0.5667,
      "step": 7391
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.8506968070252485,
      "learning_rate": 7.562162005485484e-08,
      "loss": 0.5528,
      "step": 7392
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4124423239445383,
      "learning_rate": 7.526285086797813e-08,
      "loss": 0.1469,
      "step": 7393
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1489166603636147,
      "learning_rate": 7.490492830050933e-08,
      "loss": 0.1434,
      "step": 7394
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4229736351136124,
      "learning_rate": 7.454785241398033e-08,
      "loss": 0.1748,
      "step": 7395
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.131537738804809,
      "learning_rate": 7.419162326977592e-08,
      "loss": 0.4361,
      "step": 7396
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.103820852595237,
      "learning_rate": 7.383624092913655e-08,
      "loss": 0.4941,
      "step": 7397
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4434032054253474,
      "learning_rate": 7.348170545315614e-08,
      "loss": 0.1781,
      "step": 7398
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.823788570918197,
      "learning_rate": 7.312801690278426e-08,
      "loss": 0.5068,
      "step": 7399
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.611608820267133,
      "learning_rate": 7.277517533882283e-08,
      "loss": 0.184,
      "step": 7400
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2469036672496283,
      "learning_rate": 7.242318082193e-08,
      "loss": 0.1557,
      "step": 7401
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3539756896271675,
      "learning_rate": 7.207203341261792e-08,
      "loss": 0.1348,
      "step": 7402
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.367878989845576,
      "learning_rate": 7.172173317125275e-08,
      "loss": 0.171,
      "step": 7403
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.222085720001271,
      "learning_rate": 7.137228015805519e-08,
      "loss": 0.1637,
      "step": 7404
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3423628877753788,
      "learning_rate": 7.102367443310054e-08,
      "loss": 0.1253,
      "step": 7405
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4572847796653383,
      "learning_rate": 7.067591605631752e-08,
      "loss": 0.1378,
      "step": 7406
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.072633824908139,
      "learning_rate": 7.032900508749052e-08,
      "loss": 0.5505,
      "step": 7407
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4278242642984107,
      "learning_rate": 6.998294158625684e-08,
      "loss": 0.1693,
      "step": 7408
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5085777593727034,
      "learning_rate": 6.963772561210891e-08,
      "loss": 0.1834,
      "step": 7409
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3979455461821877,
      "learning_rate": 6.929335722439367e-08,
      "loss": 0.1737,
      "step": 7410
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2598817018713209,
      "learning_rate": 6.894983648231213e-08,
      "loss": 0.1491,
      "step": 7411
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3614851118495548,
      "learning_rate": 6.86071634449198e-08,
      "loss": 0.1192,
      "step": 7412
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6779135967130423,
      "learning_rate": 6.826533817112513e-08,
      "loss": 0.1598,
      "step": 7413
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.134158430692211,
      "learning_rate": 6.792436071969277e-08,
      "loss": 0.5943,
      "step": 7414
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4052013535478112,
      "learning_rate": 6.758423114924029e-08,
      "loss": 0.1887,
      "step": 7415
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4165894192542596,
      "learning_rate": 6.724494951823979e-08,
      "loss": 0.1972,
      "step": 7416
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.621515268026614,
      "learning_rate": 6.690651588501795e-08,
      "loss": 0.1913,
      "step": 7417
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.514997996484946,
      "learning_rate": 6.656893030775546e-08,
      "loss": 0.1619,
      "step": 7418
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.663670219703273,
      "learning_rate": 6.623219284448645e-08,
      "loss": 0.6846,
      "step": 7419
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.378615395638141,
      "learning_rate": 6.589630355310128e-08,
      "loss": 0.1601,
      "step": 7420
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2687224642513935,
      "learning_rate": 6.556126249134209e-08,
      "loss": 0.1674,
      "step": 7421
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4662763892311101,
      "learning_rate": 6.522706971680726e-08,
      "loss": 0.2133,
      "step": 7422
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4869151718922742,
      "learning_rate": 6.489372528694748e-08,
      "loss": 0.2017,
      "step": 7423
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.374434892590294,
      "learning_rate": 6.456122925906971e-08,
      "loss": 0.1483,
      "step": 7424
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2135932276348984,
      "learning_rate": 6.422958169033266e-08,
      "loss": 0.1557,
      "step": 7425
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4233744210739732,
      "learning_rate": 6.389878263775129e-08,
      "loss": 0.2153,
      "step": 7426
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4695585819640937,
      "learning_rate": 6.356883215819287e-08,
      "loss": 0.1576,
      "step": 7427
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2435532995021288,
      "learning_rate": 6.323973030838037e-08,
      "loss": 0.1739,
      "step": 7428
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.759253280828082,
      "learning_rate": 6.291147714488965e-08,
      "loss": 0.2119,
      "step": 7429
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3581137099567087,
      "learning_rate": 6.258407272415223e-08,
      "loss": 0.1893,
      "step": 7430
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5767160597821097,
      "learning_rate": 6.225751710245198e-08,
      "loss": 0.1704,
      "step": 7431
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3775654810443971,
      "learning_rate": 6.193181033592788e-08,
      "loss": 0.1695,
      "step": 7432
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.363088367570273,
      "learning_rate": 6.160695248057236e-08,
      "loss": 0.5221,
      "step": 7433
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.402252013733205,
      "learning_rate": 6.128294359223297e-08,
      "loss": 0.5104,
      "step": 7434
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3862685590947474,
      "learning_rate": 6.09597837266096e-08,
      "loss": 0.1757,
      "step": 7435
    },
    {
      "epoch": 0.95,
      "grad_norm": 8.653759867273031,
      "learning_rate": 6.063747293925781e-08,
      "loss": 0.3999,
      "step": 7436
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.721134059348121,
      "learning_rate": 6.031601128558606e-08,
      "loss": 0.1665,
      "step": 7437
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3993708433716634,
      "learning_rate": 5.999539882085793e-08,
      "loss": 0.1849,
      "step": 7438
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.381567714040119,
      "learning_rate": 5.96756356001904e-08,
      "loss": 0.1739,
      "step": 7439
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.272465995100013,
      "learning_rate": 5.9356721678554554e-08,
      "loss": 0.1746,
      "step": 7440
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.646953097245007,
      "learning_rate": 5.903865711077483e-08,
      "loss": 0.4485,
      "step": 7441
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3415668366051892,
      "learning_rate": 5.872144195153029e-08,
      "loss": 0.1504,
      "step": 7442
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.303628268700507,
      "learning_rate": 5.840507625535397e-08,
      "loss": 0.1667,
      "step": 7443
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3661611200154715,
      "learning_rate": 5.808956007663291e-08,
      "loss": 0.2034,
      "step": 7444
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3203281014454287,
      "learning_rate": 5.777489346960874e-08,
      "loss": 0.1653,
      "step": 7445
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3801890627222977,
      "learning_rate": 5.7461076488374844e-08,
      "loss": 0.1821,
      "step": 7446
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.378895750439561,
      "learning_rate": 5.7148109186880854e-08,
      "loss": 0.1898,
      "step": 7447
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.43720378075161,
      "learning_rate": 5.683599161892928e-08,
      "loss": 0.1654,
      "step": 7448
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4520167684640948,
      "learning_rate": 5.65247238381772e-08,
      "loss": 0.1673,
      "step": 7449
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3754453001066538,
      "learning_rate": 5.621430589813459e-08,
      "loss": 0.1742,
      "step": 7450
    },
    {
      "epoch": 0.95,
      "grad_norm": 11.783964998570683,
      "learning_rate": 5.5904737852166545e-08,
      "loss": 0.4998,
      "step": 7451
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.389149629539519,
      "learning_rate": 5.559601975348994e-08,
      "loss": 0.1838,
      "step": 7452
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5550025773280598,
      "learning_rate": 5.5288151655178427e-08,
      "loss": 0.1781,
      "step": 7453
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.562911426972293,
      "learning_rate": 5.4981133610158e-08,
      "loss": 0.1858,
      "step": 7454
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4222741785199768,
      "learning_rate": 5.4674965671208115e-08,
      "loss": 0.1693,
      "step": 7455
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.506612887031328,
      "learning_rate": 5.436964789096222e-08,
      "loss": 0.2117,
      "step": 7456
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3588618506007029,
      "learning_rate": 5.406518032190944e-08,
      "loss": 0.1701,
      "step": 7457
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.337636100975719,
      "learning_rate": 5.3761563016389576e-08,
      "loss": 0.1725,
      "step": 7458
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3878509207945482,
      "learning_rate": 5.34587960265992e-08,
      "loss": 0.1529,
      "step": 7459
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4682616033241473,
      "learning_rate": 5.315687940458669e-08,
      "loss": 0.1911,
      "step": 7460
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4452078493646416,
      "learning_rate": 5.285581320225552e-08,
      "loss": 0.2034,
      "step": 7461
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2936674741066674,
      "learning_rate": 5.255559747136263e-08,
      "loss": 0.1621,
      "step": 7462
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4249648032261522,
      "learning_rate": 5.2256232263517835e-08,
      "loss": 0.1933,
      "step": 7463
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.373356925611359,
      "learning_rate": 5.195771763018609e-08,
      "loss": 0.5741,
      "step": 7464
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3892442426750662,
      "learning_rate": 5.166005362268467e-08,
      "loss": 0.1587,
      "step": 7465
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.582156984113007,
      "learning_rate": 5.1363240292186535e-08,
      "loss": 0.2229,
      "step": 7466
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5606311778904463,
      "learning_rate": 5.1067277689716974e-08,
      "loss": 0.2326,
      "step": 7467
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5555994892506084,
      "learning_rate": 5.07721658661553e-08,
      "loss": 0.1756,
      "step": 7468
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4770495648038318,
      "learning_rate": 5.0477904872234804e-08,
      "loss": 0.2028,
      "step": 7469
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.213507713058505,
      "learning_rate": 5.018449475854226e-08,
      "loss": 0.6122,
      "step": 7470
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3815338778664086,
      "learning_rate": 4.9891935575517856e-08,
      "loss": 0.1462,
      "step": 7471
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5872181223962927,
      "learning_rate": 4.9600227373456936e-08,
      "loss": 0.1915,
      "step": 7472
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3407518012268023,
      "learning_rate": 4.930937020250604e-08,
      "loss": 0.1671,
      "step": 7473
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.605889682774534,
      "learning_rate": 4.90193641126685e-08,
      "loss": 0.3506,
      "step": 7474
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.387204253204569,
      "learning_rate": 4.873020915379834e-08,
      "loss": 0.1766,
      "step": 7475
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5382637089860693,
      "learning_rate": 4.844190537560578e-08,
      "loss": 0.1926,
      "step": 7476
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.672603845146077,
      "learning_rate": 4.8154452827652854e-08,
      "loss": 0.2239,
      "step": 7477
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.192388367252599,
      "learning_rate": 4.7867851559356694e-08,
      "loss": 0.1364,
      "step": 7478
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6290450938322092,
      "learning_rate": 4.758210161998622e-08,
      "loss": 0.1853,
      "step": 7479
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.942990510234324,
      "learning_rate": 4.729720305866603e-08,
      "loss": 0.3866,
      "step": 7480
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.869325377800084,
      "learning_rate": 4.701315592437361e-08,
      "loss": 0.4668,
      "step": 7481
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4117047381280632,
      "learning_rate": 4.6729960265939344e-08,
      "loss": 0.1686,
      "step": 7482
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.408780945297235,
      "learning_rate": 4.644761613204818e-08,
      "loss": 0.2042,
      "step": 7483
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5673276298483472,
      "learning_rate": 4.6166123571237955e-08,
      "loss": 0.1374,
      "step": 7484
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4873753935057001,
      "learning_rate": 4.588548263190107e-08,
      "loss": 0.1801,
      "step": 7485
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2898599015937307,
      "learning_rate": 4.560569336228338e-08,
      "loss": 0.1684,
      "step": 7486
    },
    {
      "epoch": 0.96,
      "grad_norm": 9.109787473266058,
      "learning_rate": 4.5326755810482514e-08,
      "loss": 0.6744,
      "step": 7487
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4485145613914676,
      "learning_rate": 4.504867002445179e-08,
      "loss": 0.1579,
      "step": 7488
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.738859829633786,
      "learning_rate": 4.477143605199796e-08,
      "loss": 0.491,
      "step": 7489
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3123194250848842,
      "learning_rate": 4.449505394078013e-08,
      "loss": 0.1772,
      "step": 7490
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4588826114190259,
      "learning_rate": 4.4219523738311396e-08,
      "loss": 0.1776,
      "step": 7491
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3498717566928928,
      "learning_rate": 4.3944845491958874e-08,
      "loss": 0.204,
      "step": 7492
    },
    {
      "epoch": 0.96,
      "grad_norm": 8.415264214438416,
      "learning_rate": 4.3671019248943126e-08,
      "loss": 0.613,
      "step": 7493
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4857941204420835,
      "learning_rate": 4.3398045056337604e-08,
      "loss": 0.1917,
      "step": 7494
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2739577547669165,
      "learning_rate": 4.312592296106977e-08,
      "loss": 0.129,
      "step": 7495
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2869584911157184,
      "learning_rate": 4.285465300992164e-08,
      "loss": 0.1573,
      "step": 7496
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.405778597793974,
      "learning_rate": 4.258423524952648e-08,
      "loss": 0.1911,
      "step": 7497
    },
    {
      "epoch": 0.96,
      "grad_norm": 8.874487291137314,
      "learning_rate": 4.231466972637211e-08,
      "loss": 0.7176,
      "step": 7498
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3971653350984328,
      "learning_rate": 4.2045956486800877e-08,
      "loss": 0.1866,
      "step": 7499
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4618620694326565,
      "learning_rate": 4.177809557700752e-08,
      "loss": 0.1421,
      "step": 7500
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.848382532836559,
      "learning_rate": 4.1511087043039635e-08,
      "loss": 0.4446,
      "step": 7501
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6540980455984582,
      "learning_rate": 4.12449309307994e-08,
      "loss": 0.224,
      "step": 7502
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.778330560358318,
      "learning_rate": 4.097962728604299e-08,
      "loss": 0.584,
      "step": 7503
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3607489242220259,
      "learning_rate": 4.0715176154378366e-08,
      "loss": 0.1719,
      "step": 7504
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3443619046680364,
      "learning_rate": 4.04515775812675e-08,
      "loss": 0.1467,
      "step": 7505
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4362370916835214,
      "learning_rate": 4.018883161202691e-08,
      "loss": 0.1739,
      "step": 7506
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.910663940531946,
      "learning_rate": 3.992693829182548e-08,
      "loss": 0.6649,
      "step": 7507
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.156318311868486,
      "learning_rate": 3.966589766568496e-08,
      "loss": 0.6223,
      "step": 7508
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5310105480509193,
      "learning_rate": 3.940570977848168e-08,
      "loss": 0.1462,
      "step": 7509
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.500858414047923,
      "learning_rate": 3.914637467494542e-08,
      "loss": 0.1468,
      "step": 7510
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5800344348822355,
      "learning_rate": 3.888789239965885e-08,
      "loss": 0.1575,
      "step": 7511
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3695380788335305,
      "learning_rate": 3.863026299705697e-08,
      "loss": 0.157,
      "step": 7512
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2513011376744685,
      "learning_rate": 3.8373486511429916e-08,
      "loss": 0.128,
      "step": 7513
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4068598796898117,
      "learning_rate": 3.811756298692126e-08,
      "loss": 0.2151,
      "step": 7514
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.413704624461423,
      "learning_rate": 3.7862492467526376e-08,
      "loss": 0.1972,
      "step": 7515
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.626410126874353,
      "learning_rate": 3.7608274997095187e-08,
      "loss": 0.5926,
      "step": 7516
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4209187439128848,
      "learning_rate": 3.735491061932994e-08,
      "loss": 0.1967,
      "step": 7517
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3729446320971208,
      "learning_rate": 3.710239937778803e-08,
      "loss": 0.1516,
      "step": 7518
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8695689218079135,
      "learning_rate": 3.685074131587863e-08,
      "loss": 0.165,
      "step": 7519
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2855975981416292,
      "learning_rate": 3.6599936476864325e-08,
      "loss": 0.1968,
      "step": 7520
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.637455130585908,
      "learning_rate": 3.6349984903861214e-08,
      "loss": 0.4185,
      "step": 7521
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.441722229227608,
      "learning_rate": 3.6100886639839904e-08,
      "loss": 0.1615,
      "step": 7522
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2685223714687446,
      "learning_rate": 3.5852641727622264e-08,
      "loss": 0.1801,
      "step": 7523
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4939435055510566,
      "learning_rate": 3.5605250209885256e-08,
      "loss": 0.1807,
      "step": 7524
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2422122365740593,
      "learning_rate": 3.535871212915765e-08,
      "loss": 0.1582,
      "step": 7525
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8947618485588764,
      "learning_rate": 3.51130275278222e-08,
      "loss": 0.2131,
      "step": 7526
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3714757690911965,
      "learning_rate": 3.4868196448115675e-08,
      "loss": 0.1511,
      "step": 7527
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.414691583685449,
      "learning_rate": 3.46242189321272e-08,
      "loss": 0.1943,
      "step": 7528
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3902972859449925,
      "learning_rate": 3.4381095021798203e-08,
      "loss": 0.1658,
      "step": 7529
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4498055065163709,
      "learning_rate": 3.4138824758925826e-08,
      "loss": 0.1764,
      "step": 7530
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.528688740294079,
      "learning_rate": 3.389740818515841e-08,
      "loss": 0.1875,
      "step": 7531
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.497016582383866,
      "learning_rate": 3.3656845341998865e-08,
      "loss": 0.5536,
      "step": 7532
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4155887309675574,
      "learning_rate": 3.341713627080245e-08,
      "loss": 0.2388,
      "step": 7533
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3929782341870893,
      "learning_rate": 3.317828101277787e-08,
      "loss": 0.1694,
      "step": 7534
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3730301011622297,
      "learning_rate": 3.2940279608986714e-08,
      "loss": 0.1597,
      "step": 7535
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5470094952354037,
      "learning_rate": 3.270313210034515e-08,
      "loss": 0.2105,
      "step": 7536
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.518433895794553,
      "learning_rate": 3.246683852762056e-08,
      "loss": 0.1477,
      "step": 7537
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5419575247199058,
      "learning_rate": 3.22313989314349e-08,
      "loss": 0.1639,
      "step": 7538
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4478571427222966,
      "learning_rate": 3.199681335226357e-08,
      "loss": 0.2054,
      "step": 7539
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.372924921123363,
      "learning_rate": 3.176308183043375e-08,
      "loss": 0.5561,
      "step": 7540
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3009706345001384,
      "learning_rate": 3.1530204406127196e-08,
      "loss": 0.1462,
      "step": 7541
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3038017430921336,
      "learning_rate": 3.129818111937744e-08,
      "loss": 0.1539,
      "step": 7542
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.605627677228853,
      "learning_rate": 3.1067012010073114e-08,
      "loss": 0.651,
      "step": 7543
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3791988266830084,
      "learning_rate": 3.0836697117954115e-08,
      "loss": 0.1583,
      "step": 7544
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3097020295586324,
      "learning_rate": 3.0607236482613764e-08,
      "loss": 0.1557,
      "step": 7545
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.873683751511482,
      "learning_rate": 3.037863014349995e-08,
      "loss": 0.55,
      "step": 7546
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8231498242772162,
      "learning_rate": 3.0150878139912906e-08,
      "loss": 0.1738,
      "step": 7547
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5617621904669108,
      "learning_rate": 2.9923980511004645e-08,
      "loss": 0.1483,
      "step": 7548
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4026984086912717,
      "learning_rate": 2.96979372957823e-08,
      "loss": 0.1966,
      "step": 7549
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.442191242959137,
      "learning_rate": 2.947274853310589e-08,
      "loss": 0.5808,
      "step": 7550
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3288779519489888,
      "learning_rate": 2.9248414261686674e-08,
      "loss": 0.1453,
      "step": 7551
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3567507513302866,
      "learning_rate": 2.902493452009103e-08,
      "loss": 0.1722,
      "step": 7552
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.343875853430142,
      "learning_rate": 2.8802309346737666e-08,
      "loss": 0.6156,
      "step": 7553
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.597161278416525,
      "learning_rate": 2.8580538779898192e-08,
      "loss": 0.2135,
      "step": 7554
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2735164810376092,
      "learning_rate": 2.8359622857698223e-08,
      "loss": 0.1501,
      "step": 7555
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.507699894465373,
      "learning_rate": 2.813956161811515e-08,
      "loss": 0.481,
      "step": 7556
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3791146760476398,
      "learning_rate": 2.7920355098979835e-08,
      "loss": 0.1899,
      "step": 7557
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.426482698702042,
      "learning_rate": 2.7702003337977124e-08,
      "loss": 0.1867,
      "step": 7558
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3754947821119063,
      "learning_rate": 2.748450637264366e-08,
      "loss": 0.2062,
      "step": 7559
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.622249787810376,
      "learning_rate": 2.7267864240369533e-08,
      "loss": 0.1904,
      "step": 7560
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.729699748559469,
      "learning_rate": 2.7052076978398844e-08,
      "loss": 0.486,
      "step": 7561
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3829214291577632,
      "learning_rate": 2.683714462382747e-08,
      "loss": 0.1839,
      "step": 7562
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5347921588824103,
      "learning_rate": 2.6623067213605302e-08,
      "loss": 0.1716,
      "step": 7563
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4745198646773323,
      "learning_rate": 2.6409844784533456e-08,
      "loss": 0.1916,
      "step": 7564
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4622405709621555,
      "learning_rate": 2.6197477373268722e-08,
      "loss": 0.1755,
      "step": 7565
    },
    {
      "epoch": 0.97,
      "grad_norm": 8.181817631728896,
      "learning_rate": 2.5985965016318004e-08,
      "loss": 0.5385,
      "step": 7566
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6364892884557922,
      "learning_rate": 2.577530775004389e-08,
      "loss": 0.1885,
      "step": 7567
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4006495992146055,
      "learning_rate": 2.5565505610660734e-08,
      "loss": 0.1528,
      "step": 7568
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2731289543419806,
      "learning_rate": 2.5356558634235806e-08,
      "loss": 0.1481,
      "step": 7569
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2696772940880399,
      "learning_rate": 2.5148466856689812e-08,
      "loss": 0.1407,
      "step": 7570
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.529312821278451,
      "learning_rate": 2.4941230313795252e-08,
      "loss": 0.1828,
      "step": 7571
    },
    {
      "epoch": 0.97,
      "grad_norm": 8.819566622903533,
      "learning_rate": 2.4734849041179176e-08,
      "loss": 0.5027,
      "step": 7572
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.5173817367710996,
      "learning_rate": 2.4529323074320988e-08,
      "loss": 0.4854,
      "step": 7573
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.442479252774683,
      "learning_rate": 2.432465244855242e-08,
      "loss": 0.15,
      "step": 7574
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4381619727671393,
      "learning_rate": 2.412083719905922e-08,
      "loss": 0.1929,
      "step": 7575
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.7294234216728355,
      "learning_rate": 2.3917877360879472e-08,
      "loss": 0.4577,
      "step": 7576
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5380739613175998,
      "learning_rate": 2.371577296890415e-08,
      "loss": 0.1732,
      "step": 7577
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.560208218431367,
      "learning_rate": 2.3514524057877685e-08,
      "loss": 0.1811,
      "step": 7578
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3844607310030426,
      "learning_rate": 2.331413066239685e-08,
      "loss": 0.1869,
      "step": 7579
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2473660704962868,
      "learning_rate": 2.311459281691186e-08,
      "loss": 0.149,
      "step": 7580
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.561186245554992,
      "learning_rate": 2.2915910555725286e-08,
      "loss": 0.1907,
      "step": 7581
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4226048779526121,
      "learning_rate": 2.2718083912992573e-08,
      "loss": 0.1968,
      "step": 7582
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3917390084668382,
      "learning_rate": 2.2521112922723186e-08,
      "loss": 0.1826,
      "step": 7583
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6126376017233908,
      "learning_rate": 2.2324997618778375e-08,
      "loss": 0.2189,
      "step": 7584
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2533189228883388,
      "learning_rate": 2.212973803487284e-08,
      "loss": 0.1574,
      "step": 7585
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.419753990718296,
      "learning_rate": 2.1935334204573056e-08,
      "loss": 0.1917,
      "step": 7586
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.602438996223301,
      "learning_rate": 2.1741786161300628e-08,
      "loss": 0.6179,
      "step": 7587
    },
    {
      "epoch": 0.97,
      "grad_norm": 12.062103412398832,
      "learning_rate": 2.1549093938327826e-08,
      "loss": 0.4655,
      "step": 7588
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6025136994212106,
      "learning_rate": 2.135725756878093e-08,
      "loss": 0.1688,
      "step": 7589
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3896920554322236,
      "learning_rate": 2.116627708563912e-08,
      "loss": 0.189,
      "step": 7590
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4468841087105608,
      "learning_rate": 2.0976152521733905e-08,
      "loss": 0.1325,
      "step": 7591
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5932298852397586,
      "learning_rate": 2.07868839097497e-08,
      "loss": 0.1946,
      "step": 7592
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3050608279306903,
      "learning_rate": 2.059847128222381e-08,
      "loss": 0.1892,
      "step": 7593
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2065310987661018,
      "learning_rate": 2.041091467154699e-08,
      "loss": 0.142,
      "step": 7594
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3954370219405763,
      "learning_rate": 2.022421410996234e-08,
      "loss": 0.1828,
      "step": 7595
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.059984539702558,
      "learning_rate": 2.0038369629565846e-08,
      "loss": 0.6903,
      "step": 7596
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5804098527470256,
      "learning_rate": 1.9853381262306405e-08,
      "loss": 0.2037,
      "step": 7597
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6210744103591115,
      "learning_rate": 1.9669249039985794e-08,
      "loss": 0.1748,
      "step": 7598
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.330600706366939,
      "learning_rate": 1.9485972994257584e-08,
      "loss": 0.1682,
      "step": 7599
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4479382105309966,
      "learning_rate": 1.9303553156630462e-08,
      "loss": 0.1911,
      "step": 7600
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6305505029509615,
      "learning_rate": 1.912198955846323e-08,
      "loss": 0.1716,
      "step": 7601
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5312870146821875,
      "learning_rate": 1.894128223096925e-08,
      "loss": 0.1865,
      "step": 7602
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3692951044931703,
      "learning_rate": 1.8761431205214232e-08,
      "loss": 0.1564,
      "step": 7603
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.729293417720451,
      "learning_rate": 1.8582436512116776e-08,
      "loss": 0.1539,
      "step": 7604
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.535570954649474,
      "learning_rate": 1.8404298182447823e-08,
      "loss": 0.4416,
      "step": 7605
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.596918632132728,
      "learning_rate": 1.8227016246831764e-08,
      "loss": 0.3938,
      "step": 7606
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.979617614287663,
      "learning_rate": 1.8050590735745334e-08,
      "loss": 0.5936,
      "step": 7607
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5258115314146992,
      "learning_rate": 1.7875021679518156e-08,
      "loss": 0.1947,
      "step": 7608
    },
    {
      "epoch": 0.97,
      "grad_norm": 11.1253649874171,
      "learning_rate": 1.7700309108332204e-08,
      "loss": 0.59,
      "step": 7609
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.47765873991636,
      "learning_rate": 1.7526453052223446e-08,
      "loss": 0.4718,
      "step": 7610
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.586283678263755,
      "learning_rate": 1.7353453541078534e-08,
      "loss": 0.1751,
      "step": 7611
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.303310938129498,
      "learning_rate": 1.7181310604639236e-08,
      "loss": 0.155,
      "step": 7612
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4499386731711925,
      "learning_rate": 1.701002427249854e-08,
      "loss": 0.1954,
      "step": 7613
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5375248393018623,
      "learning_rate": 1.683959457410178e-08,
      "loss": 0.1636,
      "step": 7614
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.211951803734724,
      "learning_rate": 1.667002153874886e-08,
      "loss": 0.1407,
      "step": 7615
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3750260608412064,
      "learning_rate": 1.65013051955909e-08,
      "loss": 0.1725,
      "step": 7616
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3687176438969437,
      "learning_rate": 1.6333445573632478e-08,
      "loss": 0.1568,
      "step": 7617
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.231452887543637,
      "learning_rate": 1.6166442701730513e-08,
      "loss": 0.1663,
      "step": 7618
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3410252909770923,
      "learning_rate": 1.6000296608594257e-08,
      "loss": 0.1658,
      "step": 7619
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.549112328566621,
      "learning_rate": 1.5835007322786424e-08,
      "loss": 0.2078,
      "step": 7620
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6575390194793531,
      "learning_rate": 1.567057487272261e-08,
      "loss": 0.185,
      "step": 7621
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4742243585972972,
      "learning_rate": 1.5506999286669656e-08,
      "loss": 0.1739,
      "step": 7622
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2903939693903792,
      "learning_rate": 1.534428059274895e-08,
      "loss": 0.1719,
      "step": 7623
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5403513128764612,
      "learning_rate": 1.5182418818933676e-08,
      "loss": 0.1742,
      "step": 7624
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.823317885359091,
      "learning_rate": 1.502141399304935e-08,
      "loss": 0.5157,
      "step": 7625
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.345409410059383,
      "learning_rate": 1.4861266142775498e-08,
      "loss": 0.4768,
      "step": 7626
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5566259004717413,
      "learning_rate": 1.4701975295641768e-08,
      "loss": 0.1421,
      "step": 7627
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.260790600315459,
      "learning_rate": 1.4543541479033473e-08,
      "loss": 0.1545,
      "step": 7628
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4244275318376558,
      "learning_rate": 1.4385964720187162e-08,
      "loss": 0.1537,
      "step": 7629
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4203801164517182,
      "learning_rate": 1.4229245046190609e-08,
      "loss": 0.1456,
      "step": 7630
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4372443019742263,
      "learning_rate": 1.4073382483987819e-08,
      "loss": 0.1863,
      "step": 7631
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3806431188123613,
      "learning_rate": 1.3918377060371802e-08,
      "loss": 0.1644,
      "step": 7632
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.109025690699392,
      "learning_rate": 1.3764228801990686e-08,
      "loss": 0.4525,
      "step": 7633
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5567796885375982,
      "learning_rate": 1.3610937735344387e-08,
      "loss": 0.156,
      "step": 7634
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.218927670914461,
      "learning_rate": 1.3458503886784603e-08,
      "loss": 0.1454,
      "step": 7635
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5314834469575618,
      "learning_rate": 1.3306927282517034e-08,
      "loss": 0.1582,
      "step": 7636
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4014825117782284,
      "learning_rate": 1.3156207948599176e-08,
      "loss": 0.1859,
      "step": 7637
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4298675743851725,
      "learning_rate": 1.3006345910941964e-08,
      "loss": 0.1761,
      "step": 7638
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2576577940061358,
      "learning_rate": 1.2857341195308126e-08,
      "loss": 0.1685,
      "step": 7639
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7229977068604696,
      "learning_rate": 1.2709193827312727e-08,
      "loss": 0.1839,
      "step": 7640
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2724760886901783,
      "learning_rate": 1.2561903832424837e-08,
      "loss": 0.1518,
      "step": 7641
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.227684894065117,
      "learning_rate": 1.2415471235965315e-08,
      "loss": 0.5798,
      "step": 7642
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7170143812820202,
      "learning_rate": 1.2269896063107356e-08,
      "loss": 0.2051,
      "step": 7643
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3333634850203553,
      "learning_rate": 1.2125178338876498e-08,
      "loss": 0.1739,
      "step": 7644
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.170140818420045,
      "learning_rate": 1.1981318088152283e-08,
      "loss": 0.7196,
      "step": 7645
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.629572813832975,
      "learning_rate": 1.1838315335664929e-08,
      "loss": 0.226,
      "step": 7646
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2448959271026676,
      "learning_rate": 1.1696170105999772e-08,
      "loss": 0.1637,
      "step": 7647
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.7243721125155025,
      "learning_rate": 1.1554882423591706e-08,
      "loss": 0.4445,
      "step": 7648
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.509620588113506,
      "learning_rate": 1.141445231273075e-08,
      "loss": 0.1693,
      "step": 7649
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4195933320499716,
      "learning_rate": 1.1274879797558148e-08,
      "loss": 0.1847,
      "step": 7650
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5412265467365016,
      "learning_rate": 1.1136164902067481e-08,
      "loss": 0.2225,
      "step": 7651
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.365680351365322,
      "learning_rate": 1.0998307650106344e-08,
      "loss": 0.1561,
      "step": 7652
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6078406042577449,
      "learning_rate": 1.0861308065373556e-08,
      "loss": 0.224,
      "step": 7653
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.227049034951721,
      "learning_rate": 1.072516617142083e-08,
      "loss": 0.1565,
      "step": 7654
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6346933856783021,
      "learning_rate": 1.0589881991652784e-08,
      "loss": 0.2092,
      "step": 7655
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5608433838592608,
      "learning_rate": 1.0455455549326366e-08,
      "loss": 0.1666,
      "step": 7656
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4148967158137697,
      "learning_rate": 1.0321886867550868e-08,
      "loss": 0.158,
      "step": 7657
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.9004831508374735,
      "learning_rate": 1.0189175969287923e-08,
      "loss": 0.4855,
      "step": 7658
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5020948821041957,
      "learning_rate": 1.0057322877352616e-08,
      "loss": 0.1925,
      "step": 7659
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4653987192804516,
      "learning_rate": 9.926327614411813e-09,
      "loss": 0.1685,
      "step": 7660
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.568414783652443,
      "learning_rate": 9.796190202985834e-09,
      "loss": 0.1749,
      "step": 7661
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2075443145993248,
      "learning_rate": 9.666910665445673e-09,
      "loss": 0.1437,
      "step": 7662
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.439000044287341,
      "learning_rate": 9.538489024016328e-09,
      "loss": 0.2176,
      "step": 7663
    },
    {
      "epoch": 0.98,
      "grad_norm": 9.025414324820268,
      "learning_rate": 9.410925300775697e-09,
      "loss": 0.5355,
      "step": 7664
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3936593336804446,
      "learning_rate": 9.284219517652348e-09,
      "loss": 0.1662,
      "step": 7665
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.37926165665394,
      "learning_rate": 9.158371696428859e-09,
      "loss": 0.1444,
      "step": 7666
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.443783594077947,
      "learning_rate": 9.033381858740697e-09,
      "loss": 0.1822,
      "step": 7667
    },
    {
      "epoch": 0.98,
      "grad_norm": 20.941613129932293,
      "learning_rate": 8.909250026074013e-09,
      "loss": 0.4705,
      "step": 7668
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.801607671290678,
      "learning_rate": 8.785976219768954e-09,
      "loss": 0.5624,
      "step": 7669
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3403565890179605,
      "learning_rate": 8.663560461018016e-09,
      "loss": 0.1548,
      "step": 7670
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2296962552561548,
      "learning_rate": 8.542002770865477e-09,
      "loss": 0.169,
      "step": 7671
    },
    {
      "epoch": 0.98,
      "grad_norm": 8.897643359074278,
      "learning_rate": 8.421303170208505e-09,
      "loss": 0.5029,
      "step": 7672
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4240506679841456,
      "learning_rate": 8.301461679796619e-09,
      "loss": 0.1801,
      "step": 7673
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3561484405704873,
      "learning_rate": 8.182478320232223e-09,
      "loss": 0.1628,
      "step": 7674
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5199280328750229,
      "learning_rate": 8.06435311197007e-09,
      "loss": 0.1817,
      "step": 7675
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4866578456462691,
      "learning_rate": 7.947086075317246e-09,
      "loss": 0.1923,
      "step": 7676
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5280536023745375,
      "learning_rate": 7.830677230433181e-09,
      "loss": 0.174,
      "step": 7677
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3956272830439662,
      "learning_rate": 7.715126597329648e-09,
      "loss": 0.1839,
      "step": 7678
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.886159050119079,
      "learning_rate": 7.600434195871864e-09,
      "loss": 0.6513,
      "step": 7679
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4902353729091513,
      "learning_rate": 7.486600045775728e-09,
      "loss": 0.1885,
      "step": 7680
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3971724805917807,
      "learning_rate": 7.373624166611137e-09,
      "loss": 0.172,
      "step": 7681
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3758846429166427,
      "learning_rate": 7.261506577800892e-09,
      "loss": 0.1873,
      "step": 7682
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.544486835646046,
      "learning_rate": 7.1502472986179075e-09,
      "loss": 0.2253,
      "step": 7683
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4926524758479363,
      "learning_rate": 7.039846348189105e-09,
      "loss": 0.2191,
      "step": 7684
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4560950687471756,
      "learning_rate": 6.93030374549486e-09,
      "loss": 0.1705,
      "step": 7685
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.920351689793749,
      "learning_rate": 6.8216195093656624e-09,
      "loss": 0.5087,
      "step": 7686
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4302006831979275,
      "learning_rate": 6.713793658486012e-09,
      "loss": 0.1665,
      "step": 7687
    },
    {
      "epoch": 0.98,
      "grad_norm": 15.55323676921827,
      "learning_rate": 6.606826211392192e-09,
      "loss": 0.6275,
      "step": 7688
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.536637445009589,
      "learning_rate": 6.5007171864733824e-09,
      "loss": 0.5954,
      "step": 7689
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3624441573844361,
      "learning_rate": 6.395466601971101e-09,
      "loss": 0.1691,
      "step": 7690
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4509277741910742,
      "learning_rate": 6.291074475978653e-09,
      "loss": 0.165,
      "step": 7691
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5679405335825944,
      "learning_rate": 6.187540826442795e-09,
      "loss": 0.1736,
      "step": 7692
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4877196797939782,
      "learning_rate": 6.084865671162066e-09,
      "loss": 0.1957,
      "step": 7693
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.711202298683294,
      "learning_rate": 5.983049027786791e-09,
      "loss": 0.6063,
      "step": 7694
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.08610583906763,
      "learning_rate": 5.882090913821303e-09,
      "loss": 0.5397,
      "step": 7695
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4128589804959297,
      "learning_rate": 5.781991346621163e-09,
      "loss": 0.1778,
      "step": 7696
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.475809538280694,
      "learning_rate": 5.682750343394827e-09,
      "loss": 0.1446,
      "step": 7697
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.557250669564808,
      "learning_rate": 5.584367921202538e-09,
      "loss": 0.1896,
      "step": 7698
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1640217934200827,
      "learning_rate": 5.486844096957988e-09,
      "loss": 0.1656,
      "step": 7699
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4026611961986375,
      "learning_rate": 5.390178887426101e-09,
      "loss": 0.1517,
      "step": 7700
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.09997926605155,
      "learning_rate": 5.294372309224693e-09,
      "loss": 0.5869,
      "step": 7701
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2844894305928924,
      "learning_rate": 5.19942437882448e-09,
      "loss": 0.1421,
      "step": 7702
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4100118323009,
      "learning_rate": 5.10533511254796e-09,
      "loss": 0.1669,
      "step": 7703
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.237990410286268,
      "learning_rate": 5.012104526569972e-09,
      "loss": 0.496,
      "step": 7704
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.397867573456471,
      "learning_rate": 4.919732636918251e-09,
      "loss": 0.1539,
      "step": 7705
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.275265727138,
      "learning_rate": 4.8282194594723164e-09,
      "loss": 0.1438,
      "step": 7706
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4720771409711924,
      "learning_rate": 4.737565009964584e-09,
      "loss": 0.1758,
      "step": 7707
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4158451755021293,
      "learning_rate": 4.647769303979255e-09,
      "loss": 0.1849,
      "step": 7708
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.464524793339406,
      "learning_rate": 4.558832356953979e-09,
      "loss": 0.1783,
      "step": 7709
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7726800446406055,
      "learning_rate": 4.470754184177084e-09,
      "loss": 0.1776,
      "step": 7710
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5450403910012755,
      "learning_rate": 4.383534800790901e-09,
      "loss": 0.2243,
      "step": 7711
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.394359182155885,
      "learning_rate": 4.297174221788991e-09,
      "loss": 0.1497,
      "step": 7712
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.841470338078144,
      "learning_rate": 4.211672462018368e-09,
      "loss": 0.5349,
      "step": 7713
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4348975643176893,
      "learning_rate": 4.127029536177274e-09,
      "loss": 0.1952,
      "step": 7714
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2941318369326174,
      "learning_rate": 4.043245458816847e-09,
      "loss": 0.1952,
      "step": 7715
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3421192465078315,
      "learning_rate": 3.960320244340565e-09,
      "loss": 0.1793,
      "step": 7716
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.5939925411645515,
      "learning_rate": 3.878253907004248e-09,
      "loss": 0.5207,
      "step": 7717
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5503286605021556,
      "learning_rate": 3.797046460916054e-09,
      "loss": 0.2027,
      "step": 7718
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2488279802919795,
      "learning_rate": 3.716697920036483e-09,
      "loss": 0.1631,
      "step": 7719
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5673347183762933,
      "learning_rate": 3.637208298178929e-09,
      "loss": 0.2066,
      "step": 7720
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.712086018334653,
      "learning_rate": 3.558577609007463e-09,
      "loss": 0.2059,
      "step": 7721
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.487641616461558,
      "learning_rate": 3.480805866040715e-09,
      "loss": 0.1951,
      "step": 7722
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4114972924702687,
      "learning_rate": 3.403893082647991e-09,
      "loss": 0.1694,
      "step": 7723
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4027270320650962,
      "learning_rate": 3.3278392720514917e-09,
      "loss": 0.1882,
      "step": 7724
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.5806409518244005,
      "learning_rate": 3.252644447325759e-09,
      "loss": 0.7524,
      "step": 7725
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2950392460894715,
      "learning_rate": 3.1783086213976743e-09,
      "loss": 0.1902,
      "step": 7726
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3256722264798926,
      "learning_rate": 3.10483180704646e-09,
      "loss": 0.13,
      "step": 7727
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.299635982309893,
      "learning_rate": 3.0322140169036784e-09,
      "loss": 0.1632,
      "step": 7728
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2262500654725303,
      "learning_rate": 2.9604552634532325e-09,
      "loss": 0.1399,
      "step": 7729
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5748158920069217,
      "learning_rate": 2.88955555903081e-09,
      "loss": 0.2076,
      "step": 7730
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.520152709368854,
      "learning_rate": 2.81951491582555e-09,
      "loss": 0.2037,
      "step": 7731
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5220130888261236,
      "learning_rate": 2.7503333458778205e-09,
      "loss": 0.2163,
      "step": 7732
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.809482792218896,
      "learning_rate": 2.682010861080886e-09,
      "loss": 0.6854,
      "step": 7733
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4844443192327736,
      "learning_rate": 2.614547473180351e-09,
      "loss": 0.1838,
      "step": 7734
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.695527044573771,
      "learning_rate": 2.5479431937736055e-09,
      "loss": 0.2045,
      "step": 7735
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3691127692461516,
      "learning_rate": 2.482198034310934e-09,
      "loss": 0.1558,
      "step": 7736
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5555858643126679,
      "learning_rate": 2.417312006094963e-09,
      "loss": 0.1783,
      "step": 7737
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.335865451478352,
      "learning_rate": 2.353285120279547e-09,
      "loss": 0.1648,
      "step": 7738
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.478906955891998,
      "learning_rate": 2.290117387872548e-09,
      "loss": 0.1902,
      "step": 7739
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3588538926847198,
      "learning_rate": 2.227808819732502e-09,
      "loss": 0.1883,
      "step": 7740
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4204023404462878,
      "learning_rate": 2.166359426570841e-09,
      "loss": 0.1529,
      "step": 7741
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5556894542491826,
      "learning_rate": 2.105769218952447e-09,
      "loss": 0.1692,
      "step": 7742
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4908764022221348,
      "learning_rate": 2.0460382072928775e-09,
      "loss": 0.1946,
      "step": 7743
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6092817009998697,
      "learning_rate": 1.98716640186003e-09,
      "loss": 0.165,
      "step": 7744
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2880020282733904,
      "learning_rate": 1.929153812775808e-09,
      "loss": 0.1632,
      "step": 7745
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6384733377160374,
      "learning_rate": 1.8720004500122347e-09,
      "loss": 0.1959,
      "step": 7746
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5879796339622894,
      "learning_rate": 1.8157063233953387e-09,
      "loss": 0.1718,
      "step": 7747
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6035988520181288,
      "learning_rate": 1.7602714426023792e-09,
      "loss": 0.2091,
      "step": 7748
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4873226993714326,
      "learning_rate": 1.7056958171635108e-09,
      "loss": 0.1512,
      "step": 7749
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.760393587503681,
      "learning_rate": 1.6519794564606728e-09,
      "loss": 0.5941,
      "step": 7750
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.460667516219266,
      "learning_rate": 1.5991223697281454e-09,
      "loss": 0.4823,
      "step": 7751
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.473445950719236,
      "learning_rate": 1.5471245660531042e-09,
      "loss": 0.1786,
      "step": 7752
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3243311038374956,
      "learning_rate": 1.4959860543739546e-09,
      "loss": 0.1684,
      "step": 7753
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2710728398694973,
      "learning_rate": 1.4457068434831078e-09,
      "loss": 0.1505,
      "step": 7754
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.495056206990965,
      "learning_rate": 1.3962869420230951e-09,
      "loss": 0.1502,
      "step": 7755
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4253803876544946,
      "learning_rate": 1.3477263584904533e-09,
      "loss": 0.1408,
      "step": 7756
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.649955725734688,
      "learning_rate": 1.3000251012323939e-09,
      "loss": 0.165,
      "step": 7757
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5937296590123875,
      "learning_rate": 1.2531831784506898e-09,
      "loss": 0.2154,
      "step": 7758
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.278588029391801,
      "learning_rate": 1.2072005981966783e-09,
      "loss": 0.1542,
      "step": 7759
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4274030635592754,
      "learning_rate": 1.1620773683757025e-09,
      "loss": 0.1976,
      "step": 7760
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3379695502019775,
      "learning_rate": 1.1178134967454456e-09,
      "loss": 0.154,
      "step": 7761
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5177603825991177,
      "learning_rate": 1.0744089909153765e-09,
      "loss": 0.2141,
      "step": 7762
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.419799663925607,
      "learning_rate": 1.0318638583467489e-09,
      "loss": 0.1705,
      "step": 7763
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.736168081468135,
      "learning_rate": 9.901781063531568e-10,
      "loss": 0.4016,
      "step": 7764
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4154201619388165,
      "learning_rate": 9.493517421022003e-10,
      "loss": 0.2058,
      "step": 7765
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3557630932542826,
      "learning_rate": 9.093847726110439e-10,
      "loss": 0.1385,
      "step": 7766
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.223041918534062,
      "learning_rate": 8.70277204751413e-10,
      "loss": 0.5514,
      "step": 7767
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.600382719023386,
      "learning_rate": 8.320290452462632e-10,
      "loss": 0.6293,
      "step": 7768
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2976069974400182,
      "learning_rate": 7.946403006703351e-10,
      "loss": 0.1441,
      "step": 7769
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.923114802855086,
      "learning_rate": 7.581109774512651e-10,
      "loss": 0.5101,
      "step": 7770
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4046878947643309,
      "learning_rate": 7.224410818695848e-10,
      "loss": 0.1777,
      "step": 7771
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6536153512037024,
      "learning_rate": 6.876306200565008e-10,
      "loss": 0.1883,
      "step": 7772
    },
    {
      "epoch": 0.99,
      "grad_norm": 8.509252824724918,
      "learning_rate": 6.536795979966703e-10,
      "loss": 0.6009,
      "step": 7773
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4548876027187683,
      "learning_rate": 6.20588021527091e-10,
      "loss": 0.1554,
      "step": 7774
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2976803995668587,
      "learning_rate": 5.883558963359903e-10,
      "loss": 0.1607,
      "step": 7775
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5418702647795157,
      "learning_rate": 5.569832279644915e-10,
      "loss": 0.2251,
      "step": 7776
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4834053765770177,
      "learning_rate": 5.264700218066133e-10,
      "loss": 0.1869,
      "step": 7777
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.907159193584757,
      "learning_rate": 4.968162831070489e-10,
      "loss": 0.4475,
      "step": 7778
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2750400670455393,
      "learning_rate": 4.680220169639427e-10,
      "loss": 0.1702,
      "step": 7779
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.26564128183913,
      "learning_rate": 4.4008722832722397e-10,
      "loss": 0.1768,
      "step": 7780
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4012453989361915,
      "learning_rate": 4.1301192199971753e-10,
      "loss": 0.1722,
      "step": 7781
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2273272688565975,
      "learning_rate": 3.867961026354783e-10,
      "loss": 0.154,
      "step": 7782
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6406552016749476,
      "learning_rate": 3.6143977474201175e-10,
      "loss": 0.1758,
      "step": 7783
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.61831037389536,
      "learning_rate": 3.369429426769433e-10,
      "loss": 0.6374,
      "step": 7784
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.045279799119269,
      "learning_rate": 3.1330561065301413e-10,
      "loss": 0.6333,
      "step": 7785
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3821769677620162,
      "learning_rate": 2.9052778273308546e-10,
      "loss": 0.1591,
      "step": 7786
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.800474106690071,
      "learning_rate": 2.686094628329139e-10,
      "loss": 0.5839,
      "step": 7787
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.423840508427838,
      "learning_rate": 2.475506547211515e-10,
      "loss": 0.1538,
      "step": 7788
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.560980203647584,
      "learning_rate": 2.2735136201712527e-10,
      "loss": 0.4617,
      "step": 7789
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.030948592399185,
      "learning_rate": 2.080115881936129e-10,
      "loss": 0.4528,
      "step": 7790
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.303276051471497,
      "learning_rate": 1.895313365757323e-10,
      "loss": 0.3912,
      "step": 7791
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3615058871865535,
      "learning_rate": 1.7191061033983157e-10,
      "loss": 0.1756,
      "step": 7792
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.369858025908908,
      "learning_rate": 1.5514941251570937e-10,
      "loss": 0.157,
      "step": 7793
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.353999773530398,
      "learning_rate": 1.392477459843944e-10,
      "loss": 0.2073,
      "step": 7794
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.442376201783847,
      "learning_rate": 1.2420561348036598e-10,
      "loss": 0.1964,
      "step": 7795
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3685439737553027,
      "learning_rate": 1.1002301758822331e-10,
      "loss": 0.169,
      "step": 7796
    },
    {
      "epoch": 1.0,
      "grad_norm": 9.12232491346287,
      "learning_rate": 9.66999607471264e-11,
      "loss": 0.4868,
      "step": 7797
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3432207735995771,
      "learning_rate": 8.423644524691021e-11,
      "loss": 0.1845,
      "step": 7798
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5177052799610466,
      "learning_rate": 7.263247323086032e-11,
      "loss": 0.2223,
      "step": 7799
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5473885764132536,
      "learning_rate": 6.188804669349236e-11,
      "loss": 0.1599,
      "step": 7800
    },
    {
      "epoch": 1.0,
      "grad_norm": 19.463862161450074,
      "learning_rate": 5.200316748110723e-11,
      "loss": 0.4797,
      "step": 7801
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.844599721804169,
      "learning_rate": 4.297783729456662e-11,
      "loss": 0.7412,
      "step": 7802
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.70254712658039,
      "learning_rate": 3.481205768429696e-11,
      "loss": 0.7037,
      "step": 7803
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4234266201625827,
      "learning_rate": 2.7505830054175287e-11,
      "loss": 0.174,
      "step": 7804
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2918193807176401,
      "learning_rate": 2.105915566041894e-11,
      "loss": 0.1469,
      "step": 7805
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3910457286466553,
      "learning_rate": 1.5472035611585613e-11,
      "loss": 0.155,
      "step": 7806
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.434131322367487,
      "learning_rate": 1.0744470868018219e-11,
      "loss": 0.1568,
      "step": 7807
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.947631992574119,
      "learning_rate": 6.876462241844906e-12,
      "loss": 0.5891,
      "step": 7808
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3970590636464175,
      "learning_rate": 3.868010399199484e-12,
      "loss": 0.1693,
      "step": 7809
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3701381331372249,
      "learning_rate": 1.7191158563356626e-12,
      "loss": 0.1841,
      "step": 7810
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3064910910992407,
      "learning_rate": 4.2977898240259554e-13,
      "loss": 0.1427,
      "step": 7811
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.5280469308299804,
      "learning_rate": 0.0,
      "loss": 0.3187,
      "step": 7812
    },
    {
      "epoch": 1.0,
      "step": 7812,
      "total_flos": 3398925776714752.0,
      "train_loss": 0.026347098585944938,
      "train_runtime": 2801.1626,
      "train_samples_per_second": 44.624,
      "train_steps_per_second": 2.789
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 7812,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "total_flos": 3398925776714752.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}