{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.9999107899549489,
  "eval_steps": 500,
  "global_step": 22418,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 479.181575370833,
      "learning_rate": 1.4858841010401189e-08,
      "loss": 3.4191,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 444.2897708062578,
      "learning_rate": 2.9717682020802378e-08,
      "loss": 3.4145,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 471.42589410977354,
      "learning_rate": 4.457652303120357e-08,
      "loss": 3.4776,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 447.0952927337563,
      "learning_rate": 5.9435364041604756e-08,
      "loss": 3.3658,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 453.6589550797721,
      "learning_rate": 7.429420505200595e-08,
      "loss": 3.4743,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 481.06522522053183,
      "learning_rate": 8.915304606240714e-08,
      "loss": 3.4123,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 473.5375533010743,
      "learning_rate": 1.0401188707280832e-07,
      "loss": 3.3757,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 436.315608474856,
      "learning_rate": 1.1887072808320951e-07,
      "loss": 3.3641,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 414.8792812382819,
      "learning_rate": 1.337295690936107e-07,
      "loss": 3.3456,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 333.7437359925691,
      "learning_rate": 1.485884101040119e-07,
      "loss": 3.1523,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 327.378710355412,
      "learning_rate": 1.6344725111441309e-07,
      "loss": 3.1377,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 285.7239111115323,
      "learning_rate": 1.7830609212481427e-07,
      "loss": 3.0205,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 280.7644842348515,
      "learning_rate": 1.931649331352155e-07,
      "loss": 2.9787,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 142.52095750824589,
      "learning_rate": 2.0802377414561665e-07,
      "loss": 2.6684,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 127.96254371758955,
      "learning_rate": 2.2288261515601786e-07,
      "loss": 2.6067,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 120.73572885794177,
      "learning_rate": 2.3774145616641902e-07,
      "loss": 2.6169,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 112.53917359409854,
      "learning_rate": 2.5260029717682024e-07,
      "loss": 2.6291,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 92.92813910099909,
      "learning_rate": 2.674591381872214e-07,
      "loss": 2.3937,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 54.47839605852833,
      "learning_rate": 2.823179791976226e-07,
      "loss": 2.2331,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 47.32993965473415,
      "learning_rate": 2.971768202080238e-07,
      "loss": 2.2455,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 48.7811356025536,
      "learning_rate": 3.12035661218425e-07,
      "loss": 2.2506,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 44.01857855746975,
      "learning_rate": 3.2689450222882617e-07,
      "loss": 2.2508,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 41.93631579612844,
      "learning_rate": 3.4175334323922736e-07,
      "loss": 2.2085,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 38.30011234774115,
      "learning_rate": 3.5661218424962855e-07,
      "loss": 2.1267,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 43.54188231181135,
      "learning_rate": 3.7147102526002973e-07,
      "loss": 2.1306,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 52.08026321767365,
      "learning_rate": 3.86329866270431e-07,
      "loss": 2.0643,
      "step": 26
    },
    {
      "epoch": 0.0,
      "grad_norm": 46.70090747661249,
      "learning_rate": 4.0118870728083216e-07,
      "loss": 2.0185,
      "step": 27
    },
    {
      "epoch": 0.0,
      "grad_norm": 39.41131309753315,
      "learning_rate": 4.160475482912333e-07,
      "loss": 1.9262,
      "step": 28
    },
    {
      "epoch": 0.0,
      "grad_norm": 34.41968496829119,
      "learning_rate": 4.309063893016345e-07,
      "loss": 1.9676,
      "step": 29
    },
    {
      "epoch": 0.0,
      "grad_norm": 28.570605923178842,
      "learning_rate": 4.457652303120357e-07,
      "loss": 1.9105,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 28.3141498730999,
      "learning_rate": 4.606240713224369e-07,
      "loss": 1.8976,
      "step": 31
    },
    {
      "epoch": 0.0,
      "grad_norm": 27.40745888228763,
      "learning_rate": 4.7548291233283804e-07,
      "loss": 1.8376,
      "step": 32
    },
    {
      "epoch": 0.0,
      "grad_norm": 23.918437458656047,
      "learning_rate": 4.903417533432393e-07,
      "loss": 1.8495,
      "step": 33
    },
    {
      "epoch": 0.0,
      "grad_norm": 25.05587626909775,
      "learning_rate": 5.052005943536405e-07,
      "loss": 1.7793,
      "step": 34
    },
    {
      "epoch": 0.0,
      "grad_norm": 21.955775049135106,
      "learning_rate": 5.200594353640417e-07,
      "loss": 1.7941,
      "step": 35
    },
    {
      "epoch": 0.0,
      "grad_norm": 20.231098512861866,
      "learning_rate": 5.349182763744428e-07,
      "loss": 1.7493,
      "step": 36
    },
    {
      "epoch": 0.0,
      "grad_norm": 19.279727581189874,
      "learning_rate": 5.49777117384844e-07,
      "loss": 1.8063,
      "step": 37
    },
    {
      "epoch": 0.0,
      "grad_norm": 17.966177535301533,
      "learning_rate": 5.646359583952452e-07,
      "loss": 1.7594,
      "step": 38
    },
    {
      "epoch": 0.0,
      "grad_norm": 19.846803453280362,
      "learning_rate": 5.794947994056464e-07,
      "loss": 1.6701,
      "step": 39
    },
    {
      "epoch": 0.0,
      "grad_norm": 18.30996136853287,
      "learning_rate": 5.943536404160476e-07,
      "loss": 1.6956,
      "step": 40
    },
    {
      "epoch": 0.0,
      "grad_norm": 18.010934144859558,
      "learning_rate": 6.092124814264488e-07,
      "loss": 1.7458,
      "step": 41
    },
    {
      "epoch": 0.0,
      "grad_norm": 24.168274420434788,
      "learning_rate": 6.2407132243685e-07,
      "loss": 1.682,
      "step": 42
    },
    {
      "epoch": 0.0,
      "grad_norm": 20.084756508438474,
      "learning_rate": 6.389301634472512e-07,
      "loss": 1.7363,
      "step": 43
    },
    {
      "epoch": 0.0,
      "grad_norm": 19.290902998384894,
      "learning_rate": 6.537890044576523e-07,
      "loss": 1.706,
      "step": 44
    },
    {
      "epoch": 0.0,
      "grad_norm": 15.902582389775189,
      "learning_rate": 6.686478454680535e-07,
      "loss": 1.6916,
      "step": 45
    },
    {
      "epoch": 0.0,
      "grad_norm": 13.676862017586267,
      "learning_rate": 6.835066864784547e-07,
      "loss": 1.6156,
      "step": 46
    },
    {
      "epoch": 0.0,
      "grad_norm": 13.161909789045197,
      "learning_rate": 6.983655274888559e-07,
      "loss": 1.6225,
      "step": 47
    },
    {
      "epoch": 0.0,
      "grad_norm": 17.024205871793313,
      "learning_rate": 7.132243684992571e-07,
      "loss": 1.6207,
      "step": 48
    },
    {
      "epoch": 0.0,
      "grad_norm": 15.004350315629559,
      "learning_rate": 7.280832095096584e-07,
      "loss": 1.6309,
      "step": 49
    },
    {
      "epoch": 0.0,
      "grad_norm": 13.990190132490014,
      "learning_rate": 7.429420505200595e-07,
      "loss": 1.5804,
      "step": 50
    },
    {
      "epoch": 0.0,
      "grad_norm": 18.991256193625045,
      "learning_rate": 7.578008915304607e-07,
      "loss": 1.5821,
      "step": 51
    },
    {
      "epoch": 0.0,
      "grad_norm": 15.895332622201291,
      "learning_rate": 7.72659732540862e-07,
      "loss": 1.588,
      "step": 52
    },
    {
      "epoch": 0.0,
      "grad_norm": 17.09088450280486,
      "learning_rate": 7.87518573551263e-07,
      "loss": 1.6041,
      "step": 53
    },
    {
      "epoch": 0.0,
      "grad_norm": 14.500945522403708,
      "learning_rate": 8.023774145616643e-07,
      "loss": 1.5966,
      "step": 54
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.211534781116145,
      "learning_rate": 8.172362555720654e-07,
      "loss": 1.6168,
      "step": 55
    },
    {
      "epoch": 0.0,
      "grad_norm": 14.692041429656056,
      "learning_rate": 8.320950965824666e-07,
      "loss": 1.586,
      "step": 56
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.001357004610455,
      "learning_rate": 8.469539375928679e-07,
      "loss": 1.5413,
      "step": 57
    },
    {
      "epoch": 0.01,
      "grad_norm": 16.978137841649104,
      "learning_rate": 8.61812778603269e-07,
      "loss": 1.5951,
      "step": 58
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.751436330408085,
      "learning_rate": 8.766716196136702e-07,
      "loss": 1.5421,
      "step": 59
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.004915053909961,
      "learning_rate": 8.915304606240714e-07,
      "loss": 1.5123,
      "step": 60
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.906358884049599,
      "learning_rate": 9.063893016344725e-07,
      "loss": 1.5453,
      "step": 61
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.409957010954496,
      "learning_rate": 9.212481426448738e-07,
      "loss": 1.4842,
      "step": 62
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.864082352187994,
      "learning_rate": 9.36106983655275e-07,
      "loss": 1.4757,
      "step": 63
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.701845077315442,
      "learning_rate": 9.509658246656761e-07,
      "loss": 1.4957,
      "step": 64
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.713471528137951,
      "learning_rate": 9.658246656760775e-07,
      "loss": 1.5024,
      "step": 65
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.993112651003695,
      "learning_rate": 9.806835066864786e-07,
      "loss": 1.564,
      "step": 66
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.869519451134513,
      "learning_rate": 9.955423476968797e-07,
      "loss": 1.426,
      "step": 67
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.069638413154408,
      "learning_rate": 1.010401188707281e-06,
      "loss": 1.4965,
      "step": 68
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.087880279096124,
      "learning_rate": 1.025260029717682e-06,
      "loss": 1.5382,
      "step": 69
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.92505532814705,
      "learning_rate": 1.0401188707280833e-06,
      "loss": 1.4644,
      "step": 70
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.784591415007261,
      "learning_rate": 1.0549777117384844e-06,
      "loss": 1.4451,
      "step": 71
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.459471046099068,
      "learning_rate": 1.0698365527488857e-06,
      "loss": 1.4433,
      "step": 72
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.550963441072971,
      "learning_rate": 1.084695393759287e-06,
      "loss": 1.4407,
      "step": 73
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.872072499299058,
      "learning_rate": 1.099554234769688e-06,
      "loss": 1.4875,
      "step": 74
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.328535641926827,
      "learning_rate": 1.1144130757800894e-06,
      "loss": 1.4575,
      "step": 75
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.364700899036928,
      "learning_rate": 1.1292719167904904e-06,
      "loss": 1.5537,
      "step": 76
    },
    {
      "epoch": 0.01,
      "grad_norm": 16.147898840712465,
      "learning_rate": 1.1441307578008915e-06,
      "loss": 1.482,
      "step": 77
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.815008450187952,
      "learning_rate": 1.1589895988112928e-06,
      "loss": 1.3526,
      "step": 78
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.05571210539922,
      "learning_rate": 1.173848439821694e-06,
      "loss": 1.4698,
      "step": 79
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.163143091048799,
      "learning_rate": 1.1887072808320952e-06,
      "loss": 1.377,
      "step": 80
    },
    {
      "epoch": 0.01,
      "grad_norm": 17.568085696170478,
      "learning_rate": 1.2035661218424965e-06,
      "loss": 1.39,
      "step": 81
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.806868621894173,
      "learning_rate": 1.2184249628528976e-06,
      "loss": 1.3527,
      "step": 82
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.455012804839708,
      "learning_rate": 1.2332838038632989e-06,
      "loss": 1.4197,
      "step": 83
    },
    {
      "epoch": 0.01,
      "grad_norm": 25.358931898720623,
      "learning_rate": 1.2481426448737e-06,
      "loss": 1.4604,
      "step": 84
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.58656447679682,
      "learning_rate": 1.263001485884101e-06,
      "loss": 1.4368,
      "step": 85
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.814777797788555,
      "learning_rate": 1.2778603268945023e-06,
      "loss": 1.4281,
      "step": 86
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.322638396510534,
      "learning_rate": 1.2927191679049036e-06,
      "loss": 1.3675,
      "step": 87
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.134557335800913,
      "learning_rate": 1.3075780089153047e-06,
      "loss": 1.4305,
      "step": 88
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.209688427686528,
      "learning_rate": 1.322436849925706e-06,
      "loss": 1.3816,
      "step": 89
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.840224974971235,
      "learning_rate": 1.337295690936107e-06,
      "loss": 1.3839,
      "step": 90
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.696053380271618,
      "learning_rate": 1.3521545319465081e-06,
      "loss": 1.4101,
      "step": 91
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.938588514626533,
      "learning_rate": 1.3670133729569094e-06,
      "loss": 1.4081,
      "step": 92
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.055815008077932,
      "learning_rate": 1.3818722139673107e-06,
      "loss": 1.3995,
      "step": 93
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.915638253873388,
      "learning_rate": 1.3967310549777118e-06,
      "loss": 1.4221,
      "step": 94
    },
    {
      "epoch": 0.01,
      "grad_norm": 17.0224794717339,
      "learning_rate": 1.4115898959881131e-06,
      "loss": 1.3323,
      "step": 95
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.951756075064637,
      "learning_rate": 1.4264487369985142e-06,
      "loss": 1.426,
      "step": 96
    },
    {
      "epoch": 0.01,
      "grad_norm": 18.447943115932567,
      "learning_rate": 1.4413075780089153e-06,
      "loss": 1.3683,
      "step": 97
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.907615755825423,
      "learning_rate": 1.4561664190193168e-06,
      "loss": 1.3827,
      "step": 98
    },
    {
      "epoch": 0.01,
      "grad_norm": 17.527064148724257,
      "learning_rate": 1.4710252600297179e-06,
      "loss": 1.3522,
      "step": 99
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.85463892424153,
      "learning_rate": 1.485884101040119e-06,
      "loss": 1.3583,
      "step": 100
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.979405742834658,
      "learning_rate": 1.5007429420505202e-06,
      "loss": 1.3796,
      "step": 101
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.877901544159329,
      "learning_rate": 1.5156017830609213e-06,
      "loss": 1.3206,
      "step": 102
    },
    {
      "epoch": 0.01,
      "grad_norm": 18.14171894621848,
      "learning_rate": 1.5304606240713226e-06,
      "loss": 1.3531,
      "step": 103
    },
    {
      "epoch": 0.01,
      "grad_norm": 17.139693487970053,
      "learning_rate": 1.545319465081724e-06,
      "loss": 1.3549,
      "step": 104
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.643358282574313,
      "learning_rate": 1.560178306092125e-06,
      "loss": 1.3871,
      "step": 105
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.558436135548048,
      "learning_rate": 1.575037147102526e-06,
      "loss": 1.3315,
      "step": 106
    },
    {
      "epoch": 0.01,
      "grad_norm": 16.7614529553977,
      "learning_rate": 1.5898959881129271e-06,
      "loss": 1.3925,
      "step": 107
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.871331846237426,
      "learning_rate": 1.6047548291233286e-06,
      "loss": 1.3291,
      "step": 108
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.294756570468897,
      "learning_rate": 1.6196136701337297e-06,
      "loss": 1.3492,
      "step": 109
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.405533508743769,
      "learning_rate": 1.6344725111441308e-06,
      "loss": 1.2963,
      "step": 110
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.61018279134456,
      "learning_rate": 1.649331352154532e-06,
      "loss": 1.3527,
      "step": 111
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.821120193198384,
      "learning_rate": 1.6641901931649332e-06,
      "loss": 1.2968,
      "step": 112
    },
    {
      "epoch": 0.01,
      "grad_norm": 16.918411132600838,
      "learning_rate": 1.6790490341753343e-06,
      "loss": 1.3459,
      "step": 113
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.633436837689228,
      "learning_rate": 1.6939078751857358e-06,
      "loss": 1.3415,
      "step": 114
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.249921385480322,
      "learning_rate": 1.7087667161961369e-06,
      "loss": 1.2786,
      "step": 115
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.025236355476027,
      "learning_rate": 1.723625557206538e-06,
      "loss": 1.3296,
      "step": 116
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.901094356781028,
      "learning_rate": 1.7384843982169392e-06,
      "loss": 1.3116,
      "step": 117
    },
    {
      "epoch": 0.01,
      "grad_norm": 17.865371712168688,
      "learning_rate": 1.7533432392273403e-06,
      "loss": 1.3751,
      "step": 118
    },
    {
      "epoch": 0.01,
      "grad_norm": 17.300456403688568,
      "learning_rate": 1.7682020802377416e-06,
      "loss": 1.2831,
      "step": 119
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.851391973420686,
      "learning_rate": 1.7830609212481429e-06,
      "loss": 1.3637,
      "step": 120
    },
    {
      "epoch": 0.01,
      "grad_norm": 16.204876698827395,
      "learning_rate": 1.797919762258544e-06,
      "loss": 1.3002,
      "step": 121
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.392154234251004,
      "learning_rate": 1.812778603268945e-06,
      "loss": 1.3093,
      "step": 122
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.759589482992947,
      "learning_rate": 1.8276374442793463e-06,
      "loss": 1.2945,
      "step": 123
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.620869907217457,
      "learning_rate": 1.8424962852897476e-06,
      "loss": 1.3044,
      "step": 124
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.864494310444867,
      "learning_rate": 1.8573551263001487e-06,
      "loss": 1.3197,
      "step": 125
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.05485640891095,
      "learning_rate": 1.87221396731055e-06,
      "loss": 1.3127,
      "step": 126
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.584014381877582,
      "learning_rate": 1.887072808320951e-06,
      "loss": 1.3062,
      "step": 127
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.500868635254967,
      "learning_rate": 1.9019316493313522e-06,
      "loss": 1.3076,
      "step": 128
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.806498047741435,
      "learning_rate": 1.9167904903417537e-06,
      "loss": 1.316,
      "step": 129
    },
    {
      "epoch": 0.01,
      "grad_norm": 17.9649931494059,
      "learning_rate": 1.931649331352155e-06,
      "loss": 1.2616,
      "step": 130
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.855783617695623,
      "learning_rate": 1.946508172362556e-06,
      "loss": 1.3063,
      "step": 131
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.529057403136159,
      "learning_rate": 1.961367013372957e-06,
      "loss": 1.2163,
      "step": 132
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.777468106054982,
      "learning_rate": 1.976225854383358e-06,
      "loss": 1.2762,
      "step": 133
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.24424756323446,
      "learning_rate": 1.9910846953937593e-06,
      "loss": 1.2596,
      "step": 134
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.042397353824043,
      "learning_rate": 2.0059435364041606e-06,
      "loss": 1.2565,
      "step": 135
    },
    {
      "epoch": 0.01,
      "grad_norm": 17.735895883652123,
      "learning_rate": 2.020802377414562e-06,
      "loss": 1.3068,
      "step": 136
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.707673602825954,
      "learning_rate": 2.035661218424963e-06,
      "loss": 1.3003,
      "step": 137
    },
    {
      "epoch": 0.01,
      "grad_norm": 17.48684971958427,
      "learning_rate": 2.050520059435364e-06,
      "loss": 1.3342,
      "step": 138
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.228479960755173,
      "learning_rate": 2.0653789004457653e-06,
      "loss": 1.2495,
      "step": 139
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.900078884379077,
      "learning_rate": 2.0802377414561666e-06,
      "loss": 1.2468,
      "step": 140
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.47129268314176,
      "learning_rate": 2.095096582466568e-06,
      "loss": 1.296,
      "step": 141
    },
    {
      "epoch": 0.01,
      "grad_norm": 16.459660959763685,
      "learning_rate": 2.109955423476969e-06,
      "loss": 1.342,
      "step": 142
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.760276648449938,
      "learning_rate": 2.12481426448737e-06,
      "loss": 1.2727,
      "step": 143
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.700064552390026,
      "learning_rate": 2.1396731054977714e-06,
      "loss": 1.3235,
      "step": 144
    },
    {
      "epoch": 0.01,
      "grad_norm": 18.913751170862138,
      "learning_rate": 2.1545319465081727e-06,
      "loss": 1.2435,
      "step": 145
    },
    {
      "epoch": 0.01,
      "grad_norm": 18.305401315829588,
      "learning_rate": 2.169390787518574e-06,
      "loss": 1.2308,
      "step": 146
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.771982225540329,
      "learning_rate": 2.184249628528975e-06,
      "loss": 1.2199,
      "step": 147
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.979628684242488,
      "learning_rate": 2.199108469539376e-06,
      "loss": 1.2491,
      "step": 148
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.787849766802376,
      "learning_rate": 2.213967310549777e-06,
      "loss": 1.2688,
      "step": 149
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.918227376465481,
      "learning_rate": 2.2288261515601787e-06,
      "loss": 1.245,
      "step": 150
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.170387629365358,
      "learning_rate": 2.2436849925705796e-06,
      "loss": 1.2734,
      "step": 151
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.653791558432433,
      "learning_rate": 2.258543833580981e-06,
      "loss": 1.2626,
      "step": 152
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.205828606638123,
      "learning_rate": 2.273402674591382e-06,
      "loss": 1.2133,
      "step": 153
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.6149677828453,
      "learning_rate": 2.288261515601783e-06,
      "loss": 1.2684,
      "step": 154
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.388640296290708,
      "learning_rate": 2.3031203566121843e-06,
      "loss": 1.2988,
      "step": 155
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.862643638893008,
      "learning_rate": 2.3179791976225856e-06,
      "loss": 1.2823,
      "step": 156
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.512149971227803,
      "learning_rate": 2.332838038632987e-06,
      "loss": 1.2087,
      "step": 157
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.453080063018092,
      "learning_rate": 2.347696879643388e-06,
      "loss": 1.2138,
      "step": 158
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.567949862851616,
      "learning_rate": 2.362555720653789e-06,
      "loss": 1.3151,
      "step": 159
    },
    {
      "epoch": 0.01,
      "grad_norm": 21.547222876178306,
      "learning_rate": 2.3774145616641904e-06,
      "loss": 1.2234,
      "step": 160
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.251373263860538,
      "learning_rate": 2.3922734026745917e-06,
      "loss": 1.2584,
      "step": 161
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.300628443048849,
      "learning_rate": 2.407132243684993e-06,
      "loss": 1.2608,
      "step": 162
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.98705476437207,
      "learning_rate": 2.421991084695394e-06,
      "loss": 1.2159,
      "step": 163
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.791748061089544,
      "learning_rate": 2.436849925705795e-06,
      "loss": 1.2554,
      "step": 164
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.018632449496657,
      "learning_rate": 2.4517087667161964e-06,
      "loss": 1.2341,
      "step": 165
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.094477050073278,
      "learning_rate": 2.4665676077265977e-06,
      "loss": 1.2332,
      "step": 166
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.884350776445494,
      "learning_rate": 2.4814264487369986e-06,
      "loss": 1.2469,
      "step": 167
    },
    {
      "epoch": 0.01,
      "grad_norm": 17.193021833995882,
      "learning_rate": 2.4962852897474e-06,
      "loss": 1.3187,
      "step": 168
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.961130562655457,
      "learning_rate": 2.511144130757801e-06,
      "loss": 1.2319,
      "step": 169
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.993313911240913,
      "learning_rate": 2.526002971768202e-06,
      "loss": 1.2149,
      "step": 170
    },
    {
      "epoch": 0.02,
      "grad_norm": 14.047670795991726,
      "learning_rate": 2.5408618127786033e-06,
      "loss": 1.2209,
      "step": 171
    },
    {
      "epoch": 0.02,
      "grad_norm": 21.127092907882563,
      "learning_rate": 2.5557206537890046e-06,
      "loss": 1.2408,
      "step": 172
    },
    {
      "epoch": 0.02,
      "grad_norm": 14.132470723641486,
      "learning_rate": 2.5705794947994055e-06,
      "loss": 1.203,
      "step": 173
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.844890485764699,
      "learning_rate": 2.5854383358098072e-06,
      "loss": 1.1984,
      "step": 174
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.74978526362959,
      "learning_rate": 2.6002971768202085e-06,
      "loss": 1.1639,
      "step": 175
    },
    {
      "epoch": 0.02,
      "grad_norm": 13.815068691606902,
      "learning_rate": 2.6151560178306094e-06,
      "loss": 1.3011,
      "step": 176
    },
    {
      "epoch": 0.02,
      "grad_norm": 15.764349569674762,
      "learning_rate": 2.6300148588410107e-06,
      "loss": 1.2627,
      "step": 177
    },
    {
      "epoch": 0.02,
      "grad_norm": 14.470509417280374,
      "learning_rate": 2.644873699851412e-06,
      "loss": 1.2685,
      "step": 178
    },
    {
      "epoch": 0.02,
      "grad_norm": 16.039980543166294,
      "learning_rate": 2.659732540861813e-06,
      "loss": 1.1406,
      "step": 179
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.5738469121652,
      "learning_rate": 2.674591381872214e-06,
      "loss": 1.1516,
      "step": 180
    },
    {
      "epoch": 0.02,
      "grad_norm": 18.059602492533326,
      "learning_rate": 2.6894502228826154e-06,
      "loss": 1.1666,
      "step": 181
    },
    {
      "epoch": 0.02,
      "grad_norm": 20.177117462919192,
      "learning_rate": 2.7043090638930163e-06,
      "loss": 1.2744,
      "step": 182
    },
    {
      "epoch": 0.02,
      "grad_norm": 15.536263644110688,
      "learning_rate": 2.7191679049034176e-06,
      "loss": 1.1206,
      "step": 183
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.297048918515785,
      "learning_rate": 2.734026745913819e-06,
      "loss": 1.2574,
      "step": 184
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.678652060311096,
      "learning_rate": 2.74888558692422e-06,
      "loss": 1.2799,
      "step": 185
    },
    {
      "epoch": 0.02,
      "grad_norm": 15.59313597860626,
      "learning_rate": 2.7637444279346215e-06,
      "loss": 1.2248,
      "step": 186
    },
    {
      "epoch": 0.02,
      "grad_norm": 13.457200260997297,
      "learning_rate": 2.7786032689450228e-06,
      "loss": 1.2024,
      "step": 187
    },
    {
      "epoch": 0.02,
      "grad_norm": 14.59239483647375,
      "learning_rate": 2.7934621099554236e-06,
      "loss": 1.1853,
      "step": 188
    },
    {
      "epoch": 0.02,
      "grad_norm": 18.672676931184043,
      "learning_rate": 2.808320950965825e-06,
      "loss": 1.2517,
      "step": 189
    },
    {
      "epoch": 0.02,
      "grad_norm": 15.29911091977755,
      "learning_rate": 2.8231797919762262e-06,
      "loss": 1.1466,
      "step": 190
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.149515901303058,
      "learning_rate": 2.838038632986627e-06,
      "loss": 1.1838,
      "step": 191
    },
    {
      "epoch": 0.02,
      "grad_norm": 14.051324968932548,
      "learning_rate": 2.8528974739970284e-06,
      "loss": 1.1744,
      "step": 192
    },
    {
      "epoch": 0.02,
      "grad_norm": 14.522630864590466,
      "learning_rate": 2.8677563150074297e-06,
      "loss": 1.2072,
      "step": 193
    },
    {
      "epoch": 0.02,
      "grad_norm": 13.674067310609283,
      "learning_rate": 2.8826151560178305e-06,
      "loss": 1.2121,
      "step": 194
    },
    {
      "epoch": 0.02,
      "grad_norm": 13.782367562371006,
      "learning_rate": 2.8974739970282323e-06,
      "loss": 1.1845,
      "step": 195
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.879011934178767,
      "learning_rate": 2.9123328380386335e-06,
      "loss": 1.2064,
      "step": 196
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.498092970246208,
      "learning_rate": 2.9271916790490344e-06,
      "loss": 1.1234,
      "step": 197
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.17586068791182,
      "learning_rate": 2.9420505200594357e-06,
      "loss": 1.2419,
      "step": 198
    },
    {
      "epoch": 0.02,
      "grad_norm": 13.004187552958408,
      "learning_rate": 2.956909361069837e-06,
      "loss": 1.2166,
      "step": 199
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.192285836517392,
      "learning_rate": 2.971768202080238e-06,
      "loss": 1.212,
      "step": 200
    },
    {
      "epoch": 0.02,
      "grad_norm": 17.42181289015885,
      "learning_rate": 2.986627043090639e-06,
      "loss": 1.2487,
      "step": 201
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.179845802131048,
      "learning_rate": 3.0014858841010405e-06,
      "loss": 1.1612,
      "step": 202
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.600482852462585,
      "learning_rate": 3.0163447251114413e-06,
      "loss": 1.1495,
      "step": 203
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.014264113015788,
      "learning_rate": 3.0312035661218426e-06,
      "loss": 1.2616,
      "step": 204
    },
    {
      "epoch": 0.02,
      "grad_norm": 13.107693551936135,
      "learning_rate": 3.0460624071322435e-06,
      "loss": 1.139,
      "step": 205
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.917826535458081,
      "learning_rate": 3.060921248142645e-06,
      "loss": 1.2142,
      "step": 206
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.826483910058341,
      "learning_rate": 3.0757800891530465e-06,
      "loss": 1.2569,
      "step": 207
    },
    {
      "epoch": 0.02,
      "grad_norm": 15.056638792781408,
      "learning_rate": 3.090638930163448e-06,
      "loss": 1.2198,
      "step": 208
    },
    {
      "epoch": 0.02,
      "grad_norm": 16.784629298821034,
      "learning_rate": 3.1054977711738487e-06,
      "loss": 1.1242,
      "step": 209
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.040510640244115,
      "learning_rate": 3.12035661218425e-06,
      "loss": 1.1908,
      "step": 210
    },
    {
      "epoch": 0.02,
      "grad_norm": 14.314824997667564,
      "learning_rate": 3.135215453194651e-06,
      "loss": 1.2029,
      "step": 211
    },
    {
      "epoch": 0.02,
      "grad_norm": 18.09616170922017,
      "learning_rate": 3.150074294205052e-06,
      "loss": 1.2121,
      "step": 212
    },
    {
      "epoch": 0.02,
      "grad_norm": 16.781146176840643,
      "learning_rate": 3.1649331352154534e-06,
      "loss": 1.2052,
      "step": 213
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.264561761688796,
      "learning_rate": 3.1797919762258543e-06,
      "loss": 1.1961,
      "step": 214
    },
    {
      "epoch": 0.02,
      "grad_norm": 14.781151549460386,
      "learning_rate": 3.1946508172362556e-06,
      "loss": 1.2406,
      "step": 215
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.95473417118167,
      "learning_rate": 3.2095096582466573e-06,
      "loss": 1.1866,
      "step": 216
    },
    {
      "epoch": 0.02,
      "grad_norm": 15.16509082756982,
      "learning_rate": 3.2243684992570586e-06,
      "loss": 1.2092,
      "step": 217
    },
    {
      "epoch": 0.02,
      "grad_norm": 13.786658637688276,
      "learning_rate": 3.2392273402674595e-06,
      "loss": 1.2308,
      "step": 218
    },
    {
      "epoch": 0.02,
      "grad_norm": 15.021731858126865,
      "learning_rate": 3.2540861812778607e-06,
      "loss": 1.1464,
      "step": 219
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.953107241389668,
      "learning_rate": 3.2689450222882616e-06,
      "loss": 1.1401,
      "step": 220
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.081731006475007,
      "learning_rate": 3.283803863298663e-06,
      "loss": 1.2147,
      "step": 221
    },
    {
      "epoch": 0.02,
      "grad_norm": 13.20020408897566,
      "learning_rate": 3.298662704309064e-06,
      "loss": 1.1948,
      "step": 222
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.476540961563462,
      "learning_rate": 3.313521545319465e-06,
      "loss": 1.1341,
      "step": 223
    },
    {
      "epoch": 0.02,
      "grad_norm": 13.357412307087126,
      "learning_rate": 3.3283803863298664e-06,
      "loss": 1.1894,
      "step": 224
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.042216590979454,
      "learning_rate": 3.3432392273402677e-06,
      "loss": 1.2044,
      "step": 225
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.914633679413257,
      "learning_rate": 3.3580980683506685e-06,
      "loss": 1.1855,
      "step": 226
    },
    {
      "epoch": 0.02,
      "grad_norm": 14.73825595969435,
      "learning_rate": 3.3729569093610702e-06,
      "loss": 1.1828,
      "step": 227
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.519662367164644,
      "learning_rate": 3.3878157503714715e-06,
      "loss": 1.1759,
      "step": 228
    },
    {
      "epoch": 0.02,
      "grad_norm": 13.021231922314369,
      "learning_rate": 3.4026745913818724e-06,
      "loss": 1.2459,
      "step": 229
    },
    {
      "epoch": 0.02,
      "grad_norm": 13.310598451337086,
      "learning_rate": 3.4175334323922737e-06,
      "loss": 1.1945,
      "step": 230
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.949457538230213,
      "learning_rate": 3.432392273402675e-06,
      "loss": 1.2102,
      "step": 231
    },
    {
      "epoch": 0.02,
      "grad_norm": 17.045710640024133,
      "learning_rate": 3.447251114413076e-06,
      "loss": 1.1379,
      "step": 232
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.590841904831532,
      "learning_rate": 3.462109955423477e-06,
      "loss": 1.1811,
      "step": 233
    },
    {
      "epoch": 0.02,
      "grad_norm": 13.750607612300065,
      "learning_rate": 3.4769687964338785e-06,
      "loss": 1.1857,
      "step": 234
    },
    {
      "epoch": 0.02,
      "grad_norm": 15.011852341995404,
      "learning_rate": 3.4918276374442793e-06,
      "loss": 1.203,
      "step": 235
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.39528573444235,
      "learning_rate": 3.5066864784546806e-06,
      "loss": 1.1346,
      "step": 236
    },
    {
      "epoch": 0.02,
      "grad_norm": 13.840094174504417,
      "learning_rate": 3.5215453194650823e-06,
      "loss": 1.1567,
      "step": 237
    },
    {
      "epoch": 0.02,
      "grad_norm": 14.616219974659701,
      "learning_rate": 3.536404160475483e-06,
      "loss": 1.1738,
      "step": 238
    },
    {
      "epoch": 0.02,
      "grad_norm": 15.372558864126345,
      "learning_rate": 3.5512630014858845e-06,
      "loss": 1.1568,
      "step": 239
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.455412066813167,
      "learning_rate": 3.5661218424962858e-06,
      "loss": 1.1451,
      "step": 240
    },
    {
      "epoch": 0.02,
      "grad_norm": 16.910211566269286,
      "learning_rate": 3.5809806835066867e-06,
      "loss": 1.1602,
      "step": 241
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.30636616567896,
      "learning_rate": 3.595839524517088e-06,
      "loss": 1.1565,
      "step": 242
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.111097262677804,
      "learning_rate": 3.6106983655274892e-06,
      "loss": 1.169,
      "step": 243
    },
    {
      "epoch": 0.02,
      "grad_norm": 13.50312867862547,
      "learning_rate": 3.62555720653789e-06,
      "loss": 1.2179,
      "step": 244
    },
    {
      "epoch": 0.02,
      "grad_norm": 14.32961653214507,
      "learning_rate": 3.6404160475482914e-06,
      "loss": 1.1748,
      "step": 245
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.455741519438698,
      "learning_rate": 3.6552748885586927e-06,
      "loss": 1.2008,
      "step": 246
    },
    {
      "epoch": 0.02,
      "grad_norm": 15.474695964955936,
      "learning_rate": 3.6701337295690936e-06,
      "loss": 1.1185,
      "step": 247
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.393869030707698,
      "learning_rate": 3.6849925705794953e-06,
      "loss": 1.1227,
      "step": 248
    },
    {
      "epoch": 0.02,
      "grad_norm": 15.713920176980748,
      "learning_rate": 3.6998514115898966e-06,
      "loss": 1.1721,
      "step": 249
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.743457450738655,
      "learning_rate": 3.7147102526002974e-06,
      "loss": 1.2703,
      "step": 250
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.415390507140565,
      "learning_rate": 3.7295690936106987e-06,
      "loss": 1.1371,
      "step": 251
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.65868346467697,
      "learning_rate": 3.7444279346211e-06,
      "loss": 1.1417,
      "step": 252
    },
    {
      "epoch": 0.02,
      "grad_norm": 14.358408993554516,
      "learning_rate": 3.759286775631501e-06,
      "loss": 1.1465,
      "step": 253
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.829190608866632,
      "learning_rate": 3.774145616641902e-06,
      "loss": 1.2175,
      "step": 254
    },
    {
      "epoch": 0.02,
      "grad_norm": 17.185789181969838,
      "learning_rate": 3.7890044576523035e-06,
      "loss": 1.1737,
      "step": 255
    },
    {
      "epoch": 0.02,
      "grad_norm": 18.77896936569814,
      "learning_rate": 3.8038632986627044e-06,
      "loss": 1.1712,
      "step": 256
    },
    {
      "epoch": 0.02,
      "grad_norm": 16.017119240160365,
      "learning_rate": 3.818722139673105e-06,
      "loss": 1.1675,
      "step": 257
    },
    {
      "epoch": 0.02,
      "grad_norm": 13.796672333302658,
      "learning_rate": 3.833580980683507e-06,
      "loss": 1.1254,
      "step": 258
    },
    {
      "epoch": 0.02,
      "grad_norm": 13.96226868735574,
      "learning_rate": 3.848439821693909e-06,
      "loss": 1.2487,
      "step": 259
    },
    {
      "epoch": 0.02,
      "grad_norm": 13.722954246633371,
      "learning_rate": 3.86329866270431e-06,
      "loss": 1.1899,
      "step": 260
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.719410386530718,
      "learning_rate": 3.87815750371471e-06,
      "loss": 1.2081,
      "step": 261
    },
    {
      "epoch": 0.02,
      "grad_norm": 14.749493269685129,
      "learning_rate": 3.893016344725112e-06,
      "loss": 1.1285,
      "step": 262
    },
    {
      "epoch": 0.02,
      "grad_norm": 15.68745324230534,
      "learning_rate": 3.907875185735513e-06,
      "loss": 1.1248,
      "step": 263
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.965408803921283,
      "learning_rate": 3.922734026745914e-06,
      "loss": 1.143,
      "step": 264
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.40320934750862,
      "learning_rate": 3.9375928677563156e-06,
      "loss": 1.1394,
      "step": 265
    },
    {
      "epoch": 0.02,
      "grad_norm": 13.771517992861996,
      "learning_rate": 3.952451708766716e-06,
      "loss": 1.1301,
      "step": 266
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.475975059438115,
      "learning_rate": 3.967310549777117e-06,
      "loss": 1.1739,
      "step": 267
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.686715681645756,
      "learning_rate": 3.982169390787519e-06,
      "loss": 1.1226,
      "step": 268
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.101843245998369,
      "learning_rate": 3.99702823179792e-06,
      "loss": 1.1672,
      "step": 269
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.913564075048646,
      "learning_rate": 4.011887072808321e-06,
      "loss": 1.1486,
      "step": 270
    },
    {
      "epoch": 0.02,
      "grad_norm": 16.69823654863612,
      "learning_rate": 4.0267459138187225e-06,
      "loss": 1.1327,
      "step": 271
    },
    {
      "epoch": 0.02,
      "grad_norm": 16.085969793714913,
      "learning_rate": 4.041604754829124e-06,
      "loss": 1.1343,
      "step": 272
    },
    {
      "epoch": 0.02,
      "grad_norm": 15.494657069121349,
      "learning_rate": 4.056463595839525e-06,
      "loss": 1.1289,
      "step": 273
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.440177381474532,
      "learning_rate": 4.071322436849926e-06,
      "loss": 1.1858,
      "step": 274
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.8334650428448,
      "learning_rate": 4.086181277860327e-06,
      "loss": 1.1907,
      "step": 275
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.74151253320919,
      "learning_rate": 4.101040118870728e-06,
      "loss": 1.1603,
      "step": 276
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.475525789423399,
      "learning_rate": 4.115898959881129e-06,
      "loss": 1.2043,
      "step": 277
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.829299651075278,
      "learning_rate": 4.130757800891531e-06,
      "loss": 1.133,
      "step": 278
    },
    {
      "epoch": 0.02,
      "grad_norm": 17.329680635075192,
      "learning_rate": 4.145616641901932e-06,
      "loss": 1.1493,
      "step": 279
    },
    {
      "epoch": 0.02,
      "grad_norm": 14.690556911955909,
      "learning_rate": 4.160475482912333e-06,
      "loss": 1.1623,
      "step": 280
    },
    {
      "epoch": 0.03,
      "grad_norm": 13.206246265946543,
      "learning_rate": 4.1753343239227346e-06,
      "loss": 1.1234,
      "step": 281
    },
    {
      "epoch": 0.03,
      "grad_norm": 16.712358755807497,
      "learning_rate": 4.190193164933136e-06,
      "loss": 1.1051,
      "step": 282
    },
    {
      "epoch": 0.03,
      "grad_norm": 16.073673365591848,
      "learning_rate": 4.205052005943537e-06,
      "loss": 1.0933,
      "step": 283
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.508584279157313,
      "learning_rate": 4.219910846953938e-06,
      "loss": 1.1764,
      "step": 284
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.991423548643269,
      "learning_rate": 4.234769687964339e-06,
      "loss": 1.0992,
      "step": 285
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.748056613621992,
      "learning_rate": 4.24962852897474e-06,
      "loss": 1.1243,
      "step": 286
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.322306237293285,
      "learning_rate": 4.2644873699851415e-06,
      "loss": 1.1757,
      "step": 287
    },
    {
      "epoch": 0.03,
      "grad_norm": 13.988645394552838,
      "learning_rate": 4.279346210995543e-06,
      "loss": 1.1257,
      "step": 288
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.736277700046065,
      "learning_rate": 4.294205052005943e-06,
      "loss": 1.1539,
      "step": 289
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.002667620401283,
      "learning_rate": 4.309063893016345e-06,
      "loss": 1.0836,
      "step": 290
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.808712495874166,
      "learning_rate": 4.323922734026747e-06,
      "loss": 1.09,
      "step": 291
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.936288983067914,
      "learning_rate": 4.338781575037148e-06,
      "loss": 1.1336,
      "step": 292
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.145686145809051,
      "learning_rate": 4.353640416047548e-06,
      "loss": 1.1435,
      "step": 293
    },
    {
      "epoch": 0.03,
      "grad_norm": 14.456789448369179,
      "learning_rate": 4.36849925705795e-06,
      "loss": 1.1687,
      "step": 294
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.769863508672417,
      "learning_rate": 4.383358098068351e-06,
      "loss": 1.1446,
      "step": 295
    },
    {
      "epoch": 0.03,
      "grad_norm": 17.589132702641255,
      "learning_rate": 4.398216939078752e-06,
      "loss": 1.1772,
      "step": 296
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.278357014011094,
      "learning_rate": 4.4130757800891536e-06,
      "loss": 1.1981,
      "step": 297
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.20754151376717,
      "learning_rate": 4.427934621099554e-06,
      "loss": 1.114,
      "step": 298
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.029876134130642,
      "learning_rate": 4.442793462109955e-06,
      "loss": 1.1267,
      "step": 299
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.418290211193492,
      "learning_rate": 4.4576523031203574e-06,
      "loss": 1.1366,
      "step": 300
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.003956497140376,
      "learning_rate": 4.472511144130759e-06,
      "loss": 1.0846,
      "step": 301
    },
    {
      "epoch": 0.03,
      "grad_norm": 14.175925642107874,
      "learning_rate": 4.487369985141159e-06,
      "loss": 1.1238,
      "step": 302
    },
    {
      "epoch": 0.03,
      "grad_norm": 15.182664928368428,
      "learning_rate": 4.5022288261515605e-06,
      "loss": 1.1619,
      "step": 303
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.878505417883948,
      "learning_rate": 4.517087667161962e-06,
      "loss": 1.1158,
      "step": 304
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.952130330860887,
      "learning_rate": 4.531946508172363e-06,
      "loss": 1.128,
      "step": 305
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.773052293623806,
      "learning_rate": 4.546805349182764e-06,
      "loss": 1.1744,
      "step": 306
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.331953786365242,
      "learning_rate": 4.561664190193165e-06,
      "loss": 1.0938,
      "step": 307
    },
    {
      "epoch": 0.03,
      "grad_norm": 17.665432727588627,
      "learning_rate": 4.576523031203566e-06,
      "loss": 1.1313,
      "step": 308
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.84762038498788,
      "learning_rate": 4.591381872213967e-06,
      "loss": 1.1476,
      "step": 309
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.600148427775235,
      "learning_rate": 4.606240713224369e-06,
      "loss": 1.1333,
      "step": 310
    },
    {
      "epoch": 0.03,
      "grad_norm": 14.501259732471242,
      "learning_rate": 4.62109955423477e-06,
      "loss": 1.1369,
      "step": 311
    },
    {
      "epoch": 0.03,
      "grad_norm": 16.55911266314763,
      "learning_rate": 4.635958395245171e-06,
      "loss": 1.1324,
      "step": 312
    },
    {
      "epoch": 0.03,
      "grad_norm": 16.4446652146636,
      "learning_rate": 4.6508172362555726e-06,
      "loss": 1.1611,
      "step": 313
    },
    {
      "epoch": 0.03,
      "grad_norm": 14.275935419911963,
      "learning_rate": 4.665676077265974e-06,
      "loss": 1.1249,
      "step": 314
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.518630160559079,
      "learning_rate": 4.680534918276375e-06,
      "loss": 1.1577,
      "step": 315
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.777739076374935,
      "learning_rate": 4.695393759286776e-06,
      "loss": 1.1522,
      "step": 316
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.349262763923035,
      "learning_rate": 4.710252600297177e-06,
      "loss": 1.1568,
      "step": 317
    },
    {
      "epoch": 0.03,
      "grad_norm": 13.241273809687511,
      "learning_rate": 4.725111441307578e-06,
      "loss": 1.1747,
      "step": 318
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.896098906136915,
      "learning_rate": 4.7399702823179795e-06,
      "loss": 1.1553,
      "step": 319
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.798390363313116,
      "learning_rate": 4.754829123328381e-06,
      "loss": 1.1057,
      "step": 320
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.382811173244301,
      "learning_rate": 4.769687964338782e-06,
      "loss": 1.1664,
      "step": 321
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.925649919559495,
      "learning_rate": 4.784546805349183e-06,
      "loss": 1.1351,
      "step": 322
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.990226126529167,
      "learning_rate": 4.799405646359585e-06,
      "loss": 1.1642,
      "step": 323
    },
    {
      "epoch": 0.03,
      "grad_norm": 13.138448205829322,
      "learning_rate": 4.814264487369986e-06,
      "loss": 1.1545,
      "step": 324
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.669573960480008,
      "learning_rate": 4.829123328380386e-06,
      "loss": 1.1882,
      "step": 325
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.260299621981513,
      "learning_rate": 4.843982169390788e-06,
      "loss": 1.097,
      "step": 326
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.807511934158953,
      "learning_rate": 4.858841010401189e-06,
      "loss": 1.1386,
      "step": 327
    },
    {
      "epoch": 0.03,
      "grad_norm": 13.363941323925244,
      "learning_rate": 4.87369985141159e-06,
      "loss": 1.1141,
      "step": 328
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.217746968184022,
      "learning_rate": 4.8885586924219916e-06,
      "loss": 1.1128,
      "step": 329
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.48202238565019,
      "learning_rate": 4.903417533432393e-06,
      "loss": 1.161,
      "step": 330
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.752973799420971,
      "learning_rate": 4.918276374442793e-06,
      "loss": 1.1064,
      "step": 331
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.576794435696424,
      "learning_rate": 4.9331352154531954e-06,
      "loss": 1.1232,
      "step": 332
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.97070575833564,
      "learning_rate": 4.947994056463597e-06,
      "loss": 1.117,
      "step": 333
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.770311580299774,
      "learning_rate": 4.962852897473997e-06,
      "loss": 1.1629,
      "step": 334
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.591697542732224,
      "learning_rate": 4.9777117384843985e-06,
      "loss": 1.1523,
      "step": 335
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.144720072202839,
      "learning_rate": 4.9925705794948e-06,
      "loss": 1.1279,
      "step": 336
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.492500918353,
      "learning_rate": 5.007429420505201e-06,
      "loss": 1.1576,
      "step": 337
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.283862620597446,
      "learning_rate": 5.022288261515602e-06,
      "loss": 1.0454,
      "step": 338
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.894127321918003,
      "learning_rate": 5.037147102526004e-06,
      "loss": 1.015,
      "step": 339
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.881864241565623,
      "learning_rate": 5.052005943536404e-06,
      "loss": 1.1712,
      "step": 340
    },
    {
      "epoch": 0.03,
      "grad_norm": 13.812211780810685,
      "learning_rate": 5.066864784546805e-06,
      "loss": 1.1233,
      "step": 341
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.664980533391345,
      "learning_rate": 5.081723625557207e-06,
      "loss": 1.1308,
      "step": 342
    },
    {
      "epoch": 0.03,
      "grad_norm": 13.219826431829576,
      "learning_rate": 5.096582466567608e-06,
      "loss": 1.1132,
      "step": 343
    },
    {
      "epoch": 0.03,
      "grad_norm": 14.366829061611822,
      "learning_rate": 5.111441307578009e-06,
      "loss": 1.1141,
      "step": 344
    },
    {
      "epoch": 0.03,
      "grad_norm": 15.252527209826669,
      "learning_rate": 5.12630014858841e-06,
      "loss": 1.0661,
      "step": 345
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.001826088026796,
      "learning_rate": 5.141158989598811e-06,
      "loss": 1.127,
      "step": 346
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.711817410896405,
      "learning_rate": 5.156017830609212e-06,
      "loss": 1.102,
      "step": 347
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.129342182134883,
      "learning_rate": 5.1708766716196144e-06,
      "loss": 1.0366,
      "step": 348
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.803651277071285,
      "learning_rate": 5.185735512630016e-06,
      "loss": 1.1489,
      "step": 349
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.627983348990764,
      "learning_rate": 5.200594353640417e-06,
      "loss": 1.1832,
      "step": 350
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.682764159326739,
      "learning_rate": 5.215453194650818e-06,
      "loss": 1.1601,
      "step": 351
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.006312614977624,
      "learning_rate": 5.230312035661219e-06,
      "loss": 1.1131,
      "step": 352
    },
    {
      "epoch": 0.03,
      "grad_norm": 16.68769698032886,
      "learning_rate": 5.24517087667162e-06,
      "loss": 1.1326,
      "step": 353
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.357079970804948,
      "learning_rate": 5.260029717682021e-06,
      "loss": 1.0554,
      "step": 354
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.49207319213776,
      "learning_rate": 5.274888558692423e-06,
      "loss": 1.0998,
      "step": 355
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.574872920554217,
      "learning_rate": 5.289747399702824e-06,
      "loss": 1.1145,
      "step": 356
    },
    {
      "epoch": 0.03,
      "grad_norm": 13.838195094509109,
      "learning_rate": 5.304606240713225e-06,
      "loss": 1.1592,
      "step": 357
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.171830536264517,
      "learning_rate": 5.319465081723626e-06,
      "loss": 1.0931,
      "step": 358
    },
    {
      "epoch": 0.03,
      "grad_norm": 17.82660079430687,
      "learning_rate": 5.334323922734027e-06,
      "loss": 1.0801,
      "step": 359
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.164632727151705,
      "learning_rate": 5.349182763744428e-06,
      "loss": 1.0847,
      "step": 360
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.333894099834296,
      "learning_rate": 5.3640416047548295e-06,
      "loss": 1.0902,
      "step": 361
    },
    {
      "epoch": 0.03,
      "grad_norm": 13.118399448488406,
      "learning_rate": 5.378900445765231e-06,
      "loss": 1.1339,
      "step": 362
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.872355798308535,
      "learning_rate": 5.393759286775631e-06,
      "loss": 1.0995,
      "step": 363
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.901371635416284,
      "learning_rate": 5.408618127786033e-06,
      "loss": 1.1166,
      "step": 364
    },
    {
      "epoch": 0.03,
      "grad_norm": 13.570112246183713,
      "learning_rate": 5.423476968796434e-06,
      "loss": 1.1612,
      "step": 365
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.577306247838742,
      "learning_rate": 5.438335809806835e-06,
      "loss": 1.1298,
      "step": 366
    },
    {
      "epoch": 0.03,
      "grad_norm": 16.016028292625748,
      "learning_rate": 5.4531946508172365e-06,
      "loss": 1.0656,
      "step": 367
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.31447410430925,
      "learning_rate": 5.468053491827638e-06,
      "loss": 1.1652,
      "step": 368
    },
    {
      "epoch": 0.03,
      "grad_norm": 13.096617481720974,
      "learning_rate": 5.48291233283804e-06,
      "loss": 1.1162,
      "step": 369
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.915602711290008,
      "learning_rate": 5.49777117384844e-06,
      "loss": 1.0988,
      "step": 370
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.359593642520789,
      "learning_rate": 5.512630014858842e-06,
      "loss": 1.0883,
      "step": 371
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.041169748367924,
      "learning_rate": 5.527488855869243e-06,
      "loss": 1.0486,
      "step": 372
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.157720689523385,
      "learning_rate": 5.542347696879644e-06,
      "loss": 1.098,
      "step": 373
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.989415528203331,
      "learning_rate": 5.5572065378900455e-06,
      "loss": 1.0264,
      "step": 374
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.928899024622885,
      "learning_rate": 5.572065378900447e-06,
      "loss": 1.1299,
      "step": 375
    },
    {
      "epoch": 0.03,
      "grad_norm": 14.830538006019573,
      "learning_rate": 5.586924219910847e-06,
      "loss": 1.1004,
      "step": 376
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.298892518913483,
      "learning_rate": 5.6017830609212485e-06,
      "loss": 1.1726,
      "step": 377
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.597504879147783,
      "learning_rate": 5.61664190193165e-06,
      "loss": 1.1402,
      "step": 378
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.315359962449914,
      "learning_rate": 5.631500742942051e-06,
      "loss": 1.0969,
      "step": 379
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.474500926448806,
      "learning_rate": 5.6463595839524524e-06,
      "loss": 1.0829,
      "step": 380
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.143376052591305,
      "learning_rate": 5.661218424962853e-06,
      "loss": 1.0613,
      "step": 381
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.026457750192046,
      "learning_rate": 5.676077265973254e-06,
      "loss": 1.0886,
      "step": 382
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.34071087714645,
      "learning_rate": 5.6909361069836555e-06,
      "loss": 1.1038,
      "step": 383
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.087189805639952,
      "learning_rate": 5.705794947994057e-06,
      "loss": 1.1075,
      "step": 384
    },
    {
      "epoch": 0.03,
      "grad_norm": 18.182979736482697,
      "learning_rate": 5.720653789004458e-06,
      "loss": 1.1291,
      "step": 385
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.850618220765881,
      "learning_rate": 5.735512630014859e-06,
      "loss": 1.147,
      "step": 386
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.655433025474341,
      "learning_rate": 5.75037147102526e-06,
      "loss": 1.0633,
      "step": 387
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.466120735041697,
      "learning_rate": 5.765230312035661e-06,
      "loss": 1.1737,
      "step": 388
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.874822433638904,
      "learning_rate": 5.780089153046062e-06,
      "loss": 1.1781,
      "step": 389
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.235618857025864,
      "learning_rate": 5.7949479940564645e-06,
      "loss": 1.1743,
      "step": 390
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.672164991604085,
      "learning_rate": 5.809806835066866e-06,
      "loss": 1.0779,
      "step": 391
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.214118010924405,
      "learning_rate": 5.824665676077267e-06,
      "loss": 1.1064,
      "step": 392
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.186190132450484,
      "learning_rate": 5.839524517087668e-06,
      "loss": 1.1405,
      "step": 393
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.544449788417802,
      "learning_rate": 5.854383358098069e-06,
      "loss": 1.1041,
      "step": 394
    },
    {
      "epoch": 0.04,
      "grad_norm": 14.711154397535067,
      "learning_rate": 5.86924219910847e-06,
      "loss": 1.1502,
      "step": 395
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.694145647518669,
      "learning_rate": 5.884101040118871e-06,
      "loss": 1.0788,
      "step": 396
    },
    {
      "epoch": 0.04,
      "grad_norm": 12.940970512349567,
      "learning_rate": 5.898959881129273e-06,
      "loss": 1.092,
      "step": 397
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.884556149567274,
      "learning_rate": 5.913818722139674e-06,
      "loss": 1.0877,
      "step": 398
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.262113589977268,
      "learning_rate": 5.9286775631500745e-06,
      "loss": 1.1076,
      "step": 399
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.51325697539055,
      "learning_rate": 5.943536404160476e-06,
      "loss": 1.0636,
      "step": 400
    },
    {
      "epoch": 0.04,
      "grad_norm": 13.446050393774,
      "learning_rate": 5.958395245170877e-06,
      "loss": 1.0826,
      "step": 401
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.379205158608055,
      "learning_rate": 5.973254086181278e-06,
      "loss": 1.0193,
      "step": 402
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.999284403289655,
      "learning_rate": 5.98811292719168e-06,
      "loss": 1.1267,
      "step": 403
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.917812043293901,
      "learning_rate": 6.002971768202081e-06,
      "loss": 1.0823,
      "step": 404
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.449861350114748,
      "learning_rate": 6.017830609212481e-06,
      "loss": 1.0346,
      "step": 405
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.651418729054114,
      "learning_rate": 6.032689450222883e-06,
      "loss": 1.0463,
      "step": 406
    },
    {
      "epoch": 0.04,
      "grad_norm": 12.621962037936385,
      "learning_rate": 6.047548291233284e-06,
      "loss": 1.1162,
      "step": 407
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.764463028967802,
      "learning_rate": 6.062407132243685e-06,
      "loss": 1.0848,
      "step": 408
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.642134858716629,
      "learning_rate": 6.0772659732540865e-06,
      "loss": 1.0604,
      "step": 409
    },
    {
      "epoch": 0.04,
      "grad_norm": 12.631888644208388,
      "learning_rate": 6.092124814264487e-06,
      "loss": 1.0383,
      "step": 410
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.927327038251738,
      "learning_rate": 6.10698365527489e-06,
      "loss": 1.0587,
      "step": 411
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.813346491467549,
      "learning_rate": 6.12184249628529e-06,
      "loss": 1.0199,
      "step": 412
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.976474295949616,
      "learning_rate": 6.136701337295692e-06,
      "loss": 1.0789,
      "step": 413
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.218846154551258,
      "learning_rate": 6.151560178306093e-06,
      "loss": 1.085,
      "step": 414
    },
    {
      "epoch": 0.04,
      "grad_norm": 12.52370423732885,
      "learning_rate": 6.166419019316494e-06,
      "loss": 1.117,
      "step": 415
    },
    {
      "epoch": 0.04,
      "grad_norm": 13.333590316540821,
      "learning_rate": 6.181277860326896e-06,
      "loss": 1.0873,
      "step": 416
    },
    {
      "epoch": 0.04,
      "grad_norm": 13.700627178771182,
      "learning_rate": 6.196136701337296e-06,
      "loss": 1.0547,
      "step": 417
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.713914393188082,
      "learning_rate": 6.210995542347697e-06,
      "loss": 1.115,
      "step": 418
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.537414893672029,
      "learning_rate": 6.225854383358099e-06,
      "loss": 1.0873,
      "step": 419
    },
    {
      "epoch": 0.04,
      "grad_norm": 14.047764998940131,
      "learning_rate": 6.2407132243685e-06,
      "loss": 1.1282,
      "step": 420
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.996923076730589,
      "learning_rate": 6.255572065378901e-06,
      "loss": 1.0704,
      "step": 421
    },
    {
      "epoch": 0.04,
      "grad_norm": 14.469191468092976,
      "learning_rate": 6.270430906389302e-06,
      "loss": 1.1288,
      "step": 422
    },
    {
      "epoch": 0.04,
      "grad_norm": 12.62057914801523,
      "learning_rate": 6.285289747399703e-06,
      "loss": 1.0763,
      "step": 423
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.279374554299574,
      "learning_rate": 6.300148588410104e-06,
      "loss": 1.0355,
      "step": 424
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.962173335077091,
      "learning_rate": 6.3150074294205055e-06,
      "loss": 1.0891,
      "step": 425
    },
    {
      "epoch": 0.04,
      "grad_norm": 12.042352441265631,
      "learning_rate": 6.329866270430907e-06,
      "loss": 1.1005,
      "step": 426
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.0343757648233,
      "learning_rate": 6.344725111441308e-06,
      "loss": 1.0459,
      "step": 427
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.545088790676903,
      "learning_rate": 6.3595839524517086e-06,
      "loss": 1.112,
      "step": 428
    },
    {
      "epoch": 0.04,
      "grad_norm": 15.433696255392372,
      "learning_rate": 6.37444279346211e-06,
      "loss": 1.1623,
      "step": 429
    },
    {
      "epoch": 0.04,
      "grad_norm": 12.103402093460588,
      "learning_rate": 6.389301634472511e-06,
      "loss": 1.039,
      "step": 430
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.601063807102944,
      "learning_rate": 6.4041604754829124e-06,
      "loss": 1.131,
      "step": 431
    },
    {
      "epoch": 0.04,
      "grad_norm": 12.692039398190387,
      "learning_rate": 6.419019316493315e-06,
      "loss": 1.0846,
      "step": 432
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.198569465131731,
      "learning_rate": 6.433878157503716e-06,
      "loss": 1.1596,
      "step": 433
    },
    {
      "epoch": 0.04,
      "grad_norm": 13.3797476064304,
      "learning_rate": 6.448736998514117e-06,
      "loss": 1.12,
      "step": 434
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.063453876032748,
      "learning_rate": 6.463595839524518e-06,
      "loss": 1.1252,
      "step": 435
    },
    {
      "epoch": 0.04,
      "grad_norm": 13.038468465734933,
      "learning_rate": 6.478454680534919e-06,
      "loss": 1.0631,
      "step": 436
    },
    {
      "epoch": 0.04,
      "grad_norm": 15.035972964047915,
      "learning_rate": 6.49331352154532e-06,
      "loss": 1.0719,
      "step": 437
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.133314064314307,
      "learning_rate": 6.5081723625557215e-06,
      "loss": 1.0336,
      "step": 438
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.06964302939007,
      "learning_rate": 6.523031203566123e-06,
      "loss": 1.0922,
      "step": 439
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.37830849371796,
      "learning_rate": 6.537890044576523e-06,
      "loss": 1.0918,
      "step": 440
    },
    {
      "epoch": 0.04,
      "grad_norm": 12.067596500017343,
      "learning_rate": 6.5527488855869245e-06,
      "loss": 1.0846,
      "step": 441
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.298040611922534,
      "learning_rate": 6.567607726597326e-06,
      "loss": 1.1023,
      "step": 442
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.774846456763964,
      "learning_rate": 6.582466567607727e-06,
      "loss": 1.024,
      "step": 443
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.169866048657992,
      "learning_rate": 6.597325408618128e-06,
      "loss": 1.1111,
      "step": 444
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.638399799046015,
      "learning_rate": 6.61218424962853e-06,
      "loss": 1.1371,
      "step": 445
    },
    {
      "epoch": 0.04,
      "grad_norm": 12.062016188197536,
      "learning_rate": 6.62704309063893e-06,
      "loss": 1.0947,
      "step": 446
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.588075423302197,
      "learning_rate": 6.6419019316493314e-06,
      "loss": 1.0951,
      "step": 447
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.561749521728126,
      "learning_rate": 6.656760772659733e-06,
      "loss": 1.0623,
      "step": 448
    },
    {
      "epoch": 0.04,
      "grad_norm": 12.058674443345739,
      "learning_rate": 6.671619613670134e-06,
      "loss": 1.065,
      "step": 449
    },
    {
      "epoch": 0.04,
      "grad_norm": 12.537991187831732,
      "learning_rate": 6.686478454680535e-06,
      "loss": 1.0648,
      "step": 450
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.537193397979344,
      "learning_rate": 6.701337295690936e-06,
      "loss": 1.1165,
      "step": 451
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.635958391067676,
      "learning_rate": 6.716196136701337e-06,
      "loss": 1.0481,
      "step": 452
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.82577974690559,
      "learning_rate": 6.731054977711739e-06,
      "loss": 1.0411,
      "step": 453
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.435756314570446,
      "learning_rate": 6.7459138187221405e-06,
      "loss": 1.1841,
      "step": 454
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.568503818929775,
      "learning_rate": 6.760772659732542e-06,
      "loss": 1.0145,
      "step": 455
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.947804702475315,
      "learning_rate": 6.775631500742943e-06,
      "loss": 1.0216,
      "step": 456
    },
    {
      "epoch": 0.04,
      "grad_norm": 12.758707295194826,
      "learning_rate": 6.790490341753344e-06,
      "loss": 1.0554,
      "step": 457
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.916852313395255,
      "learning_rate": 6.805349182763745e-06,
      "loss": 1.0418,
      "step": 458
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.891134571046292,
      "learning_rate": 6.820208023774146e-06,
      "loss": 1.0529,
      "step": 459
    },
    {
      "epoch": 0.04,
      "grad_norm": 12.702494665572685,
      "learning_rate": 6.835066864784547e-06,
      "loss": 1.0675,
      "step": 460
    },
    {
      "epoch": 0.04,
      "grad_norm": 12.235738744844763,
      "learning_rate": 6.849925705794949e-06,
      "loss": 1.119,
      "step": 461
    },
    {
      "epoch": 0.04,
      "grad_norm": 13.204578913554466,
      "learning_rate": 6.86478454680535e-06,
      "loss": 1.0392,
      "step": 462
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.935780373604464,
      "learning_rate": 6.879643387815751e-06,
      "loss": 1.1044,
      "step": 463
    },
    {
      "epoch": 0.04,
      "grad_norm": 14.766522457944976,
      "learning_rate": 6.894502228826152e-06,
      "loss": 1.0862,
      "step": 464
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.953917063614647,
      "learning_rate": 6.909361069836553e-06,
      "loss": 1.1255,
      "step": 465
    },
    {
      "epoch": 0.04,
      "grad_norm": 14.573650841571785,
      "learning_rate": 6.924219910846954e-06,
      "loss": 1.0395,
      "step": 466
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.258706604637766,
      "learning_rate": 6.939078751857356e-06,
      "loss": 1.0577,
      "step": 467
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.03921632844517,
      "learning_rate": 6.953937592867757e-06,
      "loss": 1.1145,
      "step": 468
    },
    {
      "epoch": 0.04,
      "grad_norm": 12.06022703305555,
      "learning_rate": 6.968796433878157e-06,
      "loss": 0.9777,
      "step": 469
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.5656366673217,
      "learning_rate": 6.983655274888559e-06,
      "loss": 1.0527,
      "step": 470
    },
    {
      "epoch": 0.04,
      "grad_norm": 12.803081291319273,
      "learning_rate": 6.99851411589896e-06,
      "loss": 1.0367,
      "step": 471
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.080953557823713,
      "learning_rate": 7.013372956909361e-06,
      "loss": 1.0159,
      "step": 472
    },
    {
      "epoch": 0.04,
      "grad_norm": 12.576575443761769,
      "learning_rate": 7.0282317979197625e-06,
      "loss": 1.1302,
      "step": 473
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.389873208727682,
      "learning_rate": 7.043090638930165e-06,
      "loss": 1.0904,
      "step": 474
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.24412578383744,
      "learning_rate": 7.057949479940566e-06,
      "loss": 1.0145,
      "step": 475
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.085750076882057,
      "learning_rate": 7.072808320950966e-06,
      "loss": 1.1031,
      "step": 476
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.525127650372667,
      "learning_rate": 7.087667161961368e-06,
      "loss": 1.0408,
      "step": 477
    },
    {
      "epoch": 0.04,
      "grad_norm": 12.198103636553823,
      "learning_rate": 7.102526002971769e-06,
      "loss": 1.0692,
      "step": 478
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.363138094055538,
      "learning_rate": 7.11738484398217e-06,
      "loss": 1.0592,
      "step": 479
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.299123155036039,
      "learning_rate": 7.1322436849925716e-06,
      "loss": 1.0844,
      "step": 480
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.282021282912071,
      "learning_rate": 7.147102526002973e-06,
      "loss": 1.1159,
      "step": 481
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.246026191917027,
      "learning_rate": 7.161961367013373e-06,
      "loss": 0.9831,
      "step": 482
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.064639282097621,
      "learning_rate": 7.176820208023775e-06,
      "loss": 1.0781,
      "step": 483
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.709223791930382,
      "learning_rate": 7.191679049034176e-06,
      "loss": 1.0965,
      "step": 484
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.348270902778255,
      "learning_rate": 7.206537890044577e-06,
      "loss": 1.0928,
      "step": 485
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.328008159250846,
      "learning_rate": 7.2213967310549785e-06,
      "loss": 1.0359,
      "step": 486
    },
    {
      "epoch": 0.04,
      "grad_norm": 14.345799529536906,
      "learning_rate": 7.236255572065379e-06,
      "loss": 0.9923,
      "step": 487
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.287663906708264,
      "learning_rate": 7.25111441307578e-06,
      "loss": 1.0258,
      "step": 488
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.10281497892318,
      "learning_rate": 7.2659732540861815e-06,
      "loss": 1.1016,
      "step": 489
    },
    {
      "epoch": 0.04,
      "grad_norm": 13.917241726250158,
      "learning_rate": 7.280832095096583e-06,
      "loss": 1.1344,
      "step": 490
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.441572543164023,
      "learning_rate": 7.295690936106984e-06,
      "loss": 1.1199,
      "step": 491
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.018870027566267,
      "learning_rate": 7.310549777117385e-06,
      "loss": 1.0834,
      "step": 492
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.877156730788565,
      "learning_rate": 7.325408618127786e-06,
      "loss": 0.9573,
      "step": 493
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.21742714100454,
      "learning_rate": 7.340267459138187e-06,
      "loss": 0.9669,
      "step": 494
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.761872924756993,
      "learning_rate": 7.355126300148589e-06,
      "loss": 1.0723,
      "step": 495
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.064189396129455,
      "learning_rate": 7.3699851411589906e-06,
      "loss": 1.0928,
      "step": 496
    },
    {
      "epoch": 0.04,
      "grad_norm": 15.755575959351075,
      "learning_rate": 7.384843982169392e-06,
      "loss": 1.0533,
      "step": 497
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.66794328807516,
      "learning_rate": 7.399702823179793e-06,
      "loss": 1.0944,
      "step": 498
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.372905042944108,
      "learning_rate": 7.4145616641901944e-06,
      "loss": 1.0713,
      "step": 499
    },
    {
      "epoch": 0.04,
      "grad_norm": 13.2502854265202,
      "learning_rate": 7.429420505200595e-06,
      "loss": 1.1844,
      "step": 500
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.195471989176056,
      "learning_rate": 7.444279346210996e-06,
      "loss": 1.0479,
      "step": 501
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.949909025416147,
      "learning_rate": 7.4591381872213975e-06,
      "loss": 1.1865,
      "step": 502
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.252290027661497,
      "learning_rate": 7.473997028231799e-06,
      "loss": 1.0415,
      "step": 503
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.893804698230767,
      "learning_rate": 7.4888558692422e-06,
      "loss": 1.1058,
      "step": 504
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.41859529234815,
      "learning_rate": 7.5037147102526005e-06,
      "loss": 1.0247,
      "step": 505
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.216347496885334,
      "learning_rate": 7.518573551263002e-06,
      "loss": 1.0477,
      "step": 506
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.610736465901066,
      "learning_rate": 7.533432392273403e-06,
      "loss": 1.0016,
      "step": 507
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.669955935585106,
      "learning_rate": 7.548291233283804e-06,
      "loss": 1.1493,
      "step": 508
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.925053525678942,
      "learning_rate": 7.563150074294206e-06,
      "loss": 1.0773,
      "step": 509
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.209142854481104,
      "learning_rate": 7.578008915304607e-06,
      "loss": 1.0814,
      "step": 510
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.265542297306787,
      "learning_rate": 7.5928677563150074e-06,
      "loss": 1.0712,
      "step": 511
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.755730234614122,
      "learning_rate": 7.607726597325409e-06,
      "loss": 1.0809,
      "step": 512
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.266738188025256,
      "learning_rate": 7.62258543833581e-06,
      "loss": 1.0278,
      "step": 513
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.732289132469514,
      "learning_rate": 7.63744427934621e-06,
      "loss": 1.0468,
      "step": 514
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.259449102927174,
      "learning_rate": 7.652303120356612e-06,
      "loss": 1.1276,
      "step": 515
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.970650616986509,
      "learning_rate": 7.667161961367015e-06,
      "loss": 1.0291,
      "step": 516
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.984941535528321,
      "learning_rate": 7.682020802377416e-06,
      "loss": 1.0475,
      "step": 517
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.98840673581107,
      "learning_rate": 7.696879643387817e-06,
      "loss": 1.1277,
      "step": 518
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.351965589564253,
      "learning_rate": 7.711738484398219e-06,
      "loss": 1.0895,
      "step": 519
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.53645337733584,
      "learning_rate": 7.72659732540862e-06,
      "loss": 1.0809,
      "step": 520
    },
    {
      "epoch": 0.05,
      "grad_norm": 13.220435955460754,
      "learning_rate": 7.74145616641902e-06,
      "loss": 1.0287,
      "step": 521
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.970058713671222,
      "learning_rate": 7.75631500742942e-06,
      "loss": 1.0969,
      "step": 522
    },
    {
      "epoch": 0.05,
      "grad_norm": 13.48724598880786,
      "learning_rate": 7.771173848439822e-06,
      "loss": 1.0105,
      "step": 523
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.599097087875933,
      "learning_rate": 7.786032689450223e-06,
      "loss": 1.112,
      "step": 524
    },
    {
      "epoch": 0.05,
      "grad_norm": 14.051624029583836,
      "learning_rate": 7.800891530460625e-06,
      "loss": 1.0641,
      "step": 525
    },
    {
      "epoch": 0.05,
      "grad_norm": 14.024510106001976,
      "learning_rate": 7.815750371471026e-06,
      "loss": 1.0594,
      "step": 526
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.423018360409106,
      "learning_rate": 7.830609212481427e-06,
      "loss": 1.0375,
      "step": 527
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.611068641736255,
      "learning_rate": 7.845468053491829e-06,
      "loss": 1.0908,
      "step": 528
    },
    {
      "epoch": 0.05,
      "grad_norm": 12.885593143256525,
      "learning_rate": 7.86032689450223e-06,
      "loss": 1.1056,
      "step": 529
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.11642312546528,
      "learning_rate": 7.875185735512631e-06,
      "loss": 1.0539,
      "step": 530
    },
    {
      "epoch": 0.05,
      "grad_norm": 13.952679676317683,
      "learning_rate": 7.89004457652303e-06,
      "loss": 1.0974,
      "step": 531
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.97238125579649,
      "learning_rate": 7.904903417533432e-06,
      "loss": 1.0293,
      "step": 532
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.239216208918084,
      "learning_rate": 7.919762258543833e-06,
      "loss": 1.0133,
      "step": 533
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.887399670671932,
      "learning_rate": 7.934621099554235e-06,
      "loss": 1.0551,
      "step": 534
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.42900621886104,
      "learning_rate": 7.949479940564636e-06,
      "loss": 1.1033,
      "step": 535
    },
    {
      "epoch": 0.05,
      "grad_norm": 15.96927325842374,
      "learning_rate": 7.964338781575037e-06,
      "loss": 1.0594,
      "step": 536
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.717799099126825,
      "learning_rate": 7.97919762258544e-06,
      "loss": 1.0247,
      "step": 537
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.012322170488845,
      "learning_rate": 7.99405646359584e-06,
      "loss": 1.0061,
      "step": 538
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.942903167029972,
      "learning_rate": 8.008915304606241e-06,
      "loss": 1.05,
      "step": 539
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.690848417593793,
      "learning_rate": 8.023774145616642e-06,
      "loss": 1.0701,
      "step": 540
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.782884977548475,
      "learning_rate": 8.038632986627044e-06,
      "loss": 1.0698,
      "step": 541
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.58376454100326,
      "learning_rate": 8.053491827637445e-06,
      "loss": 1.0728,
      "step": 542
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.99206625704471,
      "learning_rate": 8.068350668647846e-06,
      "loss": 1.0914,
      "step": 543
    },
    {
      "epoch": 0.05,
      "grad_norm": 13.849014127736218,
      "learning_rate": 8.083209509658248e-06,
      "loss": 1.0262,
      "step": 544
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.343768881715658,
      "learning_rate": 8.098068350668649e-06,
      "loss": 1.0523,
      "step": 545
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.164830071550107,
      "learning_rate": 8.11292719167905e-06,
      "loss": 1.0536,
      "step": 546
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.664979640879006,
      "learning_rate": 8.127786032689451e-06,
      "loss": 1.0313,
      "step": 547
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.77106407245154,
      "learning_rate": 8.142644873699853e-06,
      "loss": 1.1136,
      "step": 548
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.62209264710672,
      "learning_rate": 8.157503714710252e-06,
      "loss": 1.0647,
      "step": 549
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.186908296221132,
      "learning_rate": 8.172362555720654e-06,
      "loss": 1.0978,
      "step": 550
    },
    {
      "epoch": 0.05,
      "grad_norm": 16.362927591768752,
      "learning_rate": 8.187221396731055e-06,
      "loss": 1.0554,
      "step": 551
    },
    {
      "epoch": 0.05,
      "grad_norm": 12.142361852107289,
      "learning_rate": 8.202080237741456e-06,
      "loss": 1.0355,
      "step": 552
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.902229141196612,
      "learning_rate": 8.216939078751858e-06,
      "loss": 1.1195,
      "step": 553
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.760229988730577,
      "learning_rate": 8.231797919762259e-06,
      "loss": 1.0914,
      "step": 554
    },
    {
      "epoch": 0.05,
      "grad_norm": 12.379092877488281,
      "learning_rate": 8.24665676077266e-06,
      "loss": 0.9086,
      "step": 555
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.573760725419241,
      "learning_rate": 8.261515601783061e-06,
      "loss": 0.9965,
      "step": 556
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.878017266700095,
      "learning_rate": 8.276374442793463e-06,
      "loss": 1.0434,
      "step": 557
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.977434223424188,
      "learning_rate": 8.291233283803864e-06,
      "loss": 1.0584,
      "step": 558
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.894103118622146,
      "learning_rate": 8.306092124814265e-06,
      "loss": 1.0357,
      "step": 559
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.846983152986644,
      "learning_rate": 8.320950965824667e-06,
      "loss": 1.1461,
      "step": 560
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.7437703543386,
      "learning_rate": 8.335809806835068e-06,
      "loss": 1.0718,
      "step": 561
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.083292178410538,
      "learning_rate": 8.350668647845469e-06,
      "loss": 0.9969,
      "step": 562
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.258467562269697,
      "learning_rate": 8.36552748885587e-06,
      "loss": 0.9757,
      "step": 563
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.530940687389796,
      "learning_rate": 8.380386329866272e-06,
      "loss": 1.0477,
      "step": 564
    },
    {
      "epoch": 0.05,
      "grad_norm": 12.44524237541492,
      "learning_rate": 8.395245170876673e-06,
      "loss": 1.1481,
      "step": 565
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.595558323591767,
      "learning_rate": 8.410104011887074e-06,
      "loss": 1.0542,
      "step": 566
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.83283229553203,
      "learning_rate": 8.424962852897474e-06,
      "loss": 1.089,
      "step": 567
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.391334696446814,
      "learning_rate": 8.439821693907875e-06,
      "loss": 1.0068,
      "step": 568
    },
    {
      "epoch": 0.05,
      "grad_norm": 13.548702971173704,
      "learning_rate": 8.454680534918276e-06,
      "loss": 1.0674,
      "step": 569
    },
    {
      "epoch": 0.05,
      "grad_norm": 13.641426146186935,
      "learning_rate": 8.469539375928678e-06,
      "loss": 0.9898,
      "step": 570
    },
    {
      "epoch": 0.05,
      "grad_norm": 12.888977843317264,
      "learning_rate": 8.484398216939079e-06,
      "loss": 1.1028,
      "step": 571
    },
    {
      "epoch": 0.05,
      "grad_norm": 12.256762954394175,
      "learning_rate": 8.49925705794948e-06,
      "loss": 1.0672,
      "step": 572
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.433856872753363,
      "learning_rate": 8.514115898959882e-06,
      "loss": 1.0657,
      "step": 573
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.954871235712872,
      "learning_rate": 8.528974739970283e-06,
      "loss": 1.0413,
      "step": 574
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.262491889811633,
      "learning_rate": 8.543833580980684e-06,
      "loss": 1.0713,
      "step": 575
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.9496586031532,
      "learning_rate": 8.558692421991086e-06,
      "loss": 1.0109,
      "step": 576
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.21498577767788,
      "learning_rate": 8.573551263001487e-06,
      "loss": 1.063,
      "step": 577
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.946231435196236,
      "learning_rate": 8.588410104011886e-06,
      "loss": 1.1269,
      "step": 578
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.298141586611257,
      "learning_rate": 8.60326894502229e-06,
      "loss": 1.0085,
      "step": 579
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.679306259663587,
      "learning_rate": 8.61812778603269e-06,
      "loss": 1.0236,
      "step": 580
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.791069919927743,
      "learning_rate": 8.632986627043092e-06,
      "loss": 1.0323,
      "step": 581
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.991060292165049,
      "learning_rate": 8.647845468053493e-06,
      "loss": 1.0712,
      "step": 582
    },
    {
      "epoch": 0.05,
      "grad_norm": 15.505209032076564,
      "learning_rate": 8.662704309063895e-06,
      "loss": 1.0781,
      "step": 583
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.075115246551459,
      "learning_rate": 8.677563150074296e-06,
      "loss": 1.0894,
      "step": 584
    },
    {
      "epoch": 0.05,
      "grad_norm": 12.61895286352917,
      "learning_rate": 8.692421991084695e-06,
      "loss": 1.0333,
      "step": 585
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.269378367666873,
      "learning_rate": 8.707280832095097e-06,
      "loss": 1.0445,
      "step": 586
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.963542249585114,
      "learning_rate": 8.722139673105498e-06,
      "loss": 1.0383,
      "step": 587
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.889542963421267,
      "learning_rate": 8.7369985141159e-06,
      "loss": 0.9926,
      "step": 588
    },
    {
      "epoch": 0.05,
      "grad_norm": 14.393390392507921,
      "learning_rate": 8.7518573551263e-06,
      "loss": 1.0115,
      "step": 589
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.988164169695649,
      "learning_rate": 8.766716196136702e-06,
      "loss": 1.0094,
      "step": 590
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.51433748687375,
      "learning_rate": 8.781575037147103e-06,
      "loss": 1.0917,
      "step": 591
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.245819635308987,
      "learning_rate": 8.796433878157505e-06,
      "loss": 1.0714,
      "step": 592
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.701221215211948,
      "learning_rate": 8.811292719167906e-06,
      "loss": 1.1008,
      "step": 593
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.336619668775048,
      "learning_rate": 8.826151560178307e-06,
      "loss": 1.075,
      "step": 594
    },
    {
      "epoch": 0.05,
      "grad_norm": 12.758621238138701,
      "learning_rate": 8.841010401188708e-06,
      "loss": 1.0117,
      "step": 595
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.660061475354045,
      "learning_rate": 8.855869242199108e-06,
      "loss": 1.0854,
      "step": 596
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.054250850552469,
      "learning_rate": 8.87072808320951e-06,
      "loss": 1.1383,
      "step": 597
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.307742987532654,
      "learning_rate": 8.88558692421991e-06,
      "loss": 1.0894,
      "step": 598
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.629190813633109,
      "learning_rate": 8.900445765230312e-06,
      "loss": 1.016,
      "step": 599
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.475523864259122,
      "learning_rate": 8.915304606240715e-06,
      "loss": 1.0546,
      "step": 600
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.846085196322862,
      "learning_rate": 8.930163447251116e-06,
      "loss": 1.0179,
      "step": 601
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.955986427794002,
      "learning_rate": 8.945022288261517e-06,
      "loss": 1.0847,
      "step": 602
    },
    {
      "epoch": 0.05,
      "grad_norm": 13.644612383693392,
      "learning_rate": 8.959881129271917e-06,
      "loss": 1.0692,
      "step": 603
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.17322004505467,
      "learning_rate": 8.974739970282318e-06,
      "loss": 1.0944,
      "step": 604
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.024198208821508,
      "learning_rate": 8.98959881129272e-06,
      "loss": 1.0187,
      "step": 605
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.366883233623614,
      "learning_rate": 9.004457652303121e-06,
      "loss": 1.0238,
      "step": 606
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.91688822096214,
      "learning_rate": 9.019316493313522e-06,
      "loss": 1.0113,
      "step": 607
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.139097874384309,
      "learning_rate": 9.034175334323924e-06,
      "loss": 1.0747,
      "step": 608
    },
    {
      "epoch": 0.05,
      "grad_norm": 13.15402019332059,
      "learning_rate": 9.049034175334325e-06,
      "loss": 1.0488,
      "step": 609
    },
    {
      "epoch": 0.05,
      "grad_norm": 12.412005995499799,
      "learning_rate": 9.063893016344726e-06,
      "loss": 0.989,
      "step": 610
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.097195969900469,
      "learning_rate": 9.078751857355127e-06,
      "loss": 1.0094,
      "step": 611
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.618506413220849,
      "learning_rate": 9.093610698365529e-06,
      "loss": 1.1212,
      "step": 612
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.47134611274406,
      "learning_rate": 9.10846953937593e-06,
      "loss": 1.0212,
      "step": 613
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.558707914610387,
      "learning_rate": 9.12332838038633e-06,
      "loss": 1.0465,
      "step": 614
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.239846394359331,
      "learning_rate": 9.138187221396731e-06,
      "loss": 1.0336,
      "step": 615
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.868584636839273,
      "learning_rate": 9.153046062407132e-06,
      "loss": 1.0395,
      "step": 616
    },
    {
      "epoch": 0.06,
      "grad_norm": 12.931118231631226,
      "learning_rate": 9.167904903417533e-06,
      "loss": 1.0187,
      "step": 617
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.246990409768058,
      "learning_rate": 9.182763744427935e-06,
      "loss": 1.0288,
      "step": 618
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.87647692590792,
      "learning_rate": 9.197622585438336e-06,
      "loss": 0.9611,
      "step": 619
    },
    {
      "epoch": 0.06,
      "grad_norm": 11.46081953158234,
      "learning_rate": 9.212481426448737e-06,
      "loss": 1.1024,
      "step": 620
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.889556633143053,
      "learning_rate": 9.227340267459139e-06,
      "loss": 1.0969,
      "step": 621
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.925013080633466,
      "learning_rate": 9.24219910846954e-06,
      "loss": 1.0522,
      "step": 622
    },
    {
      "epoch": 0.06,
      "grad_norm": 11.364528651810405,
      "learning_rate": 9.257057949479941e-06,
      "loss": 1.0758,
      "step": 623
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.302806875329254,
      "learning_rate": 9.271916790490343e-06,
      "loss": 1.0135,
      "step": 624
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.758099993632804,
      "learning_rate": 9.286775631500744e-06,
      "loss": 1.0677,
      "step": 625
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.781732195415422,
      "learning_rate": 9.301634472511145e-06,
      "loss": 1.0507,
      "step": 626
    },
    {
      "epoch": 0.06,
      "grad_norm": 13.24137316254574,
      "learning_rate": 9.316493313521546e-06,
      "loss": 1.037,
      "step": 627
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.148056374662406,
      "learning_rate": 9.331352154531948e-06,
      "loss": 1.0403,
      "step": 628
    },
    {
      "epoch": 0.06,
      "grad_norm": 11.421279834087812,
      "learning_rate": 9.346210995542349e-06,
      "loss": 0.9584,
      "step": 629
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.285767992381842,
      "learning_rate": 9.36106983655275e-06,
      "loss": 1.0439,
      "step": 630
    },
    {
      "epoch": 0.06,
      "grad_norm": 14.191651208841742,
      "learning_rate": 9.375928677563152e-06,
      "loss": 1.0583,
      "step": 631
    },
    {
      "epoch": 0.06,
      "grad_norm": 11.2131632169938,
      "learning_rate": 9.390787518573551e-06,
      "loss": 1.0128,
      "step": 632
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.877191523982757,
      "learning_rate": 9.405646359583952e-06,
      "loss": 1.0106,
      "step": 633
    },
    {
      "epoch": 0.06,
      "grad_norm": 13.289356409960144,
      "learning_rate": 9.420505200594354e-06,
      "loss": 1.0275,
      "step": 634
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.452573514775871,
      "learning_rate": 9.435364041604755e-06,
      "loss": 0.9913,
      "step": 635
    },
    {
      "epoch": 0.06,
      "grad_norm": 13.19085334689723,
      "learning_rate": 9.450222882615156e-06,
      "loss": 0.9812,
      "step": 636
    },
    {
      "epoch": 0.06,
      "grad_norm": 14.432550063786342,
      "learning_rate": 9.465081723625558e-06,
      "loss": 1.0583,
      "step": 637
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.267224293385516,
      "learning_rate": 9.479940564635959e-06,
      "loss": 0.9808,
      "step": 638
    },
    {
      "epoch": 0.06,
      "grad_norm": 11.428995053984929,
      "learning_rate": 9.49479940564636e-06,
      "loss": 1.0211,
      "step": 639
    },
    {
      "epoch": 0.06,
      "grad_norm": 10.343755396807515,
      "learning_rate": 9.509658246656762e-06,
      "loss": 0.9794,
      "step": 640
    },
    {
      "epoch": 0.06,
      "grad_norm": 10.862853573250566,
      "learning_rate": 9.524517087667163e-06,
      "loss": 0.9794,
      "step": 641
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.001965834339806,
      "learning_rate": 9.539375928677564e-06,
      "loss": 1.0429,
      "step": 642
    },
    {
      "epoch": 0.06,
      "grad_norm": 10.247742617489289,
      "learning_rate": 9.554234769687965e-06,
      "loss": 1.0699,
      "step": 643
    },
    {
      "epoch": 0.06,
      "grad_norm": 10.11654038506872,
      "learning_rate": 9.569093610698367e-06,
      "loss": 1.0915,
      "step": 644
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.277941921939588,
      "learning_rate": 9.583952451708768e-06,
      "loss": 0.9911,
      "step": 645
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.339910802415295,
      "learning_rate": 9.59881129271917e-06,
      "loss": 1.0066,
      "step": 646
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.60651880666665,
      "learning_rate": 9.61367013372957e-06,
      "loss": 0.9628,
      "step": 647
    },
    {
      "epoch": 0.06,
      "grad_norm": 10.985092683082364,
      "learning_rate": 9.628528974739972e-06,
      "loss": 1.1053,
      "step": 648
    },
    {
      "epoch": 0.06,
      "grad_norm": 10.519234349314388,
      "learning_rate": 9.643387815750373e-06,
      "loss": 1.0782,
      "step": 649
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.924695852282168,
      "learning_rate": 9.658246656760773e-06,
      "loss": 0.9985,
      "step": 650
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.737882818286664,
      "learning_rate": 9.673105497771174e-06,
      "loss": 0.9668,
      "step": 651
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.988021742516203,
      "learning_rate": 9.687964338781575e-06,
      "loss": 0.9903,
      "step": 652
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.682210911072573,
      "learning_rate": 9.702823179791977e-06,
      "loss": 1.0642,
      "step": 653
    },
    {
      "epoch": 0.06,
      "grad_norm": 10.44704879972261,
      "learning_rate": 9.717682020802378e-06,
      "loss": 1.0436,
      "step": 654
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.846456270468657,
      "learning_rate": 9.73254086181278e-06,
      "loss": 1.0275,
      "step": 655
    },
    {
      "epoch": 0.06,
      "grad_norm": 10.240281836478205,
      "learning_rate": 9.74739970282318e-06,
      "loss": 1.0986,
      "step": 656
    },
    {
      "epoch": 0.06,
      "grad_norm": 11.521280315221375,
      "learning_rate": 9.762258543833582e-06,
      "loss": 1.0837,
      "step": 657
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.922877500881096,
      "learning_rate": 9.777117384843983e-06,
      "loss": 1.054,
      "step": 658
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.869601100798858,
      "learning_rate": 9.791976225854384e-06,
      "loss": 1.0691,
      "step": 659
    },
    {
      "epoch": 0.06,
      "grad_norm": 10.812401802226132,
      "learning_rate": 9.806835066864786e-06,
      "loss": 1.0014,
      "step": 660
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.4019707347110035,
      "learning_rate": 9.821693907875185e-06,
      "loss": 1.0295,
      "step": 661
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.22565453490378,
      "learning_rate": 9.836552748885587e-06,
      "loss": 0.9849,
      "step": 662
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.668102218136031,
      "learning_rate": 9.85141158989599e-06,
      "loss": 1.1027,
      "step": 663
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.600620178550768,
      "learning_rate": 9.866270430906391e-06,
      "loss": 1.0592,
      "step": 664
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.80767286027621,
      "learning_rate": 9.881129271916792e-06,
      "loss": 1.0542,
      "step": 665
    },
    {
      "epoch": 0.06,
      "grad_norm": 13.616585560979118,
      "learning_rate": 9.895988112927193e-06,
      "loss": 1.091,
      "step": 666
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.75139719210695,
      "learning_rate": 9.910846953937595e-06,
      "loss": 0.9999,
      "step": 667
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.090000507237583,
      "learning_rate": 9.925705794947994e-06,
      "loss": 1.0477,
      "step": 668
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.80835990273545,
      "learning_rate": 9.940564635958396e-06,
      "loss": 1.0699,
      "step": 669
    },
    {
      "epoch": 0.06,
      "grad_norm": 12.0778392045012,
      "learning_rate": 9.955423476968797e-06,
      "loss": 1.0987,
      "step": 670
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.447115539065816,
      "learning_rate": 9.970282317979198e-06,
      "loss": 0.987,
      "step": 671
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.6998551159311335,
      "learning_rate": 9.9851411589896e-06,
      "loss": 1.0205,
      "step": 672
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.264205615981833,
      "learning_rate": 1e-05,
      "loss": 1.0043,
      "step": 673
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.243592653176773,
      "learning_rate": 9.999999947817975e-06,
      "loss": 1.0488,
      "step": 674
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.048871830901943,
      "learning_rate": 9.999999791271902e-06,
      "loss": 1.117,
      "step": 675
    },
    {
      "epoch": 0.06,
      "grad_norm": 10.295754683650285,
      "learning_rate": 9.999999530361782e-06,
      "loss": 0.9816,
      "step": 676
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.793308101542664,
      "learning_rate": 9.999999165087622e-06,
      "loss": 0.996,
      "step": 677
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.832588797559488,
      "learning_rate": 9.999998695449427e-06,
      "loss": 1.0481,
      "step": 678
    },
    {
      "epoch": 0.06,
      "grad_norm": 11.269392462989188,
      "learning_rate": 9.999998121447212e-06,
      "loss": 1.0856,
      "step": 679
    },
    {
      "epoch": 0.06,
      "grad_norm": 10.616875405224269,
      "learning_rate": 9.999997443080984e-06,
      "loss": 1.0914,
      "step": 680
    },
    {
      "epoch": 0.06,
      "grad_norm": 12.14873935462538,
      "learning_rate": 9.999996660350762e-06,
      "loss": 0.9706,
      "step": 681
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.280734142264352,
      "learning_rate": 9.999995773256556e-06,
      "loss": 1.0384,
      "step": 682
    },
    {
      "epoch": 0.06,
      "grad_norm": 11.511662389933731,
      "learning_rate": 9.99999478179839e-06,
      "loss": 1.0213,
      "step": 683
    },
    {
      "epoch": 0.06,
      "grad_norm": 12.122756310484402,
      "learning_rate": 9.999993685976283e-06,
      "loss": 1.0219,
      "step": 684
    },
    {
      "epoch": 0.06,
      "grad_norm": 10.605958925750578,
      "learning_rate": 9.999992485790257e-06,
      "loss": 1.004,
      "step": 685
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.00937450456748,
      "learning_rate": 9.999991181240337e-06,
      "loss": 1.0438,
      "step": 686
    },
    {
      "epoch": 0.06,
      "grad_norm": 10.773401582333547,
      "learning_rate": 9.999989772326551e-06,
      "loss": 1.0334,
      "step": 687
    },
    {
      "epoch": 0.06,
      "grad_norm": 10.199019187336722,
      "learning_rate": 9.99998825904893e-06,
      "loss": 0.9849,
      "step": 688
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.591213520881503,
      "learning_rate": 9.999986641407503e-06,
      "loss": 0.9909,
      "step": 689
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.855999649063298,
      "learning_rate": 9.999984919402304e-06,
      "loss": 1.0558,
      "step": 690
    },
    {
      "epoch": 0.06,
      "grad_norm": 11.609528934572811,
      "learning_rate": 9.99998309303337e-06,
      "loss": 1.1312,
      "step": 691
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.25822169894325,
      "learning_rate": 9.999981162300737e-06,
      "loss": 1.1013,
      "step": 692
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.473279294889037,
      "learning_rate": 9.999979127204449e-06,
      "loss": 1.0089,
      "step": 693
    },
    {
      "epoch": 0.06,
      "grad_norm": 10.801398281963607,
      "learning_rate": 9.999976987744546e-06,
      "loss": 1.0259,
      "step": 694
    },
    {
      "epoch": 0.06,
      "grad_norm": 11.02365635691635,
      "learning_rate": 9.999974743921075e-06,
      "loss": 1.0535,
      "step": 695
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.14254965528287,
      "learning_rate": 9.999972395734078e-06,
      "loss": 1.1158,
      "step": 696
    },
    {
      "epoch": 0.06,
      "grad_norm": 11.058428729146204,
      "learning_rate": 9.999969943183609e-06,
      "loss": 1.0949,
      "step": 697
    },
    {
      "epoch": 0.06,
      "grad_norm": 12.54811137290809,
      "learning_rate": 9.999967386269715e-06,
      "loss": 0.9733,
      "step": 698
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.42264648426438,
      "learning_rate": 9.999964724992454e-06,
      "loss": 1.0079,
      "step": 699
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.351788792252671,
      "learning_rate": 9.999961959351878e-06,
      "loss": 1.065,
      "step": 700
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.9013842617969,
      "learning_rate": 9.999959089348046e-06,
      "loss": 0.9206,
      "step": 701
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.898082023335332,
      "learning_rate": 9.999956114981018e-06,
      "loss": 1.0439,
      "step": 702
    },
    {
      "epoch": 0.06,
      "grad_norm": 11.092520153388849,
      "learning_rate": 9.999953036250856e-06,
      "loss": 0.9985,
      "step": 703
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.415759563340771,
      "learning_rate": 9.999949853157622e-06,
      "loss": 0.9652,
      "step": 704
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.11054503597933,
      "learning_rate": 9.999946565701386e-06,
      "loss": 1.0073,
      "step": 705
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.211414960139495,
      "learning_rate": 9.999943173882217e-06,
      "loss": 1.1216,
      "step": 706
    },
    {
      "epoch": 0.06,
      "grad_norm": 12.546861706765409,
      "learning_rate": 9.99993967770018e-06,
      "loss": 1.0782,
      "step": 707
    },
    {
      "epoch": 0.06,
      "grad_norm": 11.26090994403747,
      "learning_rate": 9.999936077155354e-06,
      "loss": 1.0074,
      "step": 708
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.410344787489873,
      "learning_rate": 9.999932372247814e-06,
      "loss": 1.008,
      "step": 709
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.320652444145278,
      "learning_rate": 9.999928562977632e-06,
      "loss": 0.9628,
      "step": 710
    },
    {
      "epoch": 0.06,
      "grad_norm": 12.06172427321282,
      "learning_rate": 9.999924649344893e-06,
      "loss": 1.0455,
      "step": 711
    },
    {
      "epoch": 0.06,
      "grad_norm": 16.605841455059778,
      "learning_rate": 9.999920631349675e-06,
      "loss": 1.0124,
      "step": 712
    },
    {
      "epoch": 0.06,
      "grad_norm": 10.328585359914742,
      "learning_rate": 9.999916508992065e-06,
      "loss": 1.0578,
      "step": 713
    },
    {
      "epoch": 0.06,
      "grad_norm": 10.181999002413107,
      "learning_rate": 9.999912282272147e-06,
      "loss": 1.0358,
      "step": 714
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.354840183862304,
      "learning_rate": 9.999907951190012e-06,
      "loss": 0.996,
      "step": 715
    },
    {
      "epoch": 0.06,
      "grad_norm": 12.100788819220488,
      "learning_rate": 9.999903515745743e-06,
      "loss": 1.0094,
      "step": 716
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.19005721409269,
      "learning_rate": 9.99989897593944e-06,
      "loss": 1.0397,
      "step": 717
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.268806035840235,
      "learning_rate": 9.999894331771199e-06,
      "loss": 1.0061,
      "step": 718
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.104417139463868,
      "learning_rate": 9.99988958324111e-06,
      "loss": 0.9958,
      "step": 719
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.943324928367094,
      "learning_rate": 9.999884730349275e-06,
      "loss": 1.0365,
      "step": 720
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.067587799986475,
      "learning_rate": 9.999879773095797e-06,
      "loss": 1.0724,
      "step": 721
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.455956516572698,
      "learning_rate": 9.999874711480777e-06,
      "loss": 1.1609,
      "step": 722
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.02512357440648,
      "learning_rate": 9.999869545504323e-06,
      "loss": 0.9997,
      "step": 723
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.348670272927727,
      "learning_rate": 9.999864275166543e-06,
      "loss": 1.0419,
      "step": 724
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.365832459444025,
      "learning_rate": 9.999858900467544e-06,
      "loss": 1.0545,
      "step": 725
    },
    {
      "epoch": 0.06,
      "grad_norm": 10.724433725506643,
      "learning_rate": 9.99985342140744e-06,
      "loss": 1.1011,
      "step": 726
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.38999422106584,
      "learning_rate": 9.999847837986346e-06,
      "loss": 0.9414,
      "step": 727
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.355783246357326,
      "learning_rate": 9.999842150204378e-06,
      "loss": 1.036,
      "step": 728
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.500280506504824,
      "learning_rate": 9.999836358061654e-06,
      "loss": 1.0393,
      "step": 729
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.87688369430259,
      "learning_rate": 9.999830461558294e-06,
      "loss": 0.9788,
      "step": 730
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.487285433517528,
      "learning_rate": 9.999824460694426e-06,
      "loss": 1.0261,
      "step": 731
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.278259370395507,
      "learning_rate": 9.99981835547017e-06,
      "loss": 1.0198,
      "step": 732
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.353512627340146,
      "learning_rate": 9.999812145885655e-06,
      "loss": 1.0128,
      "step": 733
    },
    {
      "epoch": 0.07,
      "grad_norm": 12.194894501542754,
      "learning_rate": 9.999805831941011e-06,
      "loss": 1.0391,
      "step": 734
    },
    {
      "epoch": 0.07,
      "grad_norm": 12.278488452410807,
      "learning_rate": 9.99979941363637e-06,
      "loss": 1.0658,
      "step": 735
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.7849860943382,
      "learning_rate": 9.999792890971867e-06,
      "loss": 1.112,
      "step": 736
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.113042960238992,
      "learning_rate": 9.999786263947634e-06,
      "loss": 1.0553,
      "step": 737
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.295082946096922,
      "learning_rate": 9.999779532563813e-06,
      "loss": 1.0102,
      "step": 738
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.011148920919268,
      "learning_rate": 9.999772696820545e-06,
      "loss": 1.0312,
      "step": 739
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.284219622396297,
      "learning_rate": 9.999765756717969e-06,
      "loss": 1.1135,
      "step": 740
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.154364594552943,
      "learning_rate": 9.999758712256233e-06,
      "loss": 1.0912,
      "step": 741
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.411272701943217,
      "learning_rate": 9.999751563435484e-06,
      "loss": 1.0212,
      "step": 742
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.233692824869369,
      "learning_rate": 9.99974431025587e-06,
      "loss": 1.0986,
      "step": 743
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.337589319745028,
      "learning_rate": 9.999736952717543e-06,
      "loss": 1.0235,
      "step": 744
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.85153960598842,
      "learning_rate": 9.999729490820656e-06,
      "loss": 1.0359,
      "step": 745
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.694478713750327,
      "learning_rate": 9.999721924565364e-06,
      "loss": 1.0541,
      "step": 746
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.384072501678186,
      "learning_rate": 9.999714253951828e-06,
      "loss": 1.0083,
      "step": 747
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.872780624083503,
      "learning_rate": 9.999706478980204e-06,
      "loss": 0.9994,
      "step": 748
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.931982827221065,
      "learning_rate": 9.999698599650658e-06,
      "loss": 1.002,
      "step": 749
    },
    {
      "epoch": 0.07,
      "grad_norm": 12.545860358133144,
      "learning_rate": 9.999690615963353e-06,
      "loss": 1.0254,
      "step": 750
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.453842638810778,
      "learning_rate": 9.999682527918456e-06,
      "loss": 0.9963,
      "step": 751
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.064690516333482,
      "learning_rate": 9.999674335516134e-06,
      "loss": 1.0396,
      "step": 752
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.622988431681986,
      "learning_rate": 9.999666038756559e-06,
      "loss": 1.0164,
      "step": 753
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.651091173457328,
      "learning_rate": 9.999657637639905e-06,
      "loss": 0.969,
      "step": 754
    },
    {
      "epoch": 0.07,
      "grad_norm": 13.024906087390393,
      "learning_rate": 9.999649132166346e-06,
      "loss": 1.0161,
      "step": 755
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.093115796782326,
      "learning_rate": 9.999640522336063e-06,
      "loss": 0.9684,
      "step": 756
    },
    {
      "epoch": 0.07,
      "grad_norm": 12.654367805598046,
      "learning_rate": 9.999631808149232e-06,
      "loss": 0.9895,
      "step": 757
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.795490344263026,
      "learning_rate": 9.999622989606035e-06,
      "loss": 1.0439,
      "step": 758
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.57024042655462,
      "learning_rate": 9.999614066706658e-06,
      "loss": 1.0264,
      "step": 759
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.140255222045084,
      "learning_rate": 9.999605039451287e-06,
      "loss": 1.0599,
      "step": 760
    },
    {
      "epoch": 0.07,
      "grad_norm": 12.054863774305609,
      "learning_rate": 9.999595907840109e-06,
      "loss": 1.0146,
      "step": 761
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.042345375129704,
      "learning_rate": 9.999586671873313e-06,
      "loss": 1.0749,
      "step": 762
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.213088850421277,
      "learning_rate": 9.999577331551098e-06,
      "loss": 1.049,
      "step": 763
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.598328543746009,
      "learning_rate": 9.999567886873653e-06,
      "loss": 1.0024,
      "step": 764
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.016189945664944,
      "learning_rate": 9.999558337841178e-06,
      "loss": 0.9824,
      "step": 765
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.658363974695595,
      "learning_rate": 9.99954868445387e-06,
      "loss": 1.0759,
      "step": 766
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.540760676020042,
      "learning_rate": 9.999538926711934e-06,
      "loss": 0.931,
      "step": 767
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.116209465766385,
      "learning_rate": 9.999529064615572e-06,
      "loss": 0.9968,
      "step": 768
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.765539189998801,
      "learning_rate": 9.999519098164988e-06,
      "loss": 0.9805,
      "step": 769
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.129203744777987,
      "learning_rate": 9.999509027360392e-06,
      "loss": 0.9941,
      "step": 770
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.199921314915212,
      "learning_rate": 9.999498852201994e-06,
      "loss": 0.9723,
      "step": 771
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.65799959388264,
      "learning_rate": 9.999488572690006e-06,
      "loss": 1.0301,
      "step": 772
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.474276731752648,
      "learning_rate": 9.999478188824643e-06,
      "loss": 0.9889,
      "step": 773
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.516014951082246,
      "learning_rate": 9.99946770060612e-06,
      "loss": 0.9742,
      "step": 774
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.131651491301703,
      "learning_rate": 9.99945710803466e-06,
      "loss": 1.006,
      "step": 775
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.597963201989241,
      "learning_rate": 9.999446411110479e-06,
      "loss": 0.9896,
      "step": 776
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.892245534996906,
      "learning_rate": 9.999435609833805e-06,
      "loss": 1.0729,
      "step": 777
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.54211489802198,
      "learning_rate": 9.99942470420486e-06,
      "loss": 1.0505,
      "step": 778
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.403306720055143,
      "learning_rate": 9.999413694223874e-06,
      "loss": 1.1076,
      "step": 779
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.110372965685114,
      "learning_rate": 9.999402579891073e-06,
      "loss": 1.0224,
      "step": 780
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.650474474171203,
      "learning_rate": 9.999391361206694e-06,
      "loss": 0.9987,
      "step": 781
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.842257697809902,
      "learning_rate": 9.999380038170967e-06,
      "loss": 1.0388,
      "step": 782
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.956807976120622,
      "learning_rate": 9.999368610784132e-06,
      "loss": 0.9939,
      "step": 783
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.823582455309587,
      "learning_rate": 9.999357079046426e-06,
      "loss": 1.0487,
      "step": 784
    },
    {
      "epoch": 0.07,
      "grad_norm": 12.690994313193038,
      "learning_rate": 9.999345442958088e-06,
      "loss": 1.0757,
      "step": 785
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.154045596999365,
      "learning_rate": 9.999333702519363e-06,
      "loss": 0.9684,
      "step": 786
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.605625628572112,
      "learning_rate": 9.999321857730494e-06,
      "loss": 1.0364,
      "step": 787
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.150052320692485,
      "learning_rate": 9.99930990859173e-06,
      "loss": 1.0275,
      "step": 788
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.44157761578933,
      "learning_rate": 9.99929785510332e-06,
      "loss": 0.9377,
      "step": 789
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.182123953158637,
      "learning_rate": 9.999285697265518e-06,
      "loss": 1.0636,
      "step": 790
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.996675277542571,
      "learning_rate": 9.999273435078572e-06,
      "loss": 1.0844,
      "step": 791
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.47072888980035,
      "learning_rate": 9.999261068542742e-06,
      "loss": 0.9886,
      "step": 792
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.004159944319575,
      "learning_rate": 9.999248597658286e-06,
      "loss": 1.0563,
      "step": 793
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.730669225740804,
      "learning_rate": 9.999236022425464e-06,
      "loss": 0.9809,
      "step": 794
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.866075027213889,
      "learning_rate": 9.999223342844538e-06,
      "loss": 1.0083,
      "step": 795
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.101345986166228,
      "learning_rate": 9.999210558915773e-06,
      "loss": 1.0798,
      "step": 796
    },
    {
      "epoch": 0.07,
      "grad_norm": 13.14234019942694,
      "learning_rate": 9.999197670639436e-06,
      "loss": 1.0139,
      "step": 797
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.938183447367406,
      "learning_rate": 9.999184678015793e-06,
      "loss": 1.0893,
      "step": 798
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.531737180382077,
      "learning_rate": 9.999171581045121e-06,
      "loss": 1.0059,
      "step": 799
    },
    {
      "epoch": 0.07,
      "grad_norm": 12.890705828986743,
      "learning_rate": 9.99915837972769e-06,
      "loss": 0.9439,
      "step": 800
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.074308649026603,
      "learning_rate": 9.999145074063773e-06,
      "loss": 1.0173,
      "step": 801
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.111397352114071,
      "learning_rate": 9.999131664053655e-06,
      "loss": 0.9925,
      "step": 802
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.45684403739392,
      "learning_rate": 9.999118149697608e-06,
      "loss": 1.0679,
      "step": 803
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.75158959833569,
      "learning_rate": 9.999104530995918e-06,
      "loss": 0.9635,
      "step": 804
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.6376594499753985,
      "learning_rate": 9.99909080794887e-06,
      "loss": 1.0223,
      "step": 805
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.735355647193114,
      "learning_rate": 9.999076980556748e-06,
      "loss": 1.0069,
      "step": 806
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.637124064114406,
      "learning_rate": 9.999063048819841e-06,
      "loss": 1.0236,
      "step": 807
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.989035332072238,
      "learning_rate": 9.999049012738441e-06,
      "loss": 1.018,
      "step": 808
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.853823401855315,
      "learning_rate": 9.999034872312842e-06,
      "loss": 0.988,
      "step": 809
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.727369710415346,
      "learning_rate": 9.999020627543338e-06,
      "loss": 0.9607,
      "step": 810
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.661500215412708,
      "learning_rate": 9.999006278430226e-06,
      "loss": 1.0492,
      "step": 811
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.531914949504957,
      "learning_rate": 9.998991824973804e-06,
      "loss": 0.9437,
      "step": 812
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.155739567576108,
      "learning_rate": 9.998977267174375e-06,
      "loss": 1.0163,
      "step": 813
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.5295814885244,
      "learning_rate": 9.998962605032246e-06,
      "loss": 1.0019,
      "step": 814
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.066040399773762,
      "learning_rate": 9.998947838547717e-06,
      "loss": 0.9334,
      "step": 815
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.294368526879245,
      "learning_rate": 9.9989329677211e-06,
      "loss": 1.015,
      "step": 816
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.381955374479535,
      "learning_rate": 9.998917992552706e-06,
      "loss": 1.0307,
      "step": 817
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.237152025811778,
      "learning_rate": 9.998902913042845e-06,
      "loss": 0.9688,
      "step": 818
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.765137113143844,
      "learning_rate": 9.998887729191836e-06,
      "loss": 0.9356,
      "step": 819
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.225831107915645,
      "learning_rate": 9.99887244099999e-06,
      "loss": 1.0306,
      "step": 820
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.0452868425090065,
      "learning_rate": 9.99885704846763e-06,
      "loss": 0.9511,
      "step": 821
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.225563280331507,
      "learning_rate": 9.998841551595076e-06,
      "loss": 1.0141,
      "step": 822
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.733521065070352,
      "learning_rate": 9.998825950382654e-06,
      "loss": 0.9283,
      "step": 823
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.167960354591541,
      "learning_rate": 9.998810244830685e-06,
      "loss": 0.9941,
      "step": 824
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.078901240952447,
      "learning_rate": 9.9987944349395e-06,
      "loss": 0.9588,
      "step": 825
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.475670378710102,
      "learning_rate": 9.99877852070943e-06,
      "loss": 1.0018,
      "step": 826
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.574197989227159,
      "learning_rate": 9.998762502140802e-06,
      "loss": 0.9755,
      "step": 827
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.825195363997548,
      "learning_rate": 9.998746379233958e-06,
      "loss": 0.9232,
      "step": 828
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.24960207617306,
      "learning_rate": 9.998730151989227e-06,
      "loss": 0.9687,
      "step": 829
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.130660165568324,
      "learning_rate": 9.998713820406953e-06,
      "loss": 1.0033,
      "step": 830
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.801358023593332,
      "learning_rate": 9.998697384487475e-06,
      "loss": 1.0019,
      "step": 831
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.516486898398215,
      "learning_rate": 9.998680844231135e-06,
      "loss": 1.0064,
      "step": 832
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.091949697729644,
      "learning_rate": 9.998664199638279e-06,
      "loss": 0.9589,
      "step": 833
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.00163035050054,
      "learning_rate": 9.998647450709257e-06,
      "loss": 0.9503,
      "step": 834
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.4881404341583195,
      "learning_rate": 9.998630597444415e-06,
      "loss": 0.972,
      "step": 835
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.784424487511476,
      "learning_rate": 9.998613639844104e-06,
      "loss": 0.9762,
      "step": 836
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.640234309768429,
      "learning_rate": 9.998596577908684e-06,
      "loss": 0.9857,
      "step": 837
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.326280822648059,
      "learning_rate": 9.998579411638506e-06,
      "loss": 1.1463,
      "step": 838
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.079300819160272,
      "learning_rate": 9.998562141033927e-06,
      "loss": 1.051,
      "step": 839
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.831602318197288,
      "learning_rate": 9.998544766095312e-06,
      "loss": 0.9534,
      "step": 840
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.875888004271978,
      "learning_rate": 9.998527286823021e-06,
      "loss": 0.9565,
      "step": 841
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.037118217640445,
      "learning_rate": 9.99850970321742e-06,
      "loss": 0.9761,
      "step": 842
    },
    {
      "epoch": 0.08,
      "grad_norm": 11.234180849248652,
      "learning_rate": 9.998492015278874e-06,
      "loss": 1.032,
      "step": 843
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.162553028475706,
      "learning_rate": 9.998474223007756e-06,
      "loss": 1.0046,
      "step": 844
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.006100919616056,
      "learning_rate": 9.998456326404433e-06,
      "loss": 0.9509,
      "step": 845
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.278200612900712,
      "learning_rate": 9.998438325469282e-06,
      "loss": 1.0242,
      "step": 846
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.707781213665642,
      "learning_rate": 9.998420220202678e-06,
      "loss": 0.9787,
      "step": 847
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.155020684922716,
      "learning_rate": 9.998402010604998e-06,
      "loss": 1.0536,
      "step": 848
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.321189351701641,
      "learning_rate": 9.998383696676621e-06,
      "loss": 1.0116,
      "step": 849
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.436086532330405,
      "learning_rate": 9.998365278417929e-06,
      "loss": 1.0244,
      "step": 850
    },
    {
      "epoch": 0.08,
      "grad_norm": 11.062238640879263,
      "learning_rate": 9.998346755829311e-06,
      "loss": 1.0074,
      "step": 851
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.365784590412884,
      "learning_rate": 9.99832812891115e-06,
      "loss": 1.0249,
      "step": 852
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.993906303087318,
      "learning_rate": 9.998309397663836e-06,
      "loss": 1.0291,
      "step": 853
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.959739139175088,
      "learning_rate": 9.998290562087758e-06,
      "loss": 1.0051,
      "step": 854
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.877176139961323,
      "learning_rate": 9.998271622183313e-06,
      "loss": 0.9566,
      "step": 855
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.657500736562499,
      "learning_rate": 9.998252577950891e-06,
      "loss": 0.9721,
      "step": 856
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.279961408586907,
      "learning_rate": 9.998233429390893e-06,
      "loss": 0.9345,
      "step": 857
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.6492197981166505,
      "learning_rate": 9.99821417650372e-06,
      "loss": 0.995,
      "step": 858
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.207463039174929,
      "learning_rate": 9.998194819289771e-06,
      "loss": 1.0159,
      "step": 859
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.417261353648795,
      "learning_rate": 9.998175357749451e-06,
      "loss": 1.0428,
      "step": 860
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.03448636351223,
      "learning_rate": 9.998155791883166e-06,
      "loss": 0.971,
      "step": 861
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.536641322969025,
      "learning_rate": 9.998136121691325e-06,
      "loss": 0.9238,
      "step": 862
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.227101715895324,
      "learning_rate": 9.998116347174336e-06,
      "loss": 1.0306,
      "step": 863
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.727707075132388,
      "learning_rate": 9.998096468332617e-06,
      "loss": 1.0229,
      "step": 864
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.915024409991847,
      "learning_rate": 9.998076485166579e-06,
      "loss": 1.0229,
      "step": 865
    },
    {
      "epoch": 0.08,
      "grad_norm": 12.062271159379488,
      "learning_rate": 9.998056397676639e-06,
      "loss": 1.0375,
      "step": 866
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.055285114291404,
      "learning_rate": 9.998036205863218e-06,
      "loss": 0.9396,
      "step": 867
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.718616704882963,
      "learning_rate": 9.998015909726737e-06,
      "loss": 0.9402,
      "step": 868
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.97907063207494,
      "learning_rate": 9.997995509267619e-06,
      "loss": 0.9725,
      "step": 869
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.458216468025302,
      "learning_rate": 9.99797500448629e-06,
      "loss": 0.9991,
      "step": 870
    },
    {
      "epoch": 0.08,
      "grad_norm": 12.035208430367854,
      "learning_rate": 9.997954395383178e-06,
      "loss": 0.954,
      "step": 871
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.0667790566975,
      "learning_rate": 9.997933681958714e-06,
      "loss": 0.9475,
      "step": 872
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.78850028611461,
      "learning_rate": 9.997912864213327e-06,
      "loss": 0.9803,
      "step": 873
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.761347585011826,
      "learning_rate": 9.99789194214746e-06,
      "loss": 0.9012,
      "step": 874
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.212051971180096,
      "learning_rate": 9.997870915761539e-06,
      "loss": 1.0111,
      "step": 875
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.956251670008232,
      "learning_rate": 9.997849785056009e-06,
      "loss": 0.9894,
      "step": 876
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.261481843759254,
      "learning_rate": 9.99782855003131e-06,
      "loss": 0.9911,
      "step": 877
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.400987475573911,
      "learning_rate": 9.997807210687884e-06,
      "loss": 1.0265,
      "step": 878
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.06650923897517,
      "learning_rate": 9.997785767026181e-06,
      "loss": 0.9558,
      "step": 879
    },
    {
      "epoch": 0.08,
      "grad_norm": 15.783081219499563,
      "learning_rate": 9.997764219046641e-06,
      "loss": 1.0335,
      "step": 880
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.489237644381799,
      "learning_rate": 9.99774256674972e-06,
      "loss": 0.9928,
      "step": 881
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.93508368946788,
      "learning_rate": 9.997720810135867e-06,
      "loss": 1.0215,
      "step": 882
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.858578107029446,
      "learning_rate": 9.997698949205536e-06,
      "loss": 0.9998,
      "step": 883
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.351754877748318,
      "learning_rate": 9.997676983959188e-06,
      "loss": 1.0109,
      "step": 884
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.09382991863049,
      "learning_rate": 9.997654914397274e-06,
      "loss": 0.9956,
      "step": 885
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.776792049680456,
      "learning_rate": 9.99763274052026e-06,
      "loss": 1.0228,
      "step": 886
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.450315848583875,
      "learning_rate": 9.997610462328607e-06,
      "loss": 0.9731,
      "step": 887
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.799075565391173,
      "learning_rate": 9.997588079822778e-06,
      "loss": 0.9824,
      "step": 888
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.147937211235977,
      "learning_rate": 9.997565593003245e-06,
      "loss": 0.9904,
      "step": 889
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.941964904194568,
      "learning_rate": 9.997543001870474e-06,
      "loss": 0.9687,
      "step": 890
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.19685619005091,
      "learning_rate": 9.997520306424936e-06,
      "loss": 0.9839,
      "step": 891
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.280410608907395,
      "learning_rate": 9.997497506667108e-06,
      "loss": 0.9352,
      "step": 892
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.931891914920883,
      "learning_rate": 9.997474602597463e-06,
      "loss": 1.0402,
      "step": 893
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.541957455184061,
      "learning_rate": 9.99745159421648e-06,
      "loss": 1.0096,
      "step": 894
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.482572192689705,
      "learning_rate": 9.997428481524639e-06,
      "loss": 1.008,
      "step": 895
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.772837817630702,
      "learning_rate": 9.997405264522423e-06,
      "loss": 1.0143,
      "step": 896
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.793025104270116,
      "learning_rate": 9.997381943210316e-06,
      "loss": 0.9671,
      "step": 897
    },
    {
      "epoch": 0.08,
      "grad_norm": 11.701773131723733,
      "learning_rate": 9.997358517588805e-06,
      "loss": 0.9655,
      "step": 898
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.727101413057323,
      "learning_rate": 9.99733498765838e-06,
      "loss": 0.9884,
      "step": 899
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.093472865831096,
      "learning_rate": 9.99731135341953e-06,
      "loss": 1.0063,
      "step": 900
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.88420339791517,
      "learning_rate": 9.997287614872749e-06,
      "loss": 1.0019,
      "step": 901
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.872546322008035,
      "learning_rate": 9.997263772018533e-06,
      "loss": 0.9245,
      "step": 902
    },
    {
      "epoch": 0.08,
      "grad_norm": 11.015538360963271,
      "learning_rate": 9.99723982485738e-06,
      "loss": 1.0939,
      "step": 903
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.590160878618287,
      "learning_rate": 9.99721577338979e-06,
      "loss": 0.9454,
      "step": 904
    },
    {
      "epoch": 0.08,
      "grad_norm": 11.36996702539662,
      "learning_rate": 9.997191617616264e-06,
      "loss": 0.9516,
      "step": 905
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.611965931153284,
      "learning_rate": 9.997167357537306e-06,
      "loss": 0.9522,
      "step": 906
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.380929859210196,
      "learning_rate": 9.997142993153423e-06,
      "loss": 0.9539,
      "step": 907
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.362117930071422,
      "learning_rate": 9.997118524465126e-06,
      "loss": 0.9432,
      "step": 908
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.629403569424733,
      "learning_rate": 9.99709395147292e-06,
      "loss": 0.9524,
      "step": 909
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.784298503072442,
      "learning_rate": 9.997069274177323e-06,
      "loss": 1.0483,
      "step": 910
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.08787003949281,
      "learning_rate": 9.997044492578847e-06,
      "loss": 0.9112,
      "step": 911
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.536530726793412,
      "learning_rate": 9.997019606678014e-06,
      "loss": 0.9864,
      "step": 912
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.023570588211143,
      "learning_rate": 9.996994616475336e-06,
      "loss": 0.99,
      "step": 913
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.34347507234705,
      "learning_rate": 9.996969521971341e-06,
      "loss": 0.9845,
      "step": 914
    },
    {
      "epoch": 0.08,
      "grad_norm": 12.132376942691971,
      "learning_rate": 9.996944323166549e-06,
      "loss": 1.0304,
      "step": 915
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.210421513737613,
      "learning_rate": 9.996919020061489e-06,
      "loss": 0.9442,
      "step": 916
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.70461688922444,
      "learning_rate": 9.996893612656684e-06,
      "loss": 0.9708,
      "step": 917
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.270826096425844,
      "learning_rate": 9.996868100952671e-06,
      "loss": 0.9563,
      "step": 918
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.41158915972332,
      "learning_rate": 9.99684248494998e-06,
      "loss": 1.0305,
      "step": 919
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.527136962549797,
      "learning_rate": 9.996816764649143e-06,
      "loss": 1.0226,
      "step": 920
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.923163325254967,
      "learning_rate": 9.9967909400507e-06,
      "loss": 0.9901,
      "step": 921
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.489260378612245,
      "learning_rate": 9.996765011155187e-06,
      "loss": 0.9287,
      "step": 922
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.730763089215504,
      "learning_rate": 9.996738977963149e-06,
      "loss": 0.959,
      "step": 923
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.570788783522387,
      "learning_rate": 9.996712840475127e-06,
      "loss": 1.0125,
      "step": 924
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.914867656292998,
      "learning_rate": 9.996686598691668e-06,
      "loss": 1.018,
      "step": 925
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.012072153148338,
      "learning_rate": 9.996660252613317e-06,
      "loss": 1.0536,
      "step": 926
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.628037688052148,
      "learning_rate": 9.996633802240627e-06,
      "loss": 0.9862,
      "step": 927
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.599329495026513,
      "learning_rate": 9.996607247574148e-06,
      "loss": 1.048,
      "step": 928
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.06679572032971,
      "learning_rate": 9.996580588614437e-06,
      "loss": 1.0354,
      "step": 929
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.748633377020838,
      "learning_rate": 9.996553825362046e-06,
      "loss": 1.0221,
      "step": 930
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.100222886869438,
      "learning_rate": 9.996526957817538e-06,
      "loss": 1.0547,
      "step": 931
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.878659049340408,
      "learning_rate": 9.99649998598147e-06,
      "loss": 1.003,
      "step": 932
    },
    {
      "epoch": 0.08,
      "grad_norm": 17.56335431343529,
      "learning_rate": 9.99647290985441e-06,
      "loss": 0.956,
      "step": 933
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.07651413400778,
      "learning_rate": 9.996445729436918e-06,
      "loss": 0.9476,
      "step": 934
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.709840946161437,
      "learning_rate": 9.996418444729562e-06,
      "loss": 0.9497,
      "step": 935
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.487039521473529,
      "learning_rate": 9.996391055732916e-06,
      "loss": 0.9808,
      "step": 936
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.5830926198785775,
      "learning_rate": 9.996363562447548e-06,
      "loss": 1.0021,
      "step": 937
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.176799916541752,
      "learning_rate": 9.996335964874031e-06,
      "loss": 1.0003,
      "step": 938
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.134740072491587,
      "learning_rate": 9.996308263012944e-06,
      "loss": 0.9795,
      "step": 939
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.219140081754393,
      "learning_rate": 9.996280456864862e-06,
      "loss": 0.9343,
      "step": 940
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.7292352429680955,
      "learning_rate": 9.99625254643037e-06,
      "loss": 0.9485,
      "step": 941
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.851053910202773,
      "learning_rate": 9.996224531710045e-06,
      "loss": 0.95,
      "step": 942
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.910862164118473,
      "learning_rate": 9.996196412704476e-06,
      "loss": 0.8952,
      "step": 943
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.638752770057254,
      "learning_rate": 9.996168189414246e-06,
      "loss": 1.0271,
      "step": 944
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.735001452753258,
      "learning_rate": 9.996139861839948e-06,
      "loss": 0.9979,
      "step": 945
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.47907691906831,
      "learning_rate": 9.996111429982173e-06,
      "loss": 0.9693,
      "step": 946
    },
    {
      "epoch": 0.08,
      "grad_norm": 11.569904804829509,
      "learning_rate": 9.996082893841512e-06,
      "loss": 1.009,
      "step": 947
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.591808252095188,
      "learning_rate": 9.996054253418561e-06,
      "loss": 1.0038,
      "step": 948
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.845054822540564,
      "learning_rate": 9.99602550871392e-06,
      "loss": 1.0059,
      "step": 949
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.457469461067877,
      "learning_rate": 9.995996659728187e-06,
      "loss": 0.9651,
      "step": 950
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.309180622295747,
      "learning_rate": 9.995967706461965e-06,
      "loss": 1.0216,
      "step": 951
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.206846805290429,
      "learning_rate": 9.995938648915858e-06,
      "loss": 1.0205,
      "step": 952
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.567442708562054,
      "learning_rate": 9.995909487090472e-06,
      "loss": 1.0346,
      "step": 953
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.906991452894864,
      "learning_rate": 9.995880220986417e-06,
      "loss": 1.0321,
      "step": 954
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.974302840235083,
      "learning_rate": 9.995850850604304e-06,
      "loss": 1.0519,
      "step": 955
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.6733007572190814,
      "learning_rate": 9.995821375944744e-06,
      "loss": 0.9607,
      "step": 956
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.33287573867397,
      "learning_rate": 9.995791797008354e-06,
      "loss": 0.9939,
      "step": 957
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.334985897257653,
      "learning_rate": 9.99576211379575e-06,
      "loss": 1.062,
      "step": 958
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.495972847324932,
      "learning_rate": 9.995732326307555e-06,
      "loss": 1.0429,
      "step": 959
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.054117600563726,
      "learning_rate": 9.995702434544386e-06,
      "loss": 0.9841,
      "step": 960
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.332252818361094,
      "learning_rate": 9.99567243850687e-06,
      "loss": 1.0064,
      "step": 961
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.669724934460178,
      "learning_rate": 9.99564233819563e-06,
      "loss": 0.9798,
      "step": 962
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.879399023141424,
      "learning_rate": 9.995612133611298e-06,
      "loss": 0.9366,
      "step": 963
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.055374751446326,
      "learning_rate": 9.995581824754503e-06,
      "loss": 0.9377,
      "step": 964
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.889184047874464,
      "learning_rate": 9.995551411625877e-06,
      "loss": 0.9907,
      "step": 965
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.858808648356668,
      "learning_rate": 9.995520894226056e-06,
      "loss": 0.9629,
      "step": 966
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.342147609876978,
      "learning_rate": 9.995490272555677e-06,
      "loss": 0.934,
      "step": 967
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.16091061706455,
      "learning_rate": 9.995459546615376e-06,
      "loss": 1.0047,
      "step": 968
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.295426492436262,
      "learning_rate": 9.995428716405799e-06,
      "loss": 0.9277,
      "step": 969
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.415259242933013,
      "learning_rate": 9.995397781927587e-06,
      "loss": 0.9295,
      "step": 970
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.646342171555547,
      "learning_rate": 9.995366743181385e-06,
      "loss": 0.9686,
      "step": 971
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.972979944484932,
      "learning_rate": 9.995335600167844e-06,
      "loss": 0.9972,
      "step": 972
    },
    {
      "epoch": 0.09,
      "grad_norm": 12.108019848529077,
      "learning_rate": 9.995304352887609e-06,
      "loss": 0.9574,
      "step": 973
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.096087381507601,
      "learning_rate": 9.995273001341338e-06,
      "loss": 0.9558,
      "step": 974
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.197171999669669,
      "learning_rate": 9.995241545529682e-06,
      "loss": 0.9846,
      "step": 975
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.256180287677908,
      "learning_rate": 9.995209985453296e-06,
      "loss": 0.9579,
      "step": 976
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.724295615278792,
      "learning_rate": 9.995178321112843e-06,
      "loss": 0.9875,
      "step": 977
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.340082744011708,
      "learning_rate": 9.995146552508981e-06,
      "loss": 0.971,
      "step": 978
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.404431209629506,
      "learning_rate": 9.995114679642374e-06,
      "loss": 1.0256,
      "step": 979
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.6436059677923085,
      "learning_rate": 9.995082702513688e-06,
      "loss": 0.986,
      "step": 980
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.109019610112588,
      "learning_rate": 9.995050621123589e-06,
      "loss": 0.9629,
      "step": 981
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.196770569742844,
      "learning_rate": 9.995018435472746e-06,
      "loss": 0.9324,
      "step": 982
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.048181166353656,
      "learning_rate": 9.994986145561833e-06,
      "loss": 0.9582,
      "step": 983
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.95397075644332,
      "learning_rate": 9.994953751391525e-06,
      "loss": 1.0406,
      "step": 984
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.048399361843256,
      "learning_rate": 9.994921252962492e-06,
      "loss": 0.9553,
      "step": 985
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.34785946920711,
      "learning_rate": 9.99488865027542e-06,
      "loss": 1.0136,
      "step": 986
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.318231643650801,
      "learning_rate": 9.994855943330986e-06,
      "loss": 1.001,
      "step": 987
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.306159909055284,
      "learning_rate": 9.994823132129871e-06,
      "loss": 0.9336,
      "step": 988
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.280946909804913,
      "learning_rate": 9.994790216672763e-06,
      "loss": 0.9473,
      "step": 989
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.316331266857857,
      "learning_rate": 9.994757196960348e-06,
      "loss": 0.938,
      "step": 990
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.323957963735286,
      "learning_rate": 9.994724072993313e-06,
      "loss": 0.9509,
      "step": 991
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.201511993466811,
      "learning_rate": 9.994690844772352e-06,
      "loss": 0.9711,
      "step": 992
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.087015668346059,
      "learning_rate": 9.994657512298158e-06,
      "loss": 1.0163,
      "step": 993
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.12000217774927,
      "learning_rate": 9.994624075571426e-06,
      "loss": 0.9576,
      "step": 994
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.591411847605315,
      "learning_rate": 9.994590534592854e-06,
      "loss": 0.9179,
      "step": 995
    },
    {
      "epoch": 0.09,
      "grad_norm": 14.8839240075901,
      "learning_rate": 9.994556889363145e-06,
      "loss": 0.938,
      "step": 996
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.344409484816145,
      "learning_rate": 9.994523139882997e-06,
      "loss": 0.9651,
      "step": 997
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.154111922309497,
      "learning_rate": 9.994489286153116e-06,
      "loss": 0.9531,
      "step": 998
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.514055631924581,
      "learning_rate": 9.994455328174209e-06,
      "loss": 0.9537,
      "step": 999
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.484204683552593,
      "learning_rate": 9.994421265946984e-06,
      "loss": 0.9449,
      "step": 1000
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.124542686530342,
      "learning_rate": 9.994387099472156e-06,
      "loss": 0.9988,
      "step": 1001
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.258402495003683,
      "learning_rate": 9.994352828750431e-06,
      "loss": 1.0371,
      "step": 1002
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.658579577391695,
      "learning_rate": 9.99431845378253e-06,
      "loss": 0.9525,
      "step": 1003
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.680866866367063,
      "learning_rate": 9.99428397456917e-06,
      "loss": 0.9774,
      "step": 1004
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.478438877493712,
      "learning_rate": 9.994249391111067e-06,
      "loss": 0.9784,
      "step": 1005
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.534440059744917,
      "learning_rate": 9.994214703408946e-06,
      "loss": 0.9398,
      "step": 1006
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.180300218505205,
      "learning_rate": 9.99417991146353e-06,
      "loss": 0.9068,
      "step": 1007
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.076194451869712,
      "learning_rate": 9.994145015275544e-06,
      "loss": 0.9686,
      "step": 1008
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.10532080872228,
      "learning_rate": 9.994110014845722e-06,
      "loss": 1.0107,
      "step": 1009
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.3995517598974585,
      "learning_rate": 9.994074910174787e-06,
      "loss": 0.9759,
      "step": 1010
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.207171532098636,
      "learning_rate": 9.994039701263476e-06,
      "loss": 1.0228,
      "step": 1011
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.610920064319663,
      "learning_rate": 9.994004388112521e-06,
      "loss": 0.972,
      "step": 1012
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.6119914533557695,
      "learning_rate": 9.993968970722663e-06,
      "loss": 0.9694,
      "step": 1013
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.43786462970772,
      "learning_rate": 9.99393344909464e-06,
      "loss": 0.9366,
      "step": 1014
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.794334023099357,
      "learning_rate": 9.993897823229193e-06,
      "loss": 0.9848,
      "step": 1015
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.690814839424235,
      "learning_rate": 9.993862093127063e-06,
      "loss": 0.9278,
      "step": 1016
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.19692281914169,
      "learning_rate": 9.993826258789e-06,
      "loss": 0.9573,
      "step": 1017
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.571044230516712,
      "learning_rate": 9.99379032021575e-06,
      "loss": 0.9935,
      "step": 1018
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.11484207188465,
      "learning_rate": 9.993754277408062e-06,
      "loss": 0.9836,
      "step": 1019
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.663368628584689,
      "learning_rate": 9.993718130366692e-06,
      "loss": 1.0417,
      "step": 1020
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.060058159335902,
      "learning_rate": 9.993681879092389e-06,
      "loss": 0.9284,
      "step": 1021
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.331192381249381,
      "learning_rate": 9.993645523585915e-06,
      "loss": 1.0282,
      "step": 1022
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.125837177813874,
      "learning_rate": 9.993609063848027e-06,
      "loss": 0.9738,
      "step": 1023
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.684131653502822,
      "learning_rate": 9.993572499879484e-06,
      "loss": 1.0177,
      "step": 1024
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.127310264994257,
      "learning_rate": 9.993535831681052e-06,
      "loss": 0.9843,
      "step": 1025
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.378210588730779,
      "learning_rate": 9.993499059253494e-06,
      "loss": 0.9597,
      "step": 1026
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.6256205508552135,
      "learning_rate": 9.99346218259758e-06,
      "loss": 0.9693,
      "step": 1027
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.23502643876236,
      "learning_rate": 9.993425201714078e-06,
      "loss": 0.9693,
      "step": 1028
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.378916986998767,
      "learning_rate": 9.993388116603759e-06,
      "loss": 0.9604,
      "step": 1029
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.902511319087857,
      "learning_rate": 9.9933509272674e-06,
      "loss": 0.9816,
      "step": 1030
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.096033612758523,
      "learning_rate": 9.993313633705775e-06,
      "loss": 0.978,
      "step": 1031
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.160736057075026,
      "learning_rate": 9.993276235919663e-06,
      "loss": 0.9933,
      "step": 1032
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.77309957280692,
      "learning_rate": 9.993238733909844e-06,
      "loss": 0.8811,
      "step": 1033
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.631498981222252,
      "learning_rate": 9.993201127677104e-06,
      "loss": 1.0114,
      "step": 1034
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.087523316616698,
      "learning_rate": 9.993163417222223e-06,
      "loss": 0.9418,
      "step": 1035
    },
    {
      "epoch": 0.09,
      "grad_norm": 14.897457268990928,
      "learning_rate": 9.99312560254599e-06,
      "loss": 1.0099,
      "step": 1036
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.818179185956172,
      "learning_rate": 9.993087683649196e-06,
      "loss": 0.9763,
      "step": 1037
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.787630368835707,
      "learning_rate": 9.993049660532633e-06,
      "loss": 1.0357,
      "step": 1038
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.424558187525257,
      "learning_rate": 9.99301153319709e-06,
      "loss": 1.0173,
      "step": 1039
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.820221460440834,
      "learning_rate": 9.992973301643368e-06,
      "loss": 0.9546,
      "step": 1040
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.194490458945435,
      "learning_rate": 9.992934965872261e-06,
      "loss": 0.9475,
      "step": 1041
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.17858339549725,
      "learning_rate": 9.992896525884572e-06,
      "loss": 0.9834,
      "step": 1042
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.872104660303865,
      "learning_rate": 9.9928579816811e-06,
      "loss": 0.9579,
      "step": 1043
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.064506078442822,
      "learning_rate": 9.992819333262656e-06,
      "loss": 1.0656,
      "step": 1044
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.092237797117757,
      "learning_rate": 9.992780580630039e-06,
      "loss": 1.0115,
      "step": 1045
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.514881133196614,
      "learning_rate": 9.992741723784064e-06,
      "loss": 0.9526,
      "step": 1046
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.7504479707131715,
      "learning_rate": 9.992702762725538e-06,
      "loss": 0.974,
      "step": 1047
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.972734810693535,
      "learning_rate": 9.992663697455274e-06,
      "loss": 1.0129,
      "step": 1048
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.894894067139228,
      "learning_rate": 9.99262452797409e-06,
      "loss": 1.0223,
      "step": 1049
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.5192801905960955,
      "learning_rate": 9.992585254282805e-06,
      "loss": 0.9669,
      "step": 1050
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.483212836508495,
      "learning_rate": 9.992545876382235e-06,
      "loss": 0.9536,
      "step": 1051
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.309198741285574,
      "learning_rate": 9.992506394273203e-06,
      "loss": 0.9619,
      "step": 1052
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.238895653840148,
      "learning_rate": 9.992466807956534e-06,
      "loss": 0.9521,
      "step": 1053
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.249532184885036,
      "learning_rate": 9.992427117433053e-06,
      "loss": 1.0104,
      "step": 1054
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.469745764211813,
      "learning_rate": 9.99238732270359e-06,
      "loss": 1.0179,
      "step": 1055
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.005452879324764,
      "learning_rate": 9.992347423768974e-06,
      "loss": 0.937,
      "step": 1056
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.357685359828577,
      "learning_rate": 9.992307420630038e-06,
      "loss": 1.0259,
      "step": 1057
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.991609413542076,
      "learning_rate": 9.992267313287618e-06,
      "loss": 0.9328,
      "step": 1058
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.439572687771475,
      "learning_rate": 9.992227101742551e-06,
      "loss": 0.9814,
      "step": 1059
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.40603838700476,
      "learning_rate": 9.992186785995678e-06,
      "loss": 1.0265,
      "step": 1060
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.647546453756519,
      "learning_rate": 9.992146366047834e-06,
      "loss": 0.9508,
      "step": 1061
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.542083744297042,
      "learning_rate": 9.99210584189987e-06,
      "loss": 0.9441,
      "step": 1062
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.753833685031594,
      "learning_rate": 9.992065213552628e-06,
      "loss": 0.9985,
      "step": 1063
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.222084465789717,
      "learning_rate": 9.992024481006959e-06,
      "loss": 0.9937,
      "step": 1064
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.385555053065362,
      "learning_rate": 9.99198364426371e-06,
      "loss": 0.9857,
      "step": 1065
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.005047827689797,
      "learning_rate": 9.991942703323736e-06,
      "loss": 0.9752,
      "step": 1066
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.364364180372897,
      "learning_rate": 9.991901658187888e-06,
      "loss": 0.9678,
      "step": 1067
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.994366622104884,
      "learning_rate": 9.991860508857027e-06,
      "loss": 0.9869,
      "step": 1068
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.131050189862517,
      "learning_rate": 9.991819255332009e-06,
      "loss": 0.9861,
      "step": 1069
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.835108516660802,
      "learning_rate": 9.991777897613696e-06,
      "loss": 0.9401,
      "step": 1070
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.945187528534511,
      "learning_rate": 9.991736435702951e-06,
      "loss": 0.9114,
      "step": 1071
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.285703586762205,
      "learning_rate": 9.99169486960064e-06,
      "loss": 1.0164,
      "step": 1072
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.307220579657872,
      "learning_rate": 9.99165319930763e-06,
      "loss": 0.9134,
      "step": 1073
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.768162382944341,
      "learning_rate": 9.991611424824792e-06,
      "loss": 1.0036,
      "step": 1074
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.325942651797224,
      "learning_rate": 9.991569546152995e-06,
      "loss": 0.9295,
      "step": 1075
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.961607266135269,
      "learning_rate": 9.991527563293117e-06,
      "loss": 0.9372,
      "step": 1076
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.300526537914138,
      "learning_rate": 9.99148547624603e-06,
      "loss": 1.0138,
      "step": 1077
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.158509872853647,
      "learning_rate": 9.991443285012616e-06,
      "loss": 0.9964,
      "step": 1078
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.1548880254718785,
      "learning_rate": 9.991400989593756e-06,
      "loss": 0.9657,
      "step": 1079
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.58402681642553,
      "learning_rate": 9.991358589990329e-06,
      "loss": 0.997,
      "step": 1080
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.911764367428903,
      "learning_rate": 9.991316086203225e-06,
      "loss": 1.0246,
      "step": 1081
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.916636586964335,
      "learning_rate": 9.991273478233325e-06,
      "loss": 0.9299,
      "step": 1082
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.265002041438526,
      "learning_rate": 9.991230766081526e-06,
      "loss": 0.9729,
      "step": 1083
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.26210632416989,
      "learning_rate": 9.991187949748712e-06,
      "loss": 0.9798,
      "step": 1084
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.08383816682079,
      "learning_rate": 9.99114502923578e-06,
      "loss": 0.9042,
      "step": 1085
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.969261617575683,
      "learning_rate": 9.991102004543629e-06,
      "loss": 0.9507,
      "step": 1086
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.233904234480644,
      "learning_rate": 9.991058875673152e-06,
      "loss": 0.9114,
      "step": 1087
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.4482507053481966,
      "learning_rate": 9.99101564262525e-06,
      "loss": 0.9864,
      "step": 1088
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.837422677259048,
      "learning_rate": 9.990972305400827e-06,
      "loss": 0.9562,
      "step": 1089
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.491255786958795,
      "learning_rate": 9.99092886400079e-06,
      "loss": 0.9381,
      "step": 1090
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.347360729507525,
      "learning_rate": 9.99088531842604e-06,
      "loss": 0.9955,
      "step": 1091
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.186263516867967,
      "learning_rate": 9.990841668677492e-06,
      "loss": 1.0879,
      "step": 1092
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.602142781683266,
      "learning_rate": 9.99079791475605e-06,
      "loss": 0.9565,
      "step": 1093
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.913049088861985,
      "learning_rate": 9.990754056662633e-06,
      "loss": 0.9948,
      "step": 1094
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.924370182133304,
      "learning_rate": 9.990710094398156e-06,
      "loss": 0.9707,
      "step": 1095
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.194990054620504,
      "learning_rate": 9.990666027963532e-06,
      "loss": 0.9848,
      "step": 1096
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.5567424319790435,
      "learning_rate": 9.990621857359686e-06,
      "loss": 1.0207,
      "step": 1097
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.133354823971637,
      "learning_rate": 9.99057758258754e-06,
      "loss": 1.0312,
      "step": 1098
    },
    {
      "epoch": 0.1,
      "grad_norm": 12.10945516692436,
      "learning_rate": 9.990533203648012e-06,
      "loss": 0.9866,
      "step": 1099
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.2079617724930305,
      "learning_rate": 9.990488720542035e-06,
      "loss": 0.976,
      "step": 1100
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.926600534418078,
      "learning_rate": 9.990444133270533e-06,
      "loss": 0.9907,
      "step": 1101
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.115293934503965,
      "learning_rate": 9.990399441834441e-06,
      "loss": 1.0513,
      "step": 1102
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.551324131734745,
      "learning_rate": 9.99035464623469e-06,
      "loss": 1.0111,
      "step": 1103
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.63877848176835,
      "learning_rate": 9.990309746472211e-06,
      "loss": 0.9567,
      "step": 1104
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.898827206499991,
      "learning_rate": 9.990264742547947e-06,
      "loss": 1.0162,
      "step": 1105
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.631221936025121,
      "learning_rate": 9.990219634462834e-06,
      "loss": 0.9557,
      "step": 1106
    },
    {
      "epoch": 0.1,
      "grad_norm": 10.954341859754555,
      "learning_rate": 9.990174422217815e-06,
      "loss": 0.9339,
      "step": 1107
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.929891941545815,
      "learning_rate": 9.990129105813835e-06,
      "loss": 0.9533,
      "step": 1108
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.194863491065385,
      "learning_rate": 9.990083685251836e-06,
      "loss": 0.9565,
      "step": 1109
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.4403315885118,
      "learning_rate": 9.990038160532767e-06,
      "loss": 1.01,
      "step": 1110
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.65077841893017,
      "learning_rate": 9.989992531657582e-06,
      "loss": 0.9383,
      "step": 1111
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.899303213142547,
      "learning_rate": 9.98994679862723e-06,
      "loss": 0.9888,
      "step": 1112
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.434933395462747,
      "learning_rate": 9.989900961442666e-06,
      "loss": 1.0015,
      "step": 1113
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.532893412512905,
      "learning_rate": 9.989855020104848e-06,
      "loss": 1.0047,
      "step": 1114
    },
    {
      "epoch": 0.1,
      "grad_norm": 10.300686901885404,
      "learning_rate": 9.989808974614733e-06,
      "loss": 0.9721,
      "step": 1115
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.685013457120647,
      "learning_rate": 9.989762824973284e-06,
      "loss": 0.9542,
      "step": 1116
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.2956331083394526,
      "learning_rate": 9.989716571181463e-06,
      "loss": 0.9746,
      "step": 1117
    },
    {
      "epoch": 0.1,
      "grad_norm": 12.977407482464809,
      "learning_rate": 9.989670213240234e-06,
      "loss": 1.0154,
      "step": 1118
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.630693281254286,
      "learning_rate": 9.989623751150569e-06,
      "loss": 1.0001,
      "step": 1119
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.365617509049695,
      "learning_rate": 9.989577184913436e-06,
      "loss": 0.9486,
      "step": 1120
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.278454299311035,
      "learning_rate": 9.989530514529803e-06,
      "loss": 0.9455,
      "step": 1121
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.405380323066382,
      "learning_rate": 9.98948374000065e-06,
      "loss": 0.9442,
      "step": 1122
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.301957194625947,
      "learning_rate": 9.989436861326948e-06,
      "loss": 1.025,
      "step": 1123
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.828943429354764,
      "learning_rate": 9.98938987850968e-06,
      "loss": 0.9831,
      "step": 1124
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.989599638233167,
      "learning_rate": 9.989342791549827e-06,
      "loss": 0.9351,
      "step": 1125
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.085517422343752,
      "learning_rate": 9.989295600448367e-06,
      "loss": 0.9816,
      "step": 1126
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.156246058101658,
      "learning_rate": 9.989248305206288e-06,
      "loss": 0.908,
      "step": 1127
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.165876708522559,
      "learning_rate": 9.989200905824578e-06,
      "loss": 0.9664,
      "step": 1128
    },
    {
      "epoch": 0.1,
      "grad_norm": 10.521264460833743,
      "learning_rate": 9.989153402304223e-06,
      "loss": 0.9988,
      "step": 1129
    },
    {
      "epoch": 0.1,
      "grad_norm": 12.531573057344238,
      "learning_rate": 9.989105794646219e-06,
      "loss": 0.9731,
      "step": 1130
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.102079991088457,
      "learning_rate": 9.989058082851558e-06,
      "loss": 0.9432,
      "step": 1131
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.467146681329364,
      "learning_rate": 9.989010266921233e-06,
      "loss": 0.9925,
      "step": 1132
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.637783861205074,
      "learning_rate": 9.988962346856246e-06,
      "loss": 0.9592,
      "step": 1133
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.326447599078782,
      "learning_rate": 9.988914322657594e-06,
      "loss": 0.9377,
      "step": 1134
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.545377702360122,
      "learning_rate": 9.988866194326284e-06,
      "loss": 0.8623,
      "step": 1135
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.314916041169402,
      "learning_rate": 9.988817961863315e-06,
      "loss": 0.9273,
      "step": 1136
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.606474257623923,
      "learning_rate": 9.988769625269698e-06,
      "loss": 0.9468,
      "step": 1137
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.8801302965003845,
      "learning_rate": 9.98872118454644e-06,
      "loss": 0.9928,
      "step": 1138
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.025325171025866,
      "learning_rate": 9.98867263969455e-06,
      "loss": 0.9994,
      "step": 1139
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.507025610536579,
      "learning_rate": 9.988623990715047e-06,
      "loss": 1.0042,
      "step": 1140
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.793646076210973,
      "learning_rate": 9.988575237608942e-06,
      "loss": 0.9272,
      "step": 1141
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.4003827006897,
      "learning_rate": 9.988526380377252e-06,
      "loss": 0.9636,
      "step": 1142
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.743702067695908,
      "learning_rate": 9.988477419021e-06,
      "loss": 0.9635,
      "step": 1143
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.514234122875846,
      "learning_rate": 9.988428353541207e-06,
      "loss": 0.9296,
      "step": 1144
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.8210409063994195,
      "learning_rate": 9.988379183938897e-06,
      "loss": 0.929,
      "step": 1145
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.4074527685742835,
      "learning_rate": 9.988329910215095e-06,
      "loss": 0.9513,
      "step": 1146
    },
    {
      "epoch": 0.1,
      "grad_norm": 10.054495137599615,
      "learning_rate": 9.988280532370828e-06,
      "loss": 0.9424,
      "step": 1147
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.722207552765884,
      "learning_rate": 9.988231050407134e-06,
      "loss": 0.8753,
      "step": 1148
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.486262518441393,
      "learning_rate": 9.988181464325037e-06,
      "loss": 0.9144,
      "step": 1149
    },
    {
      "epoch": 0.1,
      "grad_norm": 10.72661355953756,
      "learning_rate": 9.988131774125577e-06,
      "loss": 0.9415,
      "step": 1150
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.29639097222841,
      "learning_rate": 9.98808197980979e-06,
      "loss": 0.938,
      "step": 1151
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.7972277802652155,
      "learning_rate": 9.988032081378715e-06,
      "loss": 1.0217,
      "step": 1152
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.85497993379049,
      "learning_rate": 9.987982078833394e-06,
      "loss": 0.9604,
      "step": 1153
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.1269828936951445,
      "learning_rate": 9.987931972174872e-06,
      "loss": 0.9681,
      "step": 1154
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.702938777086002,
      "learning_rate": 9.987881761404192e-06,
      "loss": 0.9376,
      "step": 1155
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.707591098429688,
      "learning_rate": 9.987831446522404e-06,
      "loss": 0.9367,
      "step": 1156
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.559553583996403,
      "learning_rate": 9.987781027530557e-06,
      "loss": 1.0019,
      "step": 1157
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.505387412356555,
      "learning_rate": 9.987730504429704e-06,
      "loss": 1.0041,
      "step": 1158
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.20425924440906,
      "learning_rate": 9.987679877220902e-06,
      "loss": 0.9574,
      "step": 1159
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.672025359563292,
      "learning_rate": 9.987629145905203e-06,
      "loss": 0.9325,
      "step": 1160
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.92358605095803,
      "learning_rate": 9.98757831048367e-06,
      "loss": 0.9326,
      "step": 1161
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.1888861635689745,
      "learning_rate": 9.987527370957362e-06,
      "loss": 0.9666,
      "step": 1162
    },
    {
      "epoch": 0.1,
      "grad_norm": 10.089329785158665,
      "learning_rate": 9.98747632732734e-06,
      "loss": 0.9748,
      "step": 1163
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.620397293353196,
      "learning_rate": 9.987425179594675e-06,
      "loss": 0.9464,
      "step": 1164
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.869897310980044,
      "learning_rate": 9.987373927760432e-06,
      "loss": 0.9556,
      "step": 1165
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.894540920338943,
      "learning_rate": 9.987322571825678e-06,
      "loss": 0.8911,
      "step": 1166
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.4162187240772575,
      "learning_rate": 9.987271111791489e-06,
      "loss": 1.0573,
      "step": 1167
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.881055160147344,
      "learning_rate": 9.987219547658937e-06,
      "loss": 0.9865,
      "step": 1168
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.5866986148995,
      "learning_rate": 9.987167879429097e-06,
      "loss": 0.8932,
      "step": 1169
    },
    {
      "epoch": 0.1,
      "grad_norm": 11.757663189525715,
      "learning_rate": 9.987116107103052e-06,
      "loss": 0.97,
      "step": 1170
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.529205159834894,
      "learning_rate": 9.98706423068188e-06,
      "loss": 0.8876,
      "step": 1171
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.436513016982534,
      "learning_rate": 9.987012250166662e-06,
      "loss": 0.9587,
      "step": 1172
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.099617321458918,
      "learning_rate": 9.986960165558487e-06,
      "loss": 0.9109,
      "step": 1173
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.49854323931392,
      "learning_rate": 9.986907976858436e-06,
      "loss": 0.9095,
      "step": 1174
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.001144446369654,
      "learning_rate": 9.986855684067604e-06,
      "loss": 0.9766,
      "step": 1175
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.379721310870202,
      "learning_rate": 9.986803287187082e-06,
      "loss": 0.973,
      "step": 1176
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.510805098844974,
      "learning_rate": 9.986750786217961e-06,
      "loss": 0.9332,
      "step": 1177
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.41910804398586,
      "learning_rate": 9.986698181161338e-06,
      "loss": 1.0484,
      "step": 1178
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.2917170055154035,
      "learning_rate": 9.986645472018312e-06,
      "loss": 0.9096,
      "step": 1179
    },
    {
      "epoch": 0.11,
      "grad_norm": 12.894847404801544,
      "learning_rate": 9.986592658789982e-06,
      "loss": 0.948,
      "step": 1180
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.57405215872667,
      "learning_rate": 9.98653974147745e-06,
      "loss": 0.9816,
      "step": 1181
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.267550264745159,
      "learning_rate": 9.986486720081821e-06,
      "loss": 0.8923,
      "step": 1182
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.622565694635163,
      "learning_rate": 9.986433594604202e-06,
      "loss": 0.9461,
      "step": 1183
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.152602289372348,
      "learning_rate": 9.986380365045703e-06,
      "loss": 0.9147,
      "step": 1184
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.115420942282809,
      "learning_rate": 9.986327031407433e-06,
      "loss": 0.9088,
      "step": 1185
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.863773129733878,
      "learning_rate": 9.986273593690506e-06,
      "loss": 1.0022,
      "step": 1186
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.456631837738072,
      "learning_rate": 9.986220051896039e-06,
      "loss": 0.9723,
      "step": 1187
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.633662348171267,
      "learning_rate": 9.986166406025146e-06,
      "loss": 0.8882,
      "step": 1188
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.769937035758776,
      "learning_rate": 9.98611265607895e-06,
      "loss": 0.941,
      "step": 1189
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.569560007246109,
      "learning_rate": 9.986058802058572e-06,
      "loss": 0.9329,
      "step": 1190
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.419436645120479,
      "learning_rate": 9.986004843965134e-06,
      "loss": 0.9194,
      "step": 1191
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.641382482363346,
      "learning_rate": 9.985950781799766e-06,
      "loss": 0.9732,
      "step": 1192
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.498753855881156,
      "learning_rate": 9.985896615563593e-06,
      "loss": 0.9586,
      "step": 1193
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.074609922092675,
      "learning_rate": 9.985842345257748e-06,
      "loss": 0.937,
      "step": 1194
    },
    {
      "epoch": 0.11,
      "grad_norm": 10.705637069858234,
      "learning_rate": 9.985787970883363e-06,
      "loss": 1.0124,
      "step": 1195
    },
    {
      "epoch": 0.11,
      "grad_norm": 12.639772993029775,
      "learning_rate": 9.985733492441571e-06,
      "loss": 1.0221,
      "step": 1196
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.252758959482215,
      "learning_rate": 9.985678909933514e-06,
      "loss": 0.9501,
      "step": 1197
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.64825405339732,
      "learning_rate": 9.985624223360327e-06,
      "loss": 0.9106,
      "step": 1198
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.109733217191102,
      "learning_rate": 9.98556943272315e-06,
      "loss": 0.9475,
      "step": 1199
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.213791084877371,
      "learning_rate": 9.985514538023134e-06,
      "loss": 1.0137,
      "step": 1200
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.045456399696434,
      "learning_rate": 9.985459539261417e-06,
      "loss": 0.9416,
      "step": 1201
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.335525722020186,
      "learning_rate": 9.98540443643915e-06,
      "loss": 0.9938,
      "step": 1202
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.304753291064538,
      "learning_rate": 9.985349229557484e-06,
      "loss": 0.9436,
      "step": 1203
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.522355577872712,
      "learning_rate": 9.985293918617572e-06,
      "loss": 0.9849,
      "step": 1204
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.162614462254137,
      "learning_rate": 9.985238503620565e-06,
      "loss": 0.8583,
      "step": 1205
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.111004371572633,
      "learning_rate": 9.985182984567621e-06,
      "loss": 0.9467,
      "step": 1206
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.065856884940847,
      "learning_rate": 9.985127361459902e-06,
      "loss": 0.9344,
      "step": 1207
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.605132651039941,
      "learning_rate": 9.985071634298564e-06,
      "loss": 0.9179,
      "step": 1208
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.353934107653108,
      "learning_rate": 9.985015803084774e-06,
      "loss": 0.9645,
      "step": 1209
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.432296894031921,
      "learning_rate": 9.984959867819696e-06,
      "loss": 0.9286,
      "step": 1210
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.401606631111568,
      "learning_rate": 9.9849038285045e-06,
      "loss": 0.9698,
      "step": 1211
    },
    {
      "epoch": 0.11,
      "grad_norm": 13.193834510115998,
      "learning_rate": 9.984847685140351e-06,
      "loss": 1.0228,
      "step": 1212
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.940343155388753,
      "learning_rate": 9.984791437728423e-06,
      "loss": 1.0495,
      "step": 1213
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.028310604630057,
      "learning_rate": 9.984735086269889e-06,
      "loss": 0.8997,
      "step": 1214
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.26749580704144,
      "learning_rate": 9.98467863076593e-06,
      "loss": 0.9317,
      "step": 1215
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.22083249056773,
      "learning_rate": 9.984622071217718e-06,
      "loss": 0.9859,
      "step": 1216
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.19080625515515,
      "learning_rate": 9.984565407626439e-06,
      "loss": 0.939,
      "step": 1217
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.47814206451388,
      "learning_rate": 9.984508639993271e-06,
      "loss": 1.0215,
      "step": 1218
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.01660523838273,
      "learning_rate": 9.9844517683194e-06,
      "loss": 0.9814,
      "step": 1219
    },
    {
      "epoch": 0.11,
      "grad_norm": 11.732328774115322,
      "learning_rate": 9.984394792606018e-06,
      "loss": 0.9729,
      "step": 1220
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.314766515585315,
      "learning_rate": 9.984337712854307e-06,
      "loss": 0.9282,
      "step": 1221
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.978357107338396,
      "learning_rate": 9.984280529065462e-06,
      "loss": 0.9217,
      "step": 1222
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.130992415129934,
      "learning_rate": 9.984223241240678e-06,
      "loss": 0.9311,
      "step": 1223
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.520005998889693,
      "learning_rate": 9.984165849381149e-06,
      "loss": 1.023,
      "step": 1224
    },
    {
      "epoch": 0.11,
      "grad_norm": 12.930753635384601,
      "learning_rate": 9.98410835348807e-06,
      "loss": 1.0149,
      "step": 1225
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.051072034871773,
      "learning_rate": 9.984050753562648e-06,
      "loss": 1.0296,
      "step": 1226
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.37327872859455,
      "learning_rate": 9.98399304960608e-06,
      "loss": 0.9703,
      "step": 1227
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.508041502599887,
      "learning_rate": 9.98393524161957e-06,
      "loss": 1.003,
      "step": 1228
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.663663508143976,
      "learning_rate": 9.98387732960433e-06,
      "loss": 0.9686,
      "step": 1229
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.243696429590202,
      "learning_rate": 9.983819313561563e-06,
      "loss": 0.9904,
      "step": 1230
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.170779312599446,
      "learning_rate": 9.983761193492482e-06,
      "loss": 0.9254,
      "step": 1231
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.463172699851384,
      "learning_rate": 9.983702969398301e-06,
      "loss": 0.9612,
      "step": 1232
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.011931850030773,
      "learning_rate": 9.983644641280234e-06,
      "loss": 0.9266,
      "step": 1233
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.98027260352625,
      "learning_rate": 9.9835862091395e-06,
      "loss": 1.0495,
      "step": 1234
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.458429028401512,
      "learning_rate": 9.983527672977318e-06,
      "loss": 1.0018,
      "step": 1235
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.088941194700503,
      "learning_rate": 9.983469032794909e-06,
      "loss": 0.9319,
      "step": 1236
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.861913589131415,
      "learning_rate": 9.983410288593497e-06,
      "loss": 0.9352,
      "step": 1237
    },
    {
      "epoch": 0.11,
      "grad_norm": 11.688252377321225,
      "learning_rate": 9.983351440374309e-06,
      "loss": 0.9319,
      "step": 1238
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.18913119948747,
      "learning_rate": 9.983292488138573e-06,
      "loss": 0.8321,
      "step": 1239
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.418702550186432,
      "learning_rate": 9.98323343188752e-06,
      "loss": 0.9929,
      "step": 1240
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.8987301047324,
      "learning_rate": 9.983174271622382e-06,
      "loss": 0.9516,
      "step": 1241
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.651851622040894,
      "learning_rate": 9.983115007344394e-06,
      "loss": 1.004,
      "step": 1242
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.559996202245957,
      "learning_rate": 9.983055639054793e-06,
      "loss": 0.9011,
      "step": 1243
    },
    {
      "epoch": 0.11,
      "grad_norm": 15.212841122125589,
      "learning_rate": 9.982996166754819e-06,
      "loss": 1.02,
      "step": 1244
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.906746773294869,
      "learning_rate": 9.982936590445711e-06,
      "loss": 0.9467,
      "step": 1245
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.258063866301292,
      "learning_rate": 9.982876910128716e-06,
      "loss": 0.8862,
      "step": 1246
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.20897419131393,
      "learning_rate": 9.982817125805076e-06,
      "loss": 1.0499,
      "step": 1247
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.93139806355661,
      "learning_rate": 9.982757237476042e-06,
      "loss": 0.9842,
      "step": 1248
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.474674055697095,
      "learning_rate": 9.982697245142863e-06,
      "loss": 0.9798,
      "step": 1249
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.69138732584315,
      "learning_rate": 9.98263714880679e-06,
      "loss": 0.9866,
      "step": 1250
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.121282559565715,
      "learning_rate": 9.982576948469079e-06,
      "loss": 0.9382,
      "step": 1251
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.193833491172351,
      "learning_rate": 9.982516644130984e-06,
      "loss": 0.9638,
      "step": 1252
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.747625268753172,
      "learning_rate": 9.982456235793767e-06,
      "loss": 1.0191,
      "step": 1253
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.898614454753023,
      "learning_rate": 9.982395723458688e-06,
      "loss": 0.9298,
      "step": 1254
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.4215107898928405,
      "learning_rate": 9.982335107127008e-06,
      "loss": 0.9039,
      "step": 1255
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.286866785737138,
      "learning_rate": 9.982274386799996e-06,
      "loss": 0.9811,
      "step": 1256
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.136200755864387,
      "learning_rate": 9.982213562478915e-06,
      "loss": 0.9247,
      "step": 1257
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.743672423386007,
      "learning_rate": 9.982152634165037e-06,
      "loss": 1.0043,
      "step": 1258
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.5032247859144325,
      "learning_rate": 9.982091601859634e-06,
      "loss": 1.0311,
      "step": 1259
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.066509189847997,
      "learning_rate": 9.982030465563981e-06,
      "loss": 0.9259,
      "step": 1260
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.550812327149089,
      "learning_rate": 9.981969225279351e-06,
      "loss": 0.8951,
      "step": 1261
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.270534072059098,
      "learning_rate": 9.981907881007023e-06,
      "loss": 0.9649,
      "step": 1262
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.96328432718407,
      "learning_rate": 9.98184643274828e-06,
      "loss": 1.0126,
      "step": 1263
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.504304250387007,
      "learning_rate": 9.981784880504402e-06,
      "loss": 0.9901,
      "step": 1264
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.739176256622612,
      "learning_rate": 9.981723224276673e-06,
      "loss": 0.9448,
      "step": 1265
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.694588704503909,
      "learning_rate": 9.981661464066383e-06,
      "loss": 1.0384,
      "step": 1266
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.077271730312509,
      "learning_rate": 9.98159959987482e-06,
      "loss": 0.9606,
      "step": 1267
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.057522544848359,
      "learning_rate": 9.981537631703273e-06,
      "loss": 0.9693,
      "step": 1268
    },
    {
      "epoch": 0.11,
      "grad_norm": 13.543450741620813,
      "learning_rate": 9.981475559553039e-06,
      "loss": 0.9713,
      "step": 1269
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.691933529154625,
      "learning_rate": 9.981413383425411e-06,
      "loss": 0.9754,
      "step": 1270
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.75677864545581,
      "learning_rate": 9.981351103321689e-06,
      "loss": 0.9403,
      "step": 1271
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.838569790617246,
      "learning_rate": 9.981288719243171e-06,
      "loss": 0.9842,
      "step": 1272
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.62095799062756,
      "learning_rate": 9.98122623119116e-06,
      "loss": 0.9548,
      "step": 1273
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.209254676221442,
      "learning_rate": 9.981163639166958e-06,
      "loss": 0.9037,
      "step": 1274
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.849038535512064,
      "learning_rate": 9.981100943171876e-06,
      "loss": 0.9603,
      "step": 1275
    },
    {
      "epoch": 0.11,
      "grad_norm": 13.734334579243923,
      "learning_rate": 9.98103814320722e-06,
      "loss": 1.0152,
      "step": 1276
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.669537240922854,
      "learning_rate": 9.9809752392743e-06,
      "loss": 0.9393,
      "step": 1277
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.448754496315086,
      "learning_rate": 9.980912231374431e-06,
      "loss": 0.9366,
      "step": 1278
    },
    {
      "epoch": 0.11,
      "grad_norm": 10.907337796447488,
      "learning_rate": 9.980849119508927e-06,
      "loss": 0.9386,
      "step": 1279
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.965369016016487,
      "learning_rate": 9.980785903679106e-06,
      "loss": 0.9068,
      "step": 1280
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.549956437511614,
      "learning_rate": 9.980722583886286e-06,
      "loss": 0.969,
      "step": 1281
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.743954925855676,
      "learning_rate": 9.98065916013179e-06,
      "loss": 0.9311,
      "step": 1282
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.401469494581274,
      "learning_rate": 9.980595632416942e-06,
      "loss": 0.9175,
      "step": 1283
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.101171664745694,
      "learning_rate": 9.980532000743066e-06,
      "loss": 0.8783,
      "step": 1284
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.890382506358185,
      "learning_rate": 9.980468265111493e-06,
      "loss": 0.9905,
      "step": 1285
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.259861453659001,
      "learning_rate": 9.980404425523552e-06,
      "loss": 0.9246,
      "step": 1286
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.155485946474958,
      "learning_rate": 9.980340481980574e-06,
      "loss": 0.9348,
      "step": 1287
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.221275704072247,
      "learning_rate": 9.980276434483898e-06,
      "loss": 0.9977,
      "step": 1288
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.677039877850333,
      "learning_rate": 9.980212283034856e-06,
      "loss": 0.9292,
      "step": 1289
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.709714414439395,
      "learning_rate": 9.980148027634791e-06,
      "loss": 0.9038,
      "step": 1290
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.797949864259427,
      "learning_rate": 9.98008366828504e-06,
      "loss": 0.906,
      "step": 1291
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.085238797674527,
      "learning_rate": 9.98001920498695e-06,
      "loss": 0.9509,
      "step": 1292
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.656538276521108,
      "learning_rate": 9.979954637741865e-06,
      "loss": 0.9772,
      "step": 1293
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.5257275472863,
      "learning_rate": 9.97988996655113e-06,
      "loss": 0.9157,
      "step": 1294
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.264016564336823,
      "learning_rate": 9.979825191416101e-06,
      "loss": 0.9801,
      "step": 1295
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.51206960813078,
      "learning_rate": 9.979760312338127e-06,
      "loss": 0.9331,
      "step": 1296
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.585030904735335,
      "learning_rate": 9.97969532931856e-06,
      "loss": 0.9002,
      "step": 1297
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.865124609143631,
      "learning_rate": 9.97963024235876e-06,
      "loss": 0.9378,
      "step": 1298
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.140483739434602,
      "learning_rate": 9.979565051460083e-06,
      "loss": 0.9278,
      "step": 1299
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.210233560495285,
      "learning_rate": 9.97949975662389e-06,
      "loss": 0.9754,
      "step": 1300
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.524364412868774,
      "learning_rate": 9.979434357851545e-06,
      "loss": 0.9832,
      "step": 1301
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.205446824122753,
      "learning_rate": 9.979368855144413e-06,
      "loss": 0.9434,
      "step": 1302
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.07719120486523,
      "learning_rate": 9.97930324850386e-06,
      "loss": 0.9433,
      "step": 1303
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.311973327765198,
      "learning_rate": 9.979237537931257e-06,
      "loss": 0.9017,
      "step": 1304
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.8831141599387555,
      "learning_rate": 9.979171723427975e-06,
      "loss": 0.9155,
      "step": 1305
    },
    {
      "epoch": 0.12,
      "grad_norm": 10.79220419173369,
      "learning_rate": 9.979105804995388e-06,
      "loss": 0.8773,
      "step": 1306
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.1573271876080184,
      "learning_rate": 9.97903978263487e-06,
      "loss": 0.9072,
      "step": 1307
    },
    {
      "epoch": 0.12,
      "grad_norm": 11.310367111165748,
      "learning_rate": 9.978973656347801e-06,
      "loss": 0.9492,
      "step": 1308
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.283892069314969,
      "learning_rate": 9.97890742613556e-06,
      "loss": 0.9458,
      "step": 1309
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.16561713730703,
      "learning_rate": 9.97884109199953e-06,
      "loss": 0.9782,
      "step": 1310
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.827234941328474,
      "learning_rate": 9.978774653941096e-06,
      "loss": 0.8969,
      "step": 1311
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.786008277869296,
      "learning_rate": 9.978708111961645e-06,
      "loss": 1.0277,
      "step": 1312
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.758791677964346,
      "learning_rate": 9.978641466062566e-06,
      "loss": 0.9825,
      "step": 1313
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.572434089479751,
      "learning_rate": 9.978574716245248e-06,
      "loss": 1.0085,
      "step": 1314
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.93791478341313,
      "learning_rate": 9.978507862511087e-06,
      "loss": 0.9712,
      "step": 1315
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.01848007793686,
      "learning_rate": 9.978440904861474e-06,
      "loss": 0.9305,
      "step": 1316
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.315182319147407,
      "learning_rate": 9.978373843297814e-06,
      "loss": 0.9261,
      "step": 1317
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.371892337072602,
      "learning_rate": 9.9783066778215e-06,
      "loss": 0.9379,
      "step": 1318
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.55806994215191,
      "learning_rate": 9.978239408433936e-06,
      "loss": 0.8601,
      "step": 1319
    },
    {
      "epoch": 0.12,
      "grad_norm": 11.011020774129388,
      "learning_rate": 9.978172035136528e-06,
      "loss": 0.9355,
      "step": 1320
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.810276556662753,
      "learning_rate": 9.97810455793068e-06,
      "loss": 0.9866,
      "step": 1321
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.426495676264583,
      "learning_rate": 9.978036976817802e-06,
      "loss": 0.8887,
      "step": 1322
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.084751337251286,
      "learning_rate": 9.977969291799303e-06,
      "loss": 0.9734,
      "step": 1323
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.481151930344522,
      "learning_rate": 9.977901502876597e-06,
      "loss": 0.9483,
      "step": 1324
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.121779377136498,
      "learning_rate": 9.977833610051097e-06,
      "loss": 0.8844,
      "step": 1325
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.883010817170145,
      "learning_rate": 9.977765613324225e-06,
      "loss": 0.967,
      "step": 1326
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.701406343132003,
      "learning_rate": 9.977697512697394e-06,
      "loss": 0.983,
      "step": 1327
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.83416916297677,
      "learning_rate": 9.97762930817203e-06,
      "loss": 0.9384,
      "step": 1328
    },
    {
      "epoch": 0.12,
      "grad_norm": 10.10298692444333,
      "learning_rate": 9.977560999749553e-06,
      "loss": 0.9397,
      "step": 1329
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.10547203011475,
      "learning_rate": 9.977492587431391e-06,
      "loss": 0.9897,
      "step": 1330
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.005537602418642,
      "learning_rate": 9.977424071218975e-06,
      "loss": 0.9393,
      "step": 1331
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.031776983725786,
      "learning_rate": 9.97735545111373e-06,
      "loss": 0.8998,
      "step": 1332
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.546304655292217,
      "learning_rate": 9.977286727117089e-06,
      "loss": 1.0029,
      "step": 1333
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.607986167446821,
      "learning_rate": 9.977217899230488e-06,
      "loss": 0.9635,
      "step": 1334
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.2417322055113456,
      "learning_rate": 9.977148967455365e-06,
      "loss": 0.9524,
      "step": 1335
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.766439290215281,
      "learning_rate": 9.977079931793156e-06,
      "loss": 0.965,
      "step": 1336
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.259837233298295,
      "learning_rate": 9.977010792245303e-06,
      "loss": 0.9461,
      "step": 1337
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.581119162621205,
      "learning_rate": 9.97694154881325e-06,
      "loss": 0.9406,
      "step": 1338
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.629874706365678,
      "learning_rate": 9.97687220149844e-06,
      "loss": 0.9116,
      "step": 1339
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.023039510668053,
      "learning_rate": 9.976802750302323e-06,
      "loss": 0.9557,
      "step": 1340
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.350161042865411,
      "learning_rate": 9.976733195226347e-06,
      "loss": 0.8974,
      "step": 1341
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.428181666680423,
      "learning_rate": 9.976663536271964e-06,
      "loss": 0.979,
      "step": 1342
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.4830350675783,
      "learning_rate": 9.976593773440629e-06,
      "loss": 0.9363,
      "step": 1343
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.2872145602477,
      "learning_rate": 9.976523906733798e-06,
      "loss": 0.9839,
      "step": 1344
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.4572992459170235,
      "learning_rate": 9.976453936152927e-06,
      "loss": 0.9499,
      "step": 1345
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.568250627843918,
      "learning_rate": 9.976383861699482e-06,
      "loss": 0.9154,
      "step": 1346
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.787210614509,
      "learning_rate": 9.976313683374918e-06,
      "loss": 0.9099,
      "step": 1347
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.668733736831303,
      "learning_rate": 9.976243401180708e-06,
      "loss": 0.8941,
      "step": 1348
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.67169873013758,
      "learning_rate": 9.97617301511831e-06,
      "loss": 0.9313,
      "step": 1349
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.338279321264206,
      "learning_rate": 9.976102525189202e-06,
      "loss": 0.9329,
      "step": 1350
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.153217958801919,
      "learning_rate": 9.97603193139485e-06,
      "loss": 0.9762,
      "step": 1351
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.445429429087869,
      "learning_rate": 9.975961233736727e-06,
      "loss": 0.9491,
      "step": 1352
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.861502126813018,
      "learning_rate": 9.975890432216312e-06,
      "loss": 0.897,
      "step": 1353
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.032212918560607,
      "learning_rate": 9.97581952683508e-06,
      "loss": 1.0228,
      "step": 1354
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.712342143413182,
      "learning_rate": 9.975748517594515e-06,
      "loss": 0.9654,
      "step": 1355
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.665911040455217,
      "learning_rate": 9.975677404496093e-06,
      "loss": 1.0147,
      "step": 1356
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.9408618294008395,
      "learning_rate": 9.975606187541305e-06,
      "loss": 0.9692,
      "step": 1357
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.08487712606646,
      "learning_rate": 9.975534866731631e-06,
      "loss": 0.9315,
      "step": 1358
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.54238476281645,
      "learning_rate": 9.975463442068565e-06,
      "loss": 1.0039,
      "step": 1359
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.142342608252962,
      "learning_rate": 9.975391913553595e-06,
      "loss": 0.9865,
      "step": 1360
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.272058921016833,
      "learning_rate": 9.975320281188215e-06,
      "loss": 1.0235,
      "step": 1361
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.41415030257654,
      "learning_rate": 9.975248544973919e-06,
      "loss": 0.9616,
      "step": 1362
    },
    {
      "epoch": 0.12,
      "grad_norm": 12.859808225315065,
      "learning_rate": 9.975176704912206e-06,
      "loss": 0.8953,
      "step": 1363
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.225301879452562,
      "learning_rate": 9.975104761004574e-06,
      "loss": 0.9341,
      "step": 1364
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.676258468386849,
      "learning_rate": 9.975032713252527e-06,
      "loss": 0.8688,
      "step": 1365
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.4963208564464745,
      "learning_rate": 9.974960561657567e-06,
      "loss": 0.8833,
      "step": 1366
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.238044518885136,
      "learning_rate": 9.9748883062212e-06,
      "loss": 0.9369,
      "step": 1367
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.378147187315362,
      "learning_rate": 9.974815946944933e-06,
      "loss": 0.9544,
      "step": 1368
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.3452465843525605,
      "learning_rate": 9.97474348383028e-06,
      "loss": 0.9083,
      "step": 1369
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.527799805884163,
      "learning_rate": 9.97467091687875e-06,
      "loss": 0.9858,
      "step": 1370
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.089482221535015,
      "learning_rate": 9.974598246091859e-06,
      "loss": 0.9156,
      "step": 1371
    },
    {
      "epoch": 0.12,
      "grad_norm": 10.045131168120767,
      "learning_rate": 9.974525471471125e-06,
      "loss": 0.99,
      "step": 1372
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.11140258072261,
      "learning_rate": 9.974452593018064e-06,
      "loss": 0.9926,
      "step": 1373
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.026847242409577,
      "learning_rate": 9.974379610734199e-06,
      "loss": 0.9397,
      "step": 1374
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.599693175612462,
      "learning_rate": 9.974306524621054e-06,
      "loss": 0.9533,
      "step": 1375
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.312198526422322,
      "learning_rate": 9.974233334680155e-06,
      "loss": 0.9174,
      "step": 1376
    },
    {
      "epoch": 0.12,
      "grad_norm": 10.070508827867345,
      "learning_rate": 9.974160040913025e-06,
      "loss": 0.9779,
      "step": 1377
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.718744347070423,
      "learning_rate": 9.974086643321202e-06,
      "loss": 0.941,
      "step": 1378
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.752182428921662,
      "learning_rate": 9.97401314190621e-06,
      "loss": 0.9223,
      "step": 1379
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.888891295129369,
      "learning_rate": 9.973939536669589e-06,
      "loss": 0.9426,
      "step": 1380
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.936479854797045,
      "learning_rate": 9.973865827612869e-06,
      "loss": 0.9073,
      "step": 1381
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.536103427839438,
      "learning_rate": 9.973792014737596e-06,
      "loss": 0.9239,
      "step": 1382
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.534424925651377,
      "learning_rate": 9.973718098045307e-06,
      "loss": 0.8731,
      "step": 1383
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.395004677931334,
      "learning_rate": 9.973644077537543e-06,
      "loss": 0.972,
      "step": 1384
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.062003110464344,
      "learning_rate": 9.973569953215853e-06,
      "loss": 0.9652,
      "step": 1385
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.61661016289107,
      "learning_rate": 9.973495725081782e-06,
      "loss": 0.9774,
      "step": 1386
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.225084141819947,
      "learning_rate": 9.973421393136877e-06,
      "loss": 0.8932,
      "step": 1387
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.174980563139468,
      "learning_rate": 9.973346957382694e-06,
      "loss": 0.9222,
      "step": 1388
    },
    {
      "epoch": 0.12,
      "grad_norm": 10.493626974999602,
      "learning_rate": 9.973272417820785e-06,
      "loss": 0.9373,
      "step": 1389
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.748764182554102,
      "learning_rate": 9.973197774452705e-06,
      "loss": 0.9231,
      "step": 1390
    },
    {
      "epoch": 0.12,
      "grad_norm": 10.271093409145028,
      "learning_rate": 9.973123027280013e-06,
      "loss": 0.8543,
      "step": 1391
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.436258019637284,
      "learning_rate": 9.973048176304267e-06,
      "loss": 0.8816,
      "step": 1392
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.422706784420768,
      "learning_rate": 9.972973221527034e-06,
      "loss": 0.9169,
      "step": 1393
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.92028105455892,
      "learning_rate": 9.972898162949873e-06,
      "loss": 1.0274,
      "step": 1394
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.786742245681121,
      "learning_rate": 9.972823000574356e-06,
      "loss": 0.9764,
      "step": 1395
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.2748910591766975,
      "learning_rate": 9.972747734402047e-06,
      "loss": 0.9217,
      "step": 1396
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.456598129428456,
      "learning_rate": 9.97267236443452e-06,
      "loss": 0.874,
      "step": 1397
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.560215297114147,
      "learning_rate": 9.972596890673348e-06,
      "loss": 0.9331,
      "step": 1398
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.382915011542906,
      "learning_rate": 9.972521313120104e-06,
      "loss": 0.8948,
      "step": 1399
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.8005493625211395,
      "learning_rate": 9.972445631776369e-06,
      "loss": 0.9445,
      "step": 1400
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.193045312205014,
      "learning_rate": 9.972369846643718e-06,
      "loss": 0.9673,
      "step": 1401
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.489185159993385,
      "learning_rate": 9.972293957723738e-06,
      "loss": 0.964,
      "step": 1402
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.124306280574945,
      "learning_rate": 9.97221796501801e-06,
      "loss": 0.964,
      "step": 1403
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.014632279641974,
      "learning_rate": 9.972141868528122e-06,
      "loss": 0.9243,
      "step": 1404
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.00376633299006,
      "learning_rate": 9.972065668255661e-06,
      "loss": 0.9367,
      "step": 1405
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.608011079071936,
      "learning_rate": 9.971989364202218e-06,
      "loss": 0.9971,
      "step": 1406
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.539719960602782,
      "learning_rate": 9.971912956369384e-06,
      "loss": 0.9041,
      "step": 1407
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.581733229210309,
      "learning_rate": 9.971836444758757e-06,
      "loss": 0.9185,
      "step": 1408
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.591490397915814,
      "learning_rate": 9.971759829371932e-06,
      "loss": 0.8614,
      "step": 1409
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.85093714956467,
      "learning_rate": 9.971683110210508e-06,
      "loss": 0.9125,
      "step": 1410
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.793398650791511,
      "learning_rate": 9.971606287276087e-06,
      "loss": 0.9388,
      "step": 1411
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.042171884680501,
      "learning_rate": 9.971529360570274e-06,
      "loss": 0.9215,
      "step": 1412
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.764365488674578,
      "learning_rate": 9.97145233009467e-06,
      "loss": 0.8917,
      "step": 1413
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.278725636110836,
      "learning_rate": 9.971375195850888e-06,
      "loss": 0.9283,
      "step": 1414
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.308213058636161,
      "learning_rate": 9.971297957840536e-06,
      "loss": 1.0105,
      "step": 1415
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.1044021065483465,
      "learning_rate": 9.971220616065225e-06,
      "loss": 0.8685,
      "step": 1416
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.24260011826315,
      "learning_rate": 9.971143170526571e-06,
      "loss": 0.9139,
      "step": 1417
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.910762315653145,
      "learning_rate": 9.97106562122619e-06,
      "loss": 0.9503,
      "step": 1418
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.875247249659718,
      "learning_rate": 9.9709879681657e-06,
      "loss": 0.9025,
      "step": 1419
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.101539209121077,
      "learning_rate": 9.970910211346723e-06,
      "loss": 0.9489,
      "step": 1420
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.794269038346682,
      "learning_rate": 9.970832350770882e-06,
      "loss": 0.9511,
      "step": 1421
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.259308646964794,
      "learning_rate": 9.9707543864398e-06,
      "loss": 0.8548,
      "step": 1422
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.07995805562377,
      "learning_rate": 9.970676318355108e-06,
      "loss": 0.9039,
      "step": 1423
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.334089375380962,
      "learning_rate": 9.970598146518431e-06,
      "loss": 0.938,
      "step": 1424
    },
    {
      "epoch": 0.13,
      "grad_norm": 10.429980010464089,
      "learning_rate": 9.970519870931405e-06,
      "loss": 0.9694,
      "step": 1425
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.517376976859172,
      "learning_rate": 9.970441491595662e-06,
      "loss": 0.9931,
      "step": 1426
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.15408167098142,
      "learning_rate": 9.970363008512836e-06,
      "loss": 0.9004,
      "step": 1427
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.870985848401764,
      "learning_rate": 9.97028442168457e-06,
      "loss": 0.9558,
      "step": 1428
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.458755578375824,
      "learning_rate": 9.970205731112499e-06,
      "loss": 0.9081,
      "step": 1429
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.695345012298683,
      "learning_rate": 9.97012693679827e-06,
      "loss": 0.9563,
      "step": 1430
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.929847616312393,
      "learning_rate": 9.970048038743523e-06,
      "loss": 0.873,
      "step": 1431
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.089376367160769,
      "learning_rate": 9.96996903694991e-06,
      "loss": 0.8906,
      "step": 1432
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.106364765616947,
      "learning_rate": 9.969889931419076e-06,
      "loss": 0.9084,
      "step": 1433
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.977237367091928,
      "learning_rate": 9.969810722152672e-06,
      "loss": 0.9273,
      "step": 1434
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.461515941192997,
      "learning_rate": 9.969731409152357e-06,
      "loss": 0.9659,
      "step": 1435
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.092915242835346,
      "learning_rate": 9.96965199241978e-06,
      "loss": 0.9358,
      "step": 1436
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.873865237724646,
      "learning_rate": 9.969572471956601e-06,
      "loss": 0.9388,
      "step": 1437
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.160620534612267,
      "learning_rate": 9.969492847764479e-06,
      "loss": 0.9129,
      "step": 1438
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.8410005413856245,
      "learning_rate": 9.969413119845078e-06,
      "loss": 0.9585,
      "step": 1439
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.915623791878984,
      "learning_rate": 9.96933328820006e-06,
      "loss": 0.9305,
      "step": 1440
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.66773988434155,
      "learning_rate": 9.969253352831092e-06,
      "loss": 0.9756,
      "step": 1441
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.575876506793106,
      "learning_rate": 9.969173313739845e-06,
      "loss": 0.9462,
      "step": 1442
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.320628445842298,
      "learning_rate": 9.969093170927986e-06,
      "loss": 0.946,
      "step": 1443
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.3208368325851705,
      "learning_rate": 9.96901292439719e-06,
      "loss": 0.9195,
      "step": 1444
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.521625704810516,
      "learning_rate": 9.96893257414913e-06,
      "loss": 0.886,
      "step": 1445
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.529356995429448,
      "learning_rate": 9.968852120185484e-06,
      "loss": 0.8844,
      "step": 1446
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.469608780727466,
      "learning_rate": 9.968771562507934e-06,
      "loss": 0.9914,
      "step": 1447
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.505761818307187,
      "learning_rate": 9.968690901118155e-06,
      "loss": 0.9024,
      "step": 1448
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.558244684036303,
      "learning_rate": 9.968610136017839e-06,
      "loss": 0.9485,
      "step": 1449
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.265197907561143,
      "learning_rate": 9.968529267208665e-06,
      "loss": 0.9706,
      "step": 1450
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.933033012533065,
      "learning_rate": 9.968448294692323e-06,
      "loss": 0.9901,
      "step": 1451
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.101708079944538,
      "learning_rate": 9.968367218470504e-06,
      "loss": 0.8873,
      "step": 1452
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.865341625333042,
      "learning_rate": 9.9682860385449e-06,
      "loss": 0.9304,
      "step": 1453
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.279785009111402,
      "learning_rate": 9.968204754917206e-06,
      "loss": 0.8705,
      "step": 1454
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.683729958828887,
      "learning_rate": 9.968123367589117e-06,
      "loss": 0.9749,
      "step": 1455
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.15217496545845,
      "learning_rate": 9.968041876562334e-06,
      "loss": 0.9322,
      "step": 1456
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.565612783489023,
      "learning_rate": 9.967960281838555e-06,
      "loss": 0.9083,
      "step": 1457
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.068144033561165,
      "learning_rate": 9.967878583419487e-06,
      "loss": 0.9526,
      "step": 1458
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.569752827626449,
      "learning_rate": 9.967796781306832e-06,
      "loss": 0.9701,
      "step": 1459
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.894134901307497,
      "learning_rate": 9.967714875502298e-06,
      "loss": 0.8823,
      "step": 1460
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.615266358061982,
      "learning_rate": 9.967632866007595e-06,
      "loss": 0.8834,
      "step": 1461
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.317984399215114,
      "learning_rate": 9.967550752824436e-06,
      "loss": 0.8883,
      "step": 1462
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.417259047434122,
      "learning_rate": 9.967468535954532e-06,
      "loss": 0.9067,
      "step": 1463
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.577994419642581,
      "learning_rate": 9.967386215399601e-06,
      "loss": 0.8895,
      "step": 1464
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.889024129548776,
      "learning_rate": 9.967303791161362e-06,
      "loss": 0.9679,
      "step": 1465
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.705117004352962,
      "learning_rate": 9.967221263241535e-06,
      "loss": 1.0101,
      "step": 1466
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.457160401140723,
      "learning_rate": 9.967138631641842e-06,
      "loss": 0.9524,
      "step": 1467
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.614198265902633,
      "learning_rate": 9.967055896364007e-06,
      "loss": 0.9569,
      "step": 1468
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.710900416824499,
      "learning_rate": 9.96697305740976e-06,
      "loss": 0.97,
      "step": 1469
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.457975005706524,
      "learning_rate": 9.966890114780825e-06,
      "loss": 0.9078,
      "step": 1470
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.701785151874144,
      "learning_rate": 9.966807068478938e-06,
      "loss": 0.9453,
      "step": 1471
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.662837493129242,
      "learning_rate": 9.96672391850583e-06,
      "loss": 1.0557,
      "step": 1472
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.984656379870067,
      "learning_rate": 9.966640664863236e-06,
      "loss": 1.043,
      "step": 1473
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.265190386228888,
      "learning_rate": 9.966557307552897e-06,
      "loss": 0.9516,
      "step": 1474
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.631319048933607,
      "learning_rate": 9.96647384657655e-06,
      "loss": 0.9557,
      "step": 1475
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.672550121272586,
      "learning_rate": 9.966390281935939e-06,
      "loss": 0.9713,
      "step": 1476
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.479244067560822,
      "learning_rate": 9.966306613632805e-06,
      "loss": 0.9785,
      "step": 1477
    },
    {
      "epoch": 0.13,
      "grad_norm": 10.245573691690058,
      "learning_rate": 9.966222841668897e-06,
      "loss": 0.8925,
      "step": 1478
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.057061889179519,
      "learning_rate": 9.966138966045964e-06,
      "loss": 0.8694,
      "step": 1479
    },
    {
      "epoch": 0.13,
      "grad_norm": 10.76239027831692,
      "learning_rate": 9.966054986765754e-06,
      "loss": 0.9517,
      "step": 1480
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.840527642161367,
      "learning_rate": 9.965970903830022e-06,
      "loss": 0.9371,
      "step": 1481
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.334577670001729,
      "learning_rate": 9.965886717240524e-06,
      "loss": 0.9405,
      "step": 1482
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.809989489007398,
      "learning_rate": 9.965802426999015e-06,
      "loss": 0.9381,
      "step": 1483
    },
    {
      "epoch": 0.13,
      "grad_norm": 14.77528387800605,
      "learning_rate": 9.965718033107256e-06,
      "loss": 0.9434,
      "step": 1484
    },
    {
      "epoch": 0.13,
      "grad_norm": 10.273239490917932,
      "learning_rate": 9.965633535567007e-06,
      "loss": 0.9476,
      "step": 1485
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.756966387729549,
      "learning_rate": 9.965548934380033e-06,
      "loss": 0.9606,
      "step": 1486
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.99997759198704,
      "learning_rate": 9.965464229548098e-06,
      "loss": 0.9341,
      "step": 1487
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.731272228521683,
      "learning_rate": 9.965379421072974e-06,
      "loss": 1.0007,
      "step": 1488
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.662229386857797,
      "learning_rate": 9.965294508956427e-06,
      "loss": 0.9046,
      "step": 1489
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.988912540664809,
      "learning_rate": 9.96520949320023e-06,
      "loss": 0.9646,
      "step": 1490
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.2273173756626425,
      "learning_rate": 9.96512437380616e-06,
      "loss": 0.95,
      "step": 1491
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.8307106945305,
      "learning_rate": 9.965039150775992e-06,
      "loss": 0.977,
      "step": 1492
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.750281797996829,
      "learning_rate": 9.964953824111502e-06,
      "loss": 0.9012,
      "step": 1493
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.494776625352568,
      "learning_rate": 9.964868393814478e-06,
      "loss": 0.8817,
      "step": 1494
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.487876752870589,
      "learning_rate": 9.964782859886698e-06,
      "loss": 0.9434,
      "step": 1495
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.89335746750403,
      "learning_rate": 9.964697222329948e-06,
      "loss": 0.8869,
      "step": 1496
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.9075575136972835,
      "learning_rate": 9.964611481146017e-06,
      "loss": 0.9677,
      "step": 1497
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.111916567288766,
      "learning_rate": 9.964525636336693e-06,
      "loss": 0.9889,
      "step": 1498
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.095137933468349,
      "learning_rate": 9.964439687903767e-06,
      "loss": 0.9234,
      "step": 1499
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.816236430818839,
      "learning_rate": 9.964353635849035e-06,
      "loss": 0.9633,
      "step": 1500
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.886778649423889,
      "learning_rate": 9.964267480174294e-06,
      "loss": 0.861,
      "step": 1501
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.545404470312661,
      "learning_rate": 9.964181220881342e-06,
      "loss": 0.9093,
      "step": 1502
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.9030707338731645,
      "learning_rate": 9.964094857971975e-06,
      "loss": 0.931,
      "step": 1503
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.171015532527012,
      "learning_rate": 9.964008391448001e-06,
      "loss": 0.9152,
      "step": 1504
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.51890505584007,
      "learning_rate": 9.963921821311222e-06,
      "loss": 0.9268,
      "step": 1505
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.367841449567285,
      "learning_rate": 9.963835147563448e-06,
      "loss": 0.8673,
      "step": 1506
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.984130528331638,
      "learning_rate": 9.963748370206484e-06,
      "loss": 0.9503,
      "step": 1507
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.815962981905127,
      "learning_rate": 9.963661489242142e-06,
      "loss": 0.9149,
      "step": 1508
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.695890120393516,
      "learning_rate": 9.963574504672238e-06,
      "loss": 0.9357,
      "step": 1509
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.71650306074139,
      "learning_rate": 9.963487416498587e-06,
      "loss": 0.9137,
      "step": 1510
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.198996380186198,
      "learning_rate": 9.963400224723006e-06,
      "loss": 0.8921,
      "step": 1511
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.4371197132936935,
      "learning_rate": 9.963312929347314e-06,
      "loss": 0.9446,
      "step": 1512
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.496143406043688,
      "learning_rate": 9.963225530373334e-06,
      "loss": 0.9714,
      "step": 1513
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.211123153090494,
      "learning_rate": 9.96313802780289e-06,
      "loss": 0.9901,
      "step": 1514
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.777507095592017,
      "learning_rate": 9.96305042163781e-06,
      "loss": 0.9115,
      "step": 1515
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.694649435461505,
      "learning_rate": 9.96296271187992e-06,
      "loss": 0.8718,
      "step": 1516
    },
    {
      "epoch": 0.14,
      "grad_norm": 10.029009928319788,
      "learning_rate": 9.962874898531054e-06,
      "loss": 0.965,
      "step": 1517
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.118782686602256,
      "learning_rate": 9.96278698159304e-06,
      "loss": 0.9467,
      "step": 1518
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.734848318165714,
      "learning_rate": 9.962698961067719e-06,
      "loss": 0.9685,
      "step": 1519
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.714221434940324,
      "learning_rate": 9.962610836956925e-06,
      "loss": 0.9952,
      "step": 1520
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.56942945370318,
      "learning_rate": 9.962522609262496e-06,
      "loss": 0.94,
      "step": 1521
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.943309027910723,
      "learning_rate": 9.962434277986277e-06,
      "loss": 0.8906,
      "step": 1522
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.881272689063244,
      "learning_rate": 9.962345843130108e-06,
      "loss": 0.9553,
      "step": 1523
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.293231607912387,
      "learning_rate": 9.962257304695839e-06,
      "loss": 0.8877,
      "step": 1524
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.061176286345948,
      "learning_rate": 9.962168662685315e-06,
      "loss": 0.9676,
      "step": 1525
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.091752423691399,
      "learning_rate": 9.962079917100387e-06,
      "loss": 0.944,
      "step": 1526
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.810813299229871,
      "learning_rate": 9.961991067942909e-06,
      "loss": 0.9197,
      "step": 1527
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.867019895092896,
      "learning_rate": 9.961902115214732e-06,
      "loss": 0.9805,
      "step": 1528
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.905705460898345,
      "learning_rate": 9.961813058917715e-06,
      "loss": 0.9057,
      "step": 1529
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.238106648905651,
      "learning_rate": 9.961723899053717e-06,
      "loss": 0.9803,
      "step": 1530
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.30321980059641,
      "learning_rate": 9.9616346356246e-06,
      "loss": 0.9365,
      "step": 1531
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.5463167300035705,
      "learning_rate": 9.961545268632223e-06,
      "loss": 0.9197,
      "step": 1532
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.462861864393142,
      "learning_rate": 9.961455798078456e-06,
      "loss": 0.8785,
      "step": 1533
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.272348750390195,
      "learning_rate": 9.961366223965164e-06,
      "loss": 0.971,
      "step": 1534
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.29975833566798,
      "learning_rate": 9.961276546294218e-06,
      "loss": 0.972,
      "step": 1535
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.155416331614127,
      "learning_rate": 9.961186765067487e-06,
      "loss": 0.8926,
      "step": 1536
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.467988968228203,
      "learning_rate": 9.961096880286848e-06,
      "loss": 0.9183,
      "step": 1537
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.436221445441653,
      "learning_rate": 9.961006891954179e-06,
      "loss": 0.9927,
      "step": 1538
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.895549255783218,
      "learning_rate": 9.960916800071354e-06,
      "loss": 0.8846,
      "step": 1539
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.088360463767982,
      "learning_rate": 9.960826604640252e-06,
      "loss": 0.8782,
      "step": 1540
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.346497818674137,
      "learning_rate": 9.960736305662762e-06,
      "loss": 0.9009,
      "step": 1541
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.46873268186375,
      "learning_rate": 9.960645903140764e-06,
      "loss": 0.8629,
      "step": 1542
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.3758426789104945,
      "learning_rate": 9.960555397076148e-06,
      "loss": 0.9068,
      "step": 1543
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.960273536470398,
      "learning_rate": 9.9604647874708e-06,
      "loss": 0.8841,
      "step": 1544
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.14881031273148,
      "learning_rate": 9.960374074326614e-06,
      "loss": 0.9864,
      "step": 1545
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.44895562633785,
      "learning_rate": 9.96028325764548e-06,
      "loss": 0.9877,
      "step": 1546
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.462630464251822,
      "learning_rate": 9.960192337429298e-06,
      "loss": 0.9334,
      "step": 1547
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.710264637556267,
      "learning_rate": 9.960101313679962e-06,
      "loss": 0.859,
      "step": 1548
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.826475716263829,
      "learning_rate": 9.960010186399377e-06,
      "loss": 0.8924,
      "step": 1549
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.9226946013126245,
      "learning_rate": 9.959918955589437e-06,
      "loss": 0.9228,
      "step": 1550
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.614143607064812,
      "learning_rate": 9.959827621252054e-06,
      "loss": 0.9732,
      "step": 1551
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.815543882227329,
      "learning_rate": 9.959736183389129e-06,
      "loss": 0.8627,
      "step": 1552
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.609969470958643,
      "learning_rate": 9.959644642002575e-06,
      "loss": 0.8736,
      "step": 1553
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.742106783433274,
      "learning_rate": 9.959552997094298e-06,
      "loss": 0.9201,
      "step": 1554
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.525962518450011,
      "learning_rate": 9.959461248666215e-06,
      "loss": 0.9146,
      "step": 1555
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.625844750777356,
      "learning_rate": 9.959369396720239e-06,
      "loss": 0.8562,
      "step": 1556
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.161582689846505,
      "learning_rate": 9.959277441258289e-06,
      "loss": 0.8629,
      "step": 1557
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.400325484654861,
      "learning_rate": 9.959185382282281e-06,
      "loss": 0.9723,
      "step": 1558
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.769015325508441,
      "learning_rate": 9.95909321979414e-06,
      "loss": 0.9574,
      "step": 1559
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.232166618886888,
      "learning_rate": 9.959000953795788e-06,
      "loss": 0.8452,
      "step": 1560
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.346072118665611,
      "learning_rate": 9.95890858428915e-06,
      "loss": 0.9311,
      "step": 1561
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.191176639869863,
      "learning_rate": 9.958816111276157e-06,
      "loss": 0.9526,
      "step": 1562
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.030434895306083,
      "learning_rate": 9.958723534758737e-06,
      "loss": 0.9506,
      "step": 1563
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.984700349716613,
      "learning_rate": 9.958630854738823e-06,
      "loss": 0.8884,
      "step": 1564
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.896457587783347,
      "learning_rate": 9.958538071218346e-06,
      "loss": 0.8595,
      "step": 1565
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.416638824952019,
      "learning_rate": 9.95844518419925e-06,
      "loss": 0.9243,
      "step": 1566
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.452565730530738,
      "learning_rate": 9.958352193683467e-06,
      "loss": 0.9087,
      "step": 1567
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.530453787778004,
      "learning_rate": 9.95825909967294e-06,
      "loss": 0.9537,
      "step": 1568
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.062685225886606,
      "learning_rate": 9.958165902169615e-06,
      "loss": 0.9331,
      "step": 1569
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.958940142387215,
      "learning_rate": 9.958072601175433e-06,
      "loss": 0.9014,
      "step": 1570
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.319760098709922,
      "learning_rate": 9.957979196692344e-06,
      "loss": 0.9265,
      "step": 1571
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.174085759901779,
      "learning_rate": 9.957885688722297e-06,
      "loss": 0.948,
      "step": 1572
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.944361662258861,
      "learning_rate": 9.957792077267246e-06,
      "loss": 0.9295,
      "step": 1573
    },
    {
      "epoch": 0.14,
      "grad_norm": 11.250885120584293,
      "learning_rate": 9.95769836232914e-06,
      "loss": 0.9107,
      "step": 1574
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.821025433340639,
      "learning_rate": 9.957604543909937e-06,
      "loss": 0.8638,
      "step": 1575
    },
    {
      "epoch": 0.14,
      "grad_norm": 10.005432621058242,
      "learning_rate": 9.957510622011599e-06,
      "loss": 0.9291,
      "step": 1576
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.235087986577309,
      "learning_rate": 9.957416596636082e-06,
      "loss": 0.9453,
      "step": 1577
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.853508811337052,
      "learning_rate": 9.95732246778535e-06,
      "loss": 0.9367,
      "step": 1578
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.582689286431996,
      "learning_rate": 9.95722823546137e-06,
      "loss": 0.872,
      "step": 1579
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.088243063544556,
      "learning_rate": 9.957133899666104e-06,
      "loss": 0.9022,
      "step": 1580
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.681169994464124,
      "learning_rate": 9.957039460401526e-06,
      "loss": 0.9655,
      "step": 1581
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.365774528830001,
      "learning_rate": 9.956944917669603e-06,
      "loss": 0.9053,
      "step": 1582
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.232399000410088,
      "learning_rate": 9.95685027147231e-06,
      "loss": 0.9203,
      "step": 1583
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.22323845760724,
      "learning_rate": 9.956755521811624e-06,
      "loss": 0.9181,
      "step": 1584
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.215650437765172,
      "learning_rate": 9.956660668689525e-06,
      "loss": 0.9024,
      "step": 1585
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.012589820645012,
      "learning_rate": 9.956565712107985e-06,
      "loss": 0.916,
      "step": 1586
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.745181256208862,
      "learning_rate": 9.956470652068992e-06,
      "loss": 0.8777,
      "step": 1587
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.111189612432613,
      "learning_rate": 9.956375488574528e-06,
      "loss": 0.9405,
      "step": 1588
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.456324170831776,
      "learning_rate": 9.956280221626579e-06,
      "loss": 0.9381,
      "step": 1589
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.12957436599661,
      "learning_rate": 9.956184851227137e-06,
      "loss": 0.8379,
      "step": 1590
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.791346112392386,
      "learning_rate": 9.956089377378189e-06,
      "loss": 0.9704,
      "step": 1591
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.2236583279503375,
      "learning_rate": 9.955993800081726e-06,
      "loss": 0.9037,
      "step": 1592
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.4382925624230385,
      "learning_rate": 9.955898119339751e-06,
      "loss": 0.8835,
      "step": 1593
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.788875846954354,
      "learning_rate": 9.955802335154252e-06,
      "loss": 0.8519,
      "step": 1594
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.904733605378186,
      "learning_rate": 9.955706447527234e-06,
      "loss": 0.951,
      "step": 1595
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.530877175412536,
      "learning_rate": 9.955610456460696e-06,
      "loss": 0.8934,
      "step": 1596
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.848030672654852,
      "learning_rate": 9.955514361956645e-06,
      "loss": 0.9058,
      "step": 1597
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.269765210311887,
      "learning_rate": 9.95541816401708e-06,
      "loss": 0.9655,
      "step": 1598
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.572324435623046,
      "learning_rate": 9.955321862644015e-06,
      "loss": 0.8627,
      "step": 1599
    },
    {
      "epoch": 0.14,
      "grad_norm": 10.211032360126362,
      "learning_rate": 9.955225457839457e-06,
      "loss": 0.9562,
      "step": 1600
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.100305627570337,
      "learning_rate": 9.955128949605422e-06,
      "loss": 0.934,
      "step": 1601
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.836818277027663,
      "learning_rate": 9.95503233794392e-06,
      "loss": 0.8672,
      "step": 1602
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.878640492697954,
      "learning_rate": 9.95493562285697e-06,
      "loss": 0.8796,
      "step": 1603
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.748222446868447,
      "learning_rate": 9.954838804346589e-06,
      "loss": 0.9703,
      "step": 1604
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.893872397135734,
      "learning_rate": 9.954741882414798e-06,
      "loss": 0.9156,
      "step": 1605
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.324674758949067,
      "learning_rate": 9.954644857063621e-06,
      "loss": 0.9503,
      "step": 1606
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.588664667772502,
      "learning_rate": 9.954547728295086e-06,
      "loss": 0.9713,
      "step": 1607
    },
    {
      "epoch": 0.14,
      "grad_norm": 11.109978895096539,
      "learning_rate": 9.954450496111215e-06,
      "loss": 0.8839,
      "step": 1608
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.403282203101645,
      "learning_rate": 9.954353160514042e-06,
      "loss": 0.8763,
      "step": 1609
    },
    {
      "epoch": 0.14,
      "grad_norm": 11.536474803876654,
      "learning_rate": 9.954255721505594e-06,
      "loss": 0.922,
      "step": 1610
    },
    {
      "epoch": 0.14,
      "grad_norm": 10.042886649749127,
      "learning_rate": 9.954158179087909e-06,
      "loss": 0.9614,
      "step": 1611
    },
    {
      "epoch": 0.14,
      "grad_norm": 10.278404577916737,
      "learning_rate": 9.954060533263022e-06,
      "loss": 0.9188,
      "step": 1612
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.570848072092514,
      "learning_rate": 9.953962784032969e-06,
      "loss": 0.8783,
      "step": 1613
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.404222508147017,
      "learning_rate": 9.953864931399792e-06,
      "loss": 0.9393,
      "step": 1614
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.592356539149444,
      "learning_rate": 9.953766975365533e-06,
      "loss": 0.9946,
      "step": 1615
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.780505305779121,
      "learning_rate": 9.953668915932239e-06,
      "loss": 0.8763,
      "step": 1616
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.711717506942648,
      "learning_rate": 9.953570753101954e-06,
      "loss": 0.9725,
      "step": 1617
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.401967868599291,
      "learning_rate": 9.953472486876726e-06,
      "loss": 0.9136,
      "step": 1618
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.884798886408169,
      "learning_rate": 9.953374117258609e-06,
      "loss": 0.9631,
      "step": 1619
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.628192231018277,
      "learning_rate": 9.953275644249656e-06,
      "loss": 0.8703,
      "step": 1620
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.97171141187945,
      "learning_rate": 9.95317706785192e-06,
      "loss": 0.9298,
      "step": 1621
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.962987921613148,
      "learning_rate": 9.95307838806746e-06,
      "loss": 0.9245,
      "step": 1622
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.918746859730616,
      "learning_rate": 9.952979604898336e-06,
      "loss": 0.8748,
      "step": 1623
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.088328665839549,
      "learning_rate": 9.952880718346609e-06,
      "loss": 0.9175,
      "step": 1624
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.934004194409862,
      "learning_rate": 9.952781728414345e-06,
      "loss": 0.9133,
      "step": 1625
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.295567832750349,
      "learning_rate": 9.952682635103608e-06,
      "loss": 0.929,
      "step": 1626
    },
    {
      "epoch": 0.15,
      "grad_norm": 9.612628122857485,
      "learning_rate": 9.952583438416468e-06,
      "loss": 0.9739,
      "step": 1627
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.07815200572916,
      "learning_rate": 9.952484138354993e-06,
      "loss": 0.8954,
      "step": 1628
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.107446932202551,
      "learning_rate": 9.952384734921258e-06,
      "loss": 0.9275,
      "step": 1629
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.755773832239196,
      "learning_rate": 9.952285228117338e-06,
      "loss": 0.9083,
      "step": 1630
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.598739022702816,
      "learning_rate": 9.952185617945309e-06,
      "loss": 0.9246,
      "step": 1631
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.660094529278459,
      "learning_rate": 9.952085904407249e-06,
      "loss": 0.8872,
      "step": 1632
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.5431509215149015,
      "learning_rate": 9.951986087505243e-06,
      "loss": 0.9102,
      "step": 1633
    },
    {
      "epoch": 0.15,
      "grad_norm": 10.867852850266456,
      "learning_rate": 9.95188616724137e-06,
      "loss": 0.9218,
      "step": 1634
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.141830856404113,
      "learning_rate": 9.95178614361772e-06,
      "loss": 0.8333,
      "step": 1635
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.618740354305997,
      "learning_rate": 9.951686016636376e-06,
      "loss": 1.0055,
      "step": 1636
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.580377414884051,
      "learning_rate": 9.951585786299432e-06,
      "loss": 0.8796,
      "step": 1637
    },
    {
      "epoch": 0.15,
      "grad_norm": 11.313401299055668,
      "learning_rate": 9.951485452608977e-06,
      "loss": 0.8887,
      "step": 1638
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.019186887040177,
      "learning_rate": 9.951385015567107e-06,
      "loss": 0.8666,
      "step": 1639
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.257089896155504,
      "learning_rate": 9.951284475175919e-06,
      "loss": 0.9352,
      "step": 1640
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.206331596020261,
      "learning_rate": 9.951183831437511e-06,
      "loss": 0.9336,
      "step": 1641
    },
    {
      "epoch": 0.15,
      "grad_norm": 11.0335023142606,
      "learning_rate": 9.951083084353982e-06,
      "loss": 0.9119,
      "step": 1642
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.787881356412708,
      "learning_rate": 9.950982233927438e-06,
      "loss": 0.8965,
      "step": 1643
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.534586493182818,
      "learning_rate": 9.950881280159982e-06,
      "loss": 0.8833,
      "step": 1644
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.297411512513467,
      "learning_rate": 9.950780223053721e-06,
      "loss": 0.8603,
      "step": 1645
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.530969771725447,
      "learning_rate": 9.950679062610765e-06,
      "loss": 0.9355,
      "step": 1646
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.438774427964296,
      "learning_rate": 9.950577798833225e-06,
      "loss": 0.9604,
      "step": 1647
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.57751516707807,
      "learning_rate": 9.950476431723215e-06,
      "loss": 0.9704,
      "step": 1648
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.060286671113943,
      "learning_rate": 9.950374961282851e-06,
      "loss": 0.8842,
      "step": 1649
    },
    {
      "epoch": 0.15,
      "grad_norm": 9.465632701451721,
      "learning_rate": 9.950273387514251e-06,
      "loss": 0.9383,
      "step": 1650
    },
    {
      "epoch": 0.15,
      "grad_norm": 9.394179935959775,
      "learning_rate": 9.950171710419536e-06,
      "loss": 0.8442,
      "step": 1651
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.395904798073917,
      "learning_rate": 9.950069930000826e-06,
      "loss": 0.9532,
      "step": 1652
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.385200628482545,
      "learning_rate": 9.949968046260247e-06,
      "loss": 0.9614,
      "step": 1653
    },
    {
      "epoch": 0.15,
      "grad_norm": 10.814686917170802,
      "learning_rate": 9.949866059199924e-06,
      "loss": 0.9036,
      "step": 1654
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.8487332424652365,
      "learning_rate": 9.949763968821988e-06,
      "loss": 0.901,
      "step": 1655
    },
    {
      "epoch": 0.15,
      "grad_norm": 10.025135776580573,
      "learning_rate": 9.94966177512857e-06,
      "loss": 0.9594,
      "step": 1656
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.62001875585656,
      "learning_rate": 9.949559478121803e-06,
      "loss": 0.944,
      "step": 1657
    },
    {
      "epoch": 0.15,
      "grad_norm": 9.11919480741434,
      "learning_rate": 9.949457077803818e-06,
      "loss": 0.946,
      "step": 1658
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.915650163655938,
      "learning_rate": 9.949354574176758e-06,
      "loss": 0.8696,
      "step": 1659
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.091343227222517,
      "learning_rate": 9.949251967242758e-06,
      "loss": 0.9585,
      "step": 1660
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.6338414515675,
      "learning_rate": 9.949149257003962e-06,
      "loss": 0.9497,
      "step": 1661
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.9403887689123485,
      "learning_rate": 9.949046443462516e-06,
      "loss": 0.874,
      "step": 1662
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.033071413391915,
      "learning_rate": 9.948943526620563e-06,
      "loss": 0.9472,
      "step": 1663
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.879510918371288,
      "learning_rate": 9.94884050648025e-06,
      "loss": 0.8576,
      "step": 1664
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.553250545870903,
      "learning_rate": 9.948737383043731e-06,
      "loss": 0.9467,
      "step": 1665
    },
    {
      "epoch": 0.15,
      "grad_norm": 9.298021151931284,
      "learning_rate": 9.948634156313155e-06,
      "loss": 0.9417,
      "step": 1666
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.679930869260088,
      "learning_rate": 9.948530826290681e-06,
      "loss": 0.9032,
      "step": 1667
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.67615368403085,
      "learning_rate": 9.948427392978461e-06,
      "loss": 0.8964,
      "step": 1668
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.14373243843311,
      "learning_rate": 9.948323856378657e-06,
      "loss": 0.9366,
      "step": 1669
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.46552524471624,
      "learning_rate": 9.94822021649343e-06,
      "loss": 0.9038,
      "step": 1670
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.266437386319904,
      "learning_rate": 9.948116473324941e-06,
      "loss": 0.9006,
      "step": 1671
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.898731506656267,
      "learning_rate": 9.948012626875357e-06,
      "loss": 1.0095,
      "step": 1672
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.3668682307498905,
      "learning_rate": 9.947908677146844e-06,
      "loss": 0.8666,
      "step": 1673
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.4966152024561685,
      "learning_rate": 9.947804624141574e-06,
      "loss": 0.9298,
      "step": 1674
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.603147347741379,
      "learning_rate": 9.94770046786172e-06,
      "loss": 0.952,
      "step": 1675
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.520447343137617,
      "learning_rate": 9.947596208309451e-06,
      "loss": 0.8984,
      "step": 1676
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.041433439476444,
      "learning_rate": 9.947491845486949e-06,
      "loss": 1.0086,
      "step": 1677
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.548263325205756,
      "learning_rate": 9.947387379396388e-06,
      "loss": 0.9611,
      "step": 1678
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.376305733970497,
      "learning_rate": 9.947282810039952e-06,
      "loss": 0.8724,
      "step": 1679
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.88473640583211,
      "learning_rate": 9.947178137419818e-06,
      "loss": 0.9513,
      "step": 1680
    },
    {
      "epoch": 0.15,
      "grad_norm": 10.15334400772932,
      "learning_rate": 9.947073361538178e-06,
      "loss": 0.968,
      "step": 1681
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.5419835993865325,
      "learning_rate": 9.946968482397212e-06,
      "loss": 0.9032,
      "step": 1682
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.556415358310082,
      "learning_rate": 9.946863499999115e-06,
      "loss": 0.9255,
      "step": 1683
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.045787642531568,
      "learning_rate": 9.946758414346076e-06,
      "loss": 0.9172,
      "step": 1684
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.18339175995122,
      "learning_rate": 9.946653225440288e-06,
      "loss": 0.9054,
      "step": 1685
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.257034861838449,
      "learning_rate": 9.946547933283946e-06,
      "loss": 0.883,
      "step": 1686
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.018608651371892,
      "learning_rate": 9.94644253787925e-06,
      "loss": 0.889,
      "step": 1687
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.033089154995128,
      "learning_rate": 9.946337039228398e-06,
      "loss": 0.9152,
      "step": 1688
    },
    {
      "epoch": 0.15,
      "grad_norm": 9.024412305894637,
      "learning_rate": 9.946231437333591e-06,
      "loss": 0.936,
      "step": 1689
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.980561338240078,
      "learning_rate": 9.946125732197038e-06,
      "loss": 0.9654,
      "step": 1690
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.847790590680537,
      "learning_rate": 9.94601992382094e-06,
      "loss": 0.9473,
      "step": 1691
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.869862480320677,
      "learning_rate": 9.945914012207507e-06,
      "loss": 0.935,
      "step": 1692
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.387289337860115,
      "learning_rate": 9.945807997358952e-06,
      "loss": 0.9288,
      "step": 1693
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.852107908778715,
      "learning_rate": 9.945701879277487e-06,
      "loss": 0.9103,
      "step": 1694
    },
    {
      "epoch": 0.15,
      "grad_norm": 11.069794500256458,
      "learning_rate": 9.945595657965325e-06,
      "loss": 0.914,
      "step": 1695
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.100434649482077,
      "learning_rate": 9.945489333424684e-06,
      "loss": 0.9481,
      "step": 1696
    },
    {
      "epoch": 0.15,
      "grad_norm": 9.102953555515983,
      "learning_rate": 9.945382905657784e-06,
      "loss": 0.8937,
      "step": 1697
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.859250402005038,
      "learning_rate": 9.945276374666846e-06,
      "loss": 0.8706,
      "step": 1698
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.170524301478496,
      "learning_rate": 9.945169740454093e-06,
      "loss": 0.9892,
      "step": 1699
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.60665034666195,
      "learning_rate": 9.945063003021752e-06,
      "loss": 0.873,
      "step": 1700
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.153761801261673,
      "learning_rate": 9.94495616237205e-06,
      "loss": 0.9607,
      "step": 1701
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.845610578743642,
      "learning_rate": 9.944849218507218e-06,
      "loss": 0.9081,
      "step": 1702
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.996019886605271,
      "learning_rate": 9.944742171429487e-06,
      "loss": 0.9512,
      "step": 1703
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.652224529152166,
      "learning_rate": 9.944635021141092e-06,
      "loss": 0.9158,
      "step": 1704
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.286106931876958,
      "learning_rate": 9.94452776764427e-06,
      "loss": 0.8864,
      "step": 1705
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.082729459693538,
      "learning_rate": 9.944420410941259e-06,
      "loss": 0.904,
      "step": 1706
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.524322349595408,
      "learning_rate": 9.9443129510343e-06,
      "loss": 0.907,
      "step": 1707
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.135708507307406,
      "learning_rate": 9.944205387925636e-06,
      "loss": 0.9169,
      "step": 1708
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.396832705431355,
      "learning_rate": 9.944097721617512e-06,
      "loss": 0.9884,
      "step": 1709
    },
    {
      "epoch": 0.15,
      "grad_norm": 10.013758210589762,
      "learning_rate": 9.943989952112175e-06,
      "loss": 0.9621,
      "step": 1710
    },
    {
      "epoch": 0.15,
      "grad_norm": 9.870494631658731,
      "learning_rate": 9.943882079411875e-06,
      "loss": 0.9017,
      "step": 1711
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.036521946912087,
      "learning_rate": 9.943774103518865e-06,
      "loss": 1.0531,
      "step": 1712
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.687615130350311,
      "learning_rate": 9.943666024435396e-06,
      "loss": 0.9067,
      "step": 1713
    },
    {
      "epoch": 0.15,
      "grad_norm": 10.441340833719739,
      "learning_rate": 9.943557842163724e-06,
      "loss": 0.8934,
      "step": 1714
    },
    {
      "epoch": 0.15,
      "grad_norm": 10.11224638978164,
      "learning_rate": 9.94344955670611e-06,
      "loss": 0.9045,
      "step": 1715
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.713886352289906,
      "learning_rate": 9.943341168064814e-06,
      "loss": 0.8881,
      "step": 1716
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.085050871851386,
      "learning_rate": 9.943232676242093e-06,
      "loss": 0.9092,
      "step": 1717
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.7501780656307435,
      "learning_rate": 9.943124081240218e-06,
      "loss": 0.935,
      "step": 1718
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.249628563876324,
      "learning_rate": 9.943015383061452e-06,
      "loss": 0.9152,
      "step": 1719
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.110192158442351,
      "learning_rate": 9.942906581708065e-06,
      "loss": 0.8407,
      "step": 1720
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.091247509880226,
      "learning_rate": 9.94279767718233e-06,
      "loss": 0.885,
      "step": 1721
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.643072655517878,
      "learning_rate": 9.942688669486516e-06,
      "loss": 0.9057,
      "step": 1722
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.203657265005804,
      "learning_rate": 9.9425795586229e-06,
      "loss": 0.9195,
      "step": 1723
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.382953068547134,
      "learning_rate": 9.94247034459376e-06,
      "loss": 0.9226,
      "step": 1724
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.928244709776375,
      "learning_rate": 9.942361027401377e-06,
      "loss": 0.9165,
      "step": 1725
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.173564018957629,
      "learning_rate": 9.942251607048028e-06,
      "loss": 0.9384,
      "step": 1726
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.517349645697361,
      "learning_rate": 9.942142083536001e-06,
      "loss": 0.8512,
      "step": 1727
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.593011436539596,
      "learning_rate": 9.942032456867583e-06,
      "loss": 0.85,
      "step": 1728
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.031208553051318,
      "learning_rate": 9.94192272704506e-06,
      "loss": 0.9005,
      "step": 1729
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.393408853043171,
      "learning_rate": 9.941812894070721e-06,
      "loss": 0.9366,
      "step": 1730
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.143431683291103,
      "learning_rate": 9.941702957946862e-06,
      "loss": 0.8755,
      "step": 1731
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.165204469012367,
      "learning_rate": 9.941592918675776e-06,
      "loss": 0.9885,
      "step": 1732
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.859869829834116,
      "learning_rate": 9.941482776259759e-06,
      "loss": 0.9207,
      "step": 1733
    },
    {
      "epoch": 0.15,
      "grad_norm": 10.075930869996785,
      "learning_rate": 9.94137253070111e-06,
      "loss": 0.9505,
      "step": 1734
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.739330753387975,
      "learning_rate": 9.941262182002132e-06,
      "loss": 0.8816,
      "step": 1735
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.316348404513737,
      "learning_rate": 9.941151730165126e-06,
      "loss": 0.8949,
      "step": 1736
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.6237514632247185,
      "learning_rate": 9.941041175192398e-06,
      "loss": 0.9201,
      "step": 1737
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.3332960734469035,
      "learning_rate": 9.940930517086257e-06,
      "loss": 0.9457,
      "step": 1738
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.472547554711808,
      "learning_rate": 9.940819755849014e-06,
      "loss": 0.8611,
      "step": 1739
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.210100322273938,
      "learning_rate": 9.940708891482976e-06,
      "loss": 0.9845,
      "step": 1740
    },
    {
      "epoch": 0.16,
      "grad_norm": 10.003344880230083,
      "learning_rate": 9.94059792399046e-06,
      "loss": 0.8689,
      "step": 1741
    },
    {
      "epoch": 0.16,
      "grad_norm": 11.720424237016914,
      "learning_rate": 9.940486853373784e-06,
      "loss": 0.8544,
      "step": 1742
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.494506023182325,
      "learning_rate": 9.940375679635263e-06,
      "loss": 0.909,
      "step": 1743
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.929783731476001,
      "learning_rate": 9.94026440277722e-06,
      "loss": 0.9753,
      "step": 1744
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.088974893253955,
      "learning_rate": 9.940153022801974e-06,
      "loss": 1.0166,
      "step": 1745
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.177373254500309,
      "learning_rate": 9.940041539711856e-06,
      "loss": 0.921,
      "step": 1746
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.915353698236475,
      "learning_rate": 9.939929953509187e-06,
      "loss": 0.9182,
      "step": 1747
    },
    {
      "epoch": 0.16,
      "grad_norm": 9.147677884431397,
      "learning_rate": 9.939818264196298e-06,
      "loss": 0.9032,
      "step": 1748
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.888899738992613,
      "learning_rate": 9.939706471775522e-06,
      "loss": 1.0015,
      "step": 1749
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.057088963816157,
      "learning_rate": 9.939594576249191e-06,
      "loss": 0.9629,
      "step": 1750
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.70960381311623,
      "learning_rate": 9.939482577619642e-06,
      "loss": 0.8987,
      "step": 1751
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.920418830419786,
      "learning_rate": 9.93937047588921e-06,
      "loss": 0.9154,
      "step": 1752
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.834139175094775,
      "learning_rate": 9.939258271060237e-06,
      "loss": 0.9288,
      "step": 1753
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.671382658879792,
      "learning_rate": 9.939145963135065e-06,
      "loss": 0.9393,
      "step": 1754
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.996948101096514,
      "learning_rate": 9.939033552116036e-06,
      "loss": 0.8532,
      "step": 1755
    },
    {
      "epoch": 0.16,
      "grad_norm": 9.643435811130972,
      "learning_rate": 9.938921038005497e-06,
      "loss": 0.975,
      "step": 1756
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.506197130723651,
      "learning_rate": 9.9388084208058e-06,
      "loss": 0.8805,
      "step": 1757
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.293251260862998,
      "learning_rate": 9.938695700519292e-06,
      "loss": 0.9772,
      "step": 1758
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.1752799019602245,
      "learning_rate": 9.938582877148327e-06,
      "loss": 0.9273,
      "step": 1759
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.321742424173082,
      "learning_rate": 9.938469950695258e-06,
      "loss": 0.9175,
      "step": 1760
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.832853998502766,
      "learning_rate": 9.938356921162446e-06,
      "loss": 0.9648,
      "step": 1761
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.38881770548638,
      "learning_rate": 9.938243788552246e-06,
      "loss": 0.8909,
      "step": 1762
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.878616423231119,
      "learning_rate": 9.938130552867025e-06,
      "loss": 0.9131,
      "step": 1763
    },
    {
      "epoch": 0.16,
      "grad_norm": 10.222039142772411,
      "learning_rate": 9.93801721410914e-06,
      "loss": 0.9377,
      "step": 1764
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.6978143321121255,
      "learning_rate": 9.937903772280962e-06,
      "loss": 0.9057,
      "step": 1765
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.649518989820997,
      "learning_rate": 9.937790227384854e-06,
      "loss": 0.8694,
      "step": 1766
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.010692958809571,
      "learning_rate": 9.93767657942319e-06,
      "loss": 0.9183,
      "step": 1767
    },
    {
      "epoch": 0.16,
      "grad_norm": 9.173506448218816,
      "learning_rate": 9.93756282839834e-06,
      "loss": 0.8333,
      "step": 1768
    },
    {
      "epoch": 0.16,
      "grad_norm": 9.38566092052211,
      "learning_rate": 9.93744897431268e-06,
      "loss": 0.9245,
      "step": 1769
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.083496806377413,
      "learning_rate": 9.937335017168585e-06,
      "loss": 0.924,
      "step": 1770
    },
    {
      "epoch": 0.16,
      "grad_norm": 12.676180583987946,
      "learning_rate": 9.937220956968434e-06,
      "loss": 0.9565,
      "step": 1771
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.848290000227272,
      "learning_rate": 9.937106793714608e-06,
      "loss": 0.9675,
      "step": 1772
    },
    {
      "epoch": 0.16,
      "grad_norm": 9.360101372723747,
      "learning_rate": 9.93699252740949e-06,
      "loss": 0.8847,
      "step": 1773
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.219843094041831,
      "learning_rate": 9.936878158055463e-06,
      "loss": 0.924,
      "step": 1774
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.280625889415871,
      "learning_rate": 9.936763685654917e-06,
      "loss": 0.9191,
      "step": 1775
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.257874447531805,
      "learning_rate": 9.936649110210243e-06,
      "loss": 0.9347,
      "step": 1776
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.466223301226329,
      "learning_rate": 9.936534431723828e-06,
      "loss": 0.8684,
      "step": 1777
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.070416857126578,
      "learning_rate": 9.936419650198066e-06,
      "loss": 0.9398,
      "step": 1778
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.051678861781091,
      "learning_rate": 9.936304765635355e-06,
      "loss": 0.9605,
      "step": 1779
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.848459080243438,
      "learning_rate": 9.936189778038095e-06,
      "loss": 0.8847,
      "step": 1780
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.798703419517447,
      "learning_rate": 9.93607468740868e-06,
      "loss": 0.9223,
      "step": 1781
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.280889318500933,
      "learning_rate": 9.935959493749519e-06,
      "loss": 0.9165,
      "step": 1782
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.952845267276942,
      "learning_rate": 9.93584419706301e-06,
      "loss": 0.8565,
      "step": 1783
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.115539773139738,
      "learning_rate": 9.935728797351565e-06,
      "loss": 0.8972,
      "step": 1784
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.772440394887406,
      "learning_rate": 9.935613294617589e-06,
      "loss": 0.9451,
      "step": 1785
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.900180483950491,
      "learning_rate": 9.935497688863495e-06,
      "loss": 0.8711,
      "step": 1786
    },
    {
      "epoch": 0.16,
      "grad_norm": 10.058514416182714,
      "learning_rate": 9.935381980091696e-06,
      "loss": 0.8858,
      "step": 1787
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.74635600102307,
      "learning_rate": 9.935266168304605e-06,
      "loss": 0.9124,
      "step": 1788
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.543819299755475,
      "learning_rate": 9.935150253504641e-06,
      "loss": 0.9816,
      "step": 1789
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.71051336583126,
      "learning_rate": 9.935034235694224e-06,
      "loss": 0.9064,
      "step": 1790
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.157460895670773,
      "learning_rate": 9.934918114875775e-06,
      "loss": 0.9326,
      "step": 1791
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.982347172311036,
      "learning_rate": 9.934801891051718e-06,
      "loss": 0.9103,
      "step": 1792
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.761216800206074,
      "learning_rate": 9.934685564224476e-06,
      "loss": 0.9176,
      "step": 1793
    },
    {
      "epoch": 0.16,
      "grad_norm": 15.761723177451964,
      "learning_rate": 9.934569134396481e-06,
      "loss": 0.8476,
      "step": 1794
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.887121635298372,
      "learning_rate": 9.934452601570163e-06,
      "loss": 0.985,
      "step": 1795
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.551831083264154,
      "learning_rate": 9.934335965747952e-06,
      "loss": 1.0044,
      "step": 1796
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.545950066831496,
      "learning_rate": 9.934219226932284e-06,
      "loss": 0.8444,
      "step": 1797
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.984871545978071,
      "learning_rate": 9.934102385125597e-06,
      "loss": 0.9725,
      "step": 1798
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.00606220749479,
      "learning_rate": 9.933985440330326e-06,
      "loss": 0.9604,
      "step": 1799
    },
    {
      "epoch": 0.16,
      "grad_norm": 9.63832991000302,
      "learning_rate": 9.933868392548916e-06,
      "loss": 0.8603,
      "step": 1800
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.396207497244717,
      "learning_rate": 9.933751241783807e-06,
      "loss": 0.952,
      "step": 1801
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.680716023126336,
      "learning_rate": 9.933633988037446e-06,
      "loss": 0.9466,
      "step": 1802
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.573240544586446,
      "learning_rate": 9.93351663131228e-06,
      "loss": 0.8866,
      "step": 1803
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.6730128401360025,
      "learning_rate": 9.933399171610757e-06,
      "loss": 0.8836,
      "step": 1804
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.126686317194952,
      "learning_rate": 9.933281608935332e-06,
      "loss": 0.8863,
      "step": 1805
    },
    {
      "epoch": 0.16,
      "grad_norm": 10.110240002277981,
      "learning_rate": 9.933163943288456e-06,
      "loss": 1.0113,
      "step": 1806
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.273503558260192,
      "learning_rate": 9.933046174672587e-06,
      "loss": 0.9297,
      "step": 1807
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.235582858420247,
      "learning_rate": 9.932928303090182e-06,
      "loss": 0.9864,
      "step": 1808
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.816108483676732,
      "learning_rate": 9.932810328543702e-06,
      "loss": 0.8795,
      "step": 1809
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.876782758717439,
      "learning_rate": 9.932692251035609e-06,
      "loss": 0.9412,
      "step": 1810
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.404128718579789,
      "learning_rate": 9.932574070568366e-06,
      "loss": 0.9329,
      "step": 1811
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.709091974940604,
      "learning_rate": 9.932455787144444e-06,
      "loss": 0.9167,
      "step": 1812
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.364741106400842,
      "learning_rate": 9.932337400766308e-06,
      "loss": 0.8257,
      "step": 1813
    },
    {
      "epoch": 0.16,
      "grad_norm": 10.94237148338265,
      "learning_rate": 9.93221891143643e-06,
      "loss": 0.957,
      "step": 1814
    },
    {
      "epoch": 0.16,
      "grad_norm": 9.160605191826752,
      "learning_rate": 9.932100319157286e-06,
      "loss": 0.908,
      "step": 1815
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.191472244948973,
      "learning_rate": 9.931981623931346e-06,
      "loss": 0.9232,
      "step": 1816
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.550798514247578,
      "learning_rate": 9.931862825761092e-06,
      "loss": 0.9624,
      "step": 1817
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.70307094046196,
      "learning_rate": 9.931743924649e-06,
      "loss": 0.9524,
      "step": 1818
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.8865632625895925,
      "learning_rate": 9.931624920597556e-06,
      "loss": 0.9107,
      "step": 1819
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.418871189570078,
      "learning_rate": 9.93150581360924e-06,
      "loss": 0.8916,
      "step": 1820
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.514776149416491,
      "learning_rate": 9.931386603686544e-06,
      "loss": 0.9075,
      "step": 1821
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.185962051438692,
      "learning_rate": 9.931267290831948e-06,
      "loss": 0.8462,
      "step": 1822
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.341480267939516,
      "learning_rate": 9.931147875047948e-06,
      "loss": 0.8568,
      "step": 1823
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.776046214120354,
      "learning_rate": 9.931028356337035e-06,
      "loss": 0.9167,
      "step": 1824
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.943871078296796,
      "learning_rate": 9.930908734701705e-06,
      "loss": 0.8797,
      "step": 1825
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.050885725217994,
      "learning_rate": 9.930789010144453e-06,
      "loss": 0.8178,
      "step": 1826
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.788748465974695,
      "learning_rate": 9.930669182667778e-06,
      "loss": 0.8805,
      "step": 1827
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.500139943187679,
      "learning_rate": 9.930549252274182e-06,
      "loss": 0.9023,
      "step": 1828
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.285187248469305,
      "learning_rate": 9.930429218966169e-06,
      "loss": 0.8776,
      "step": 1829
    },
    {
      "epoch": 0.16,
      "grad_norm": 9.700379898311727,
      "learning_rate": 9.930309082746244e-06,
      "loss": 0.9606,
      "step": 1830
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.677150029203911,
      "learning_rate": 9.930188843616913e-06,
      "loss": 0.8879,
      "step": 1831
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.769367180082794,
      "learning_rate": 9.930068501580686e-06,
      "loss": 0.9091,
      "step": 1832
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.950325365742314,
      "learning_rate": 9.929948056640076e-06,
      "loss": 0.8786,
      "step": 1833
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.977420432526979,
      "learning_rate": 9.929827508797599e-06,
      "loss": 0.8691,
      "step": 1834
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.459009814426587,
      "learning_rate": 9.929706858055765e-06,
      "loss": 0.8706,
      "step": 1835
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.833810649229092,
      "learning_rate": 9.9295861044171e-06,
      "loss": 0.8892,
      "step": 1836
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.677945432990281,
      "learning_rate": 9.929465247884119e-06,
      "loss": 0.9004,
      "step": 1837
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.4757094775664825,
      "learning_rate": 9.929344288459346e-06,
      "loss": 0.8786,
      "step": 1838
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.759768845458339,
      "learning_rate": 9.929223226145306e-06,
      "loss": 0.8523,
      "step": 1839
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.289840448734264,
      "learning_rate": 9.929102060944525e-06,
      "loss": 0.8634,
      "step": 1840
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.4787952227336145,
      "learning_rate": 9.928980792859535e-06,
      "loss": 0.9583,
      "step": 1841
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.382608806931879,
      "learning_rate": 9.928859421892864e-06,
      "loss": 0.9281,
      "step": 1842
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.794575501427869,
      "learning_rate": 9.928737948047047e-06,
      "loss": 0.8746,
      "step": 1843
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.042826174845755,
      "learning_rate": 9.92861637132462e-06,
      "loss": 0.8881,
      "step": 1844
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.8871961000326,
      "learning_rate": 9.928494691728118e-06,
      "loss": 0.8743,
      "step": 1845
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.2265327253727065,
      "learning_rate": 9.928372909260085e-06,
      "loss": 0.9684,
      "step": 1846
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.097632110478237,
      "learning_rate": 9.928251023923059e-06,
      "loss": 0.9012,
      "step": 1847
    },
    {
      "epoch": 0.16,
      "grad_norm": 10.435580718285049,
      "learning_rate": 9.928129035719584e-06,
      "loss": 0.9562,
      "step": 1848
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.387948799425206,
      "learning_rate": 9.92800694465221e-06,
      "loss": 0.9084,
      "step": 1849
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.767053791447127,
      "learning_rate": 9.927884750723482e-06,
      "loss": 0.922,
      "step": 1850
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.744215818627964,
      "learning_rate": 9.927762453935954e-06,
      "loss": 0.9167,
      "step": 1851
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.576825531094879,
      "learning_rate": 9.927640054292173e-06,
      "loss": 0.878,
      "step": 1852
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.96017431411553,
      "learning_rate": 9.927517551794698e-06,
      "loss": 0.901,
      "step": 1853
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.011882537527502,
      "learning_rate": 9.927394946446087e-06,
      "loss": 0.899,
      "step": 1854
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.51646628272277,
      "learning_rate": 9.927272238248897e-06,
      "loss": 0.9859,
      "step": 1855
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.490810594206654,
      "learning_rate": 9.92714942720569e-06,
      "loss": 0.9102,
      "step": 1856
    },
    {
      "epoch": 0.17,
      "grad_norm": 11.666435860647674,
      "learning_rate": 9.927026513319026e-06,
      "loss": 0.9393,
      "step": 1857
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.17489488136651,
      "learning_rate": 9.926903496591476e-06,
      "loss": 0.9371,
      "step": 1858
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.111503436322434,
      "learning_rate": 9.926780377025605e-06,
      "loss": 0.907,
      "step": 1859
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.334113458524133,
      "learning_rate": 9.926657154623982e-06,
      "loss": 0.9594,
      "step": 1860
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.951764168478322,
      "learning_rate": 9.92653382938918e-06,
      "loss": 0.8548,
      "step": 1861
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.56410017346514,
      "learning_rate": 9.926410401323774e-06,
      "loss": 0.9511,
      "step": 1862
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.154232913908849,
      "learning_rate": 9.92628687043034e-06,
      "loss": 0.9197,
      "step": 1863
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.092390652436593,
      "learning_rate": 9.926163236711457e-06,
      "loss": 0.9148,
      "step": 1864
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.142839655365062,
      "learning_rate": 9.926039500169702e-06,
      "loss": 0.9152,
      "step": 1865
    },
    {
      "epoch": 0.17,
      "grad_norm": 11.439279821569576,
      "learning_rate": 9.925915660807662e-06,
      "loss": 0.8971,
      "step": 1866
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.363104876581712,
      "learning_rate": 9.925791718627919e-06,
      "loss": 0.9054,
      "step": 1867
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.101905569038976,
      "learning_rate": 9.925667673633062e-06,
      "loss": 0.9095,
      "step": 1868
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.517695386863341,
      "learning_rate": 9.925543525825678e-06,
      "loss": 0.8681,
      "step": 1869
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.528612035711372,
      "learning_rate": 9.92541927520836e-06,
      "loss": 0.9151,
      "step": 1870
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.596313011679014,
      "learning_rate": 9.925294921783703e-06,
      "loss": 0.8948,
      "step": 1871
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.986210726300345,
      "learning_rate": 9.9251704655543e-06,
      "loss": 0.8871,
      "step": 1872
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.07407802467284,
      "learning_rate": 9.925045906522748e-06,
      "loss": 0.9302,
      "step": 1873
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.450319992669869,
      "learning_rate": 9.92492124469165e-06,
      "loss": 0.9303,
      "step": 1874
    },
    {
      "epoch": 0.17,
      "grad_norm": 11.568036017642704,
      "learning_rate": 9.924796480063605e-06,
      "loss": 0.9168,
      "step": 1875
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.134079275798095,
      "learning_rate": 9.924671612641222e-06,
      "loss": 0.8691,
      "step": 1876
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.475980810739225,
      "learning_rate": 9.9245466424271e-06,
      "loss": 0.893,
      "step": 1877
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.401800382519264,
      "learning_rate": 9.924421569423852e-06,
      "loss": 0.8866,
      "step": 1878
    },
    {
      "epoch": 0.17,
      "grad_norm": 11.791024924165079,
      "learning_rate": 9.92429639363409e-06,
      "loss": 0.9232,
      "step": 1879
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.038795508962569,
      "learning_rate": 9.924171115060423e-06,
      "loss": 0.8589,
      "step": 1880
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.786422823953024,
      "learning_rate": 9.924045733705467e-06,
      "loss": 0.8787,
      "step": 1881
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.958520736653043,
      "learning_rate": 9.923920249571842e-06,
      "loss": 0.9571,
      "step": 1882
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.700412820806753,
      "learning_rate": 9.923794662662166e-06,
      "loss": 0.8854,
      "step": 1883
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.265128094951468,
      "learning_rate": 9.923668972979056e-06,
      "loss": 0.9124,
      "step": 1884
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.228660306821579,
      "learning_rate": 9.92354318052514e-06,
      "loss": 0.8981,
      "step": 1885
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.405093126120839,
      "learning_rate": 9.923417285303043e-06,
      "loss": 0.864,
      "step": 1886
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.873532244534752,
      "learning_rate": 9.923291287315392e-06,
      "loss": 0.8779,
      "step": 1887
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.178450270825086,
      "learning_rate": 9.923165186564817e-06,
      "loss": 0.8535,
      "step": 1888
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.9617608829561135,
      "learning_rate": 9.92303898305395e-06,
      "loss": 0.8498,
      "step": 1889
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.959421014875193,
      "learning_rate": 9.922912676785428e-06,
      "loss": 0.867,
      "step": 1890
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.916917540446551,
      "learning_rate": 9.922786267761881e-06,
      "loss": 0.961,
      "step": 1891
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.029072540901147,
      "learning_rate": 9.922659755985954e-06,
      "loss": 0.9609,
      "step": 1892
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.071687110048709,
      "learning_rate": 9.922533141460285e-06,
      "loss": 0.8697,
      "step": 1893
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.638039774159249,
      "learning_rate": 9.922406424187515e-06,
      "loss": 0.9471,
      "step": 1894
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.45587084533974,
      "learning_rate": 9.922279604170293e-06,
      "loss": 0.9574,
      "step": 1895
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.290059975080132,
      "learning_rate": 9.922152681411262e-06,
      "loss": 0.9148,
      "step": 1896
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.181058830881105,
      "learning_rate": 9.922025655913074e-06,
      "loss": 0.8844,
      "step": 1897
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.321380880243945,
      "learning_rate": 9.92189852767838e-06,
      "loss": 0.8734,
      "step": 1898
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.760217216419728,
      "learning_rate": 9.921771296709832e-06,
      "loss": 0.9711,
      "step": 1899
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.055354464722107,
      "learning_rate": 9.921643963010085e-06,
      "loss": 0.8882,
      "step": 1900
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.836520447813757,
      "learning_rate": 9.921516526581803e-06,
      "loss": 0.8764,
      "step": 1901
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.32581548456697,
      "learning_rate": 9.921388987427639e-06,
      "loss": 0.9239,
      "step": 1902
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.254615236010352,
      "learning_rate": 9.921261345550256e-06,
      "loss": 0.8584,
      "step": 1903
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.779936389155573,
      "learning_rate": 9.921133600952321e-06,
      "loss": 0.8893,
      "step": 1904
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.627847045567178,
      "learning_rate": 9.9210057536365e-06,
      "loss": 0.9519,
      "step": 1905
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.9604796296887566,
      "learning_rate": 9.92087780360546e-06,
      "loss": 0.8856,
      "step": 1906
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.244701130694521,
      "learning_rate": 9.920749750861874e-06,
      "loss": 0.8529,
      "step": 1907
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.491034716598624,
      "learning_rate": 9.92062159540841e-06,
      "loss": 0.9026,
      "step": 1908
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.120924476461528,
      "learning_rate": 9.920493337247749e-06,
      "loss": 0.9109,
      "step": 1909
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.43822864710141,
      "learning_rate": 9.920364976382563e-06,
      "loss": 0.8757,
      "step": 1910
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.4086210836062,
      "learning_rate": 9.920236512815536e-06,
      "loss": 0.8666,
      "step": 1911
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.518059251736505,
      "learning_rate": 9.920107946549345e-06,
      "loss": 0.8669,
      "step": 1912
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.202682029561717,
      "learning_rate": 9.919979277586677e-06,
      "loss": 0.8566,
      "step": 1913
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.6661192463369074,
      "learning_rate": 9.919850505930216e-06,
      "loss": 0.9361,
      "step": 1914
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.815692049731296,
      "learning_rate": 9.919721631582648e-06,
      "loss": 0.9425,
      "step": 1915
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.697392440671447,
      "learning_rate": 9.919592654546665e-06,
      "loss": 0.8469,
      "step": 1916
    },
    {
      "epoch": 0.17,
      "grad_norm": 10.025702154733576,
      "learning_rate": 9.919463574824962e-06,
      "loss": 0.8919,
      "step": 1917
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.783534581188596,
      "learning_rate": 9.919334392420229e-06,
      "loss": 0.9603,
      "step": 1918
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.123191100170658,
      "learning_rate": 9.919205107335163e-06,
      "loss": 0.9162,
      "step": 1919
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.040096213134516,
      "learning_rate": 9.919075719572464e-06,
      "loss": 0.941,
      "step": 1920
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.509762656935784,
      "learning_rate": 9.918946229134831e-06,
      "loss": 0.8527,
      "step": 1921
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.696737954557049,
      "learning_rate": 9.918816636024968e-06,
      "loss": 0.8829,
      "step": 1922
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.567100658400628,
      "learning_rate": 9.918686940245581e-06,
      "loss": 0.8782,
      "step": 1923
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.351512861535879,
      "learning_rate": 9.918557141799374e-06,
      "loss": 0.9349,
      "step": 1924
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.42830355070083,
      "learning_rate": 9.918427240689059e-06,
      "loss": 0.89,
      "step": 1925
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.876763625626068,
      "learning_rate": 9.918297236917346e-06,
      "loss": 0.9172,
      "step": 1926
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.597408151383452,
      "learning_rate": 9.918167130486951e-06,
      "loss": 0.9205,
      "step": 1927
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.369350234967008,
      "learning_rate": 9.918036921400587e-06,
      "loss": 0.9008,
      "step": 1928
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.173214508315758,
      "learning_rate": 9.917906609660971e-06,
      "loss": 0.8335,
      "step": 1929
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.040322187744046,
      "learning_rate": 9.917776195270825e-06,
      "loss": 0.9115,
      "step": 1930
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.740263102581305,
      "learning_rate": 9.917645678232872e-06,
      "loss": 0.8874,
      "step": 1931
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.705295795198285,
      "learning_rate": 9.917515058549833e-06,
      "loss": 0.9233,
      "step": 1932
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.773984162868728,
      "learning_rate": 9.917384336224438e-06,
      "loss": 0.9101,
      "step": 1933
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.6283725286476525,
      "learning_rate": 9.917253511259414e-06,
      "loss": 0.9035,
      "step": 1934
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.654361041352256,
      "learning_rate": 9.91712258365749e-06,
      "loss": 0.9339,
      "step": 1935
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.832164352205454,
      "learning_rate": 9.916991553421403e-06,
      "loss": 0.9133,
      "step": 1936
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.96423653404186,
      "learning_rate": 9.916860420553881e-06,
      "loss": 0.8127,
      "step": 1937
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.713232759298602,
      "learning_rate": 9.91672918505767e-06,
      "loss": 0.9149,
      "step": 1938
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.684821338614426,
      "learning_rate": 9.916597846935502e-06,
      "loss": 0.8879,
      "step": 1939
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.967809386412753,
      "learning_rate": 9.916466406190121e-06,
      "loss": 0.8744,
      "step": 1940
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.667198955023833,
      "learning_rate": 9.916334862824272e-06,
      "loss": 0.9152,
      "step": 1941
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.135679782415234,
      "learning_rate": 9.9162032168407e-06,
      "loss": 0.9429,
      "step": 1942
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.910917507211552,
      "learning_rate": 9.916071468242152e-06,
      "loss": 0.9173,
      "step": 1943
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.439059393930484,
      "learning_rate": 9.915939617031375e-06,
      "loss": 0.9142,
      "step": 1944
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.371772749151809,
      "learning_rate": 9.915807663211128e-06,
      "loss": 0.9456,
      "step": 1945
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.800079047212957,
      "learning_rate": 9.915675606784158e-06,
      "loss": 0.9046,
      "step": 1946
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.0617666201185,
      "learning_rate": 9.915543447753228e-06,
      "loss": 0.8866,
      "step": 1947
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.452949616337805,
      "learning_rate": 9.915411186121092e-06,
      "loss": 0.9017,
      "step": 1948
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.805725858739714,
      "learning_rate": 9.915278821890512e-06,
      "loss": 0.9173,
      "step": 1949
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.286654577296225,
      "learning_rate": 9.915146355064252e-06,
      "loss": 0.9681,
      "step": 1950
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.061557601425344,
      "learning_rate": 9.915013785645073e-06,
      "loss": 0.9548,
      "step": 1951
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.08752451587432,
      "learning_rate": 9.914881113635748e-06,
      "loss": 0.8549,
      "step": 1952
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.83571411592925,
      "learning_rate": 9.914748339039042e-06,
      "loss": 0.8799,
      "step": 1953
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.740089928393752,
      "learning_rate": 9.914615461857728e-06,
      "loss": 0.8745,
      "step": 1954
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.094155507273022,
      "learning_rate": 9.91448248209458e-06,
      "loss": 0.8806,
      "step": 1955
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.408256607257896,
      "learning_rate": 9.914349399752368e-06,
      "loss": 0.9337,
      "step": 1956
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.007507891567109,
      "learning_rate": 9.914216214833879e-06,
      "loss": 0.9435,
      "step": 1957
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.403628454508832,
      "learning_rate": 9.914082927341887e-06,
      "loss": 0.9746,
      "step": 1958
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.25357044320242,
      "learning_rate": 9.913949537279176e-06,
      "loss": 0.8987,
      "step": 1959
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.621670635165892,
      "learning_rate": 9.91381604464853e-06,
      "loss": 0.8934,
      "step": 1960
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.358817982254543,
      "learning_rate": 9.913682449452735e-06,
      "loss": 0.9296,
      "step": 1961
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.525071767820268,
      "learning_rate": 9.913548751694578e-06,
      "loss": 0.8718,
      "step": 1962
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.529690337827119,
      "learning_rate": 9.913414951376853e-06,
      "loss": 0.8605,
      "step": 1963
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.496647781045088,
      "learning_rate": 9.91328104850235e-06,
      "loss": 0.8859,
      "step": 1964
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.8734741299726005,
      "learning_rate": 9.913147043073867e-06,
      "loss": 0.8372,
      "step": 1965
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.153102687426184,
      "learning_rate": 9.913012935094197e-06,
      "loss": 0.9078,
      "step": 1966
    },
    {
      "epoch": 0.18,
      "grad_norm": 10.119948417401723,
      "learning_rate": 9.912878724566141e-06,
      "loss": 0.8316,
      "step": 1967
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.7170432941270795,
      "learning_rate": 9.912744411492503e-06,
      "loss": 0.9331,
      "step": 1968
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.42798279232322,
      "learning_rate": 9.912609995876084e-06,
      "loss": 0.8632,
      "step": 1969
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.21176447176481,
      "learning_rate": 9.912475477719688e-06,
      "loss": 0.873,
      "step": 1970
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.799431989132538,
      "learning_rate": 9.912340857026124e-06,
      "loss": 0.9177,
      "step": 1971
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.737126987515644,
      "learning_rate": 9.912206133798205e-06,
      "loss": 0.9201,
      "step": 1972
    },
    {
      "epoch": 0.18,
      "grad_norm": 9.166344749335119,
      "learning_rate": 9.912071308038737e-06,
      "loss": 0.9408,
      "step": 1973
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.995732501781593,
      "learning_rate": 9.911936379750541e-06,
      "loss": 0.8739,
      "step": 1974
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.2185525870880705,
      "learning_rate": 9.911801348936429e-06,
      "loss": 0.8424,
      "step": 1975
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.318702610216175,
      "learning_rate": 9.91166621559922e-06,
      "loss": 0.926,
      "step": 1976
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.235385437874636,
      "learning_rate": 9.911530979741736e-06,
      "loss": 0.8553,
      "step": 1977
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.009170187282513,
      "learning_rate": 9.9113956413668e-06,
      "loss": 0.9196,
      "step": 1978
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.239851164289183,
      "learning_rate": 9.911260200477234e-06,
      "loss": 0.9056,
      "step": 1979
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.936474483853389,
      "learning_rate": 9.911124657075866e-06,
      "loss": 0.8739,
      "step": 1980
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.723968955679734,
      "learning_rate": 9.910989011165526e-06,
      "loss": 0.8838,
      "step": 1981
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.638872907467477,
      "learning_rate": 9.910853262749048e-06,
      "loss": 0.9227,
      "step": 1982
    },
    {
      "epoch": 0.18,
      "grad_norm": 11.803980294848449,
      "learning_rate": 9.91071741182926e-06,
      "loss": 0.967,
      "step": 1983
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.463895986294813,
      "learning_rate": 9.910581458409e-06,
      "loss": 0.8865,
      "step": 1984
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.071897909744786,
      "learning_rate": 9.910445402491107e-06,
      "loss": 0.9097,
      "step": 1985
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.921383347447673,
      "learning_rate": 9.91030924407842e-06,
      "loss": 0.9381,
      "step": 1986
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.330717435511584,
      "learning_rate": 9.910172983173782e-06,
      "loss": 0.9083,
      "step": 1987
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.89734972069986,
      "learning_rate": 9.910036619780035e-06,
      "loss": 0.9549,
      "step": 1988
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.854682930859191,
      "learning_rate": 9.909900153900027e-06,
      "loss": 0.9571,
      "step": 1989
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.037619282926741,
      "learning_rate": 9.909763585536605e-06,
      "loss": 0.9013,
      "step": 1990
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.749472341718931,
      "learning_rate": 9.909626914692621e-06,
      "loss": 0.9237,
      "step": 1991
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.36227730819618,
      "learning_rate": 9.909490141370927e-06,
      "loss": 0.8414,
      "step": 1992
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.126348501517217,
      "learning_rate": 9.909353265574377e-06,
      "loss": 0.8621,
      "step": 1993
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.85785431612523,
      "learning_rate": 9.909216287305829e-06,
      "loss": 0.9117,
      "step": 1994
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.573583223064552,
      "learning_rate": 9.909079206568143e-06,
      "loss": 0.9056,
      "step": 1995
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.592444379852677,
      "learning_rate": 9.908942023364178e-06,
      "loss": 0.8578,
      "step": 1996
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.490470127359818,
      "learning_rate": 9.9088047376968e-06,
      "loss": 0.8741,
      "step": 1997
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.005209099826573,
      "learning_rate": 9.908667349568873e-06,
      "loss": 0.9036,
      "step": 1998
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.433089662182811,
      "learning_rate": 9.908529858983265e-06,
      "loss": 0.9159,
      "step": 1999
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.491019034307945,
      "learning_rate": 9.908392265942844e-06,
      "loss": 0.8495,
      "step": 2000
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.796467806634283,
      "learning_rate": 9.908254570450485e-06,
      "loss": 0.8601,
      "step": 2001
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.70479080609709,
      "learning_rate": 9.90811677250906e-06,
      "loss": 0.9967,
      "step": 2002
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.5418846906631805,
      "learning_rate": 9.907978872121447e-06,
      "loss": 0.8818,
      "step": 2003
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.573150870365698,
      "learning_rate": 9.907840869290523e-06,
      "loss": 0.9338,
      "step": 2004
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.162570029168391,
      "learning_rate": 9.907702764019169e-06,
      "loss": 0.9462,
      "step": 2005
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.200153322542364,
      "learning_rate": 9.907564556310267e-06,
      "loss": 0.9661,
      "step": 2006
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.275668764211943,
      "learning_rate": 9.907426246166703e-06,
      "loss": 0.9807,
      "step": 2007
    },
    {
      "epoch": 0.18,
      "grad_norm": 10.138663621820154,
      "learning_rate": 9.907287833591361e-06,
      "loss": 0.9129,
      "step": 2008
    },
    {
      "epoch": 0.18,
      "grad_norm": 11.888579991935643,
      "learning_rate": 9.907149318587134e-06,
      "loss": 0.9234,
      "step": 2009
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.959712115797687,
      "learning_rate": 9.907010701156912e-06,
      "loss": 0.8569,
      "step": 2010
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.450134610293375,
      "learning_rate": 9.906871981303587e-06,
      "loss": 0.8922,
      "step": 2011
    },
    {
      "epoch": 0.18,
      "grad_norm": 9.853139847198163,
      "learning_rate": 9.906733159030054e-06,
      "loss": 0.9756,
      "step": 2012
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.747838577177867,
      "learning_rate": 9.906594234339215e-06,
      "loss": 0.9275,
      "step": 2013
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.925840277332531,
      "learning_rate": 9.906455207233966e-06,
      "loss": 0.9427,
      "step": 2014
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.376629033433445,
      "learning_rate": 9.906316077717207e-06,
      "loss": 0.9468,
      "step": 2015
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.312578615230302,
      "learning_rate": 9.906176845791847e-06,
      "loss": 0.8663,
      "step": 2016
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.096714087856581,
      "learning_rate": 9.906037511460788e-06,
      "loss": 0.8374,
      "step": 2017
    },
    {
      "epoch": 0.18,
      "grad_norm": 9.1116856303445,
      "learning_rate": 9.905898074726941e-06,
      "loss": 0.9269,
      "step": 2018
    },
    {
      "epoch": 0.18,
      "grad_norm": 10.539006284849428,
      "learning_rate": 9.905758535593215e-06,
      "loss": 0.9391,
      "step": 2019
    },
    {
      "epoch": 0.18,
      "grad_norm": 9.864912662322192,
      "learning_rate": 9.905618894062524e-06,
      "loss": 0.9289,
      "step": 2020
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.126719407594258,
      "learning_rate": 9.90547915013778e-06,
      "loss": 0.9773,
      "step": 2021
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.871764505797942,
      "learning_rate": 9.905339303821902e-06,
      "loss": 0.9296,
      "step": 2022
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.55487488643203,
      "learning_rate": 9.905199355117808e-06,
      "loss": 0.9293,
      "step": 2023
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.734065897471375,
      "learning_rate": 9.905059304028421e-06,
      "loss": 0.8571,
      "step": 2024
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.752441005141632,
      "learning_rate": 9.904919150556662e-06,
      "loss": 0.877,
      "step": 2025
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.27453726764212,
      "learning_rate": 9.904778894705458e-06,
      "loss": 0.8732,
      "step": 2026
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.210247528261339,
      "learning_rate": 9.904638536477735e-06,
      "loss": 0.9052,
      "step": 2027
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.017030642411388,
      "learning_rate": 9.904498075876425e-06,
      "loss": 0.8758,
      "step": 2028
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.879683816183592,
      "learning_rate": 9.904357512904457e-06,
      "loss": 0.9831,
      "step": 2029
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.5354084234922505,
      "learning_rate": 9.904216847564767e-06,
      "loss": 0.9267,
      "step": 2030
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.199546629510897,
      "learning_rate": 9.904076079860289e-06,
      "loss": 0.9482,
      "step": 2031
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.4984542204914275,
      "learning_rate": 9.903935209793965e-06,
      "loss": 0.9528,
      "step": 2032
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.236196516210283,
      "learning_rate": 9.90379423736873e-06,
      "loss": 0.9108,
      "step": 2033
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.885510703640033,
      "learning_rate": 9.903653162587532e-06,
      "loss": 0.9105,
      "step": 2034
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.536666493085037,
      "learning_rate": 9.903511985453312e-06,
      "loss": 0.8732,
      "step": 2035
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.969015012190011,
      "learning_rate": 9.903370705969017e-06,
      "loss": 0.9261,
      "step": 2036
    },
    {
      "epoch": 0.18,
      "grad_norm": 9.47660247675216,
      "learning_rate": 9.9032293241376e-06,
      "loss": 0.9095,
      "step": 2037
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.072236113857404,
      "learning_rate": 9.903087839962004e-06,
      "loss": 0.9449,
      "step": 2038
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.991617589878235,
      "learning_rate": 9.90294625344519e-06,
      "loss": 0.9071,
      "step": 2039
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.970823421571839,
      "learning_rate": 9.90280456459011e-06,
      "loss": 0.8834,
      "step": 2040
    },
    {
      "epoch": 0.18,
      "grad_norm": 9.34619354321064,
      "learning_rate": 9.902662773399721e-06,
      "loss": 0.9302,
      "step": 2041
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.748297086636995,
      "learning_rate": 9.902520879876984e-06,
      "loss": 0.9265,
      "step": 2042
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.9719637613149334,
      "learning_rate": 9.90237888402486e-06,
      "loss": 0.926,
      "step": 2043
    },
    {
      "epoch": 0.18,
      "grad_norm": 9.0074753028014,
      "learning_rate": 9.902236785846311e-06,
      "loss": 0.9674,
      "step": 2044
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.62824528180294,
      "learning_rate": 9.902094585344307e-06,
      "loss": 0.8141,
      "step": 2045
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.425252219921024,
      "learning_rate": 9.901952282521812e-06,
      "loss": 0.8711,
      "step": 2046
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.015712142735464,
      "learning_rate": 9.901809877381798e-06,
      "loss": 0.8657,
      "step": 2047
    },
    {
      "epoch": 0.18,
      "grad_norm": 9.15660922824645,
      "learning_rate": 9.901667369927239e-06,
      "loss": 0.8774,
      "step": 2048
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.703872341778651,
      "learning_rate": 9.901524760161108e-06,
      "loss": 0.8865,
      "step": 2049
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.891815699843743,
      "learning_rate": 9.901382048086383e-06,
      "loss": 0.8812,
      "step": 2050
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.149839929928816,
      "learning_rate": 9.901239233706039e-06,
      "loss": 0.9314,
      "step": 2051
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.190037027773378,
      "learning_rate": 9.901096317023061e-06,
      "loss": 0.8652,
      "step": 2052
    },
    {
      "epoch": 0.18,
      "grad_norm": 9.121987516799603,
      "learning_rate": 9.90095329804043e-06,
      "loss": 0.897,
      "step": 2053
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.665772977942403,
      "learning_rate": 9.900810176761133e-06,
      "loss": 0.9101,
      "step": 2054
    },
    {
      "epoch": 0.18,
      "grad_norm": 9.051203965137995,
      "learning_rate": 9.900666953188155e-06,
      "loss": 0.8974,
      "step": 2055
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.436286918207099,
      "learning_rate": 9.900523627324487e-06,
      "loss": 0.8581,
      "step": 2056
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.1388914101264245,
      "learning_rate": 9.90038019917312e-06,
      "loss": 0.9535,
      "step": 2057
    },
    {
      "epoch": 0.18,
      "grad_norm": 9.609413528777043,
      "learning_rate": 9.90023666873705e-06,
      "loss": 0.9623,
      "step": 2058
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.663332567122199,
      "learning_rate": 9.900093036019267e-06,
      "loss": 0.8389,
      "step": 2059
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.133643706749837,
      "learning_rate": 9.899949301022776e-06,
      "loss": 0.861,
      "step": 2060
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.681505045115308,
      "learning_rate": 9.89980546375057e-06,
      "loss": 0.8626,
      "step": 2061
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.60572856806792,
      "learning_rate": 9.89966152420566e-06,
      "loss": 0.9483,
      "step": 2062
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.445110119790856,
      "learning_rate": 9.899517482391044e-06,
      "loss": 0.8858,
      "step": 2063
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.11196567009811,
      "learning_rate": 9.89937333830973e-06,
      "loss": 0.9117,
      "step": 2064
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.971045567996113,
      "learning_rate": 9.899229091964728e-06,
      "loss": 0.872,
      "step": 2065
    },
    {
      "epoch": 0.18,
      "grad_norm": 9.249976026800834,
      "learning_rate": 9.899084743359048e-06,
      "loss": 0.8023,
      "step": 2066
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.769827169558951,
      "learning_rate": 9.898940292495702e-06,
      "loss": 0.9284,
      "step": 2067
    },
    {
      "epoch": 0.18,
      "grad_norm": 9.054915461818512,
      "learning_rate": 9.898795739377706e-06,
      "loss": 0.9074,
      "step": 2068
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.515340806440426,
      "learning_rate": 9.898651084008077e-06,
      "loss": 0.8703,
      "step": 2069
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.118855344860218,
      "learning_rate": 9.898506326389835e-06,
      "loss": 0.9071,
      "step": 2070
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.324322347367508,
      "learning_rate": 9.898361466526e-06,
      "loss": 0.9011,
      "step": 2071
    },
    {
      "epoch": 0.18,
      "grad_norm": 9.86070396653971,
      "learning_rate": 9.898216504419597e-06,
      "loss": 0.8726,
      "step": 2072
    },
    {
      "epoch": 0.18,
      "grad_norm": 9.10279621006575,
      "learning_rate": 9.89807144007365e-06,
      "loss": 0.8909,
      "step": 2073
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.561204581920791,
      "learning_rate": 9.897926273491192e-06,
      "loss": 0.8182,
      "step": 2074
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.7401041820770455,
      "learning_rate": 9.897781004675246e-06,
      "loss": 0.9465,
      "step": 2075
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.741613920593604,
      "learning_rate": 9.897635633628848e-06,
      "loss": 0.8711,
      "step": 2076
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.626644066524354,
      "learning_rate": 9.897490160355033e-06,
      "loss": 0.8134,
      "step": 2077
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.96638494102516,
      "learning_rate": 9.897344584856836e-06,
      "loss": 0.8571,
      "step": 2078
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.19122059989182,
      "learning_rate": 9.897198907137296e-06,
      "loss": 0.8474,
      "step": 2079
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.91946256933639,
      "learning_rate": 9.897053127199451e-06,
      "loss": 0.88,
      "step": 2080
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.177540433386215,
      "learning_rate": 9.89690724504635e-06,
      "loss": 0.8844,
      "step": 2081
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.200296555131909,
      "learning_rate": 9.896761260681032e-06,
      "loss": 0.9088,
      "step": 2082
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.598899446246424,
      "learning_rate": 9.896615174106547e-06,
      "loss": 0.8758,
      "step": 2083
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.437407016073574,
      "learning_rate": 9.896468985325943e-06,
      "loss": 0.9642,
      "step": 2084
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.559428525089809,
      "learning_rate": 9.896322694342272e-06,
      "loss": 0.8794,
      "step": 2085
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.501938759191598,
      "learning_rate": 9.896176301158588e-06,
      "loss": 0.9369,
      "step": 2086
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.06374219742869,
      "learning_rate": 9.896029805777947e-06,
      "loss": 0.8526,
      "step": 2087
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.2985857634015625,
      "learning_rate": 9.895883208203405e-06,
      "loss": 0.9294,
      "step": 2088
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.847330226151448,
      "learning_rate": 9.895736508438023e-06,
      "loss": 0.9444,
      "step": 2089
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.913330447806181,
      "learning_rate": 9.895589706484862e-06,
      "loss": 0.8688,
      "step": 2090
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.5434023244214,
      "learning_rate": 9.895442802346989e-06,
      "loss": 0.9602,
      "step": 2091
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.380144999360028,
      "learning_rate": 9.895295796027467e-06,
      "loss": 0.9024,
      "step": 2092
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.828883164145811,
      "learning_rate": 9.895148687529366e-06,
      "loss": 0.987,
      "step": 2093
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.863717358786092,
      "learning_rate": 9.895001476855757e-06,
      "loss": 0.8637,
      "step": 2094
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.712884095518957,
      "learning_rate": 9.894854164009712e-06,
      "loss": 0.9229,
      "step": 2095
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.757259126770476,
      "learning_rate": 9.894706748994305e-06,
      "loss": 0.8584,
      "step": 2096
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.655174773921718,
      "learning_rate": 9.894559231812614e-06,
      "loss": 0.8641,
      "step": 2097
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.75962201639118,
      "learning_rate": 9.894411612467718e-06,
      "loss": 0.8842,
      "step": 2098
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.252388476535,
      "learning_rate": 9.894263890962698e-06,
      "loss": 0.8703,
      "step": 2099
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.600037800806456,
      "learning_rate": 9.894116067300638e-06,
      "loss": 0.8975,
      "step": 2100
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.234022941048635,
      "learning_rate": 9.893968141484622e-06,
      "loss": 0.8408,
      "step": 2101
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.368395755375345,
      "learning_rate": 9.893820113517741e-06,
      "loss": 0.9435,
      "step": 2102
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.0418403116653465,
      "learning_rate": 9.89367198340308e-06,
      "loss": 0.9464,
      "step": 2103
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.972165412324308,
      "learning_rate": 9.893523751143734e-06,
      "loss": 0.9034,
      "step": 2104
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.8138784584021685,
      "learning_rate": 9.893375416742796e-06,
      "loss": 0.9416,
      "step": 2105
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.655779242186405,
      "learning_rate": 9.893226980203364e-06,
      "loss": 0.9926,
      "step": 2106
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.290522556032731,
      "learning_rate": 9.893078441528534e-06,
      "loss": 0.8797,
      "step": 2107
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.845787361183861,
      "learning_rate": 9.892929800721406e-06,
      "loss": 0.8894,
      "step": 2108
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.462588431889658,
      "learning_rate": 9.892781057785087e-06,
      "loss": 1.0165,
      "step": 2109
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.179273219130319,
      "learning_rate": 9.892632212722675e-06,
      "loss": 0.8053,
      "step": 2110
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.470727697484738,
      "learning_rate": 9.892483265537281e-06,
      "loss": 0.9681,
      "step": 2111
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.80868888632017,
      "learning_rate": 9.892334216232014e-06,
      "loss": 0.8349,
      "step": 2112
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.694398933292079,
      "learning_rate": 9.892185064809983e-06,
      "loss": 0.8856,
      "step": 2113
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.265585010813266,
      "learning_rate": 9.892035811274303e-06,
      "loss": 0.9229,
      "step": 2114
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.466503587655123,
      "learning_rate": 9.89188645562809e-06,
      "loss": 0.9661,
      "step": 2115
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.785581204707679,
      "learning_rate": 9.891736997874459e-06,
      "loss": 0.8511,
      "step": 2116
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.443857673535124,
      "learning_rate": 9.891587438016532e-06,
      "loss": 0.8652,
      "step": 2117
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.747148904847142,
      "learning_rate": 9.891437776057429e-06,
      "loss": 0.9349,
      "step": 2118
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.870466674492571,
      "learning_rate": 9.891288012000273e-06,
      "loss": 0.837,
      "step": 2119
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.7642043858497285,
      "learning_rate": 9.891138145848194e-06,
      "loss": 0.9366,
      "step": 2120
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.867599078039895,
      "learning_rate": 9.890988177604316e-06,
      "loss": 0.8828,
      "step": 2121
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.914739682600121,
      "learning_rate": 9.890838107271772e-06,
      "loss": 0.8688,
      "step": 2122
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.049210777665678,
      "learning_rate": 9.890687934853692e-06,
      "loss": 0.8339,
      "step": 2123
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.9928549104300295,
      "learning_rate": 9.890537660353212e-06,
      "loss": 0.933,
      "step": 2124
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.844406610574125,
      "learning_rate": 9.89038728377347e-06,
      "loss": 0.9076,
      "step": 2125
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.069294992950967,
      "learning_rate": 9.8902368051176e-06,
      "loss": 0.8815,
      "step": 2126
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.686428466737412,
      "learning_rate": 9.89008622438875e-06,
      "loss": 0.9631,
      "step": 2127
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.765179628805529,
      "learning_rate": 9.889935541590057e-06,
      "loss": 0.9622,
      "step": 2128
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.096095191872828,
      "learning_rate": 9.88978475672467e-06,
      "loss": 0.8716,
      "step": 2129
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.277889721067508,
      "learning_rate": 9.889633869795732e-06,
      "loss": 0.9709,
      "step": 2130
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.485231384189294,
      "learning_rate": 9.889482880806397e-06,
      "loss": 0.883,
      "step": 2131
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.640639423598375,
      "learning_rate": 9.889331789759813e-06,
      "loss": 0.8509,
      "step": 2132
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.041268397296485,
      "learning_rate": 9.889180596659138e-06,
      "loss": 0.8243,
      "step": 2133
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.46378729526833,
      "learning_rate": 9.889029301507523e-06,
      "loss": 0.8272,
      "step": 2134
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.039869892098624,
      "learning_rate": 9.88887790430813e-06,
      "loss": 0.858,
      "step": 2135
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.744955153364569,
      "learning_rate": 9.888726405064115e-06,
      "loss": 0.9023,
      "step": 2136
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.446687239903378,
      "learning_rate": 9.888574803778643e-06,
      "loss": 1.0018,
      "step": 2137
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.802978045479446,
      "learning_rate": 9.888423100454878e-06,
      "loss": 0.8648,
      "step": 2138
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.535546873831142,
      "learning_rate": 9.888271295095986e-06,
      "loss": 0.8977,
      "step": 2139
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.64942107163137,
      "learning_rate": 9.888119387705137e-06,
      "loss": 0.8712,
      "step": 2140
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.770252942979096,
      "learning_rate": 9.8879673782855e-06,
      "loss": 0.9388,
      "step": 2141
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.400889092018879,
      "learning_rate": 9.887815266840249e-06,
      "loss": 0.8299,
      "step": 2142
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.377049640217959,
      "learning_rate": 9.887663053372556e-06,
      "loss": 0.8922,
      "step": 2143
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.544174219673894,
      "learning_rate": 9.887510737885604e-06,
      "loss": 0.9496,
      "step": 2144
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.741848711883488,
      "learning_rate": 9.887358320382565e-06,
      "loss": 0.898,
      "step": 2145
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.368122258264475,
      "learning_rate": 9.887205800866627e-06,
      "loss": 0.9581,
      "step": 2146
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.8257323440612065,
      "learning_rate": 9.88705317934097e-06,
      "loss": 0.8568,
      "step": 2147
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.082414128052759,
      "learning_rate": 9.886900455808782e-06,
      "loss": 0.8704,
      "step": 2148
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.675016533105091,
      "learning_rate": 9.886747630273245e-06,
      "loss": 0.8913,
      "step": 2149
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.537220658871868,
      "learning_rate": 9.886594702737556e-06,
      "loss": 0.9538,
      "step": 2150
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.710399039726077,
      "learning_rate": 9.886441673204903e-06,
      "loss": 0.9754,
      "step": 2151
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.489406292314961,
      "learning_rate": 9.886288541678482e-06,
      "loss": 0.8807,
      "step": 2152
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.294134460396027,
      "learning_rate": 9.886135308161488e-06,
      "loss": 0.9366,
      "step": 2153
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.406496967819656,
      "learning_rate": 9.88598197265712e-06,
      "loss": 0.8774,
      "step": 2154
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.606938911490862,
      "learning_rate": 9.885828535168578e-06,
      "loss": 0.9536,
      "step": 2155
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.277691223444727,
      "learning_rate": 9.885674995699066e-06,
      "loss": 0.8955,
      "step": 2156
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.805208067656203,
      "learning_rate": 9.885521354251789e-06,
      "loss": 0.8612,
      "step": 2157
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.625502571504582,
      "learning_rate": 9.88536761082995e-06,
      "loss": 0.9223,
      "step": 2158
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.150758315523154,
      "learning_rate": 9.885213765436763e-06,
      "loss": 0.9585,
      "step": 2159
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.3938814667532915,
      "learning_rate": 9.885059818075436e-06,
      "loss": 0.9673,
      "step": 2160
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.14903165097625,
      "learning_rate": 9.884905768749184e-06,
      "loss": 0.8109,
      "step": 2161
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.508600408396909,
      "learning_rate": 9.884751617461221e-06,
      "loss": 0.8988,
      "step": 2162
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.807297131318559,
      "learning_rate": 9.884597364214766e-06,
      "loss": 0.8734,
      "step": 2163
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.14949747545713,
      "learning_rate": 9.884443009013038e-06,
      "loss": 0.872,
      "step": 2164
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.254711629111561,
      "learning_rate": 9.88428855185926e-06,
      "loss": 0.923,
      "step": 2165
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.699474827610734,
      "learning_rate": 9.884133992756652e-06,
      "loss": 0.9219,
      "step": 2166
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.134220953608768,
      "learning_rate": 9.883979331708446e-06,
      "loss": 0.8951,
      "step": 2167
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.345800617949756,
      "learning_rate": 9.883824568717866e-06,
      "loss": 0.8778,
      "step": 2168
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.0584090883792125,
      "learning_rate": 9.883669703788143e-06,
      "loss": 0.8432,
      "step": 2169
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.028768942564992,
      "learning_rate": 9.88351473692251e-06,
      "loss": 0.8579,
      "step": 2170
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.27762074564474,
      "learning_rate": 9.883359668124204e-06,
      "loss": 0.9159,
      "step": 2171
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.448728168998883,
      "learning_rate": 9.883204497396457e-06,
      "loss": 0.9268,
      "step": 2172
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.9566970553127705,
      "learning_rate": 9.88304922474251e-06,
      "loss": 0.8428,
      "step": 2173
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.607499974286011,
      "learning_rate": 9.882893850165606e-06,
      "loss": 0.8769,
      "step": 2174
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.7372746282982545,
      "learning_rate": 9.882738373668985e-06,
      "loss": 0.8981,
      "step": 2175
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.580840921300068,
      "learning_rate": 9.882582795255893e-06,
      "loss": 0.9398,
      "step": 2176
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.062027287772669,
      "learning_rate": 9.882427114929578e-06,
      "loss": 0.9006,
      "step": 2177
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.881971178763336,
      "learning_rate": 9.882271332693289e-06,
      "loss": 0.9347,
      "step": 2178
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.547705587533227,
      "learning_rate": 9.882115448550279e-06,
      "loss": 0.9444,
      "step": 2179
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.820382408708565,
      "learning_rate": 9.8819594625038e-06,
      "loss": 0.9551,
      "step": 2180
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.484085078846652,
      "learning_rate": 9.881803374557108e-06,
      "loss": 0.9276,
      "step": 2181
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.839967762015463,
      "learning_rate": 9.881647184713463e-06,
      "loss": 0.8716,
      "step": 2182
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.631715673908413,
      "learning_rate": 9.881490892976124e-06,
      "loss": 0.9389,
      "step": 2183
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.4860422554074475,
      "learning_rate": 9.881334499348351e-06,
      "loss": 0.9412,
      "step": 2184
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.149991143304295,
      "learning_rate": 9.881178003833411e-06,
      "loss": 0.9031,
      "step": 2185
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.728348032008197,
      "learning_rate": 9.88102140643457e-06,
      "loss": 0.9259,
      "step": 2186
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.712095869398096,
      "learning_rate": 9.880864707155096e-06,
      "loss": 0.9238,
      "step": 2187
    },
    {
      "epoch": 0.2,
      "grad_norm": 10.571304371004494,
      "learning_rate": 9.88070790599826e-06,
      "loss": 0.8822,
      "step": 2188
    },
    {
      "epoch": 0.2,
      "grad_norm": 10.19300980735663,
      "learning_rate": 9.880551002967335e-06,
      "loss": 0.9791,
      "step": 2189
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.105738584884674,
      "learning_rate": 9.880393998065597e-06,
      "loss": 0.8527,
      "step": 2190
    },
    {
      "epoch": 0.2,
      "grad_norm": 9.755124440734525,
      "learning_rate": 9.880236891296322e-06,
      "loss": 0.8786,
      "step": 2191
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.571558688477606,
      "learning_rate": 9.880079682662788e-06,
      "loss": 0.8487,
      "step": 2192
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.051867144947555,
      "learning_rate": 9.87992237216828e-06,
      "loss": 0.9366,
      "step": 2193
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.027739182740559,
      "learning_rate": 9.879764959816077e-06,
      "loss": 0.8638,
      "step": 2194
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.70729429903281,
      "learning_rate": 9.87960744560947e-06,
      "loss": 0.8388,
      "step": 2195
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.847061740492873,
      "learning_rate": 9.87944982955174e-06,
      "loss": 0.8157,
      "step": 2196
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.438635538772719,
      "learning_rate": 9.879292111646183e-06,
      "loss": 0.83,
      "step": 2197
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.678539337539831,
      "learning_rate": 9.879134291896088e-06,
      "loss": 0.915,
      "step": 2198
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.218580299337293,
      "learning_rate": 9.87897637030475e-06,
      "loss": 0.8906,
      "step": 2199
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.723026803415765,
      "learning_rate": 9.878818346875466e-06,
      "loss": 0.8795,
      "step": 2200
    },
    {
      "epoch": 0.2,
      "grad_norm": 10.205048138024566,
      "learning_rate": 9.878660221611532e-06,
      "loss": 0.9256,
      "step": 2201
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.284635257029652,
      "learning_rate": 9.87850199451625e-06,
      "loss": 0.9108,
      "step": 2202
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.243702500146266,
      "learning_rate": 9.878343665592922e-06,
      "loss": 0.8748,
      "step": 2203
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.874647352544806,
      "learning_rate": 9.878185234844855e-06,
      "loss": 0.8273,
      "step": 2204
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.790231257953075,
      "learning_rate": 9.878026702275354e-06,
      "loss": 0.922,
      "step": 2205
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.985939415363083,
      "learning_rate": 9.877868067887727e-06,
      "loss": 0.8432,
      "step": 2206
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.316780932979622,
      "learning_rate": 9.877709331685288e-06,
      "loss": 0.8201,
      "step": 2207
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.478992198331348,
      "learning_rate": 9.877550493671347e-06,
      "loss": 0.9743,
      "step": 2208
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.319210115539912,
      "learning_rate": 9.877391553849223e-06,
      "loss": 0.9195,
      "step": 2209
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.523042745267375,
      "learning_rate": 9.87723251222223e-06,
      "loss": 0.8636,
      "step": 2210
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.30172163689355,
      "learning_rate": 9.87707336879369e-06,
      "loss": 0.8823,
      "step": 2211
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.4152119859920225,
      "learning_rate": 9.876914123566923e-06,
      "loss": 0.8591,
      "step": 2212
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.794097256103962,
      "learning_rate": 9.876754776545257e-06,
      "loss": 0.929,
      "step": 2213
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.194967269187306,
      "learning_rate": 9.876595327732012e-06,
      "loss": 0.9421,
      "step": 2214
    },
    {
      "epoch": 0.2,
      "grad_norm": 12.682541896666878,
      "learning_rate": 9.87643577713052e-06,
      "loss": 0.9453,
      "step": 2215
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.923221544727557,
      "learning_rate": 9.87627612474411e-06,
      "loss": 0.9197,
      "step": 2216
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.086520597046269,
      "learning_rate": 9.876116370576115e-06,
      "loss": 0.8686,
      "step": 2217
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.76674044382957,
      "learning_rate": 9.87595651462987e-06,
      "loss": 0.8878,
      "step": 2218
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.645363320544467,
      "learning_rate": 9.87579655690871e-06,
      "loss": 0.8841,
      "step": 2219
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.7774046981462766,
      "learning_rate": 9.875636497415974e-06,
      "loss": 0.9761,
      "step": 2220
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.80285561159722,
      "learning_rate": 9.875476336155005e-06,
      "loss": 0.8793,
      "step": 2221
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.378222649200347,
      "learning_rate": 9.875316073129144e-06,
      "loss": 0.8684,
      "step": 2222
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.778738359668561,
      "learning_rate": 9.875155708341738e-06,
      "loss": 0.8933,
      "step": 2223
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.962168090409637,
      "learning_rate": 9.874995241796132e-06,
      "loss": 0.8927,
      "step": 2224
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.945287664505354,
      "learning_rate": 9.874834673495675e-06,
      "loss": 0.889,
      "step": 2225
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.427987104552881,
      "learning_rate": 9.874674003443722e-06,
      "loss": 0.8549,
      "step": 2226
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.220279481633401,
      "learning_rate": 9.874513231643623e-06,
      "loss": 0.933,
      "step": 2227
    },
    {
      "epoch": 0.2,
      "grad_norm": 9.86431107617849,
      "learning_rate": 9.874352358098737e-06,
      "loss": 0.8341,
      "step": 2228
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.708657613156528,
      "learning_rate": 9.874191382812419e-06,
      "loss": 0.8966,
      "step": 2229
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.23692879973121,
      "learning_rate": 9.87403030578803e-06,
      "loss": 0.7692,
      "step": 2230
    },
    {
      "epoch": 0.2,
      "grad_norm": 9.195870018735967,
      "learning_rate": 9.873869127028932e-06,
      "loss": 0.8913,
      "step": 2231
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.880687846018852,
      "learning_rate": 9.873707846538491e-06,
      "loss": 0.8821,
      "step": 2232
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.624580216039018,
      "learning_rate": 9.87354646432007e-06,
      "loss": 0.9226,
      "step": 2233
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.443765826626277,
      "learning_rate": 9.87338498037704e-06,
      "loss": 0.9148,
      "step": 2234
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.426419461742298,
      "learning_rate": 9.873223394712773e-06,
      "loss": 0.8732,
      "step": 2235
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.3629416874023965,
      "learning_rate": 9.873061707330638e-06,
      "loss": 0.872,
      "step": 2236
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.953473113117662,
      "learning_rate": 9.872899918234014e-06,
      "loss": 0.8608,
      "step": 2237
    },
    {
      "epoch": 0.2,
      "grad_norm": 10.80696203720073,
      "learning_rate": 9.872738027426271e-06,
      "loss": 0.9278,
      "step": 2238
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.86226730184155,
      "learning_rate": 9.872576034910797e-06,
      "loss": 0.8677,
      "step": 2239
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.112985929760328,
      "learning_rate": 9.872413940690969e-06,
      "loss": 0.9453,
      "step": 2240
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.374576933699547,
      "learning_rate": 9.872251744770168e-06,
      "loss": 0.8816,
      "step": 2241
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.247891040010777,
      "learning_rate": 9.872089447151783e-06,
      "loss": 0.8528,
      "step": 2242
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.151375900206258,
      "learning_rate": 9.871927047839201e-06,
      "loss": 0.9067,
      "step": 2243
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.994878186926854,
      "learning_rate": 9.87176454683581e-06,
      "loss": 0.8545,
      "step": 2244
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.293832263666268,
      "learning_rate": 9.871601944145003e-06,
      "loss": 0.9417,
      "step": 2245
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.743929839839472,
      "learning_rate": 9.871439239770175e-06,
      "loss": 0.9278,
      "step": 2246
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.685281764469025,
      "learning_rate": 9.87127643371472e-06,
      "loss": 0.8916,
      "step": 2247
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.3551939559078345,
      "learning_rate": 9.871113525982039e-06,
      "loss": 0.9586,
      "step": 2248
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.941544694458453,
      "learning_rate": 9.87095051657553e-06,
      "loss": 0.8422,
      "step": 2249
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.824241286080517,
      "learning_rate": 9.870787405498597e-06,
      "loss": 0.8767,
      "step": 2250
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.437603587946523,
      "learning_rate": 9.870624192754643e-06,
      "loss": 0.9015,
      "step": 2251
    },
    {
      "epoch": 0.2,
      "grad_norm": 9.834935537081883,
      "learning_rate": 9.870460878347075e-06,
      "loss": 0.8989,
      "step": 2252
    },
    {
      "epoch": 0.2,
      "grad_norm": 12.089752743817384,
      "learning_rate": 9.870297462279303e-06,
      "loss": 0.8842,
      "step": 2253
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.596727508081665,
      "learning_rate": 9.870133944554736e-06,
      "loss": 0.8994,
      "step": 2254
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.238669569988212,
      "learning_rate": 9.86997032517679e-06,
      "loss": 0.9366,
      "step": 2255
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.215271913908399,
      "learning_rate": 9.869806604148877e-06,
      "loss": 0.9364,
      "step": 2256
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.839483743832036,
      "learning_rate": 9.869642781474416e-06,
      "loss": 0.8987,
      "step": 2257
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.585629507724072,
      "learning_rate": 9.869478857156826e-06,
      "loss": 0.8227,
      "step": 2258
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.445490723768435,
      "learning_rate": 9.86931483119953e-06,
      "loss": 0.9727,
      "step": 2259
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.842141317364588,
      "learning_rate": 9.869150703605951e-06,
      "loss": 0.8395,
      "step": 2260
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.565304485910724,
      "learning_rate": 9.868986474379515e-06,
      "loss": 0.853,
      "step": 2261
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.505063752897994,
      "learning_rate": 9.868822143523646e-06,
      "loss": 0.9191,
      "step": 2262
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.331840313137623,
      "learning_rate": 9.868657711041779e-06,
      "loss": 0.87,
      "step": 2263
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.739395347720917,
      "learning_rate": 9.868493176937346e-06,
      "loss": 0.87,
      "step": 2264
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.684875764632902,
      "learning_rate": 9.868328541213777e-06,
      "loss": 0.8766,
      "step": 2265
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.48164243796936,
      "learning_rate": 9.868163803874513e-06,
      "loss": 0.9624,
      "step": 2266
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.950332654875624,
      "learning_rate": 9.867998964922988e-06,
      "loss": 0.9211,
      "step": 2267
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.5554362496870295,
      "learning_rate": 9.867834024362648e-06,
      "loss": 0.8844,
      "step": 2268
    },
    {
      "epoch": 0.2,
      "grad_norm": 9.269038094449606,
      "learning_rate": 9.867668982196933e-06,
      "loss": 0.8367,
      "step": 2269
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.834102163569689,
      "learning_rate": 9.867503838429288e-06,
      "loss": 0.933,
      "step": 2270
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.591729639804242,
      "learning_rate": 9.867338593063159e-06,
      "loss": 0.8488,
      "step": 2271
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.468734554423874,
      "learning_rate": 9.867173246101998e-06,
      "loss": 0.8288,
      "step": 2272
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.488849722618598,
      "learning_rate": 9.867007797549253e-06,
      "loss": 0.8692,
      "step": 2273
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.253206815655089,
      "learning_rate": 9.866842247408378e-06,
      "loss": 1.0104,
      "step": 2274
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.901837273070414,
      "learning_rate": 9.86667659568283e-06,
      "loss": 0.9066,
      "step": 2275
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.118116528833679,
      "learning_rate": 9.866510842376067e-06,
      "loss": 0.9258,
      "step": 2276
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.073684622602918,
      "learning_rate": 9.866344987491546e-06,
      "loss": 0.8807,
      "step": 2277
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.852905260073221,
      "learning_rate": 9.866179031032732e-06,
      "loss": 0.926,
      "step": 2278
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.307703647075966,
      "learning_rate": 9.866012973003087e-06,
      "loss": 0.9008,
      "step": 2279
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.34877679475249,
      "learning_rate": 9.865846813406077e-06,
      "loss": 0.9561,
      "step": 2280
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.607567225978684,
      "learning_rate": 9.865680552245171e-06,
      "loss": 0.8679,
      "step": 2281
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.797125335745823,
      "learning_rate": 9.86551418952384e-06,
      "loss": 0.923,
      "step": 2282
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.243659696785783,
      "learning_rate": 9.865347725245553e-06,
      "loss": 0.8806,
      "step": 2283
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.088980520523614,
      "learning_rate": 9.86518115941379e-06,
      "loss": 0.889,
      "step": 2284
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.594255558188656,
      "learning_rate": 9.865014492032023e-06,
      "loss": 0.9395,
      "step": 2285
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.923838005335186,
      "learning_rate": 9.864847723103734e-06,
      "loss": 0.8839,
      "step": 2286
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.215290184471627,
      "learning_rate": 9.864680852632402e-06,
      "loss": 0.9064,
      "step": 2287
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.043336711044405,
      "learning_rate": 9.864513880621511e-06,
      "loss": 0.8744,
      "step": 2288
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.975456576005333,
      "learning_rate": 9.864346807074545e-06,
      "loss": 0.9212,
      "step": 2289
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.201425782180013,
      "learning_rate": 9.864179631994994e-06,
      "loss": 0.879,
      "step": 2290
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.805802528193533,
      "learning_rate": 9.864012355386344e-06,
      "loss": 0.8696,
      "step": 2291
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.647285538209215,
      "learning_rate": 9.863844977252088e-06,
      "loss": 0.8434,
      "step": 2292
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.497824376359516,
      "learning_rate": 9.863677497595721e-06,
      "loss": 0.8404,
      "step": 2293
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.770101581188881,
      "learning_rate": 9.863509916420737e-06,
      "loss": 0.9012,
      "step": 2294
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.858052908037306,
      "learning_rate": 9.863342233730633e-06,
      "loss": 0.8735,
      "step": 2295
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.601306347319077,
      "learning_rate": 9.863174449528911e-06,
      "loss": 0.9107,
      "step": 2296
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.515911380543793,
      "learning_rate": 9.863006563819074e-06,
      "loss": 0.9695,
      "step": 2297
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.378236506281609,
      "learning_rate": 9.862838576604623e-06,
      "loss": 0.8766,
      "step": 2298
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.2850415714147605,
      "learning_rate": 9.862670487889068e-06,
      "loss": 0.8497,
      "step": 2299
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.00133660861966,
      "learning_rate": 9.862502297675914e-06,
      "loss": 0.9888,
      "step": 2300
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.720446607959591,
      "learning_rate": 9.862334005968675e-06,
      "loss": 0.7964,
      "step": 2301
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.863406747388093,
      "learning_rate": 9.86216561277086e-06,
      "loss": 0.9389,
      "step": 2302
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.559430427816366,
      "learning_rate": 9.861997118085986e-06,
      "loss": 0.914,
      "step": 2303
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.08174428874434,
      "learning_rate": 9.861828521917571e-06,
      "loss": 0.8788,
      "step": 2304
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.517843411158368,
      "learning_rate": 9.861659824269132e-06,
      "loss": 0.898,
      "step": 2305
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.417034101187277,
      "learning_rate": 9.861491025144192e-06,
      "loss": 0.8925,
      "step": 2306
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.609235991404573,
      "learning_rate": 9.861322124546273e-06,
      "loss": 0.8763,
      "step": 2307
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.2076753982263915,
      "learning_rate": 9.861153122478899e-06,
      "loss": 0.941,
      "step": 2308
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.070982118579626,
      "learning_rate": 9.8609840189456e-06,
      "loss": 0.8778,
      "step": 2309
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.84940231916242,
      "learning_rate": 9.860814813949906e-06,
      "loss": 0.89,
      "step": 2310
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.599023406140649,
      "learning_rate": 9.860645507495348e-06,
      "loss": 0.8955,
      "step": 2311
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.44699036513979,
      "learning_rate": 9.86047609958546e-06,
      "loss": 0.8653,
      "step": 2312
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.59756759125582,
      "learning_rate": 9.860306590223777e-06,
      "loss": 0.9121,
      "step": 2313
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.290210945038987,
      "learning_rate": 9.860136979413835e-06,
      "loss": 0.9132,
      "step": 2314
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.090764215514604,
      "learning_rate": 9.85996726715918e-06,
      "loss": 0.8752,
      "step": 2315
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.090169904229091,
      "learning_rate": 9.85979745346335e-06,
      "loss": 0.8866,
      "step": 2316
    },
    {
      "epoch": 0.21,
      "grad_norm": 9.398285054298242,
      "learning_rate": 9.859627538329894e-06,
      "loss": 0.8913,
      "step": 2317
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.713513640234962,
      "learning_rate": 9.859457521762351e-06,
      "loss": 0.8637,
      "step": 2318
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.953153002490357,
      "learning_rate": 9.859287403764279e-06,
      "loss": 0.8472,
      "step": 2319
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.60179677540859,
      "learning_rate": 9.859117184339221e-06,
      "loss": 0.9034,
      "step": 2320
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.55700092714277,
      "learning_rate": 9.858946863490735e-06,
      "loss": 0.8437,
      "step": 2321
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.24452216268417,
      "learning_rate": 9.858776441222373e-06,
      "loss": 0.9415,
      "step": 2322
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.724300016411335,
      "learning_rate": 9.858605917537694e-06,
      "loss": 0.8893,
      "step": 2323
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.120602713324038,
      "learning_rate": 9.858435292440256e-06,
      "loss": 0.8686,
      "step": 2324
    },
    {
      "epoch": 0.21,
      "grad_norm": 11.871696064010134,
      "learning_rate": 9.85826456593362e-06,
      "loss": 0.84,
      "step": 2325
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.276460010330153,
      "learning_rate": 9.85809373802135e-06,
      "loss": 0.865,
      "step": 2326
    },
    {
      "epoch": 0.21,
      "grad_norm": 9.790236814581444,
      "learning_rate": 9.857922808707015e-06,
      "loss": 0.9164,
      "step": 2327
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.519435116248819,
      "learning_rate": 9.85775177799418e-06,
      "loss": 0.8836,
      "step": 2328
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.309724469387918,
      "learning_rate": 9.857580645886413e-06,
      "loss": 0.8735,
      "step": 2329
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.527244808701618,
      "learning_rate": 9.857409412387289e-06,
      "loss": 0.9415,
      "step": 2330
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.838146635962811,
      "learning_rate": 9.857238077500382e-06,
      "loss": 0.878,
      "step": 2331
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.486866181644194,
      "learning_rate": 9.857066641229265e-06,
      "loss": 0.8108,
      "step": 2332
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.013700059765665,
      "learning_rate": 9.85689510357752e-06,
      "loss": 0.8447,
      "step": 2333
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.393662713874419,
      "learning_rate": 9.856723464548726e-06,
      "loss": 0.9248,
      "step": 2334
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.297802720681074,
      "learning_rate": 9.856551724146465e-06,
      "loss": 0.8656,
      "step": 2335
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.328783292218154,
      "learning_rate": 9.856379882374325e-06,
      "loss": 0.8233,
      "step": 2336
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.248698933116842,
      "learning_rate": 9.856207939235888e-06,
      "loss": 0.842,
      "step": 2337
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.389323020738942,
      "learning_rate": 9.856035894734745e-06,
      "loss": 0.8504,
      "step": 2338
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.5183856190796545,
      "learning_rate": 9.855863748874487e-06,
      "loss": 0.8724,
      "step": 2339
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.697486409810685,
      "learning_rate": 9.85569150165871e-06,
      "loss": 0.8624,
      "step": 2340
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.588459575494302,
      "learning_rate": 9.855519153091005e-06,
      "loss": 0.8555,
      "step": 2341
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.969295164169597,
      "learning_rate": 9.85534670317497e-06,
      "loss": 0.9064,
      "step": 2342
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.585261426821832,
      "learning_rate": 9.855174151914206e-06,
      "loss": 0.9023,
      "step": 2343
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.389555040968531,
      "learning_rate": 9.855001499312315e-06,
      "loss": 0.8672,
      "step": 2344
    },
    {
      "epoch": 0.21,
      "grad_norm": 9.33887385783266,
      "learning_rate": 9.8548287453729e-06,
      "loss": 0.9304,
      "step": 2345
    },
    {
      "epoch": 0.21,
      "grad_norm": 9.596423200870415,
      "learning_rate": 9.854655890099568e-06,
      "loss": 0.8495,
      "step": 2346
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.912411501791836,
      "learning_rate": 9.854482933495923e-06,
      "loss": 0.8585,
      "step": 2347
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.482246003695389,
      "learning_rate": 9.85430987556558e-06,
      "loss": 0.9013,
      "step": 2348
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.152525189787622,
      "learning_rate": 9.854136716312148e-06,
      "loss": 0.8419,
      "step": 2349
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.588827866549158,
      "learning_rate": 9.853963455739243e-06,
      "loss": 0.9493,
      "step": 2350
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.421296248802825,
      "learning_rate": 9.853790093850481e-06,
      "loss": 0.9337,
      "step": 2351
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.967737978635837,
      "learning_rate": 9.85361663064948e-06,
      "loss": 0.9358,
      "step": 2352
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.307409848223405,
      "learning_rate": 9.853443066139862e-06,
      "loss": 0.8627,
      "step": 2353
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.620241099807115,
      "learning_rate": 9.853269400325246e-06,
      "loss": 0.9337,
      "step": 2354
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.123702760883563,
      "learning_rate": 9.853095633209265e-06,
      "loss": 0.8905,
      "step": 2355
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.7320168156037195,
      "learning_rate": 9.852921764795536e-06,
      "loss": 0.9049,
      "step": 2356
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.864477566092781,
      "learning_rate": 9.852747795087695e-06,
      "loss": 0.8907,
      "step": 2357
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.05457904005513,
      "learning_rate": 9.85257372408937e-06,
      "loss": 0.9348,
      "step": 2358
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.008355357702138,
      "learning_rate": 9.852399551804196e-06,
      "loss": 0.8791,
      "step": 2359
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.945923748720724,
      "learning_rate": 9.852225278235807e-06,
      "loss": 0.8668,
      "step": 2360
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.083010185422739,
      "learning_rate": 9.852050903387845e-06,
      "loss": 0.8906,
      "step": 2361
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.6338293926415846,
      "learning_rate": 9.851876427263943e-06,
      "loss": 0.9641,
      "step": 2362
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.476598708579082,
      "learning_rate": 9.851701849867745e-06,
      "loss": 0.8635,
      "step": 2363
    },
    {
      "epoch": 0.21,
      "grad_norm": 10.309861143195437,
      "learning_rate": 9.8515271712029e-06,
      "loss": 0.9672,
      "step": 2364
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.1919677163005,
      "learning_rate": 9.851352391273045e-06,
      "loss": 0.8999,
      "step": 2365
    },
    {
      "epoch": 0.21,
      "grad_norm": 9.406522748379095,
      "learning_rate": 9.851177510081837e-06,
      "loss": 0.8999,
      "step": 2366
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.133980021133775,
      "learning_rate": 9.85100252763292e-06,
      "loss": 0.8974,
      "step": 2367
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.707877506771987,
      "learning_rate": 9.850827443929952e-06,
      "loss": 0.836,
      "step": 2368
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.215576168305376,
      "learning_rate": 9.85065225897658e-06,
      "loss": 0.8822,
      "step": 2369
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.490967098980666,
      "learning_rate": 9.850476972776468e-06,
      "loss": 0.8416,
      "step": 2370
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.57736643156538,
      "learning_rate": 9.85030158533327e-06,
      "loss": 0.9106,
      "step": 2371
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.720586686561496,
      "learning_rate": 9.850126096650649e-06,
      "loss": 0.8353,
      "step": 2372
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.212036649452139,
      "learning_rate": 9.849950506732267e-06,
      "loss": 0.9115,
      "step": 2373
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.728007018885922,
      "learning_rate": 9.84977481558179e-06,
      "loss": 0.9988,
      "step": 2374
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.214722973818981,
      "learning_rate": 9.849599023202884e-06,
      "loss": 0.9061,
      "step": 2375
    },
    {
      "epoch": 0.21,
      "grad_norm": 10.238561102953476,
      "learning_rate": 9.849423129599218e-06,
      "loss": 0.8355,
      "step": 2376
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.0032912330738855,
      "learning_rate": 9.849247134774466e-06,
      "loss": 0.8361,
      "step": 2377
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.898534246692978,
      "learning_rate": 9.849071038732299e-06,
      "loss": 0.8906,
      "step": 2378
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.205994358332342,
      "learning_rate": 9.848894841476393e-06,
      "loss": 0.843,
      "step": 2379
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.017239443393275,
      "learning_rate": 9.848718543010426e-06,
      "loss": 0.9253,
      "step": 2380
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.9489185874866415,
      "learning_rate": 9.848542143338079e-06,
      "loss": 0.9244,
      "step": 2381
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.321721353637983,
      "learning_rate": 9.848365642463032e-06,
      "loss": 0.8595,
      "step": 2382
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.321452559370103,
      "learning_rate": 9.848189040388969e-06,
      "loss": 0.8652,
      "step": 2383
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.43385239585459,
      "learning_rate": 9.84801233711958e-06,
      "loss": 0.8651,
      "step": 2384
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.421417950961274,
      "learning_rate": 9.847835532658547e-06,
      "loss": 0.8866,
      "step": 2385
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.930191201254984,
      "learning_rate": 9.847658627009567e-06,
      "loss": 0.9224,
      "step": 2386
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.479871714192704,
      "learning_rate": 9.847481620176326e-06,
      "loss": 0.9511,
      "step": 2387
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.348485518203908,
      "learning_rate": 9.847304512162524e-06,
      "loss": 0.9154,
      "step": 2388
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.552225781575768,
      "learning_rate": 9.847127302971855e-06,
      "loss": 0.9318,
      "step": 2389
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.932647221276676,
      "learning_rate": 9.84694999260802e-06,
      "loss": 0.8355,
      "step": 2390
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.616187633565632,
      "learning_rate": 9.846772581074717e-06,
      "loss": 0.8511,
      "step": 2391
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.322110534721389,
      "learning_rate": 9.846595068375653e-06,
      "loss": 0.9187,
      "step": 2392
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.663968550547319,
      "learning_rate": 9.846417454514528e-06,
      "loss": 0.9032,
      "step": 2393
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.915759084476773,
      "learning_rate": 9.846239739495054e-06,
      "loss": 0.8325,
      "step": 2394
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.060553029584971,
      "learning_rate": 9.846061923320938e-06,
      "loss": 0.9411,
      "step": 2395
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.58227686056441,
      "learning_rate": 9.845884005995892e-06,
      "loss": 0.8313,
      "step": 2396
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.823665099351213,
      "learning_rate": 9.845705987523628e-06,
      "loss": 0.9829,
      "step": 2397
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.908340471231027,
      "learning_rate": 9.845527867907866e-06,
      "loss": 0.9222,
      "step": 2398
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.368443202961965,
      "learning_rate": 9.845349647152321e-06,
      "loss": 0.9127,
      "step": 2399
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.9412490731285645,
      "learning_rate": 9.845171325260712e-06,
      "loss": 0.8938,
      "step": 2400
    },
    {
      "epoch": 0.21,
      "grad_norm": 9.60580638752615,
      "learning_rate": 9.844992902236762e-06,
      "loss": 0.9556,
      "step": 2401
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.001069830718047,
      "learning_rate": 9.844814378084196e-06,
      "loss": 0.8143,
      "step": 2402
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.473625257422621,
      "learning_rate": 9.844635752806739e-06,
      "loss": 0.8793,
      "step": 2403
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.703567722146185,
      "learning_rate": 9.844457026408123e-06,
      "loss": 0.7995,
      "step": 2404
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.172092345863533,
      "learning_rate": 9.844278198892074e-06,
      "loss": 0.9213,
      "step": 2405
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.602670517612148,
      "learning_rate": 9.844099270262327e-06,
      "loss": 0.8602,
      "step": 2406
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.4743631989026955,
      "learning_rate": 9.843920240522616e-06,
      "loss": 0.945,
      "step": 2407
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.651098485186191,
      "learning_rate": 9.843741109676676e-06,
      "loss": 0.8964,
      "step": 2408
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.607664017096932,
      "learning_rate": 9.843561877728249e-06,
      "loss": 0.9342,
      "step": 2409
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.377297393276695,
      "learning_rate": 9.843382544681076e-06,
      "loss": 0.8614,
      "step": 2410
    },
    {
      "epoch": 0.22,
      "grad_norm": 8.419120132774292,
      "learning_rate": 9.8432031105389e-06,
      "loss": 0.9777,
      "step": 2411
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.603376204394441,
      "learning_rate": 9.843023575305464e-06,
      "loss": 0.8744,
      "step": 2412
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.411525569016264,
      "learning_rate": 9.842843938984517e-06,
      "loss": 0.9244,
      "step": 2413
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.294141609221271,
      "learning_rate": 9.842664201579807e-06,
      "loss": 0.9068,
      "step": 2414
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.864759344218264,
      "learning_rate": 9.84248436309509e-06,
      "loss": 0.8957,
      "step": 2415
    },
    {
      "epoch": 0.22,
      "grad_norm": 8.393404092070174,
      "learning_rate": 9.842304423534116e-06,
      "loss": 0.9784,
      "step": 2416
    },
    {
      "epoch": 0.22,
      "grad_norm": 8.363562667370635,
      "learning_rate": 9.84212438290064e-06,
      "loss": 0.8925,
      "step": 2417
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.286418924483991,
      "learning_rate": 9.841944241198422e-06,
      "loss": 0.8934,
      "step": 2418
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.684811563871887,
      "learning_rate": 9.841763998431223e-06,
      "loss": 0.8286,
      "step": 2419
    },
    {
      "epoch": 0.22,
      "grad_norm": 8.392467924845583,
      "learning_rate": 9.841583654602802e-06,
      "loss": 0.938,
      "step": 2420
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.797361711277601,
      "learning_rate": 9.841403209716927e-06,
      "loss": 0.9516,
      "step": 2421
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.645003866515064,
      "learning_rate": 9.84122266377736e-06,
      "loss": 0.8603,
      "step": 2422
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.857645763346228,
      "learning_rate": 9.841042016787876e-06,
      "loss": 0.9517,
      "step": 2423
    },
    {
      "epoch": 0.22,
      "grad_norm": 9.965922994671066,
      "learning_rate": 9.840861268752239e-06,
      "loss": 0.9255,
      "step": 2424
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.78194304641037,
      "learning_rate": 9.840680419674225e-06,
      "loss": 0.9601,
      "step": 2425
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.670116584123574,
      "learning_rate": 9.840499469557607e-06,
      "loss": 0.9175,
      "step": 2426
    },
    {
      "epoch": 0.22,
      "grad_norm": 11.173469189065774,
      "learning_rate": 9.840318418406163e-06,
      "loss": 0.9089,
      "step": 2427
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.975093799401943,
      "learning_rate": 9.840137266223672e-06,
      "loss": 0.8303,
      "step": 2428
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.367420553662351,
      "learning_rate": 9.839956013013918e-06,
      "loss": 0.8765,
      "step": 2429
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.31729886906235,
      "learning_rate": 9.83977465878068e-06,
      "loss": 0.8527,
      "step": 2430
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.746994675722921,
      "learning_rate": 9.839593203527744e-06,
      "loss": 0.8682,
      "step": 2431
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.6223095883821825,
      "learning_rate": 9.839411647258898e-06,
      "loss": 0.8793,
      "step": 2432
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.073531451814592,
      "learning_rate": 9.839229989977934e-06,
      "loss": 0.8131,
      "step": 2433
    },
    {
      "epoch": 0.22,
      "grad_norm": 8.331787145535653,
      "learning_rate": 9.839048231688642e-06,
      "loss": 0.8983,
      "step": 2434
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.701679796902026,
      "learning_rate": 9.838866372394813e-06,
      "loss": 0.8606,
      "step": 2435
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.6042890135484535,
      "learning_rate": 9.838684412100247e-06,
      "loss": 0.8402,
      "step": 2436
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.491210350277201,
      "learning_rate": 9.83850235080874e-06,
      "loss": 0.8741,
      "step": 2437
    },
    {
      "epoch": 0.22,
      "grad_norm": 8.451974819796494,
      "learning_rate": 9.838320188524093e-06,
      "loss": 0.8717,
      "step": 2438
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.513248423056612,
      "learning_rate": 9.838137925250107e-06,
      "loss": 0.9001,
      "step": 2439
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.707903076990253,
      "learning_rate": 9.837955560990588e-06,
      "loss": 0.9079,
      "step": 2440
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.1758720395564,
      "learning_rate": 9.837773095749342e-06,
      "loss": 0.8675,
      "step": 2441
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.115664614432461,
      "learning_rate": 9.837590529530175e-06,
      "loss": 0.903,
      "step": 2442
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.81203520424122,
      "learning_rate": 9.837407862336902e-06,
      "loss": 0.918,
      "step": 2443
    },
    {
      "epoch": 0.22,
      "grad_norm": 9.71382787349826,
      "learning_rate": 9.837225094173333e-06,
      "loss": 0.8824,
      "step": 2444
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.62204743760053,
      "learning_rate": 9.837042225043284e-06,
      "loss": 0.918,
      "step": 2445
    },
    {
      "epoch": 0.22,
      "grad_norm": 9.469075192759506,
      "learning_rate": 9.83685925495057e-06,
      "loss": 0.9045,
      "step": 2446
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.91407905973321,
      "learning_rate": 9.836676183899013e-06,
      "loss": 0.9078,
      "step": 2447
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.3392280493779385,
      "learning_rate": 9.836493011892431e-06,
      "loss": 0.9031,
      "step": 2448
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.201237849198963,
      "learning_rate": 9.83630973893465e-06,
      "loss": 0.8927,
      "step": 2449
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.563751850673228,
      "learning_rate": 9.836126365029497e-06,
      "loss": 0.874,
      "step": 2450
    },
    {
      "epoch": 0.22,
      "grad_norm": 10.97997151589386,
      "learning_rate": 9.835942890180794e-06,
      "loss": 0.9763,
      "step": 2451
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.80124493789533,
      "learning_rate": 9.835759314392375e-06,
      "loss": 0.8825,
      "step": 2452
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.357952232284233,
      "learning_rate": 9.83557563766807e-06,
      "loss": 0.861,
      "step": 2453
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.2537024621687305,
      "learning_rate": 9.835391860011714e-06,
      "loss": 0.9099,
      "step": 2454
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.17319107758288,
      "learning_rate": 9.835207981427141e-06,
      "loss": 0.9895,
      "step": 2455
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.350517457214161,
      "learning_rate": 9.835024001918191e-06,
      "loss": 0.945,
      "step": 2456
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.228178870297306,
      "learning_rate": 9.834839921488703e-06,
      "loss": 0.8597,
      "step": 2457
    },
    {
      "epoch": 0.22,
      "grad_norm": 8.357418884027526,
      "learning_rate": 9.83465574014252e-06,
      "loss": 0.9634,
      "step": 2458
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.818749598397317,
      "learning_rate": 9.834471457883488e-06,
      "loss": 0.8788,
      "step": 2459
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.446346598378385,
      "learning_rate": 9.83428707471545e-06,
      "loss": 0.8553,
      "step": 2460
    },
    {
      "epoch": 0.22,
      "grad_norm": 9.159967057807451,
      "learning_rate": 9.834102590642256e-06,
      "loss": 0.9249,
      "step": 2461
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.677027363856248,
      "learning_rate": 9.833918005667757e-06,
      "loss": 0.8261,
      "step": 2462
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.190222324147908,
      "learning_rate": 9.833733319795806e-06,
      "loss": 0.9071,
      "step": 2463
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.594179097248613,
      "learning_rate": 9.833548533030257e-06,
      "loss": 0.841,
      "step": 2464
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.6490695878304775,
      "learning_rate": 9.833363645374969e-06,
      "loss": 0.905,
      "step": 2465
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.2799054099002545,
      "learning_rate": 9.833178656833799e-06,
      "loss": 0.9011,
      "step": 2466
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.962959359357749,
      "learning_rate": 9.832993567410608e-06,
      "loss": 0.9411,
      "step": 2467
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.062598066473613,
      "learning_rate": 9.832808377109262e-06,
      "loss": 0.9132,
      "step": 2468
    },
    {
      "epoch": 0.22,
      "grad_norm": 9.334217609800442,
      "learning_rate": 9.832623085933623e-06,
      "loss": 0.8755,
      "step": 2469
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.626308702056708,
      "learning_rate": 9.832437693887561e-06,
      "loss": 0.9012,
      "step": 2470
    },
    {
      "epoch": 0.22,
      "grad_norm": 9.030800981080466,
      "learning_rate": 9.832252200974946e-06,
      "loss": 0.9034,
      "step": 2471
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.60333883872574,
      "learning_rate": 9.832066607199648e-06,
      "loss": 0.9011,
      "step": 2472
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.358043820446671,
      "learning_rate": 9.831880912565544e-06,
      "loss": 0.8921,
      "step": 2473
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.879178905798572,
      "learning_rate": 9.831695117076503e-06,
      "loss": 0.8321,
      "step": 2474
    },
    {
      "epoch": 0.22,
      "grad_norm": 8.63503572166374,
      "learning_rate": 9.831509220736412e-06,
      "loss": 0.8897,
      "step": 2475
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.991945284442245,
      "learning_rate": 9.831323223549145e-06,
      "loss": 0.8978,
      "step": 2476
    },
    {
      "epoch": 0.22,
      "grad_norm": 8.994498663596056,
      "learning_rate": 9.831137125518586e-06,
      "loss": 0.86,
      "step": 2477
    },
    {
      "epoch": 0.22,
      "grad_norm": 8.969385962828241,
      "learning_rate": 9.83095092664862e-06,
      "loss": 0.8995,
      "step": 2478
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.291466166656589,
      "learning_rate": 9.830764626943132e-06,
      "loss": 0.8826,
      "step": 2479
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.636571497043233,
      "learning_rate": 9.830578226406012e-06,
      "loss": 0.9473,
      "step": 2480
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.450213770577966,
      "learning_rate": 9.83039172504115e-06,
      "loss": 0.8736,
      "step": 2481
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.4496827588255075,
      "learning_rate": 9.83020512285244e-06,
      "loss": 0.9009,
      "step": 2482
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.5642542517204365,
      "learning_rate": 9.830018419843776e-06,
      "loss": 0.8865,
      "step": 2483
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.368464394009108,
      "learning_rate": 9.829831616019055e-06,
      "loss": 0.8515,
      "step": 2484
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.76275007670052,
      "learning_rate": 9.829644711382176e-06,
      "loss": 0.9078,
      "step": 2485
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.791807037677918,
      "learning_rate": 9.82945770593704e-06,
      "loss": 0.9295,
      "step": 2486
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.708693862665198,
      "learning_rate": 9.82927059968755e-06,
      "loss": 0.8558,
      "step": 2487
    },
    {
      "epoch": 0.22,
      "grad_norm": 9.705755020783277,
      "learning_rate": 9.829083392637614e-06,
      "loss": 0.8693,
      "step": 2488
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.692614907472564,
      "learning_rate": 9.828896084791138e-06,
      "loss": 0.8798,
      "step": 2489
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.458214927027219,
      "learning_rate": 9.828708676152029e-06,
      "loss": 0.8439,
      "step": 2490
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.627640823818022,
      "learning_rate": 9.828521166724202e-06,
      "loss": 0.8025,
      "step": 2491
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.643857323903263,
      "learning_rate": 9.82833355651157e-06,
      "loss": 0.9572,
      "step": 2492
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.81601116765085,
      "learning_rate": 9.828145845518049e-06,
      "loss": 0.9413,
      "step": 2493
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.706585895183364,
      "learning_rate": 9.827958033747558e-06,
      "loss": 0.9232,
      "step": 2494
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.75340033346612,
      "learning_rate": 9.827770121204015e-06,
      "loss": 0.8642,
      "step": 2495
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.819261753264862,
      "learning_rate": 9.827582107891343e-06,
      "loss": 0.8666,
      "step": 2496
    },
    {
      "epoch": 0.22,
      "grad_norm": 8.454850974444113,
      "learning_rate": 9.827393993813468e-06,
      "loss": 0.906,
      "step": 2497
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.222603574783349,
      "learning_rate": 9.827205778974314e-06,
      "loss": 0.922,
      "step": 2498
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.524701386095603,
      "learning_rate": 9.827017463377811e-06,
      "loss": 0.886,
      "step": 2499
    },
    {
      "epoch": 0.22,
      "grad_norm": 8.906055580114247,
      "learning_rate": 9.826829047027889e-06,
      "loss": 0.8452,
      "step": 2500
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.7685239490543205,
      "learning_rate": 9.826640529928482e-06,
      "loss": 0.8872,
      "step": 2501
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.4790939034817105,
      "learning_rate": 9.826451912083523e-06,
      "loss": 0.9619,
      "step": 2502
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.640321274877171,
      "learning_rate": 9.826263193496952e-06,
      "loss": 0.9404,
      "step": 2503
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.222953721968597,
      "learning_rate": 9.826074374172704e-06,
      "loss": 0.8954,
      "step": 2504
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.052759430257341,
      "learning_rate": 9.825885454114725e-06,
      "loss": 0.8968,
      "step": 2505
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.210615536856999,
      "learning_rate": 9.825696433326954e-06,
      "loss": 0.9112,
      "step": 2506
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.771478610895155,
      "learning_rate": 9.825507311813338e-06,
      "loss": 0.8803,
      "step": 2507
    },
    {
      "epoch": 0.22,
      "grad_norm": 8.680176138918098,
      "learning_rate": 9.825318089577825e-06,
      "loss": 0.8536,
      "step": 2508
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.723865795817744,
      "learning_rate": 9.825128766624364e-06,
      "loss": 0.8565,
      "step": 2509
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.8827621710566405,
      "learning_rate": 9.824939342956906e-06,
      "loss": 0.7971,
      "step": 2510
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.585113517542912,
      "learning_rate": 9.824749818579406e-06,
      "loss": 0.8714,
      "step": 2511
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.323047083924022,
      "learning_rate": 9.82456019349582e-06,
      "loss": 0.8322,
      "step": 2512
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.448253358987684,
      "learning_rate": 9.824370467710106e-06,
      "loss": 0.9071,
      "step": 2513
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.712828164131581,
      "learning_rate": 9.824180641226223e-06,
      "loss": 0.8673,
      "step": 2514
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.380255011357968,
      "learning_rate": 9.823990714048135e-06,
      "loss": 0.8615,
      "step": 2515
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.167816480732198,
      "learning_rate": 9.823800686179802e-06,
      "loss": 0.8581,
      "step": 2516
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.545902614413698,
      "learning_rate": 9.823610557625199e-06,
      "loss": 0.8593,
      "step": 2517
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.896918339137562,
      "learning_rate": 9.823420328388285e-06,
      "loss": 0.8843,
      "step": 2518
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.571439345385356,
      "learning_rate": 9.823229998473037e-06,
      "loss": 0.8905,
      "step": 2519
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.966851577638757,
      "learning_rate": 9.823039567883423e-06,
      "loss": 0.8712,
      "step": 2520
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.354245738473049,
      "learning_rate": 9.82284903662342e-06,
      "loss": 0.9063,
      "step": 2521
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.097093077716542,
      "learning_rate": 9.822658404697008e-06,
      "loss": 0.8213,
      "step": 2522
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.351510651658792,
      "learning_rate": 9.822467672108163e-06,
      "loss": 0.8582,
      "step": 2523
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.64901844746443,
      "learning_rate": 9.822276838860865e-06,
      "loss": 0.896,
      "step": 2524
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.832522642817914,
      "learning_rate": 9.822085904959097e-06,
      "loss": 0.9471,
      "step": 2525
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.21945147643285,
      "learning_rate": 9.821894870406849e-06,
      "loss": 0.901,
      "step": 2526
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.322050982253022,
      "learning_rate": 9.821703735208104e-06,
      "loss": 0.8344,
      "step": 2527
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.884219257854892,
      "learning_rate": 9.821512499366855e-06,
      "loss": 0.8756,
      "step": 2528
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.69454009346205,
      "learning_rate": 9.821321162887088e-06,
      "loss": 0.8584,
      "step": 2529
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.380347594797289,
      "learning_rate": 9.821129725772802e-06,
      "loss": 0.8452,
      "step": 2530
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.821666943433089,
      "learning_rate": 9.82093818802799e-06,
      "loss": 0.9732,
      "step": 2531
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.624631746290377,
      "learning_rate": 9.820746549656654e-06,
      "loss": 0.8656,
      "step": 2532
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.929477238217664,
      "learning_rate": 9.820554810662789e-06,
      "loss": 0.8849,
      "step": 2533
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.766988580938753,
      "learning_rate": 9.820362971050398e-06,
      "loss": 0.911,
      "step": 2534
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.476071076943576,
      "learning_rate": 9.820171030823489e-06,
      "loss": 0.8603,
      "step": 2535
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.794647936409016,
      "learning_rate": 9.819978989986063e-06,
      "loss": 0.8607,
      "step": 2536
    },
    {
      "epoch": 0.23,
      "grad_norm": 13.034955927417442,
      "learning_rate": 9.819786848542132e-06,
      "loss": 0.9288,
      "step": 2537
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.929182732321872,
      "learning_rate": 9.819594606495707e-06,
      "loss": 0.9203,
      "step": 2538
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.546972455933007,
      "learning_rate": 9.819402263850799e-06,
      "loss": 0.8315,
      "step": 2539
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.8591333477435885,
      "learning_rate": 9.819209820611422e-06,
      "loss": 0.8535,
      "step": 2540
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.2851734647998905,
      "learning_rate": 9.819017276781595e-06,
      "loss": 0.9678,
      "step": 2541
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.836935165628057,
      "learning_rate": 9.818824632365334e-06,
      "loss": 0.9209,
      "step": 2542
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.107025889817976,
      "learning_rate": 9.818631887366663e-06,
      "loss": 0.8734,
      "step": 2543
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.87517105528627,
      "learning_rate": 9.818439041789604e-06,
      "loss": 0.9327,
      "step": 2544
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.303846576535831,
      "learning_rate": 9.818246095638183e-06,
      "loss": 0.8514,
      "step": 2545
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.643394571077277,
      "learning_rate": 9.818053048916425e-06,
      "loss": 0.8517,
      "step": 2546
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.769305203148667,
      "learning_rate": 9.817859901628362e-06,
      "loss": 0.8678,
      "step": 2547
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.56597133725013,
      "learning_rate": 9.817666653778024e-06,
      "loss": 0.8451,
      "step": 2548
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.671769038065682,
      "learning_rate": 9.817473305369447e-06,
      "loss": 0.9007,
      "step": 2549
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.544861851518432,
      "learning_rate": 9.817279856406664e-06,
      "loss": 0.8377,
      "step": 2550
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.183065178571264,
      "learning_rate": 9.817086306893711e-06,
      "loss": 0.9428,
      "step": 2551
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.713328026630656,
      "learning_rate": 9.816892656834632e-06,
      "loss": 0.86,
      "step": 2552
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.759402543524686,
      "learning_rate": 9.816698906233469e-06,
      "loss": 0.9151,
      "step": 2553
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.369250187237468,
      "learning_rate": 9.816505055094264e-06,
      "loss": 0.8986,
      "step": 2554
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.911550838012886,
      "learning_rate": 9.816311103421064e-06,
      "loss": 0.8695,
      "step": 2555
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.33847535883093,
      "learning_rate": 9.816117051217916e-06,
      "loss": 0.9601,
      "step": 2556
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.519861875299529,
      "learning_rate": 9.815922898488873e-06,
      "loss": 0.9094,
      "step": 2557
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.616197266089012,
      "learning_rate": 9.815728645237984e-06,
      "loss": 0.9196,
      "step": 2558
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.986200245995738,
      "learning_rate": 9.815534291469308e-06,
      "loss": 0.8413,
      "step": 2559
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.231001733878972,
      "learning_rate": 9.8153398371869e-06,
      "loss": 0.9118,
      "step": 2560
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.176490814786114,
      "learning_rate": 9.815145282394818e-06,
      "loss": 0.8986,
      "step": 2561
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.599662407369803,
      "learning_rate": 9.814950627097122e-06,
      "loss": 0.9204,
      "step": 2562
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.3010769285434485,
      "learning_rate": 9.814755871297876e-06,
      "loss": 0.8472,
      "step": 2563
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.8709532051061455,
      "learning_rate": 9.814561015001147e-06,
      "loss": 0.855,
      "step": 2564
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.223104587320485,
      "learning_rate": 9.814366058211e-06,
      "loss": 0.8136,
      "step": 2565
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.046472206734374,
      "learning_rate": 9.814171000931503e-06,
      "loss": 0.8438,
      "step": 2566
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.739132864667485,
      "learning_rate": 9.813975843166733e-06,
      "loss": 0.8658,
      "step": 2567
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.488078623989257,
      "learning_rate": 9.813780584920756e-06,
      "loss": 0.8744,
      "step": 2568
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.361184877142465,
      "learning_rate": 9.813585226197653e-06,
      "loss": 0.8485,
      "step": 2569
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.46610938607044,
      "learning_rate": 9.8133897670015e-06,
      "loss": 0.9022,
      "step": 2570
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.590913314453891,
      "learning_rate": 9.813194207336376e-06,
      "loss": 0.8614,
      "step": 2571
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.177536435166868,
      "learning_rate": 9.812998547206364e-06,
      "loss": 0.8668,
      "step": 2572
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.152119205764954,
      "learning_rate": 9.812802786615547e-06,
      "loss": 0.8415,
      "step": 2573
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.212690089721803,
      "learning_rate": 9.812606925568014e-06,
      "loss": 0.8598,
      "step": 2574
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.5190849763482195,
      "learning_rate": 9.812410964067849e-06,
      "loss": 0.8702,
      "step": 2575
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.860538443925113,
      "learning_rate": 9.812214902119144e-06,
      "loss": 0.8569,
      "step": 2576
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.378638385637407,
      "learning_rate": 9.812018739725992e-06,
      "loss": 0.8168,
      "step": 2577
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.139503053603262,
      "learning_rate": 9.811822476892487e-06,
      "loss": 0.8871,
      "step": 2578
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.073651723359445,
      "learning_rate": 9.811626113622725e-06,
      "loss": 0.8613,
      "step": 2579
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.096354137737452,
      "learning_rate": 9.811429649920805e-06,
      "loss": 0.846,
      "step": 2580
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.945653633157951,
      "learning_rate": 9.811233085790828e-06,
      "loss": 0.8549,
      "step": 2581
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.730098357178783,
      "learning_rate": 9.811036421236898e-06,
      "loss": 0.8608,
      "step": 2582
    },
    {
      "epoch": 0.23,
      "grad_norm": 12.411209091442236,
      "learning_rate": 9.810839656263117e-06,
      "loss": 0.8887,
      "step": 2583
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.875537394301156,
      "learning_rate": 9.810642790873594e-06,
      "loss": 0.8624,
      "step": 2584
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.278468831963581,
      "learning_rate": 9.810445825072439e-06,
      "loss": 0.8895,
      "step": 2585
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.425862400237917,
      "learning_rate": 9.810248758863762e-06,
      "loss": 0.8461,
      "step": 2586
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.059628710133865,
      "learning_rate": 9.810051592251676e-06,
      "loss": 0.9032,
      "step": 2587
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.418639503816613,
      "learning_rate": 9.809854325240297e-06,
      "loss": 0.8826,
      "step": 2588
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.588726826963997,
      "learning_rate": 9.809656957833742e-06,
      "loss": 0.8609,
      "step": 2589
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.3305575338036295,
      "learning_rate": 9.80945949003613e-06,
      "loss": 0.8426,
      "step": 2590
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.316967537884525,
      "learning_rate": 9.809261921851585e-06,
      "loss": 0.8085,
      "step": 2591
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.232819523369361,
      "learning_rate": 9.809064253284231e-06,
      "loss": 0.8456,
      "step": 2592
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.464288723592693,
      "learning_rate": 9.80886648433819e-06,
      "loss": 0.8624,
      "step": 2593
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.258335931845298,
      "learning_rate": 9.808668615017593e-06,
      "loss": 0.8529,
      "step": 2594
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.262585936482656,
      "learning_rate": 9.808470645326568e-06,
      "loss": 0.8651,
      "step": 2595
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.832751364670795,
      "learning_rate": 9.808272575269252e-06,
      "loss": 0.9293,
      "step": 2596
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.292768028024154,
      "learning_rate": 9.808074404849774e-06,
      "loss": 0.8698,
      "step": 2597
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.3081936280850694,
      "learning_rate": 9.807876134072272e-06,
      "loss": 0.9037,
      "step": 2598
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.091178689188524,
      "learning_rate": 9.807677762940886e-06,
      "loss": 0.9077,
      "step": 2599
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.618092923512377,
      "learning_rate": 9.807479291459754e-06,
      "loss": 0.8939,
      "step": 2600
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.361214936520273,
      "learning_rate": 9.807280719633021e-06,
      "loss": 0.8382,
      "step": 2601
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.25733728611795,
      "learning_rate": 9.80708204746483e-06,
      "loss": 0.9136,
      "step": 2602
    },
    {
      "epoch": 0.23,
      "grad_norm": 9.183134623700713,
      "learning_rate": 9.806883274959329e-06,
      "loss": 0.9757,
      "step": 2603
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.569039585987264,
      "learning_rate": 9.806684402120665e-06,
      "loss": 0.8842,
      "step": 2604
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.149575956592454,
      "learning_rate": 9.806485428952993e-06,
      "loss": 0.8912,
      "step": 2605
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.59626676209339,
      "learning_rate": 9.806286355460463e-06,
      "loss": 0.878,
      "step": 2606
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.400864836100951,
      "learning_rate": 9.80608718164723e-06,
      "loss": 0.9485,
      "step": 2607
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.187529097026416,
      "learning_rate": 9.805887907517452e-06,
      "loss": 0.928,
      "step": 2608
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.908417138784767,
      "learning_rate": 9.80568853307529e-06,
      "loss": 0.8766,
      "step": 2609
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.836648206219623,
      "learning_rate": 9.805489058324904e-06,
      "loss": 1.0171,
      "step": 2610
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.341021600055184,
      "learning_rate": 9.805289483270458e-06,
      "loss": 0.9318,
      "step": 2611
    },
    {
      "epoch": 0.23,
      "grad_norm": 9.067583487890701,
      "learning_rate": 9.805089807916116e-06,
      "loss": 0.9393,
      "step": 2612
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.979457899459106,
      "learning_rate": 9.804890032266047e-06,
      "loss": 0.8221,
      "step": 2613
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.699905005815041,
      "learning_rate": 9.804690156324424e-06,
      "loss": 0.9275,
      "step": 2614
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.645119830924037,
      "learning_rate": 9.804490180095413e-06,
      "loss": 0.9306,
      "step": 2615
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.261927394111703,
      "learning_rate": 9.804290103583193e-06,
      "loss": 0.8423,
      "step": 2616
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.848621009335286,
      "learning_rate": 9.804089926791938e-06,
      "loss": 0.8989,
      "step": 2617
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.031191080038291,
      "learning_rate": 9.803889649725825e-06,
      "loss": 0.919,
      "step": 2618
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.090313306875797,
      "learning_rate": 9.803689272389037e-06,
      "loss": 0.8684,
      "step": 2619
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.451350574654388,
      "learning_rate": 9.803488794785755e-06,
      "loss": 0.8575,
      "step": 2620
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.585424331822806,
      "learning_rate": 9.803288216920166e-06,
      "loss": 0.8975,
      "step": 2621
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.30978483139431,
      "learning_rate": 9.803087538796452e-06,
      "loss": 0.9466,
      "step": 2622
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.610123823099812,
      "learning_rate": 9.802886760418804e-06,
      "loss": 0.8632,
      "step": 2623
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.624445997142006,
      "learning_rate": 9.802685881791414e-06,
      "loss": 0.8971,
      "step": 2624
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.478622766159374,
      "learning_rate": 9.802484902918474e-06,
      "loss": 0.9145,
      "step": 2625
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.989950189467666,
      "learning_rate": 9.80228382380418e-06,
      "loss": 0.8691,
      "step": 2626
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.354638344772864,
      "learning_rate": 9.802082644452725e-06,
      "loss": 0.8851,
      "step": 2627
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.861692463708176,
      "learning_rate": 9.801881364868315e-06,
      "loss": 0.9797,
      "step": 2628
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.843793005529265,
      "learning_rate": 9.801679985055145e-06,
      "loss": 0.8421,
      "step": 2629
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.470252478830073,
      "learning_rate": 9.80147850501742e-06,
      "loss": 0.8698,
      "step": 2630
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.096335299421257,
      "learning_rate": 9.801276924759348e-06,
      "loss": 0.8901,
      "step": 2631
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.255654868884059,
      "learning_rate": 9.801075244285135e-06,
      "loss": 0.793,
      "step": 2632
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.893767657643325,
      "learning_rate": 9.80087346359899e-06,
      "loss": 0.8226,
      "step": 2633
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.231606526332115,
      "learning_rate": 9.800671582705125e-06,
      "loss": 0.8213,
      "step": 2634
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.646276226033292,
      "learning_rate": 9.800469601607754e-06,
      "loss": 0.8475,
      "step": 2635
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.584779456636542,
      "learning_rate": 9.800267520311092e-06,
      "loss": 0.8084,
      "step": 2636
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.421095949206507,
      "learning_rate": 9.80006533881936e-06,
      "loss": 0.8584,
      "step": 2637
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.26842197190103,
      "learning_rate": 9.799863057136775e-06,
      "loss": 0.9493,
      "step": 2638
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.598106942769635,
      "learning_rate": 9.79966067526756e-06,
      "loss": 0.8817,
      "step": 2639
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.582426853178686,
      "learning_rate": 9.799458193215939e-06,
      "loss": 0.8687,
      "step": 2640
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.723467923519511,
      "learning_rate": 9.79925561098614e-06,
      "loss": 0.8629,
      "step": 2641
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.359909125188462,
      "learning_rate": 9.799052928582389e-06,
      "loss": 0.9083,
      "step": 2642
    },
    {
      "epoch": 0.24,
      "grad_norm": 10.572358281853177,
      "learning_rate": 9.798850146008919e-06,
      "loss": 0.9806,
      "step": 2643
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.129108012280406,
      "learning_rate": 9.798647263269961e-06,
      "loss": 0.8456,
      "step": 2644
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.745890779515605,
      "learning_rate": 9.79844428036975e-06,
      "loss": 0.8613,
      "step": 2645
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.49312904313298,
      "learning_rate": 9.798241197312523e-06,
      "loss": 0.8598,
      "step": 2646
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.9349781806210204,
      "learning_rate": 9.79803801410252e-06,
      "loss": 0.8599,
      "step": 2647
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.938414145996964,
      "learning_rate": 9.79783473074398e-06,
      "loss": 0.9016,
      "step": 2648
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.495506309611578,
      "learning_rate": 9.797631347241147e-06,
      "loss": 0.8173,
      "step": 2649
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.426418747841631,
      "learning_rate": 9.797427863598268e-06,
      "loss": 0.8708,
      "step": 2650
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.922290423735427,
      "learning_rate": 9.797224279819588e-06,
      "loss": 0.8007,
      "step": 2651
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.7449455059954175,
      "learning_rate": 9.797020595909355e-06,
      "loss": 0.8657,
      "step": 2652
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.506792026897186,
      "learning_rate": 9.796816811871823e-06,
      "loss": 0.8237,
      "step": 2653
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.266965115449891,
      "learning_rate": 9.796612927711246e-06,
      "loss": 0.8685,
      "step": 2654
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.66052021092432,
      "learning_rate": 9.796408943431878e-06,
      "loss": 0.9259,
      "step": 2655
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.451909861199742,
      "learning_rate": 9.796204859037976e-06,
      "loss": 0.9007,
      "step": 2656
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.227135945791931,
      "learning_rate": 9.796000674533804e-06,
      "loss": 0.8658,
      "step": 2657
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.8040110429027845,
      "learning_rate": 9.795796389923617e-06,
      "loss": 0.8565,
      "step": 2658
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.183202440628008,
      "learning_rate": 9.795592005211687e-06,
      "loss": 0.9607,
      "step": 2659
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.110464677544873,
      "learning_rate": 9.795387520402273e-06,
      "loss": 0.8597,
      "step": 2660
    },
    {
      "epoch": 0.24,
      "grad_norm": 9.261121158660702,
      "learning_rate": 9.795182935499648e-06,
      "loss": 0.9278,
      "step": 2661
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.268610342040886,
      "learning_rate": 9.79497825050808e-06,
      "loss": 0.8848,
      "step": 2662
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.904880851410976,
      "learning_rate": 9.794773465431842e-06,
      "loss": 0.8289,
      "step": 2663
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.189922282603601,
      "learning_rate": 9.794568580275208e-06,
      "loss": 0.8855,
      "step": 2664
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.3203383759399525,
      "learning_rate": 9.794363595042456e-06,
      "loss": 0.9353,
      "step": 2665
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.1641003428207695,
      "learning_rate": 9.794158509737862e-06,
      "loss": 0.8294,
      "step": 2666
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.88655136872046,
      "learning_rate": 9.793953324365708e-06,
      "loss": 0.8735,
      "step": 2667
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.685943365598724,
      "learning_rate": 9.793748038930277e-06,
      "loss": 0.845,
      "step": 2668
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.1844555274771835,
      "learning_rate": 9.793542653435854e-06,
      "loss": 0.8235,
      "step": 2669
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.164089872041784,
      "learning_rate": 9.793337167886727e-06,
      "loss": 0.8756,
      "step": 2670
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.748076456797995,
      "learning_rate": 9.793131582287183e-06,
      "loss": 0.8723,
      "step": 2671
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.849169786675935,
      "learning_rate": 9.792925896641513e-06,
      "loss": 0.8937,
      "step": 2672
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.199732243766565,
      "learning_rate": 9.792720110954013e-06,
      "loss": 0.8583,
      "step": 2673
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.962879215965455,
      "learning_rate": 9.792514225228974e-06,
      "loss": 0.9036,
      "step": 2674
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.166743027870251,
      "learning_rate": 9.792308239470698e-06,
      "loss": 0.8507,
      "step": 2675
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.648807479850959,
      "learning_rate": 9.792102153683482e-06,
      "loss": 0.9224,
      "step": 2676
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.716276880362775,
      "learning_rate": 9.79189596787163e-06,
      "loss": 0.8991,
      "step": 2677
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.447549417463951,
      "learning_rate": 9.79168968203944e-06,
      "loss": 0.9025,
      "step": 2678
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.975676791685224,
      "learning_rate": 9.791483296191224e-06,
      "loss": 0.8399,
      "step": 2679
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.318259501697101,
      "learning_rate": 9.791276810331287e-06,
      "loss": 0.881,
      "step": 2680
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.615465788680345,
      "learning_rate": 9.791070224463939e-06,
      "loss": 0.8976,
      "step": 2681
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.242169511125475,
      "learning_rate": 9.790863538593492e-06,
      "loss": 0.81,
      "step": 2682
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.979957733070584,
      "learning_rate": 9.79065675272426e-06,
      "loss": 0.8491,
      "step": 2683
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.007512967026943,
      "learning_rate": 9.790449866860561e-06,
      "loss": 0.8925,
      "step": 2684
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.415212261644758,
      "learning_rate": 9.790242881006713e-06,
      "loss": 0.876,
      "step": 2685
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.393105366067362,
      "learning_rate": 9.790035795167033e-06,
      "loss": 0.9045,
      "step": 2686
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.706083333124812,
      "learning_rate": 9.789828609345847e-06,
      "loss": 0.8598,
      "step": 2687
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.157613252724487,
      "learning_rate": 9.789621323547478e-06,
      "loss": 0.9179,
      "step": 2688
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.66661546407446,
      "learning_rate": 9.789413937776253e-06,
      "loss": 0.9678,
      "step": 2689
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.198605828395586,
      "learning_rate": 9.789206452036499e-06,
      "loss": 0.9159,
      "step": 2690
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.874026038134801,
      "learning_rate": 9.78899886633255e-06,
      "loss": 0.931,
      "step": 2691
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.766153563797758,
      "learning_rate": 9.788791180668738e-06,
      "loss": 0.8778,
      "step": 2692
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.5845312914360985,
      "learning_rate": 9.788583395049399e-06,
      "loss": 0.8815,
      "step": 2693
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.0238100520426086,
      "learning_rate": 9.788375509478865e-06,
      "loss": 0.8491,
      "step": 2694
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.376318353056826,
      "learning_rate": 9.78816752396148e-06,
      "loss": 0.8317,
      "step": 2695
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.786249922582333,
      "learning_rate": 9.787959438501584e-06,
      "loss": 0.8783,
      "step": 2696
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.837715439027034,
      "learning_rate": 9.787751253103518e-06,
      "loss": 0.92,
      "step": 2697
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.0183746791063415,
      "learning_rate": 9.787542967771633e-06,
      "loss": 0.9987,
      "step": 2698
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.502893365765567,
      "learning_rate": 9.78733458251027e-06,
      "loss": 0.9345,
      "step": 2699
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.862912350946509,
      "learning_rate": 9.787126097323785e-06,
      "loss": 0.851,
      "step": 2700
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.21344499833905,
      "learning_rate": 9.786917512216523e-06,
      "loss": 0.8796,
      "step": 2701
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.747656206489461,
      "learning_rate": 9.786708827192843e-06,
      "loss": 0.9877,
      "step": 2702
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.652939921240534,
      "learning_rate": 9.786500042257098e-06,
      "loss": 0.8393,
      "step": 2703
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.949785792937308,
      "learning_rate": 9.786291157413648e-06,
      "loss": 0.8889,
      "step": 2704
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.27546917660922,
      "learning_rate": 9.786082172666852e-06,
      "loss": 0.8526,
      "step": 2705
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.66508193119292,
      "learning_rate": 9.785873088021072e-06,
      "loss": 0.8913,
      "step": 2706
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.575185278857841,
      "learning_rate": 9.785663903480672e-06,
      "loss": 0.9224,
      "step": 2707
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.788937651779318,
      "learning_rate": 9.785454619050017e-06,
      "loss": 0.8804,
      "step": 2708
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.373237261785669,
      "learning_rate": 9.785245234733477e-06,
      "loss": 0.9642,
      "step": 2709
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.764861027808113,
      "learning_rate": 9.785035750535423e-06,
      "loss": 0.9576,
      "step": 2710
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.542975894528062,
      "learning_rate": 9.784826166460227e-06,
      "loss": 0.8365,
      "step": 2711
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.3224079906286725,
      "learning_rate": 9.784616482512264e-06,
      "loss": 0.8863,
      "step": 2712
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.236532906928701,
      "learning_rate": 9.784406698695908e-06,
      "loss": 0.855,
      "step": 2713
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.765668075678219,
      "learning_rate": 9.784196815015542e-06,
      "loss": 0.8568,
      "step": 2714
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.583453015972575,
      "learning_rate": 9.783986831475542e-06,
      "loss": 0.9707,
      "step": 2715
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.694189645864043,
      "learning_rate": 9.783776748080296e-06,
      "loss": 0.8477,
      "step": 2716
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.876151234173226,
      "learning_rate": 9.783566564834186e-06,
      "loss": 0.9393,
      "step": 2717
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.809707056928803,
      "learning_rate": 9.783356281741601e-06,
      "loss": 0.8605,
      "step": 2718
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.376235027172749,
      "learning_rate": 9.783145898806927e-06,
      "loss": 0.8857,
      "step": 2719
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.372740084125681,
      "learning_rate": 9.782935416034559e-06,
      "loss": 0.8796,
      "step": 2720
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.738844992370631,
      "learning_rate": 9.782724833428888e-06,
      "loss": 0.8866,
      "step": 2721
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.798585374060727,
      "learning_rate": 9.782514150994309e-06,
      "loss": 0.9193,
      "step": 2722
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.131842571200875,
      "learning_rate": 9.782303368735222e-06,
      "loss": 0.9224,
      "step": 2723
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.983329925546443,
      "learning_rate": 9.782092486656024e-06,
      "loss": 0.9616,
      "step": 2724
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.731805214562907,
      "learning_rate": 9.78188150476112e-06,
      "loss": 0.9403,
      "step": 2725
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.858053441161775,
      "learning_rate": 9.781670423054908e-06,
      "loss": 0.8966,
      "step": 2726
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.362925974161827,
      "learning_rate": 9.781459241541801e-06,
      "loss": 0.8146,
      "step": 2727
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.461919282286607,
      "learning_rate": 9.781247960226203e-06,
      "loss": 0.8671,
      "step": 2728
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.801718089682048,
      "learning_rate": 9.781036579112523e-06,
      "loss": 0.9381,
      "step": 2729
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.393150717870906,
      "learning_rate": 9.780825098205176e-06,
      "loss": 0.931,
      "step": 2730
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.602922873403676,
      "learning_rate": 9.780613517508575e-06,
      "loss": 0.8977,
      "step": 2731
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.773538450331096,
      "learning_rate": 9.780401837027137e-06,
      "loss": 0.8216,
      "step": 2732
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.229063704059266,
      "learning_rate": 9.780190056765278e-06,
      "loss": 0.8506,
      "step": 2733
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.647809568840821,
      "learning_rate": 9.77997817672742e-06,
      "loss": 0.9387,
      "step": 2734
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.868684241754241,
      "learning_rate": 9.779766196917985e-06,
      "loss": 0.8534,
      "step": 2735
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.559750276543947,
      "learning_rate": 9.7795541173414e-06,
      "loss": 0.846,
      "step": 2736
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.644125206206668,
      "learning_rate": 9.779341938002089e-06,
      "loss": 0.8436,
      "step": 2737
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.246963962046901,
      "learning_rate": 9.779129658904482e-06,
      "loss": 0.8486,
      "step": 2738
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.5637500453940145,
      "learning_rate": 9.778917280053008e-06,
      "loss": 0.8525,
      "step": 2739
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.846542709214897,
      "learning_rate": 9.778704801452103e-06,
      "loss": 0.971,
      "step": 2740
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.136444226598629,
      "learning_rate": 9.7784922231062e-06,
      "loss": 0.8691,
      "step": 2741
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.294131028760153,
      "learning_rate": 9.778279545019737e-06,
      "loss": 0.8523,
      "step": 2742
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.148059570591415,
      "learning_rate": 9.778066767197152e-06,
      "loss": 0.8954,
      "step": 2743
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.730837875508556,
      "learning_rate": 9.777853889642887e-06,
      "loss": 0.8469,
      "step": 2744
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.359756233134474,
      "learning_rate": 9.777640912361385e-06,
      "loss": 0.9201,
      "step": 2745
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.776477293125902,
      "learning_rate": 9.777427835357094e-06,
      "loss": 0.8253,
      "step": 2746
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.995346132849461,
      "learning_rate": 9.777214658634458e-06,
      "loss": 0.8858,
      "step": 2747
    },
    {
      "epoch": 0.25,
      "grad_norm": 8.574556061867451,
      "learning_rate": 9.777001382197928e-06,
      "loss": 0.8739,
      "step": 2748
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.779615281579619,
      "learning_rate": 9.776788006051954e-06,
      "loss": 0.8557,
      "step": 2749
    },
    {
      "epoch": 0.25,
      "grad_norm": 8.44790723757806,
      "learning_rate": 9.776574530200993e-06,
      "loss": 0.8574,
      "step": 2750
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.755302360797564,
      "learning_rate": 9.776360954649499e-06,
      "loss": 0.8773,
      "step": 2751
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.500506408185759,
      "learning_rate": 9.77614727940193e-06,
      "loss": 0.8729,
      "step": 2752
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.267226467781265,
      "learning_rate": 9.775933504462746e-06,
      "loss": 0.8627,
      "step": 2753
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.142928676266129,
      "learning_rate": 9.775719629836409e-06,
      "loss": 0.8825,
      "step": 2754
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.349355114397092,
      "learning_rate": 9.775505655527384e-06,
      "loss": 0.8724,
      "step": 2755
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.837113862706781,
      "learning_rate": 9.775291581540136e-06,
      "loss": 0.8468,
      "step": 2756
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.398738499103372,
      "learning_rate": 9.775077407879135e-06,
      "loss": 0.9375,
      "step": 2757
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.58356398752839,
      "learning_rate": 9.774863134548848e-06,
      "loss": 0.919,
      "step": 2758
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.298146369914531,
      "learning_rate": 9.774648761553753e-06,
      "loss": 0.8664,
      "step": 2759
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.991561455952531,
      "learning_rate": 9.77443428889832e-06,
      "loss": 0.8798,
      "step": 2760
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.468540874355288,
      "learning_rate": 9.774219716587026e-06,
      "loss": 0.7828,
      "step": 2761
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.5287219830061805,
      "learning_rate": 9.774005044624352e-06,
      "loss": 0.873,
      "step": 2762
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.0892975595159635,
      "learning_rate": 9.773790273014777e-06,
      "loss": 0.8349,
      "step": 2763
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.9526311630527875,
      "learning_rate": 9.773575401762784e-06,
      "loss": 0.904,
      "step": 2764
    },
    {
      "epoch": 0.25,
      "grad_norm": 8.322204692848214,
      "learning_rate": 9.77336043087286e-06,
      "loss": 0.9067,
      "step": 2765
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.36718290578751,
      "learning_rate": 9.77314536034949e-06,
      "loss": 0.9257,
      "step": 2766
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.317669747563812,
      "learning_rate": 9.772930190197164e-06,
      "loss": 0.9467,
      "step": 2767
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.700300870616958,
      "learning_rate": 9.772714920420372e-06,
      "loss": 0.9372,
      "step": 2768
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.5670995001177435,
      "learning_rate": 9.772499551023608e-06,
      "loss": 0.8368,
      "step": 2769
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.9836405995199256,
      "learning_rate": 9.772284082011367e-06,
      "loss": 0.7823,
      "step": 2770
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.999009212941442,
      "learning_rate": 9.772068513388148e-06,
      "loss": 0.8855,
      "step": 2771
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.099438849203958,
      "learning_rate": 9.771852845158449e-06,
      "loss": 0.8386,
      "step": 2772
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.672111585642031,
      "learning_rate": 9.771637077326772e-06,
      "loss": 0.8992,
      "step": 2773
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.791653225875142,
      "learning_rate": 9.771421209897622e-06,
      "loss": 0.8826,
      "step": 2774
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.0946047182331595,
      "learning_rate": 9.7712052428755e-06,
      "loss": 0.8019,
      "step": 2775
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.645489027390587,
      "learning_rate": 9.770989176264922e-06,
      "loss": 0.8698,
      "step": 2776
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.6865389850730255,
      "learning_rate": 9.77077301007039e-06,
      "loss": 0.786,
      "step": 2777
    },
    {
      "epoch": 0.25,
      "grad_norm": 8.493061422314831,
      "learning_rate": 9.77055674429642e-06,
      "loss": 0.8842,
      "step": 2778
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.980537048741069,
      "learning_rate": 9.770340378947526e-06,
      "loss": 0.8748,
      "step": 2779
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.925270711819601,
      "learning_rate": 9.770123914028221e-06,
      "loss": 0.844,
      "step": 2780
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.873106982715956,
      "learning_rate": 9.769907349543028e-06,
      "loss": 0.8852,
      "step": 2781
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.784975302325801,
      "learning_rate": 9.769690685496466e-06,
      "loss": 0.862,
      "step": 2782
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.599531993031499,
      "learning_rate": 9.769473921893055e-06,
      "loss": 0.9496,
      "step": 2783
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.6051755742430815,
      "learning_rate": 9.76925705873732e-06,
      "loss": 0.9054,
      "step": 2784
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.92768668168519,
      "learning_rate": 9.769040096033789e-06,
      "loss": 0.8312,
      "step": 2785
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.076069387073558,
      "learning_rate": 9.76882303378699e-06,
      "loss": 0.8656,
      "step": 2786
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.343827471498975,
      "learning_rate": 9.768605872001456e-06,
      "loss": 0.9515,
      "step": 2787
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.980252648333873,
      "learning_rate": 9.768388610681714e-06,
      "loss": 0.8992,
      "step": 2788
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.226357920448103,
      "learning_rate": 9.768171249832303e-06,
      "loss": 0.8617,
      "step": 2789
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.800945881532723,
      "learning_rate": 9.76795378945776e-06,
      "loss": 0.9122,
      "step": 2790
    },
    {
      "epoch": 0.25,
      "grad_norm": 9.905904372245397,
      "learning_rate": 9.767736229562624e-06,
      "loss": 0.9085,
      "step": 2791
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.901053884306244,
      "learning_rate": 9.767518570151435e-06,
      "loss": 0.854,
      "step": 2792
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.989533017615994,
      "learning_rate": 9.767300811228736e-06,
      "loss": 0.8354,
      "step": 2793
    },
    {
      "epoch": 0.25,
      "grad_norm": 8.453291011593855,
      "learning_rate": 9.767082952799072e-06,
      "loss": 0.8895,
      "step": 2794
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.473722835903454,
      "learning_rate": 9.766864994866993e-06,
      "loss": 0.8713,
      "step": 2795
    },
    {
      "epoch": 0.25,
      "grad_norm": 8.602226396793602,
      "learning_rate": 9.766646937437045e-06,
      "loss": 0.9775,
      "step": 2796
    },
    {
      "epoch": 0.25,
      "grad_norm": 10.045253860252014,
      "learning_rate": 9.766428780513781e-06,
      "loss": 0.9078,
      "step": 2797
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.6039178957952664,
      "learning_rate": 9.766210524101755e-06,
      "loss": 0.8625,
      "step": 2798
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.714226728403699,
      "learning_rate": 9.765992168205523e-06,
      "loss": 0.8221,
      "step": 2799
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.013580469644222,
      "learning_rate": 9.76577371282964e-06,
      "loss": 0.9081,
      "step": 2800
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.817975849668641,
      "learning_rate": 9.76555515797867e-06,
      "loss": 0.8279,
      "step": 2801
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.5432454934953865,
      "learning_rate": 9.76533650365717e-06,
      "loss": 0.8664,
      "step": 2802
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.004204562818076,
      "learning_rate": 9.765117749869709e-06,
      "loss": 0.8008,
      "step": 2803
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.932230656094097,
      "learning_rate": 9.764898896620848e-06,
      "loss": 0.8336,
      "step": 2804
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.810491546622725,
      "learning_rate": 9.764679943915159e-06,
      "loss": 0.8977,
      "step": 2805
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.735327187310551,
      "learning_rate": 9.764460891757208e-06,
      "loss": 0.8214,
      "step": 2806
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.850600389140342,
      "learning_rate": 9.764241740151573e-06,
      "loss": 0.8575,
      "step": 2807
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.586181325088457,
      "learning_rate": 9.764022489102824e-06,
      "loss": 0.918,
      "step": 2808
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.7699324345170675,
      "learning_rate": 9.76380313861554e-06,
      "loss": 0.8617,
      "step": 2809
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.944715026188692,
      "learning_rate": 9.763583688694296e-06,
      "loss": 0.8427,
      "step": 2810
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.576624296391786,
      "learning_rate": 9.763364139343674e-06,
      "loss": 0.9615,
      "step": 2811
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.969423724425987,
      "learning_rate": 9.763144490568259e-06,
      "loss": 0.8437,
      "step": 2812
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.496369947213672,
      "learning_rate": 9.762924742372633e-06,
      "loss": 0.8955,
      "step": 2813
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.933321967741968,
      "learning_rate": 9.762704894761385e-06,
      "loss": 0.8868,
      "step": 2814
    },
    {
      "epoch": 0.25,
      "grad_norm": 11.896269551557472,
      "learning_rate": 9.762484947739101e-06,
      "loss": 0.8659,
      "step": 2815
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.49834176504775,
      "learning_rate": 9.762264901310374e-06,
      "loss": 0.9035,
      "step": 2816
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.148391319083141,
      "learning_rate": 9.762044755479796e-06,
      "loss": 0.8264,
      "step": 2817
    },
    {
      "epoch": 0.25,
      "grad_norm": 8.238339876336386,
      "learning_rate": 9.761824510251962e-06,
      "loss": 0.9057,
      "step": 2818
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.970052578710862,
      "learning_rate": 9.761604165631471e-06,
      "loss": 0.9477,
      "step": 2819
    },
    {
      "epoch": 0.25,
      "grad_norm": 8.687872467926363,
      "learning_rate": 9.761383721622919e-06,
      "loss": 0.9206,
      "step": 2820
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.915848235167734,
      "learning_rate": 9.76116317823091e-06,
      "loss": 0.9,
      "step": 2821
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.068579697066319,
      "learning_rate": 9.760942535460046e-06,
      "loss": 0.9263,
      "step": 2822
    },
    {
      "epoch": 0.25,
      "grad_norm": 8.2731666311295,
      "learning_rate": 9.760721793314933e-06,
      "loss": 0.9102,
      "step": 2823
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.035729977822004,
      "learning_rate": 9.76050095180018e-06,
      "loss": 0.8776,
      "step": 2824
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.501370770699194,
      "learning_rate": 9.760280010920393e-06,
      "loss": 0.9212,
      "step": 2825
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.379689991471207,
      "learning_rate": 9.760058970680186e-06,
      "loss": 0.8326,
      "step": 2826
    },
    {
      "epoch": 0.25,
      "grad_norm": 9.397104606522134,
      "learning_rate": 9.759837831084173e-06,
      "loss": 0.913,
      "step": 2827
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.842582420718361,
      "learning_rate": 9.759616592136968e-06,
      "loss": 0.8716,
      "step": 2828
    },
    {
      "epoch": 0.25,
      "grad_norm": 8.322911114899028,
      "learning_rate": 9.75939525384319e-06,
      "loss": 0.8255,
      "step": 2829
    },
    {
      "epoch": 0.25,
      "grad_norm": 9.106312041272632,
      "learning_rate": 9.75917381620746e-06,
      "loss": 0.9267,
      "step": 2830
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.59906955614898,
      "learning_rate": 9.758952279234399e-06,
      "loss": 0.8449,
      "step": 2831
    },
    {
      "epoch": 0.25,
      "grad_norm": 8.455592523900059,
      "learning_rate": 9.758730642928631e-06,
      "loss": 0.8447,
      "step": 2832
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.020823342756915,
      "learning_rate": 9.758508907294781e-06,
      "loss": 0.862,
      "step": 2833
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.433792988007039,
      "learning_rate": 9.75828707233748e-06,
      "loss": 0.9121,
      "step": 2834
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.05886167094975,
      "learning_rate": 9.758065138061356e-06,
      "loss": 0.9358,
      "step": 2835
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.188982499603427,
      "learning_rate": 9.757843104471042e-06,
      "loss": 0.8718,
      "step": 2836
    },
    {
      "epoch": 0.25,
      "grad_norm": 8.60841006301467,
      "learning_rate": 9.757620971571174e-06,
      "loss": 0.8494,
      "step": 2837
    },
    {
      "epoch": 0.25,
      "grad_norm": 10.83398398502046,
      "learning_rate": 9.757398739366386e-06,
      "loss": 0.9197,
      "step": 2838
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.808321032052221,
      "learning_rate": 9.757176407861319e-06,
      "loss": 0.8221,
      "step": 2839
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.773297689805632,
      "learning_rate": 9.756953977060612e-06,
      "loss": 0.9085,
      "step": 2840
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.069904897518965,
      "learning_rate": 9.756731446968907e-06,
      "loss": 0.9257,
      "step": 2841
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.965376229778286,
      "learning_rate": 9.756508817590854e-06,
      "loss": 0.828,
      "step": 2842
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.257811111312696,
      "learning_rate": 9.756286088931092e-06,
      "loss": 0.8145,
      "step": 2843
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.643831391214958,
      "learning_rate": 9.756063260994275e-06,
      "loss": 0.8877,
      "step": 2844
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.087507987376631,
      "learning_rate": 9.755840333785054e-06,
      "loss": 0.8683,
      "step": 2845
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.26934372742727,
      "learning_rate": 9.75561730730808e-06,
      "loss": 0.8762,
      "step": 2846
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.7785316511649185,
      "learning_rate": 9.755394181568009e-06,
      "loss": 0.9274,
      "step": 2847
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.67807526834406,
      "learning_rate": 9.7551709565695e-06,
      "loss": 0.9736,
      "step": 2848
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.681002116778516,
      "learning_rate": 9.75494763231721e-06,
      "loss": 0.8239,
      "step": 2849
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.855690226134151,
      "learning_rate": 9.754724208815803e-06,
      "loss": 0.8363,
      "step": 2850
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.74102483362881,
      "learning_rate": 9.75450068606994e-06,
      "loss": 0.9335,
      "step": 2851
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.670280530294297,
      "learning_rate": 9.754277064084285e-06,
      "loss": 0.8493,
      "step": 2852
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.718225502667886,
      "learning_rate": 9.75405334286351e-06,
      "loss": 0.8894,
      "step": 2853
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.074470864918526,
      "learning_rate": 9.753829522412282e-06,
      "loss": 0.8212,
      "step": 2854
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.383515466506689,
      "learning_rate": 9.753605602735274e-06,
      "loss": 0.8817,
      "step": 2855
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.395338639125125,
      "learning_rate": 9.75338158383716e-06,
      "loss": 0.9058,
      "step": 2856
    },
    {
      "epoch": 0.25,
      "grad_norm": 9.420861052455948,
      "learning_rate": 9.753157465722613e-06,
      "loss": 0.9086,
      "step": 2857
    },
    {
      "epoch": 0.25,
      "grad_norm": 8.38096713872705,
      "learning_rate": 9.752933248396315e-06,
      "loss": 0.8873,
      "step": 2858
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.417842873070168,
      "learning_rate": 9.752708931862944e-06,
      "loss": 0.8344,
      "step": 2859
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.369877953894997,
      "learning_rate": 9.752484516127182e-06,
      "loss": 0.9023,
      "step": 2860
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.470930717282282,
      "learning_rate": 9.752260001193714e-06,
      "loss": 0.889,
      "step": 2861
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.624514304144448,
      "learning_rate": 9.752035387067224e-06,
      "loss": 0.8687,
      "step": 2862
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.697291337858547,
      "learning_rate": 9.751810673752403e-06,
      "loss": 0.8582,
      "step": 2863
    },
    {
      "epoch": 0.26,
      "grad_norm": 8.198871372912354,
      "learning_rate": 9.75158586125394e-06,
      "loss": 0.8397,
      "step": 2864
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.1966499117576745,
      "learning_rate": 9.75136094957653e-06,
      "loss": 0.8914,
      "step": 2865
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.076035933448314,
      "learning_rate": 9.751135938724863e-06,
      "loss": 0.8432,
      "step": 2866
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.7321615396185805,
      "learning_rate": 9.75091082870364e-06,
      "loss": 0.8619,
      "step": 2867
    },
    {
      "epoch": 0.26,
      "grad_norm": 10.236162654072077,
      "learning_rate": 9.750685619517557e-06,
      "loss": 0.8831,
      "step": 2868
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.0368359989912825,
      "learning_rate": 9.750460311171316e-06,
      "loss": 0.8882,
      "step": 2869
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.396985931217516,
      "learning_rate": 9.750234903669619e-06,
      "loss": 0.9123,
      "step": 2870
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.263403710716351,
      "learning_rate": 9.750009397017171e-06,
      "loss": 0.8641,
      "step": 2871
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.314063925399342,
      "learning_rate": 9.749783791218678e-06,
      "loss": 0.8502,
      "step": 2872
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.322122493961188,
      "learning_rate": 9.749558086278851e-06,
      "loss": 0.9168,
      "step": 2873
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.62407953546833,
      "learning_rate": 9.749332282202402e-06,
      "loss": 0.8678,
      "step": 2874
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.0998585526245375,
      "learning_rate": 9.749106378994041e-06,
      "loss": 0.8696,
      "step": 2875
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.4229610528251255,
      "learning_rate": 9.748880376658485e-06,
      "loss": 0.8164,
      "step": 2876
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.381250291008569,
      "learning_rate": 9.74865427520045e-06,
      "loss": 0.9883,
      "step": 2877
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.363132886885148,
      "learning_rate": 9.748428074624658e-06,
      "loss": 0.8567,
      "step": 2878
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.620487746608921,
      "learning_rate": 9.748201774935827e-06,
      "loss": 0.8501,
      "step": 2879
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.827970743734044,
      "learning_rate": 9.747975376138686e-06,
      "loss": 0.8584,
      "step": 2880
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.523694046270023,
      "learning_rate": 9.747748878237952e-06,
      "loss": 0.867,
      "step": 2881
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.3878216838003175,
      "learning_rate": 9.747522281238362e-06,
      "loss": 0.8426,
      "step": 2882
    },
    {
      "epoch": 0.26,
      "grad_norm": 8.137851769640685,
      "learning_rate": 9.747295585144638e-06,
      "loss": 0.868,
      "step": 2883
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.343942854787575,
      "learning_rate": 9.747068789961517e-06,
      "loss": 0.8851,
      "step": 2884
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.0798125758845565,
      "learning_rate": 9.74684189569373e-06,
      "loss": 0.9381,
      "step": 2885
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.14861414292764,
      "learning_rate": 9.746614902346014e-06,
      "loss": 0.9195,
      "step": 2886
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.888916749207589,
      "learning_rate": 9.746387809923107e-06,
      "loss": 0.7806,
      "step": 2887
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.851349533372259,
      "learning_rate": 9.746160618429748e-06,
      "loss": 0.8738,
      "step": 2888
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.264630920492703,
      "learning_rate": 9.745933327870682e-06,
      "loss": 0.8982,
      "step": 2889
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.7055387728737506,
      "learning_rate": 9.74570593825065e-06,
      "loss": 0.8632,
      "step": 2890
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.932900363596224,
      "learning_rate": 9.745478449574399e-06,
      "loss": 0.8282,
      "step": 2891
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.0599422807706285,
      "learning_rate": 9.745250861846677e-06,
      "loss": 0.8605,
      "step": 2892
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.405757695219403,
      "learning_rate": 9.745023175072237e-06,
      "loss": 0.8511,
      "step": 2893
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.610856757394888,
      "learning_rate": 9.744795389255829e-06,
      "loss": 0.8605,
      "step": 2894
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.941718061509259,
      "learning_rate": 9.744567504402207e-06,
      "loss": 0.9416,
      "step": 2895
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.955518524405605,
      "learning_rate": 9.744339520516131e-06,
      "loss": 0.926,
      "step": 2896
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.77032982974119,
      "learning_rate": 9.744111437602358e-06,
      "loss": 0.896,
      "step": 2897
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.082696542555093,
      "learning_rate": 9.743883255665645e-06,
      "loss": 0.8159,
      "step": 2898
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.779114541081023,
      "learning_rate": 9.743654974710759e-06,
      "loss": 0.9015,
      "step": 2899
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.983439664921714,
      "learning_rate": 9.743426594742464e-06,
      "loss": 0.7906,
      "step": 2900
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.193667111972571,
      "learning_rate": 9.743198115765529e-06,
      "loss": 0.8952,
      "step": 2901
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.141342924566357,
      "learning_rate": 9.742969537784718e-06,
      "loss": 0.8941,
      "step": 2902
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.628371712677069,
      "learning_rate": 9.742740860804805e-06,
      "loss": 0.833,
      "step": 2903
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.388684274794961,
      "learning_rate": 9.742512084830565e-06,
      "loss": 0.8006,
      "step": 2904
    },
    {
      "epoch": 0.26,
      "grad_norm": 8.672479523925197,
      "learning_rate": 9.74228320986677e-06,
      "loss": 0.8271,
      "step": 2905
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.091641752497882,
      "learning_rate": 9.742054235918197e-06,
      "loss": 0.8692,
      "step": 2906
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.1540331560805335,
      "learning_rate": 9.74182516298963e-06,
      "loss": 0.9671,
      "step": 2907
    },
    {
      "epoch": 0.26,
      "grad_norm": 8.056251956754588,
      "learning_rate": 9.741595991085843e-06,
      "loss": 0.9516,
      "step": 2908
    },
    {
      "epoch": 0.26,
      "grad_norm": 8.867196236690921,
      "learning_rate": 9.741366720211627e-06,
      "loss": 0.8891,
      "step": 2909
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.126255083200468,
      "learning_rate": 9.741137350371762e-06,
      "loss": 0.8397,
      "step": 2910
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.015000204456142,
      "learning_rate": 9.740907881571039e-06,
      "loss": 0.8738,
      "step": 2911
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.147275929650224,
      "learning_rate": 9.740678313814244e-06,
      "loss": 0.8892,
      "step": 2912
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.807774428538608,
      "learning_rate": 9.740448647106174e-06,
      "loss": 0.899,
      "step": 2913
    },
    {
      "epoch": 0.26,
      "grad_norm": 10.828103234112147,
      "learning_rate": 9.740218881451619e-06,
      "loss": 0.7848,
      "step": 2914
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.119381502937452,
      "learning_rate": 9.739989016855376e-06,
      "loss": 0.8818,
      "step": 2915
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.960927044661801,
      "learning_rate": 9.739759053322243e-06,
      "loss": 0.9317,
      "step": 2916
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.328862698144004,
      "learning_rate": 9.73952899085702e-06,
      "loss": 0.8599,
      "step": 2917
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.019691564053099,
      "learning_rate": 9.739298829464506e-06,
      "loss": 0.9338,
      "step": 2918
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.761833974412195,
      "learning_rate": 9.73906856914951e-06,
      "loss": 0.8485,
      "step": 2919
    },
    {
      "epoch": 0.26,
      "grad_norm": 8.208942206094564,
      "learning_rate": 9.738838209916836e-06,
      "loss": 0.8938,
      "step": 2920
    },
    {
      "epoch": 0.26,
      "grad_norm": 9.308792645723306,
      "learning_rate": 9.738607751771291e-06,
      "loss": 0.8737,
      "step": 2921
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.055625211557587,
      "learning_rate": 9.738377194717687e-06,
      "loss": 0.8509,
      "step": 2922
    },
    {
      "epoch": 0.26,
      "grad_norm": 9.437071853954981,
      "learning_rate": 9.738146538760837e-06,
      "loss": 0.8952,
      "step": 2923
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.979620390703249,
      "learning_rate": 9.737915783905553e-06,
      "loss": 0.886,
      "step": 2924
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.264675802428782,
      "learning_rate": 9.737684930156654e-06,
      "loss": 0.861,
      "step": 2925
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.97857322062904,
      "learning_rate": 9.737453977518957e-06,
      "loss": 0.9481,
      "step": 2926
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.204513346029063,
      "learning_rate": 9.737222925997283e-06,
      "loss": 0.8769,
      "step": 2927
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.403017162918406,
      "learning_rate": 9.736991775596456e-06,
      "loss": 0.8891,
      "step": 2928
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.442809919887207,
      "learning_rate": 9.736760526321296e-06,
      "loss": 0.9208,
      "step": 2929
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.763889608189788,
      "learning_rate": 9.736529178176636e-06,
      "loss": 0.8936,
      "step": 2930
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.562682083101366,
      "learning_rate": 9.736297731167304e-06,
      "loss": 0.8586,
      "step": 2931
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.888473653598032,
      "learning_rate": 9.736066185298125e-06,
      "loss": 0.8179,
      "step": 2932
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.88260428310222,
      "learning_rate": 9.735834540573939e-06,
      "loss": 0.815,
      "step": 2933
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.267593251941493,
      "learning_rate": 9.735602796999578e-06,
      "loss": 0.9035,
      "step": 2934
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.3493494664780155,
      "learning_rate": 9.73537095457988e-06,
      "loss": 0.8981,
      "step": 2935
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.909088475938786,
      "learning_rate": 9.73513901331968e-06,
      "loss": 0.8694,
      "step": 2936
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.625050847153191,
      "learning_rate": 9.734906973223826e-06,
      "loss": 0.8215,
      "step": 2937
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.5072111731740065,
      "learning_rate": 9.734674834297157e-06,
      "loss": 0.8131,
      "step": 2938
    },
    {
      "epoch": 0.26,
      "grad_norm": 8.90791703021879,
      "learning_rate": 9.734442596544521e-06,
      "loss": 0.8875,
      "step": 2939
    },
    {
      "epoch": 0.26,
      "grad_norm": 8.348527307421849,
      "learning_rate": 9.734210259970764e-06,
      "loss": 0.9028,
      "step": 2940
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.362296515600365,
      "learning_rate": 9.733977824580734e-06,
      "loss": 0.8756,
      "step": 2941
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.605867000641782,
      "learning_rate": 9.733745290379286e-06,
      "loss": 0.8215,
      "step": 2942
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.627514498670066,
      "learning_rate": 9.73351265737127e-06,
      "loss": 0.8336,
      "step": 2943
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.282500455703085,
      "learning_rate": 9.733279925561544e-06,
      "loss": 0.8803,
      "step": 2944
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.22762304400222,
      "learning_rate": 9.733047094954967e-06,
      "loss": 0.9233,
      "step": 2945
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.347144310677438,
      "learning_rate": 9.732814165556394e-06,
      "loss": 0.8143,
      "step": 2946
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.699640097327885,
      "learning_rate": 9.73258113737069e-06,
      "loss": 0.8497,
      "step": 2947
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.0083609867977374,
      "learning_rate": 9.732348010402723e-06,
      "loss": 0.8671,
      "step": 2948
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.03530284541389,
      "learning_rate": 9.732114784657352e-06,
      "loss": 0.8384,
      "step": 2949
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.216601275165391,
      "learning_rate": 9.731881460139448e-06,
      "loss": 0.9267,
      "step": 2950
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.275725666034609,
      "learning_rate": 9.731648036853882e-06,
      "loss": 0.8775,
      "step": 2951
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.600833917916779,
      "learning_rate": 9.731414514805525e-06,
      "loss": 0.8959,
      "step": 2952
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.962874566436721,
      "learning_rate": 9.731180893999252e-06,
      "loss": 0.8608,
      "step": 2953
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.975862917741863,
      "learning_rate": 9.73094717443994e-06,
      "loss": 0.8517,
      "step": 2954
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.308199314459953,
      "learning_rate": 9.730713356132464e-06,
      "loss": 0.8908,
      "step": 2955
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.9865638546773505,
      "learning_rate": 9.730479439081707e-06,
      "loss": 0.8101,
      "step": 2956
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.1604430967937684,
      "learning_rate": 9.730245423292552e-06,
      "loss": 0.8953,
      "step": 2957
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.7685934922296935,
      "learning_rate": 9.730011308769883e-06,
      "loss": 0.9055,
      "step": 2958
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.924664523432183,
      "learning_rate": 9.729777095518586e-06,
      "loss": 0.8468,
      "step": 2959
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.030315269337679,
      "learning_rate": 9.72954278354355e-06,
      "loss": 0.8628,
      "step": 2960
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.599489734621938,
      "learning_rate": 9.729308372849667e-06,
      "loss": 0.8848,
      "step": 2961
    },
    {
      "epoch": 0.26,
      "grad_norm": 8.954298784004244,
      "learning_rate": 9.729073863441828e-06,
      "loss": 0.9052,
      "step": 2962
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.327207387368721,
      "learning_rate": 9.728839255324928e-06,
      "loss": 0.8665,
      "step": 2963
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.813953664625126,
      "learning_rate": 9.728604548503865e-06,
      "loss": 0.8904,
      "step": 2964
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.192390636695897,
      "learning_rate": 9.728369742983539e-06,
      "loss": 0.7919,
      "step": 2965
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.73410436949795,
      "learning_rate": 9.728134838768847e-06,
      "loss": 0.8438,
      "step": 2966
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.266903500290609,
      "learning_rate": 9.727899835864697e-06,
      "loss": 0.8595,
      "step": 2967
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.435451006971222,
      "learning_rate": 9.72766473427599e-06,
      "loss": 0.9297,
      "step": 2968
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.826049793218305,
      "learning_rate": 9.727429534007638e-06,
      "loss": 0.8379,
      "step": 2969
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.376219537423548,
      "learning_rate": 9.727194235064544e-06,
      "loss": 0.7575,
      "step": 2970
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.52114784233041,
      "learning_rate": 9.726958837451624e-06,
      "loss": 0.9139,
      "step": 2971
    },
    {
      "epoch": 0.27,
      "grad_norm": 9.652068023351346,
      "learning_rate": 9.726723341173791e-06,
      "loss": 0.8638,
      "step": 2972
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.942138630915804,
      "learning_rate": 9.726487746235957e-06,
      "loss": 0.8446,
      "step": 2973
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.174683412330826,
      "learning_rate": 9.726252052643046e-06,
      "loss": 0.7962,
      "step": 2974
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.744745658971282,
      "learning_rate": 9.726016260399971e-06,
      "loss": 0.879,
      "step": 2975
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.416093318737307,
      "learning_rate": 9.725780369511657e-06,
      "loss": 0.7969,
      "step": 2976
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.8958048999102255,
      "learning_rate": 9.725544379983028e-06,
      "loss": 0.8167,
      "step": 2977
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.997307011734548,
      "learning_rate": 9.725308291819007e-06,
      "loss": 0.8657,
      "step": 2978
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.663988356169776,
      "learning_rate": 9.725072105024524e-06,
      "loss": 0.8249,
      "step": 2979
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.772938558538891,
      "learning_rate": 9.724835819604509e-06,
      "loss": 0.8484,
      "step": 2980
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.17723222878287,
      "learning_rate": 9.724599435563893e-06,
      "loss": 0.7995,
      "step": 2981
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.188131035102536,
      "learning_rate": 9.724362952907611e-06,
      "loss": 0.9012,
      "step": 2982
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.601561459703495,
      "learning_rate": 9.724126371640598e-06,
      "loss": 0.846,
      "step": 2983
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.339322277599898,
      "learning_rate": 9.723889691767793e-06,
      "loss": 0.9109,
      "step": 2984
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.890204103196911,
      "learning_rate": 9.723652913294138e-06,
      "loss": 0.8524,
      "step": 2985
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.417764764532755,
      "learning_rate": 9.723416036224571e-06,
      "loss": 0.9152,
      "step": 2986
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.830040889304214,
      "learning_rate": 9.72317906056404e-06,
      "loss": 0.88,
      "step": 2987
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.35729496556518,
      "learning_rate": 9.722941986317487e-06,
      "loss": 0.8499,
      "step": 2988
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.819337618132686,
      "learning_rate": 9.722704813489863e-06,
      "loss": 0.8945,
      "step": 2989
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.624524072529956,
      "learning_rate": 9.722467542086121e-06,
      "loss": 0.8033,
      "step": 2990
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.739670631633869,
      "learning_rate": 9.722230172111208e-06,
      "loss": 0.8676,
      "step": 2991
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.705037771502696,
      "learning_rate": 9.721992703570085e-06,
      "loss": 0.9331,
      "step": 2992
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.72512206209416,
      "learning_rate": 9.721755136467703e-06,
      "loss": 0.8404,
      "step": 2993
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.81601356479582,
      "learning_rate": 9.721517470809025e-06,
      "loss": 0.8172,
      "step": 2994
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.738913689577057,
      "learning_rate": 9.721279706599008e-06,
      "loss": 0.893,
      "step": 2995
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.559604397382546,
      "learning_rate": 9.721041843842617e-06,
      "loss": 0.8801,
      "step": 2996
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.289574030331662,
      "learning_rate": 9.720803882544817e-06,
      "loss": 0.849,
      "step": 2997
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.295826900174799,
      "learning_rate": 9.720565822710573e-06,
      "loss": 0.8815,
      "step": 2998
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.378654818142778,
      "learning_rate": 9.720327664344857e-06,
      "loss": 0.8586,
      "step": 2999
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.009031717970705,
      "learning_rate": 9.720089407452638e-06,
      "loss": 0.8706,
      "step": 3000
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.247181007463583,
      "learning_rate": 9.71985105203889e-06,
      "loss": 0.8761,
      "step": 3001
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.392037019565254,
      "learning_rate": 9.719612598108586e-06,
      "loss": 0.8636,
      "step": 3002
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.502236968620264,
      "learning_rate": 9.719374045666705e-06,
      "loss": 0.8564,
      "step": 3003
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.804786698984423,
      "learning_rate": 9.719135394718226e-06,
      "loss": 0.9188,
      "step": 3004
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.798845885189531,
      "learning_rate": 9.71889664526813e-06,
      "loss": 0.8413,
      "step": 3005
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.751671984973259,
      "learning_rate": 9.718657797321403e-06,
      "loss": 0.8483,
      "step": 3006
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.657856568841324,
      "learning_rate": 9.718418850883026e-06,
      "loss": 0.8498,
      "step": 3007
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.309178735831084,
      "learning_rate": 9.718179805957989e-06,
      "loss": 0.9353,
      "step": 3008
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.960153357440854,
      "learning_rate": 9.717940662551282e-06,
      "loss": 0.8651,
      "step": 3009
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.721039105715273,
      "learning_rate": 9.717701420667896e-06,
      "loss": 0.8622,
      "step": 3010
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.109305585256216,
      "learning_rate": 9.717462080312822e-06,
      "loss": 0.9345,
      "step": 3011
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.576706305775325,
      "learning_rate": 9.71722264149106e-06,
      "loss": 0.8873,
      "step": 3012
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.083828073709991,
      "learning_rate": 9.716983104207606e-06,
      "loss": 0.8397,
      "step": 3013
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.432312790722551,
      "learning_rate": 9.71674346846746e-06,
      "loss": 0.9376,
      "step": 3014
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.289393904218191,
      "learning_rate": 9.716503734275622e-06,
      "loss": 0.8611,
      "step": 3015
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.913699490982757,
      "learning_rate": 9.7162639016371e-06,
      "loss": 0.8188,
      "step": 3016
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.401470100349791,
      "learning_rate": 9.716023970556896e-06,
      "loss": 0.9262,
      "step": 3017
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.737600280104129,
      "learning_rate": 9.715783941040021e-06,
      "loss": 0.8469,
      "step": 3018
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.35588559436862,
      "learning_rate": 9.715543813091481e-06,
      "loss": 0.8515,
      "step": 3019
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.888578498117389,
      "learning_rate": 9.715303586716292e-06,
      "loss": 0.8081,
      "step": 3020
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.083627670957956,
      "learning_rate": 9.715063261919467e-06,
      "loss": 0.8631,
      "step": 3021
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.4877558159340385,
      "learning_rate": 9.714822838706021e-06,
      "loss": 0.8517,
      "step": 3022
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.423370863219031,
      "learning_rate": 9.714582317080976e-06,
      "loss": 0.9301,
      "step": 3023
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.077866947948512,
      "learning_rate": 9.714341697049348e-06,
      "loss": 0.8866,
      "step": 3024
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.7058037974219165,
      "learning_rate": 9.714100978616162e-06,
      "loss": 0.8504,
      "step": 3025
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.126247431815928,
      "learning_rate": 9.713860161786441e-06,
      "loss": 0.8418,
      "step": 3026
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.12274760390638,
      "learning_rate": 9.713619246565214e-06,
      "loss": 0.8604,
      "step": 3027
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.1009013808003365,
      "learning_rate": 9.713378232957507e-06,
      "loss": 0.9589,
      "step": 3028
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.294754227005092,
      "learning_rate": 9.713137120968353e-06,
      "loss": 0.8648,
      "step": 3029
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.72011645450055,
      "learning_rate": 9.71289591060278e-06,
      "loss": 0.7517,
      "step": 3030
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.324603758460063,
      "learning_rate": 9.712654601865827e-06,
      "loss": 0.821,
      "step": 3031
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.30609037253442,
      "learning_rate": 9.712413194762532e-06,
      "loss": 0.9233,
      "step": 3032
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.864252397438357,
      "learning_rate": 9.712171689297929e-06,
      "loss": 0.932,
      "step": 3033
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.828146196493721,
      "learning_rate": 9.71193008547706e-06,
      "loss": 0.8605,
      "step": 3034
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.226277686181857,
      "learning_rate": 9.711688383304973e-06,
      "loss": 0.8937,
      "step": 3035
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.763712090779456,
      "learning_rate": 9.711446582786709e-06,
      "loss": 0.7996,
      "step": 3036
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.323801297513233,
      "learning_rate": 9.711204683927314e-06,
      "loss": 0.8789,
      "step": 3037
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.392012742147735,
      "learning_rate": 9.71096268673184e-06,
      "loss": 0.8255,
      "step": 3038
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.506087504924075,
      "learning_rate": 9.710720591205336e-06,
      "loss": 0.8745,
      "step": 3039
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.925964866910463,
      "learning_rate": 9.710478397352855e-06,
      "loss": 0.846,
      "step": 3040
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.91284595079221,
      "learning_rate": 9.710236105179455e-06,
      "loss": 0.869,
      "step": 3041
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.0869896512749735,
      "learning_rate": 9.70999371469019e-06,
      "loss": 0.921,
      "step": 3042
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.494163483001028,
      "learning_rate": 9.709751225890122e-06,
      "loss": 0.9077,
      "step": 3043
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.2843385399789,
      "learning_rate": 9.709508638784311e-06,
      "loss": 0.859,
      "step": 3044
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.530931226316387,
      "learning_rate": 9.70926595337782e-06,
      "loss": 0.8734,
      "step": 3045
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.581384372045062,
      "learning_rate": 9.709023169675717e-06,
      "loss": 0.9032,
      "step": 3046
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.549839701851494,
      "learning_rate": 9.708780287683066e-06,
      "loss": 0.8154,
      "step": 3047
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.937033524548889,
      "learning_rate": 9.70853730740494e-06,
      "loss": 0.8937,
      "step": 3048
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.5795631402627635,
      "learning_rate": 9.708294228846408e-06,
      "loss": 0.902,
      "step": 3049
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.402778556155768,
      "learning_rate": 9.708051052012546e-06,
      "loss": 0.8409,
      "step": 3050
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.196068560860864,
      "learning_rate": 9.70780777690843e-06,
      "loss": 0.9531,
      "step": 3051
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.893785059208222,
      "learning_rate": 9.707564403539134e-06,
      "loss": 0.874,
      "step": 3052
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.09095217779118,
      "learning_rate": 9.707320931909742e-06,
      "loss": 0.8364,
      "step": 3053
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.657812298400147,
      "learning_rate": 9.707077362025334e-06,
      "loss": 0.9116,
      "step": 3054
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.825614940690465,
      "learning_rate": 9.706833693890993e-06,
      "loss": 0.8039,
      "step": 3055
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.944091454606696,
      "learning_rate": 9.706589927511809e-06,
      "loss": 0.9097,
      "step": 3056
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.713767355178581,
      "learning_rate": 9.706346062892867e-06,
      "loss": 0.8914,
      "step": 3057
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.254164684885612,
      "learning_rate": 9.706102100039257e-06,
      "loss": 0.8299,
      "step": 3058
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.470055035643572,
      "learning_rate": 9.705858038956072e-06,
      "loss": 0.8037,
      "step": 3059
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.782994962006137,
      "learning_rate": 9.705613879648404e-06,
      "loss": 0.8406,
      "step": 3060
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.674539505744804,
      "learning_rate": 9.705369622121353e-06,
      "loss": 0.8432,
      "step": 3061
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.667483778679136,
      "learning_rate": 9.705125266380017e-06,
      "loss": 0.7977,
      "step": 3062
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.317012052566255,
      "learning_rate": 9.704880812429494e-06,
      "loss": 0.8351,
      "step": 3063
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.776689852676243,
      "learning_rate": 9.704636260274887e-06,
      "loss": 0.848,
      "step": 3064
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.279969165424993,
      "learning_rate": 9.704391609921302e-06,
      "loss": 0.8754,
      "step": 3065
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.121376733645529,
      "learning_rate": 9.704146861373844e-06,
      "loss": 0.8667,
      "step": 3066
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.466646014938607,
      "learning_rate": 9.703902014637623e-06,
      "loss": 0.8281,
      "step": 3067
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.54825064934803,
      "learning_rate": 9.703657069717748e-06,
      "loss": 0.8476,
      "step": 3068
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.438999694595895,
      "learning_rate": 9.703412026619332e-06,
      "loss": 0.881,
      "step": 3069
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.793220747208214,
      "learning_rate": 9.703166885347492e-06,
      "loss": 0.98,
      "step": 3070
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.9333955266146585,
      "learning_rate": 9.702921645907341e-06,
      "loss": 0.7855,
      "step": 3071
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.538153711101955,
      "learning_rate": 9.702676308304001e-06,
      "loss": 0.8254,
      "step": 3072
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.011271229283828,
      "learning_rate": 9.702430872542592e-06,
      "loss": 0.8777,
      "step": 3073
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.732567330834858,
      "learning_rate": 9.702185338628237e-06,
      "loss": 0.7949,
      "step": 3074
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.248069808278189,
      "learning_rate": 9.70193970656606e-06,
      "loss": 0.8518,
      "step": 3075
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.312674565886622,
      "learning_rate": 9.701693976361188e-06,
      "loss": 0.8338,
      "step": 3076
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.922212125874724,
      "learning_rate": 9.701448148018751e-06,
      "loss": 0.9342,
      "step": 3077
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.5156878344543605,
      "learning_rate": 9.701202221543881e-06,
      "loss": 0.8259,
      "step": 3078
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.028429782770404,
      "learning_rate": 9.70095619694171e-06,
      "loss": 0.8851,
      "step": 3079
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.1805829835697725,
      "learning_rate": 9.700710074217372e-06,
      "loss": 0.8366,
      "step": 3080
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.8203835597586835,
      "learning_rate": 9.700463853376006e-06,
      "loss": 0.8549,
      "step": 3081
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.706546882775098,
      "learning_rate": 9.700217534422752e-06,
      "loss": 0.9334,
      "step": 3082
    },
    {
      "epoch": 0.28,
      "grad_norm": 8.0726992502605,
      "learning_rate": 9.69997111736275e-06,
      "loss": 0.8607,
      "step": 3083
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.968332838288017,
      "learning_rate": 9.699724602201142e-06,
      "loss": 0.8509,
      "step": 3084
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.393826786263297,
      "learning_rate": 9.699477988943077e-06,
      "loss": 0.9329,
      "step": 3085
    },
    {
      "epoch": 0.28,
      "grad_norm": 8.109300297285262,
      "learning_rate": 9.6992312775937e-06,
      "loss": 0.8247,
      "step": 3086
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.50054179465883,
      "learning_rate": 9.698984468158162e-06,
      "loss": 0.8819,
      "step": 3087
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.394411122300468,
      "learning_rate": 9.698737560641613e-06,
      "loss": 0.9489,
      "step": 3088
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.407657738506801,
      "learning_rate": 9.698490555049208e-06,
      "loss": 0.8444,
      "step": 3089
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.273921730997827,
      "learning_rate": 9.6982434513861e-06,
      "loss": 0.8846,
      "step": 3090
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.542645993087722,
      "learning_rate": 9.69799624965745e-06,
      "loss": 0.8066,
      "step": 3091
    },
    {
      "epoch": 0.28,
      "grad_norm": 8.059994910208939,
      "learning_rate": 9.697748949868419e-06,
      "loss": 0.8113,
      "step": 3092
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.966063802944681,
      "learning_rate": 9.697501552024165e-06,
      "loss": 0.8055,
      "step": 3093
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.60635306931067,
      "learning_rate": 9.697254056129852e-06,
      "loss": 0.9323,
      "step": 3094
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.215033859835412,
      "learning_rate": 9.69700646219065e-06,
      "loss": 0.9075,
      "step": 3095
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.950879798369478,
      "learning_rate": 9.696758770211723e-06,
      "loss": 0.8296,
      "step": 3096
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.480255455704876,
      "learning_rate": 9.696510980198243e-06,
      "loss": 0.8272,
      "step": 3097
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.210900774178484,
      "learning_rate": 9.69626309215538e-06,
      "loss": 0.826,
      "step": 3098
    },
    {
      "epoch": 0.28,
      "grad_norm": 8.818406427982387,
      "learning_rate": 9.696015106088311e-06,
      "loss": 0.8786,
      "step": 3099
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.888962304077095,
      "learning_rate": 9.69576702200221e-06,
      "loss": 0.8406,
      "step": 3100
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.758322634006026,
      "learning_rate": 9.695518839902258e-06,
      "loss": 0.8815,
      "step": 3101
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.041040975960407,
      "learning_rate": 9.69527055979363e-06,
      "loss": 0.8772,
      "step": 3102
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.493073765176016,
      "learning_rate": 9.695022181681514e-06,
      "loss": 0.8644,
      "step": 3103
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.398889080429567,
      "learning_rate": 9.69477370557109e-06,
      "loss": 0.8232,
      "step": 3104
    },
    {
      "epoch": 0.28,
      "grad_norm": 8.314973260921656,
      "learning_rate": 9.694525131467547e-06,
      "loss": 0.9314,
      "step": 3105
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.406315483476398,
      "learning_rate": 9.694276459376075e-06,
      "loss": 0.8469,
      "step": 3106
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.825182870472401,
      "learning_rate": 9.69402768930186e-06,
      "loss": 0.9465,
      "step": 3107
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.497608041587692,
      "learning_rate": 9.693778821250096e-06,
      "loss": 0.8122,
      "step": 3108
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.887614620199217,
      "learning_rate": 9.693529855225977e-06,
      "loss": 0.8395,
      "step": 3109
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.302208079550322,
      "learning_rate": 9.693280791234706e-06,
      "loss": 0.8805,
      "step": 3110
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.25617539695755,
      "learning_rate": 9.693031629281473e-06,
      "loss": 0.8314,
      "step": 3111
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.970438627994429,
      "learning_rate": 9.692782369371482e-06,
      "loss": 0.8578,
      "step": 3112
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.058051769939529,
      "learning_rate": 9.692533011509939e-06,
      "loss": 0.8264,
      "step": 3113
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.337832567677477,
      "learning_rate": 9.692283555702044e-06,
      "loss": 0.8891,
      "step": 3114
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.437364920677783,
      "learning_rate": 9.692034001953005e-06,
      "loss": 0.8312,
      "step": 3115
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.835990075673858,
      "learning_rate": 9.691784350268031e-06,
      "loss": 0.8553,
      "step": 3116
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.380677278052266,
      "learning_rate": 9.691534600652335e-06,
      "loss": 0.816,
      "step": 3117
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.787092320460224,
      "learning_rate": 9.69128475311113e-06,
      "loss": 0.8289,
      "step": 3118
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.790469621322249,
      "learning_rate": 9.691034807649626e-06,
      "loss": 0.9361,
      "step": 3119
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.920477493960096,
      "learning_rate": 9.690784764273046e-06,
      "loss": 0.8681,
      "step": 3120
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.308046893441458,
      "learning_rate": 9.690534622986606e-06,
      "loss": 0.9,
      "step": 3121
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.318040749637306,
      "learning_rate": 9.690284383795528e-06,
      "loss": 0.8635,
      "step": 3122
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.6210665613162325,
      "learning_rate": 9.690034046705034e-06,
      "loss": 0.8076,
      "step": 3123
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.945470680369091,
      "learning_rate": 9.689783611720352e-06,
      "loss": 0.8656,
      "step": 3124
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.474235876074848,
      "learning_rate": 9.689533078846707e-06,
      "loss": 0.9374,
      "step": 3125
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.175672950706755,
      "learning_rate": 9.68928244808933e-06,
      "loss": 0.8464,
      "step": 3126
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.636105792392971,
      "learning_rate": 9.689031719453448e-06,
      "loss": 0.879,
      "step": 3127
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.585011262553224,
      "learning_rate": 9.688780892944301e-06,
      "loss": 0.9177,
      "step": 3128
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.704417870731716,
      "learning_rate": 9.688529968567118e-06,
      "loss": 0.8571,
      "step": 3129
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.415263129546187,
      "learning_rate": 9.688278946327142e-06,
      "loss": 0.7728,
      "step": 3130
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.278537664929393,
      "learning_rate": 9.68802782622961e-06,
      "loss": 0.8681,
      "step": 3131
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.769998127805124,
      "learning_rate": 9.687776608279762e-06,
      "loss": 0.9236,
      "step": 3132
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.268969041925272,
      "learning_rate": 9.687525292482845e-06,
      "loss": 0.8591,
      "step": 3133
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.929442309200846,
      "learning_rate": 9.687273878844104e-06,
      "loss": 0.8484,
      "step": 3134
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.5056650256862225,
      "learning_rate": 9.687022367368783e-06,
      "loss": 0.8376,
      "step": 3135
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.158105730098482,
      "learning_rate": 9.686770758062136e-06,
      "loss": 0.8718,
      "step": 3136
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.393676039939339,
      "learning_rate": 9.686519050929413e-06,
      "loss": 0.9134,
      "step": 3137
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.055929185913235,
      "learning_rate": 9.68626724597587e-06,
      "loss": 0.8547,
      "step": 3138
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.834701431409927,
      "learning_rate": 9.686015343206757e-06,
      "loss": 0.8531,
      "step": 3139
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.415379673532301,
      "learning_rate": 9.68576334262734e-06,
      "loss": 0.9031,
      "step": 3140
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.504060836737989,
      "learning_rate": 9.685511244242871e-06,
      "loss": 0.8711,
      "step": 3141
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.460845434511139,
      "learning_rate": 9.685259048058618e-06,
      "loss": 0.8177,
      "step": 3142
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.384491072062221,
      "learning_rate": 9.685006754079842e-06,
      "loss": 0.8618,
      "step": 3143
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.6375318072833025,
      "learning_rate": 9.684754362311811e-06,
      "loss": 0.8295,
      "step": 3144
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.15981836815742,
      "learning_rate": 9.684501872759792e-06,
      "loss": 0.8692,
      "step": 3145
    },
    {
      "epoch": 0.28,
      "grad_norm": 8.043785083444916,
      "learning_rate": 9.684249285429055e-06,
      "loss": 0.7807,
      "step": 3146
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.75189633989718,
      "learning_rate": 9.683996600324874e-06,
      "loss": 0.8626,
      "step": 3147
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.568543420948107,
      "learning_rate": 9.68374381745252e-06,
      "loss": 0.8825,
      "step": 3148
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.263052597374487,
      "learning_rate": 9.68349093681727e-06,
      "loss": 0.8815,
      "step": 3149
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.067041652148552,
      "learning_rate": 9.683237958424406e-06,
      "loss": 0.8472,
      "step": 3150
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.245109202971791,
      "learning_rate": 9.682984882279204e-06,
      "loss": 0.9488,
      "step": 3151
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.022896106761805,
      "learning_rate": 9.682731708386948e-06,
      "loss": 0.8679,
      "step": 3152
    },
    {
      "epoch": 0.28,
      "grad_norm": 9.92315186851741,
      "learning_rate": 9.682478436752924e-06,
      "loss": 0.8758,
      "step": 3153
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.454874714822032,
      "learning_rate": 9.682225067382417e-06,
      "loss": 0.8318,
      "step": 3154
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.754173493299103,
      "learning_rate": 9.681971600280714e-06,
      "loss": 0.9687,
      "step": 3155
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.1088758266358125,
      "learning_rate": 9.681718035453108e-06,
      "loss": 0.9343,
      "step": 3156
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.258531043098854,
      "learning_rate": 9.681464372904892e-06,
      "loss": 0.8337,
      "step": 3157
    },
    {
      "epoch": 0.28,
      "grad_norm": 8.148949817233103,
      "learning_rate": 9.681210612641359e-06,
      "loss": 0.8502,
      "step": 3158
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.130661953832419,
      "learning_rate": 9.680956754667805e-06,
      "loss": 0.8584,
      "step": 3159
    },
    {
      "epoch": 0.28,
      "grad_norm": 8.965400212516986,
      "learning_rate": 9.680702798989529e-06,
      "loss": 0.8964,
      "step": 3160
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.8014045043185645,
      "learning_rate": 9.680448745611835e-06,
      "loss": 0.8632,
      "step": 3161
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.252974689431325,
      "learning_rate": 9.680194594540021e-06,
      "loss": 0.8341,
      "step": 3162
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.216286371943972,
      "learning_rate": 9.679940345779398e-06,
      "loss": 0.9139,
      "step": 3163
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.57946937979855,
      "learning_rate": 9.679685999335265e-06,
      "loss": 0.8534,
      "step": 3164
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.985706025320577,
      "learning_rate": 9.679431555212938e-06,
      "loss": 0.7925,
      "step": 3165
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.804563194858407,
      "learning_rate": 9.679177013417724e-06,
      "loss": 0.8916,
      "step": 3166
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.555506626811849,
      "learning_rate": 9.678922373954938e-06,
      "loss": 0.8665,
      "step": 3167
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.949576602526726,
      "learning_rate": 9.678667636829892e-06,
      "loss": 0.885,
      "step": 3168
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.292594944687893,
      "learning_rate": 9.678412802047907e-06,
      "loss": 0.8441,
      "step": 3169
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.860988032981282,
      "learning_rate": 9.678157869614299e-06,
      "loss": 0.8866,
      "step": 3170
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.408207106265668,
      "learning_rate": 9.677902839534391e-06,
      "loss": 0.8996,
      "step": 3171
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.988207734842082,
      "learning_rate": 9.677647711813508e-06,
      "loss": 0.9256,
      "step": 3172
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.809379842703858,
      "learning_rate": 9.677392486456969e-06,
      "loss": 0.8721,
      "step": 3173
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.943788628379881,
      "learning_rate": 9.677137163470107e-06,
      "loss": 0.8829,
      "step": 3174
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.384666078039953,
      "learning_rate": 9.676881742858249e-06,
      "loss": 0.8813,
      "step": 3175
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.474707675153014,
      "learning_rate": 9.676626224626728e-06,
      "loss": 0.9031,
      "step": 3176
    },
    {
      "epoch": 0.28,
      "grad_norm": 10.913266601187164,
      "learning_rate": 9.676370608780876e-06,
      "loss": 0.835,
      "step": 3177
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.3687479560024425,
      "learning_rate": 9.676114895326028e-06,
      "loss": 0.9084,
      "step": 3178
    },
    {
      "epoch": 0.28,
      "grad_norm": 8.437121011407891,
      "learning_rate": 9.675859084267521e-06,
      "loss": 0.8293,
      "step": 3179
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.751903686782551,
      "learning_rate": 9.675603175610697e-06,
      "loss": 0.8889,
      "step": 3180
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.989918664110538,
      "learning_rate": 9.675347169360896e-06,
      "loss": 0.8153,
      "step": 3181
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.675388371990813,
      "learning_rate": 9.67509106552346e-06,
      "loss": 0.8343,
      "step": 3182
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.486240394544817,
      "learning_rate": 9.67483486410374e-06,
      "loss": 0.896,
      "step": 3183
    },
    {
      "epoch": 0.28,
      "grad_norm": 9.53227553846843,
      "learning_rate": 9.674578565107077e-06,
      "loss": 0.8414,
      "step": 3184
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.656020192978489,
      "learning_rate": 9.674322168538825e-06,
      "loss": 0.8995,
      "step": 3185
    },
    {
      "epoch": 0.28,
      "grad_norm": 8.274969651396205,
      "learning_rate": 9.674065674404333e-06,
      "loss": 0.8966,
      "step": 3186
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.899276773606116,
      "learning_rate": 9.673809082708956e-06,
      "loss": 0.9166,
      "step": 3187
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.771447449385443,
      "learning_rate": 9.673552393458051e-06,
      "loss": 0.8239,
      "step": 3188
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.601009897481193,
      "learning_rate": 9.673295606656975e-06,
      "loss": 0.9053,
      "step": 3189
    },
    {
      "epoch": 0.28,
      "grad_norm": 8.421944188060367,
      "learning_rate": 9.673038722311086e-06,
      "loss": 0.9374,
      "step": 3190
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.309512076353031,
      "learning_rate": 9.672781740425748e-06,
      "loss": 0.8751,
      "step": 3191
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.256443469602372,
      "learning_rate": 9.672524661006325e-06,
      "loss": 0.8961,
      "step": 3192
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.7702887809561965,
      "learning_rate": 9.672267484058183e-06,
      "loss": 0.8463,
      "step": 3193
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.5743138489856845,
      "learning_rate": 9.672010209586687e-06,
      "loss": 0.8855,
      "step": 3194
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.832818103543249,
      "learning_rate": 9.67175283759721e-06,
      "loss": 0.8285,
      "step": 3195
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.865656925945503,
      "learning_rate": 9.671495368095125e-06,
      "loss": 0.8645,
      "step": 3196
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.735240508132219,
      "learning_rate": 9.671237801085804e-06,
      "loss": 0.8669,
      "step": 3197
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.05606042808752,
      "learning_rate": 9.670980136574623e-06,
      "loss": 0.8491,
      "step": 3198
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.728683492847892,
      "learning_rate": 9.670722374566962e-06,
      "loss": 0.8156,
      "step": 3199
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.765082700843074,
      "learning_rate": 9.670464515068199e-06,
      "loss": 0.8497,
      "step": 3200
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.160938816513407,
      "learning_rate": 9.67020655808372e-06,
      "loss": 0.9139,
      "step": 3201
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.439873484383166,
      "learning_rate": 9.669948503618904e-06,
      "loss": 0.8755,
      "step": 3202
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.467998032568611,
      "learning_rate": 9.66969035167914e-06,
      "loss": 0.8342,
      "step": 3203
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.315771730196795,
      "learning_rate": 9.669432102269818e-06,
      "loss": 0.9044,
      "step": 3204
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.095019958915507,
      "learning_rate": 9.669173755396325e-06,
      "loss": 0.8882,
      "step": 3205
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.079103404471578,
      "learning_rate": 9.668915311064055e-06,
      "loss": 0.8498,
      "step": 3206
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.512489273438903,
      "learning_rate": 9.668656769278403e-06,
      "loss": 0.8556,
      "step": 3207
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.63484293921342,
      "learning_rate": 9.668398130044767e-06,
      "loss": 0.8854,
      "step": 3208
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.660146844284628,
      "learning_rate": 9.668139393368543e-06,
      "loss": 0.8833,
      "step": 3209
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.632639072263363,
      "learning_rate": 9.667880559255131e-06,
      "loss": 0.7798,
      "step": 3210
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.162952336637108,
      "learning_rate": 9.667621627709935e-06,
      "loss": 0.8469,
      "step": 3211
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.019741071547217,
      "learning_rate": 9.667362598738362e-06,
      "loss": 0.8546,
      "step": 3212
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.595872600380877,
      "learning_rate": 9.667103472345813e-06,
      "loss": 0.9477,
      "step": 3213
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.808182812482724,
      "learning_rate": 9.6668442485377e-06,
      "loss": 0.8089,
      "step": 3214
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.373317893705277,
      "learning_rate": 9.666584927319434e-06,
      "loss": 0.9004,
      "step": 3215
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.440600601312156,
      "learning_rate": 9.666325508696429e-06,
      "loss": 0.9576,
      "step": 3216
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.698930879546522,
      "learning_rate": 9.666065992674096e-06,
      "loss": 0.7462,
      "step": 3217
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.39945366945253,
      "learning_rate": 9.665806379257853e-06,
      "loss": 0.8152,
      "step": 3218
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.117147210582255,
      "learning_rate": 9.665546668453123e-06,
      "loss": 0.8506,
      "step": 3219
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.120538698544707,
      "learning_rate": 9.665286860265322e-06,
      "loss": 0.7932,
      "step": 3220
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.836145542175159,
      "learning_rate": 9.665026954699874e-06,
      "loss": 0.9218,
      "step": 3221
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.9838692142903565,
      "learning_rate": 9.664766951762206e-06,
      "loss": 0.778,
      "step": 3222
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.121592358941899,
      "learning_rate": 9.664506851457742e-06,
      "loss": 0.8784,
      "step": 3223
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.982416383203774,
      "learning_rate": 9.664246653791913e-06,
      "loss": 0.8628,
      "step": 3224
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.095868318260642,
      "learning_rate": 9.66398635877015e-06,
      "loss": 0.8858,
      "step": 3225
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.001861291180722,
      "learning_rate": 9.663725966397886e-06,
      "loss": 0.9059,
      "step": 3226
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.1564442362735665,
      "learning_rate": 9.663465476680555e-06,
      "loss": 0.9026,
      "step": 3227
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.828267874227665,
      "learning_rate": 9.663204889623595e-06,
      "loss": 0.884,
      "step": 3228
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.719416698457259,
      "learning_rate": 9.662944205232447e-06,
      "loss": 0.8864,
      "step": 3229
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.592579387418979,
      "learning_rate": 9.662683423512548e-06,
      "loss": 0.8789,
      "step": 3230
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.70215959677774,
      "learning_rate": 9.662422544469346e-06,
      "loss": 0.8698,
      "step": 3231
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.2826388946792004,
      "learning_rate": 9.662161568108281e-06,
      "loss": 0.842,
      "step": 3232
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.27689763267672,
      "learning_rate": 9.661900494434805e-06,
      "loss": 0.8781,
      "step": 3233
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.704530347439892,
      "learning_rate": 9.661639323454367e-06,
      "loss": 0.8746,
      "step": 3234
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.540422916435023,
      "learning_rate": 9.661378055172413e-06,
      "loss": 0.9256,
      "step": 3235
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.706133546919614,
      "learning_rate": 9.661116689594403e-06,
      "loss": 0.8956,
      "step": 3236
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.879708412698,
      "learning_rate": 9.66085522672579e-06,
      "loss": 0.8957,
      "step": 3237
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.587427742006504,
      "learning_rate": 9.660593666572032e-06,
      "loss": 0.8406,
      "step": 3238
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.882327532801281,
      "learning_rate": 9.660332009138585e-06,
      "loss": 0.8733,
      "step": 3239
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.485961387937878,
      "learning_rate": 9.660070254430914e-06,
      "loss": 0.8527,
      "step": 3240
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.159733893175498,
      "learning_rate": 9.659808402454483e-06,
      "loss": 0.8797,
      "step": 3241
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.8419444495901685,
      "learning_rate": 9.659546453214757e-06,
      "loss": 0.8529,
      "step": 3242
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.341351250475801,
      "learning_rate": 9.6592844067172e-06,
      "loss": 0.9046,
      "step": 3243
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.6091916582488315,
      "learning_rate": 9.659022262967288e-06,
      "loss": 0.928,
      "step": 3244
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.675894952145962,
      "learning_rate": 9.658760021970487e-06,
      "loss": 0.8248,
      "step": 3245
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.669820888433391,
      "learning_rate": 9.658497683732274e-06,
      "loss": 0.8474,
      "step": 3246
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.731281506261774,
      "learning_rate": 9.658235248258122e-06,
      "loss": 0.8812,
      "step": 3247
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.581281009637838,
      "learning_rate": 9.657972715553515e-06,
      "loss": 0.834,
      "step": 3248
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.9748642136000605,
      "learning_rate": 9.657710085623924e-06,
      "loss": 0.7966,
      "step": 3249
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.633024215553253,
      "learning_rate": 9.657447358474837e-06,
      "loss": 0.8935,
      "step": 3250
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.891154672980486,
      "learning_rate": 9.657184534111737e-06,
      "loss": 0.9034,
      "step": 3251
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.539125200335768,
      "learning_rate": 9.656921612540107e-06,
      "loss": 0.8857,
      "step": 3252
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.153059015627392,
      "learning_rate": 9.656658593765438e-06,
      "loss": 0.8992,
      "step": 3253
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.748770334691978,
      "learning_rate": 9.656395477793218e-06,
      "loss": 0.843,
      "step": 3254
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.600675452495197,
      "learning_rate": 9.65613226462894e-06,
      "loss": 0.8954,
      "step": 3255
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.725131329293039,
      "learning_rate": 9.655868954278099e-06,
      "loss": 0.8785,
      "step": 3256
    },
    {
      "epoch": 0.29,
      "grad_norm": 9.469562012035917,
      "learning_rate": 9.655605546746188e-06,
      "loss": 0.8538,
      "step": 3257
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.521536526411392,
      "learning_rate": 9.655342042038706e-06,
      "loss": 0.9058,
      "step": 3258
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.069242427430499,
      "learning_rate": 9.655078440161157e-06,
      "loss": 0.8818,
      "step": 3259
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.31136303426663,
      "learning_rate": 9.654814741119037e-06,
      "loss": 0.8661,
      "step": 3260
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.498782823953539,
      "learning_rate": 9.654550944917856e-06,
      "loss": 0.8503,
      "step": 3261
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.301957183058016,
      "learning_rate": 9.654287051563115e-06,
      "loss": 0.8726,
      "step": 3262
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.738990029288649,
      "learning_rate": 9.654023061060324e-06,
      "loss": 0.8516,
      "step": 3263
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.654299354941623,
      "learning_rate": 9.653758973414995e-06,
      "loss": 0.8309,
      "step": 3264
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.411897694986788,
      "learning_rate": 9.65349478863264e-06,
      "loss": 0.8122,
      "step": 3265
    },
    {
      "epoch": 0.29,
      "grad_norm": 11.102965441242938,
      "learning_rate": 9.653230506718771e-06,
      "loss": 0.8418,
      "step": 3266
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.730850119903953,
      "learning_rate": 9.652966127678906e-06,
      "loss": 0.8319,
      "step": 3267
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.7429308842034335,
      "learning_rate": 9.652701651518564e-06,
      "loss": 0.799,
      "step": 3268
    },
    {
      "epoch": 0.29,
      "grad_norm": 9.32940240849665,
      "learning_rate": 9.652437078243261e-06,
      "loss": 0.8482,
      "step": 3269
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.507926132588859,
      "learning_rate": 9.652172407858525e-06,
      "loss": 0.8795,
      "step": 3270
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.327345818068146,
      "learning_rate": 9.651907640369877e-06,
      "loss": 0.8139,
      "step": 3271
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.22565246792412,
      "learning_rate": 9.651642775782846e-06,
      "loss": 0.8435,
      "step": 3272
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.393530691249871,
      "learning_rate": 9.651377814102958e-06,
      "loss": 0.9151,
      "step": 3273
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.922593234843766,
      "learning_rate": 9.651112755335745e-06,
      "loss": 0.8923,
      "step": 3274
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.950403544262727,
      "learning_rate": 9.65084759948674e-06,
      "loss": 0.8568,
      "step": 3275
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.751361811284917,
      "learning_rate": 9.650582346561475e-06,
      "loss": 0.8195,
      "step": 3276
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.636788899586379,
      "learning_rate": 9.650316996565488e-06,
      "loss": 0.8931,
      "step": 3277
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.689404616876132,
      "learning_rate": 9.650051549504317e-06,
      "loss": 0.8665,
      "step": 3278
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.803147688962341,
      "learning_rate": 9.649786005383506e-06,
      "loss": 0.8516,
      "step": 3279
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.042305161005341,
      "learning_rate": 9.649520364208591e-06,
      "loss": 0.883,
      "step": 3280
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.993775653036826,
      "learning_rate": 9.649254625985126e-06,
      "loss": 0.8943,
      "step": 3281
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.384939201549452,
      "learning_rate": 9.648988790718647e-06,
      "loss": 0.8525,
      "step": 3282
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.01121459257376,
      "learning_rate": 9.648722858414712e-06,
      "loss": 0.8279,
      "step": 3283
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.318889196851436,
      "learning_rate": 9.648456829078866e-06,
      "loss": 0.8841,
      "step": 3284
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.081013441060796,
      "learning_rate": 9.648190702716664e-06,
      "loss": 0.8501,
      "step": 3285
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.808064504684433,
      "learning_rate": 9.647924479333661e-06,
      "loss": 0.8574,
      "step": 3286
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.902832492827207,
      "learning_rate": 9.647658158935412e-06,
      "loss": 0.9167,
      "step": 3287
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.723950472690446,
      "learning_rate": 9.64739174152748e-06,
      "loss": 0.8002,
      "step": 3288
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.875158382660294,
      "learning_rate": 9.64712522711542e-06,
      "loss": 0.9007,
      "step": 3289
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.067225216481924,
      "learning_rate": 9.646858615704798e-06,
      "loss": 0.9136,
      "step": 3290
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.213739952198843,
      "learning_rate": 9.64659190730118e-06,
      "loss": 0.8295,
      "step": 3291
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.218594232852929,
      "learning_rate": 9.646325101910132e-06,
      "loss": 0.7895,
      "step": 3292
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.311624435328392,
      "learning_rate": 9.646058199537221e-06,
      "loss": 0.8508,
      "step": 3293
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.814927968197268,
      "learning_rate": 9.645791200188021e-06,
      "loss": 0.8651,
      "step": 3294
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.466737169763425,
      "learning_rate": 9.645524103868105e-06,
      "loss": 0.7959,
      "step": 3295
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.340550626756768,
      "learning_rate": 9.645256910583045e-06,
      "loss": 0.7638,
      "step": 3296
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.5786509406678,
      "learning_rate": 9.644989620338421e-06,
      "loss": 0.7974,
      "step": 3297
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.799530810684751,
      "learning_rate": 9.644722233139811e-06,
      "loss": 0.7909,
      "step": 3298
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.406752796406822,
      "learning_rate": 9.644454748992795e-06,
      "loss": 0.8319,
      "step": 3299
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.84689073541873,
      "learning_rate": 9.644187167902958e-06,
      "loss": 0.7951,
      "step": 3300
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.757869804849618,
      "learning_rate": 9.643919489875885e-06,
      "loss": 0.8163,
      "step": 3301
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.152601867040951,
      "learning_rate": 9.643651714917161e-06,
      "loss": 0.7986,
      "step": 3302
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.972063663911529,
      "learning_rate": 9.643383843032378e-06,
      "loss": 0.9055,
      "step": 3303
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.3052503304995975,
      "learning_rate": 9.643115874227127e-06,
      "loss": 0.842,
      "step": 3304
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.469812637890127,
      "learning_rate": 9.642847808507e-06,
      "loss": 0.8654,
      "step": 3305
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.546746591786719,
      "learning_rate": 9.642579645877592e-06,
      "loss": 0.8524,
      "step": 3306
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.533281032679706,
      "learning_rate": 9.642311386344501e-06,
      "loss": 0.888,
      "step": 3307
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.643224084104299,
      "learning_rate": 9.642043029913327e-06,
      "loss": 0.8424,
      "step": 3308
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.7130529852478285,
      "learning_rate": 9.641774576589671e-06,
      "loss": 0.7892,
      "step": 3309
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.896989878956347,
      "learning_rate": 9.641506026379136e-06,
      "loss": 0.8847,
      "step": 3310
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.041873234756183,
      "learning_rate": 9.641237379287325e-06,
      "loss": 0.9059,
      "step": 3311
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.258076931982257,
      "learning_rate": 9.64096863531985e-06,
      "loss": 0.884,
      "step": 3312
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.843383546446361,
      "learning_rate": 9.640699794482318e-06,
      "loss": 0.839,
      "step": 3313
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.419672283985424,
      "learning_rate": 9.64043085678034e-06,
      "loss": 0.8385,
      "step": 3314
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.627641947321581,
      "learning_rate": 9.64016182221953e-06,
      "loss": 0.8717,
      "step": 3315
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.624852147561525,
      "learning_rate": 9.639892690805506e-06,
      "loss": 0.8068,
      "step": 3316
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.864658873493239,
      "learning_rate": 9.63962346254388e-06,
      "loss": 0.8386,
      "step": 3317
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.7338139283483915,
      "learning_rate": 9.639354137440278e-06,
      "loss": 0.8599,
      "step": 3318
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.758086011234576,
      "learning_rate": 9.639084715500316e-06,
      "loss": 0.7661,
      "step": 3319
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.852930876868596,
      "learning_rate": 9.63881519672962e-06,
      "loss": 0.8511,
      "step": 3320
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.9736862068141425,
      "learning_rate": 9.638545581133819e-06,
      "loss": 0.8602,
      "step": 3321
    },
    {
      "epoch": 0.3,
      "grad_norm": 11.82276474569912,
      "learning_rate": 9.638275868718534e-06,
      "loss": 0.8198,
      "step": 3322
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.771761479629931,
      "learning_rate": 9.638006059489397e-06,
      "loss": 0.8255,
      "step": 3323
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.279806030623002,
      "learning_rate": 9.637736153452044e-06,
      "loss": 0.8593,
      "step": 3324
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.737409577622434,
      "learning_rate": 9.637466150612103e-06,
      "loss": 0.8456,
      "step": 3325
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.9483347844199566,
      "learning_rate": 9.637196050975212e-06,
      "loss": 0.8414,
      "step": 3326
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.389414333049423,
      "learning_rate": 9.63692585454701e-06,
      "loss": 0.8499,
      "step": 3327
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.556196115502422,
      "learning_rate": 9.636655561333134e-06,
      "loss": 0.8956,
      "step": 3328
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.65732393623127,
      "learning_rate": 9.63638517133923e-06,
      "loss": 0.8797,
      "step": 3329
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.022036509560413,
      "learning_rate": 9.636114684570936e-06,
      "loss": 0.9197,
      "step": 3330
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.540375557093509,
      "learning_rate": 9.635844101033902e-06,
      "loss": 0.8004,
      "step": 3331
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.312068524439566,
      "learning_rate": 9.635573420733775e-06,
      "loss": 0.8872,
      "step": 3332
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.17558454806183,
      "learning_rate": 9.635302643676203e-06,
      "loss": 0.8655,
      "step": 3333
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.654042195661868,
      "learning_rate": 9.635031769866841e-06,
      "loss": 0.9031,
      "step": 3334
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.2121930830064835,
      "learning_rate": 9.63476079931134e-06,
      "loss": 0.8693,
      "step": 3335
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.5531078763406425,
      "learning_rate": 9.63448973201536e-06,
      "loss": 0.8366,
      "step": 3336
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.567827311141023,
      "learning_rate": 9.634218567984554e-06,
      "loss": 0.8829,
      "step": 3337
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.963964174311849,
      "learning_rate": 9.633947307224584e-06,
      "loss": 0.7813,
      "step": 3338
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.331547233935041,
      "learning_rate": 9.633675949741112e-06,
      "loss": 0.892,
      "step": 3339
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.068419907924525,
      "learning_rate": 9.633404495539805e-06,
      "loss": 0.8276,
      "step": 3340
    },
    {
      "epoch": 0.3,
      "grad_norm": 9.291975998275534,
      "learning_rate": 9.633132944626322e-06,
      "loss": 0.8635,
      "step": 3341
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.026333863081252,
      "learning_rate": 9.632861297006337e-06,
      "loss": 0.8926,
      "step": 3342
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.659741342317169,
      "learning_rate": 9.63258955268552e-06,
      "loss": 0.8415,
      "step": 3343
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.7727945925817,
      "learning_rate": 9.63231771166954e-06,
      "loss": 0.9476,
      "step": 3344
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.265032495582461,
      "learning_rate": 9.632045773964072e-06,
      "loss": 0.84,
      "step": 3345
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.930967365961253,
      "learning_rate": 9.631773739574793e-06,
      "loss": 0.7982,
      "step": 3346
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.790109417668356,
      "learning_rate": 9.631501608507379e-06,
      "loss": 0.822,
      "step": 3347
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.21076479784875,
      "learning_rate": 9.631229380767513e-06,
      "loss": 0.8805,
      "step": 3348
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.664679373058769,
      "learning_rate": 9.630957056360875e-06,
      "loss": 0.8874,
      "step": 3349
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.191654599892978,
      "learning_rate": 9.63068463529315e-06,
      "loss": 0.8527,
      "step": 3350
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.928166168223627,
      "learning_rate": 9.630412117570028e-06,
      "loss": 0.8199,
      "step": 3351
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.700043591230171,
      "learning_rate": 9.63013950319719e-06,
      "loss": 0.8265,
      "step": 3352
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.966426912297472,
      "learning_rate": 9.62986679218033e-06,
      "loss": 0.8032,
      "step": 3353
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.595881450642672,
      "learning_rate": 9.62959398452514e-06,
      "loss": 0.8813,
      "step": 3354
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.642856834016924,
      "learning_rate": 9.629321080237315e-06,
      "loss": 0.8582,
      "step": 3355
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.772016770497053,
      "learning_rate": 9.629048079322549e-06,
      "loss": 0.9213,
      "step": 3356
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.440615603525215,
      "learning_rate": 9.628774981786544e-06,
      "loss": 0.9015,
      "step": 3357
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.592831518786836,
      "learning_rate": 9.628501787634996e-06,
      "loss": 0.8974,
      "step": 3358
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.278952815241524,
      "learning_rate": 9.62822849687361e-06,
      "loss": 0.8932,
      "step": 3359
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.259218372906293,
      "learning_rate": 9.627955109508089e-06,
      "loss": 0.8614,
      "step": 3360
    },
    {
      "epoch": 0.3,
      "grad_norm": 10.030660706226964,
      "learning_rate": 9.627681625544142e-06,
      "loss": 0.8497,
      "step": 3361
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.7346740908553775,
      "learning_rate": 9.627408044987474e-06,
      "loss": 0.9084,
      "step": 3362
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.470426279547836,
      "learning_rate": 9.627134367843798e-06,
      "loss": 0.8905,
      "step": 3363
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.146327641158245,
      "learning_rate": 9.626860594118826e-06,
      "loss": 0.8707,
      "step": 3364
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.543680580533257,
      "learning_rate": 9.62658672381827e-06,
      "loss": 0.9132,
      "step": 3365
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.744839366650756,
      "learning_rate": 9.626312756947848e-06,
      "loss": 0.8274,
      "step": 3366
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.231240320506806,
      "learning_rate": 9.62603869351328e-06,
      "loss": 0.8854,
      "step": 3367
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.161760805925736,
      "learning_rate": 9.625764533520285e-06,
      "loss": 1.0146,
      "step": 3368
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.2850656129913824,
      "learning_rate": 9.625490276974585e-06,
      "loss": 0.8114,
      "step": 3369
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.175369543935384,
      "learning_rate": 9.625215923881905e-06,
      "loss": 0.7775,
      "step": 3370
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.800946895838132,
      "learning_rate": 9.624941474247973e-06,
      "loss": 0.8201,
      "step": 3371
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.55102228187858,
      "learning_rate": 9.624666928078515e-06,
      "loss": 0.9652,
      "step": 3372
    },
    {
      "epoch": 0.3,
      "grad_norm": 10.55409484785092,
      "learning_rate": 9.624392285379263e-06,
      "loss": 0.8431,
      "step": 3373
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.620122896290645,
      "learning_rate": 9.62411754615595e-06,
      "loss": 0.9095,
      "step": 3374
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.473874117005483,
      "learning_rate": 9.623842710414308e-06,
      "loss": 0.8687,
      "step": 3375
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.52348434570083,
      "learning_rate": 9.623567778160077e-06,
      "loss": 0.8433,
      "step": 3376
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.0668878966377715,
      "learning_rate": 9.623292749398993e-06,
      "loss": 0.8711,
      "step": 3377
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.470576058597409,
      "learning_rate": 9.623017624136799e-06,
      "loss": 0.9129,
      "step": 3378
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.683444069021747,
      "learning_rate": 9.622742402379234e-06,
      "loss": 0.8524,
      "step": 3379
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.615756000338311,
      "learning_rate": 9.622467084132048e-06,
      "loss": 0.8258,
      "step": 3380
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.043361183688054,
      "learning_rate": 9.622191669400983e-06,
      "loss": 0.8662,
      "step": 3381
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.268358447106379,
      "learning_rate": 9.621916158191789e-06,
      "loss": 0.8419,
      "step": 3382
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.450451396666109,
      "learning_rate": 9.621640550510218e-06,
      "loss": 0.7574,
      "step": 3383
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.992651651470137,
      "learning_rate": 9.621364846362023e-06,
      "loss": 0.8132,
      "step": 3384
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.277339903337183,
      "learning_rate": 9.621089045752954e-06,
      "loss": 0.78,
      "step": 3385
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.768864616516342,
      "learning_rate": 9.620813148688773e-06,
      "loss": 0.8698,
      "step": 3386
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.972507706359331,
      "learning_rate": 9.620537155175238e-06,
      "loss": 0.8485,
      "step": 3387
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.719492046496953,
      "learning_rate": 9.620261065218106e-06,
      "loss": 0.8954,
      "step": 3388
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.9614397086320725,
      "learning_rate": 9.619984878823145e-06,
      "loss": 0.8315,
      "step": 3389
    },
    {
      "epoch": 0.3,
      "grad_norm": 10.946851612448748,
      "learning_rate": 9.619708595996116e-06,
      "loss": 0.8787,
      "step": 3390
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.541238550805727,
      "learning_rate": 9.619432216742787e-06,
      "loss": 0.8401,
      "step": 3391
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.428930741635411,
      "learning_rate": 9.619155741068929e-06,
      "loss": 0.9019,
      "step": 3392
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.7662750656522475,
      "learning_rate": 9.618879168980308e-06,
      "loss": 0.872,
      "step": 3393
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.824038273913719,
      "learning_rate": 9.6186025004827e-06,
      "loss": 0.8158,
      "step": 3394
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.0066369173077625,
      "learning_rate": 9.61832573558188e-06,
      "loss": 0.8632,
      "step": 3395
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.918052776994631,
      "learning_rate": 9.618048874283623e-06,
      "loss": 0.9108,
      "step": 3396
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.850860855466185,
      "learning_rate": 9.61777191659371e-06,
      "loss": 0.8424,
      "step": 3397
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.748728067665785,
      "learning_rate": 9.617494862517922e-06,
      "loss": 0.8977,
      "step": 3398
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.642096426084905,
      "learning_rate": 9.61721771206204e-06,
      "loss": 0.8856,
      "step": 3399
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.487060401058544,
      "learning_rate": 9.61694046523185e-06,
      "loss": 0.9016,
      "step": 3400
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.191498690134695,
      "learning_rate": 9.616663122033139e-06,
      "loss": 0.8338,
      "step": 3401
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.58416574174315,
      "learning_rate": 9.616385682471693e-06,
      "loss": 0.7772,
      "step": 3402
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.332135260966037,
      "learning_rate": 9.61610814655331e-06,
      "loss": 0.8161,
      "step": 3403
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.917316674591485,
      "learning_rate": 9.615830514283776e-06,
      "loss": 0.8997,
      "step": 3404
    },
    {
      "epoch": 0.3,
      "grad_norm": 9.39296016235748,
      "learning_rate": 9.615552785668888e-06,
      "loss": 0.9031,
      "step": 3405
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.513347492140642,
      "learning_rate": 9.615274960714445e-06,
      "loss": 0.8885,
      "step": 3406
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.051032938346747,
      "learning_rate": 9.614997039426243e-06,
      "loss": 0.8879,
      "step": 3407
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.106094900155611,
      "learning_rate": 9.614719021810085e-06,
      "loss": 0.8076,
      "step": 3408
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.55306556161794,
      "learning_rate": 9.614440907871773e-06,
      "loss": 0.7935,
      "step": 3409
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.625618214025119,
      "learning_rate": 9.614162697617116e-06,
      "loss": 0.8569,
      "step": 3410
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.773199754625026,
      "learning_rate": 9.613884391051912e-06,
      "loss": 0.9128,
      "step": 3411
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.535228390514244,
      "learning_rate": 9.61360598818198e-06,
      "loss": 0.7988,
      "step": 3412
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.293915425492607,
      "learning_rate": 9.613327489013124e-06,
      "loss": 0.859,
      "step": 3413
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.548229995068532,
      "learning_rate": 9.61304889355116e-06,
      "loss": 0.7892,
      "step": 3414
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.282683917853358,
      "learning_rate": 9.612770201801905e-06,
      "loss": 0.8482,
      "step": 3415
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.9507552079744555,
      "learning_rate": 9.612491413771172e-06,
      "loss": 0.8861,
      "step": 3416
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.657796515927321,
      "learning_rate": 9.612212529464781e-06,
      "loss": 0.8761,
      "step": 3417
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.971418912424004,
      "learning_rate": 9.611933548888557e-06,
      "loss": 0.9099,
      "step": 3418
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.173698598289152,
      "learning_rate": 9.611654472048318e-06,
      "loss": 0.8989,
      "step": 3419
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.121841069723469,
      "learning_rate": 9.611375298949892e-06,
      "loss": 0.8276,
      "step": 3420
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.114944892276505,
      "learning_rate": 9.611096029599105e-06,
      "loss": 0.8687,
      "step": 3421
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.048883322055748,
      "learning_rate": 9.610816664001785e-06,
      "loss": 0.8776,
      "step": 3422
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.384000117718312,
      "learning_rate": 9.610537202163766e-06,
      "loss": 0.8181,
      "step": 3423
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.225639309710435,
      "learning_rate": 9.610257644090881e-06,
      "loss": 0.902,
      "step": 3424
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.227230702748475,
      "learning_rate": 9.609977989788963e-06,
      "loss": 0.8019,
      "step": 3425
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.163592062823973,
      "learning_rate": 9.60969823926385e-06,
      "loss": 0.8099,
      "step": 3426
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.001670570950575,
      "learning_rate": 9.60941839252138e-06,
      "loss": 0.771,
      "step": 3427
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.895884706004711,
      "learning_rate": 9.609138449567397e-06,
      "loss": 0.8404,
      "step": 3428
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.269332821667848,
      "learning_rate": 9.608858410407743e-06,
      "loss": 0.8495,
      "step": 3429
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.792041287272132,
      "learning_rate": 9.608578275048262e-06,
      "loss": 0.8602,
      "step": 3430
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.380643228161576,
      "learning_rate": 9.608298043494803e-06,
      "loss": 0.8435,
      "step": 3431
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.229900574347504,
      "learning_rate": 9.608017715753213e-06,
      "loss": 0.7892,
      "step": 3432
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.171037572613418,
      "learning_rate": 9.607737291829346e-06,
      "loss": 0.8797,
      "step": 3433
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.751272082337137,
      "learning_rate": 9.607456771729054e-06,
      "loss": 0.7965,
      "step": 3434
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.905273700507282,
      "learning_rate": 9.60717615545819e-06,
      "loss": 0.9135,
      "step": 3435
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.56859321769154,
      "learning_rate": 9.606895443022615e-06,
      "loss": 0.8424,
      "step": 3436
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.794086058787489,
      "learning_rate": 9.606614634428186e-06,
      "loss": 0.8683,
      "step": 3437
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.079769686661954,
      "learning_rate": 9.606333729680766e-06,
      "loss": 0.8742,
      "step": 3438
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.970446584919136,
      "learning_rate": 9.606052728786216e-06,
      "loss": 0.9455,
      "step": 3439
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.342584215525004,
      "learning_rate": 9.605771631750402e-06,
      "loss": 0.9009,
      "step": 3440
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.3417769737565965,
      "learning_rate": 9.605490438579192e-06,
      "loss": 0.8446,
      "step": 3441
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.950888431712724,
      "learning_rate": 9.605209149278456e-06,
      "loss": 0.8703,
      "step": 3442
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.591338590840988,
      "learning_rate": 9.604927763854063e-06,
      "loss": 0.8676,
      "step": 3443
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.438493641844304,
      "learning_rate": 9.604646282311887e-06,
      "loss": 0.8816,
      "step": 3444
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.725969056923775,
      "learning_rate": 9.604364704657806e-06,
      "loss": 0.8778,
      "step": 3445
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.543729130245039,
      "learning_rate": 9.604083030897694e-06,
      "loss": 0.8939,
      "step": 3446
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.499654552495176,
      "learning_rate": 9.603801261037432e-06,
      "loss": 0.7936,
      "step": 3447
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.748152230023837,
      "learning_rate": 9.603519395082898e-06,
      "loss": 0.8896,
      "step": 3448
    },
    {
      "epoch": 0.31,
      "grad_norm": 9.781860167701735,
      "learning_rate": 9.603237433039981e-06,
      "loss": 0.9457,
      "step": 3449
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.956923969345379,
      "learning_rate": 9.602955374914563e-06,
      "loss": 0.9545,
      "step": 3450
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.412439508867281,
      "learning_rate": 9.602673220712532e-06,
      "loss": 0.8682,
      "step": 3451
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.689709767197531,
      "learning_rate": 9.602390970439777e-06,
      "loss": 0.9145,
      "step": 3452
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.846444710629745,
      "learning_rate": 9.60210862410219e-06,
      "loss": 0.7755,
      "step": 3453
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.06589211119371,
      "learning_rate": 9.601826181705662e-06,
      "loss": 0.8784,
      "step": 3454
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.4469908659287345,
      "learning_rate": 9.601543643256092e-06,
      "loss": 0.8999,
      "step": 3455
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.8900432115264225,
      "learning_rate": 9.601261008759377e-06,
      "loss": 0.8767,
      "step": 3456
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.689289012532748,
      "learning_rate": 9.600978278221412e-06,
      "loss": 0.8465,
      "step": 3457
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.352669505701338,
      "learning_rate": 9.600695451648104e-06,
      "loss": 0.9459,
      "step": 3458
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.720708121195964,
      "learning_rate": 9.60041252904535e-06,
      "loss": 0.802,
      "step": 3459
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.288847250301714,
      "learning_rate": 9.600129510419063e-06,
      "loss": 0.8834,
      "step": 3460
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.298504657229926,
      "learning_rate": 9.599846395775145e-06,
      "loss": 0.8282,
      "step": 3461
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.901691860386791,
      "learning_rate": 9.599563185119509e-06,
      "loss": 0.8662,
      "step": 3462
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.305908656057918,
      "learning_rate": 9.599279878458062e-06,
      "loss": 0.8779,
      "step": 3463
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.470330919245296,
      "learning_rate": 9.59899647579672e-06,
      "loss": 0.8925,
      "step": 3464
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.041278595778447,
      "learning_rate": 9.598712977141399e-06,
      "loss": 0.8543,
      "step": 3465
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.772698524322452,
      "learning_rate": 9.598429382498016e-06,
      "loss": 0.8951,
      "step": 3466
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.363549125634146,
      "learning_rate": 9.598145691872488e-06,
      "loss": 0.8617,
      "step": 3467
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.013020003109283,
      "learning_rate": 9.59786190527074e-06,
      "loss": 0.8716,
      "step": 3468
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.276986824280207,
      "learning_rate": 9.597578022698695e-06,
      "loss": 0.9078,
      "step": 3469
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.899679535170715,
      "learning_rate": 9.597294044162276e-06,
      "loss": 0.841,
      "step": 3470
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.917423273737217,
      "learning_rate": 9.597009969667412e-06,
      "loss": 0.8716,
      "step": 3471
    },
    {
      "epoch": 0.31,
      "grad_norm": 10.93843433813374,
      "learning_rate": 9.59672579922003e-06,
      "loss": 0.8852,
      "step": 3472
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.781614828753245,
      "learning_rate": 9.596441532826066e-06,
      "loss": 0.8981,
      "step": 3473
    },
    {
      "epoch": 0.31,
      "grad_norm": 10.721843403661758,
      "learning_rate": 9.596157170491451e-06,
      "loss": 0.8888,
      "step": 3474
    },
    {
      "epoch": 0.31,
      "grad_norm": 9.009056533090284,
      "learning_rate": 9.59587271222212e-06,
      "loss": 0.9025,
      "step": 3475
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.540411852855542,
      "learning_rate": 9.595588158024012e-06,
      "loss": 0.8564,
      "step": 3476
    },
    {
      "epoch": 0.31,
      "grad_norm": 10.481661549739771,
      "learning_rate": 9.595303507903064e-06,
      "loss": 0.8731,
      "step": 3477
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.9004372436782,
      "learning_rate": 9.595018761865217e-06,
      "loss": 0.8173,
      "step": 3478
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.3169315298902795,
      "learning_rate": 9.594733919916419e-06,
      "loss": 0.8674,
      "step": 3479
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.712873194231846,
      "learning_rate": 9.594448982062614e-06,
      "loss": 0.9338,
      "step": 3480
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.935733517204914,
      "learning_rate": 9.594163948309745e-06,
      "loss": 0.9382,
      "step": 3481
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.4088970447402,
      "learning_rate": 9.593878818663767e-06,
      "loss": 0.8779,
      "step": 3482
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.686290836717649,
      "learning_rate": 9.593593593130627e-06,
      "loss": 0.8311,
      "step": 3483
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.493249908380524,
      "learning_rate": 9.593308271716283e-06,
      "loss": 0.8745,
      "step": 3484
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.034599733549633,
      "learning_rate": 9.593022854426686e-06,
      "loss": 0.8647,
      "step": 3485
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.431445491480448,
      "learning_rate": 9.592737341267796e-06,
      "loss": 0.8671,
      "step": 3486
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.143320709843091,
      "learning_rate": 9.59245173224557e-06,
      "loss": 0.8548,
      "step": 3487
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.892591827684415,
      "learning_rate": 9.592166027365974e-06,
      "loss": 0.8646,
      "step": 3488
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.62270628164334,
      "learning_rate": 9.591880226634969e-06,
      "loss": 0.8788,
      "step": 3489
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.567567247826698,
      "learning_rate": 9.591594330058517e-06,
      "loss": 0.8644,
      "step": 3490
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.2697946448700055,
      "learning_rate": 9.591308337642592e-06,
      "loss": 0.8208,
      "step": 3491
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.135293718266525,
      "learning_rate": 9.591022249393161e-06,
      "loss": 0.8274,
      "step": 3492
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.051272736692218,
      "learning_rate": 9.590736065316193e-06,
      "loss": 0.8949,
      "step": 3493
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.83517831624625,
      "learning_rate": 9.590449785417665e-06,
      "loss": 0.8228,
      "step": 3494
    },
    {
      "epoch": 0.31,
      "grad_norm": 11.528558542606188,
      "learning_rate": 9.590163409703547e-06,
      "loss": 0.8224,
      "step": 3495
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.23527683112531,
      "learning_rate": 9.589876938179824e-06,
      "loss": 0.8564,
      "step": 3496
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.773637246293891,
      "learning_rate": 9.58959037085247e-06,
      "loss": 0.8302,
      "step": 3497
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.226228439480646,
      "learning_rate": 9.589303707727466e-06,
      "loss": 0.8571,
      "step": 3498
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.064724556539138,
      "learning_rate": 9.5890169488108e-06,
      "loss": 0.8032,
      "step": 3499
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.066756078546219,
      "learning_rate": 9.588730094108453e-06,
      "loss": 0.9226,
      "step": 3500
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.561285042621237,
      "learning_rate": 9.588443143626415e-06,
      "loss": 0.9036,
      "step": 3501
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.0780105970633365,
      "learning_rate": 9.588156097370676e-06,
      "loss": 0.86,
      "step": 3502
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.370910609109139,
      "learning_rate": 9.587868955347225e-06,
      "loss": 0.8648,
      "step": 3503
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.685044625824857,
      "learning_rate": 9.58758171756206e-06,
      "loss": 0.8796,
      "step": 3504
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.454638883754281,
      "learning_rate": 9.58729438402117e-06,
      "loss": 0.7542,
      "step": 3505
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.207949946284218,
      "learning_rate": 9.587006954730557e-06,
      "loss": 0.8774,
      "step": 3506
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.18903139943031,
      "learning_rate": 9.586719429696219e-06,
      "loss": 0.8676,
      "step": 3507
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.407761451978195,
      "learning_rate": 9.586431808924157e-06,
      "loss": 0.8257,
      "step": 3508
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.6814902192315815,
      "learning_rate": 9.586144092420375e-06,
      "loss": 0.8532,
      "step": 3509
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.065459929800284,
      "learning_rate": 9.58585628019088e-06,
      "loss": 0.8995,
      "step": 3510
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.349327067724237,
      "learning_rate": 9.585568372241677e-06,
      "loss": 0.8536,
      "step": 3511
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.116770283818196,
      "learning_rate": 9.585280368578777e-06,
      "loss": 0.7847,
      "step": 3512
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.447525895038339,
      "learning_rate": 9.584992269208191e-06,
      "loss": 0.876,
      "step": 3513
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.160070681561613,
      "learning_rate": 9.584704074135931e-06,
      "loss": 0.8786,
      "step": 3514
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.991508798298955,
      "learning_rate": 9.584415783368015e-06,
      "loss": 0.8439,
      "step": 3515
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.078877639677518,
      "learning_rate": 9.584127396910458e-06,
      "loss": 0.8369,
      "step": 3516
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.986532875413086,
      "learning_rate": 9.583838914769283e-06,
      "loss": 0.8342,
      "step": 3517
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.734577723806855,
      "learning_rate": 9.583550336950507e-06,
      "loss": 0.8001,
      "step": 3518
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.078435101495751,
      "learning_rate": 9.583261663460157e-06,
      "loss": 0.9061,
      "step": 3519
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.104943835138198,
      "learning_rate": 9.582972894304255e-06,
      "loss": 0.8464,
      "step": 3520
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.760774102832016,
      "learning_rate": 9.582684029488832e-06,
      "loss": 0.8567,
      "step": 3521
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.577183366463439,
      "learning_rate": 9.582395069019915e-06,
      "loss": 0.8804,
      "step": 3522
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.083501739213018,
      "learning_rate": 9.582106012903536e-06,
      "loss": 0.9069,
      "step": 3523
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.178221645753052,
      "learning_rate": 9.58181686114573e-06,
      "loss": 0.8398,
      "step": 3524
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.657797459968471,
      "learning_rate": 9.58152761375253e-06,
      "loss": 0.8762,
      "step": 3525
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.0415751617880105,
      "learning_rate": 9.581238270729976e-06,
      "loss": 0.8285,
      "step": 3526
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.658327797583851,
      "learning_rate": 9.580948832084103e-06,
      "loss": 0.8062,
      "step": 3527
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.398126484986254,
      "learning_rate": 9.580659297820958e-06,
      "loss": 0.801,
      "step": 3528
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.335328780702546,
      "learning_rate": 9.580369667946578e-06,
      "loss": 0.8987,
      "step": 3529
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.775932283323086,
      "learning_rate": 9.580079942467015e-06,
      "loss": 0.8799,
      "step": 3530
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.02179323497502,
      "learning_rate": 9.579790121388312e-06,
      "loss": 0.8606,
      "step": 3531
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.448164874991882,
      "learning_rate": 9.579500204716522e-06,
      "loss": 0.878,
      "step": 3532
    },
    {
      "epoch": 0.32,
      "grad_norm": 9.465945541540695,
      "learning_rate": 9.579210192457694e-06,
      "loss": 0.8331,
      "step": 3533
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.092231595299681,
      "learning_rate": 9.578920084617879e-06,
      "loss": 0.8322,
      "step": 3534
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.960821820322847,
      "learning_rate": 9.578629881203137e-06,
      "loss": 0.8479,
      "step": 3535
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.185038301974877,
      "learning_rate": 9.57833958221952e-06,
      "loss": 0.8046,
      "step": 3536
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.913534608168803,
      "learning_rate": 9.578049187673093e-06,
      "loss": 0.8453,
      "step": 3537
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.117089176969306,
      "learning_rate": 9.577758697569916e-06,
      "loss": 0.8244,
      "step": 3538
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.8022435831414265,
      "learning_rate": 9.577468111916051e-06,
      "loss": 0.7794,
      "step": 3539
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.7594133363091276,
      "learning_rate": 9.577177430717562e-06,
      "loss": 0.8762,
      "step": 3540
    },
    {
      "epoch": 0.32,
      "grad_norm": 9.365907390698078,
      "learning_rate": 9.57688665398052e-06,
      "loss": 0.8585,
      "step": 3541
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.228988452911693,
      "learning_rate": 9.576595781710992e-06,
      "loss": 0.8735,
      "step": 3542
    },
    {
      "epoch": 0.32,
      "grad_norm": 9.117395832446263,
      "learning_rate": 9.576304813915048e-06,
      "loss": 0.8257,
      "step": 3543
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.7967095319313255,
      "learning_rate": 9.576013750598763e-06,
      "loss": 0.7856,
      "step": 3544
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.879915063171007,
      "learning_rate": 9.575722591768212e-06,
      "loss": 0.8825,
      "step": 3545
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.127377578624026,
      "learning_rate": 9.575431337429476e-06,
      "loss": 0.8597,
      "step": 3546
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.494929632308291,
      "learning_rate": 9.575139987588628e-06,
      "loss": 0.8052,
      "step": 3547
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.071826386100276,
      "learning_rate": 9.574848542251751e-06,
      "loss": 0.8832,
      "step": 3548
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.909086387468175,
      "learning_rate": 9.574557001424932e-06,
      "loss": 0.8079,
      "step": 3549
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.543895788913973,
      "learning_rate": 9.574265365114253e-06,
      "loss": 0.9742,
      "step": 3550
    },
    {
      "epoch": 0.32,
      "grad_norm": 10.261507511817912,
      "learning_rate": 9.5739736333258e-06,
      "loss": 0.8194,
      "step": 3551
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.178597692573573,
      "learning_rate": 9.573681806065666e-06,
      "loss": 0.8389,
      "step": 3552
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.136957961370826,
      "learning_rate": 9.573389883339943e-06,
      "loss": 0.8928,
      "step": 3553
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.103130609146673,
      "learning_rate": 9.573097865154717e-06,
      "loss": 0.7965,
      "step": 3554
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.531100683969125,
      "learning_rate": 9.57280575151609e-06,
      "loss": 0.9365,
      "step": 3555
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.438707807336985,
      "learning_rate": 9.572513542430158e-06,
      "loss": 0.9404,
      "step": 3556
    },
    {
      "epoch": 0.32,
      "grad_norm": 12.510956189121323,
      "learning_rate": 9.572221237903019e-06,
      "loss": 0.9042,
      "step": 3557
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.71511973466415,
      "learning_rate": 9.571928837940774e-06,
      "loss": 0.8138,
      "step": 3558
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.530865009003993,
      "learning_rate": 9.571636342549528e-06,
      "loss": 0.8626,
      "step": 3559
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.3570619655240455,
      "learning_rate": 9.571343751735385e-06,
      "loss": 0.8249,
      "step": 3560
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.5808934796384735,
      "learning_rate": 9.571051065504451e-06,
      "loss": 0.8639,
      "step": 3561
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.5879357594156325,
      "learning_rate": 9.570758283862838e-06,
      "loss": 0.8361,
      "step": 3562
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.434025515774722,
      "learning_rate": 9.570465406816653e-06,
      "loss": 0.8593,
      "step": 3563
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.080492888510507,
      "learning_rate": 9.570172434372014e-06,
      "loss": 0.8955,
      "step": 3564
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.817578576454439,
      "learning_rate": 9.569879366535033e-06,
      "loss": 0.9307,
      "step": 3565
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.4998311905496235,
      "learning_rate": 9.56958620331183e-06,
      "loss": 0.7963,
      "step": 3566
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.9537449062458885,
      "learning_rate": 9.56929294470852e-06,
      "loss": 0.931,
      "step": 3567
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.349300873220086,
      "learning_rate": 9.568999590731226e-06,
      "loss": 0.8261,
      "step": 3568
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.821407050283483,
      "learning_rate": 9.568706141386072e-06,
      "loss": 0.9506,
      "step": 3569
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.383183807424078,
      "learning_rate": 9.568412596679184e-06,
      "loss": 0.894,
      "step": 3570
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.541549333089124,
      "learning_rate": 9.568118956616688e-06,
      "loss": 0.8499,
      "step": 3571
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.981076109066628,
      "learning_rate": 9.567825221204711e-06,
      "loss": 0.8281,
      "step": 3572
    },
    {
      "epoch": 0.32,
      "grad_norm": 11.476418726736789,
      "learning_rate": 9.567531390449387e-06,
      "loss": 0.872,
      "step": 3573
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.82834843583265,
      "learning_rate": 9.567237464356847e-06,
      "loss": 0.8483,
      "step": 3574
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.752497297451695,
      "learning_rate": 9.566943442933228e-06,
      "loss": 0.8874,
      "step": 3575
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.661386617341038,
      "learning_rate": 9.566649326184665e-06,
      "loss": 0.8321,
      "step": 3576
    },
    {
      "epoch": 0.32,
      "grad_norm": 9.408791812347566,
      "learning_rate": 9.566355114117299e-06,
      "loss": 0.8781,
      "step": 3577
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.105298346332253,
      "learning_rate": 9.56606080673727e-06,
      "loss": 0.8238,
      "step": 3578
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.3069398204274565,
      "learning_rate": 9.56576640405072e-06,
      "loss": 0.8696,
      "step": 3579
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.785982028185541,
      "learning_rate": 9.565471906063797e-06,
      "loss": 0.889,
      "step": 3580
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.628102150510248,
      "learning_rate": 9.565177312782645e-06,
      "loss": 0.8361,
      "step": 3581
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.254412522562947,
      "learning_rate": 9.564882624213414e-06,
      "loss": 0.9199,
      "step": 3582
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.811736623323117,
      "learning_rate": 9.564587840362255e-06,
      "loss": 0.7547,
      "step": 3583
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.902416913932424,
      "learning_rate": 9.564292961235322e-06,
      "loss": 0.9145,
      "step": 3584
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.710749511392299,
      "learning_rate": 9.56399798683877e-06,
      "loss": 0.8359,
      "step": 3585
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.317837393414583,
      "learning_rate": 9.563702917178753e-06,
      "loss": 0.82,
      "step": 3586
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.379340663770807,
      "learning_rate": 9.563407752261432e-06,
      "loss": 0.8441,
      "step": 3587
    },
    {
      "epoch": 0.32,
      "grad_norm": 12.964687228627342,
      "learning_rate": 9.56311249209297e-06,
      "loss": 0.9542,
      "step": 3588
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.874905957816876,
      "learning_rate": 9.562817136679526e-06,
      "loss": 0.7692,
      "step": 3589
    },
    {
      "epoch": 0.32,
      "grad_norm": 9.550593555768234,
      "learning_rate": 9.562521686027266e-06,
      "loss": 0.8194,
      "step": 3590
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.340986892956327,
      "learning_rate": 9.56222614014236e-06,
      "loss": 0.8947,
      "step": 3591
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.444066094587974,
      "learning_rate": 9.561930499030973e-06,
      "loss": 0.9147,
      "step": 3592
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.535042506238342,
      "learning_rate": 9.561634762699277e-06,
      "loss": 0.8043,
      "step": 3593
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.351937957246336,
      "learning_rate": 9.561338931153446e-06,
      "loss": 0.8672,
      "step": 3594
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.730848813544495,
      "learning_rate": 9.561043004399653e-06,
      "loss": 0.8443,
      "step": 3595
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.256581815710109,
      "learning_rate": 9.560746982444075e-06,
      "loss": 0.9205,
      "step": 3596
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.545959349596419,
      "learning_rate": 9.560450865292895e-06,
      "loss": 0.8474,
      "step": 3597
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.3934775079334045,
      "learning_rate": 9.560154652952288e-06,
      "loss": 0.8384,
      "step": 3598
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.750547701404012,
      "learning_rate": 9.559858345428438e-06,
      "loss": 0.8978,
      "step": 3599
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.317195834458686,
      "learning_rate": 9.559561942727534e-06,
      "loss": 0.9396,
      "step": 3600
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.285283119095217,
      "learning_rate": 9.559265444855757e-06,
      "loss": 0.8576,
      "step": 3601
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.50972893020817,
      "learning_rate": 9.558968851819299e-06,
      "loss": 0.8267,
      "step": 3602
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.35375508649762,
      "learning_rate": 9.558672163624352e-06,
      "loss": 0.8351,
      "step": 3603
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.357580069865971,
      "learning_rate": 9.558375380277105e-06,
      "loss": 0.872,
      "step": 3604
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.516381986570741,
      "learning_rate": 9.558078501783754e-06,
      "loss": 0.8076,
      "step": 3605
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.53746657289533,
      "learning_rate": 9.557781528150498e-06,
      "loss": 0.8097,
      "step": 3606
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.145415457239077,
      "learning_rate": 9.557484459383534e-06,
      "loss": 0.9258,
      "step": 3607
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.018001378222648,
      "learning_rate": 9.557187295489061e-06,
      "loss": 0.8673,
      "step": 3608
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.102729899828129,
      "learning_rate": 9.556890036473285e-06,
      "loss": 0.7436,
      "step": 3609
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.739893018175834,
      "learning_rate": 9.55659268234241e-06,
      "loss": 0.9437,
      "step": 3610
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.11421551326104,
      "learning_rate": 9.556295233102639e-06,
      "loss": 0.8122,
      "step": 3611
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.011533995494063,
      "learning_rate": 9.555997688760184e-06,
      "loss": 0.7962,
      "step": 3612
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.955880461251037,
      "learning_rate": 9.555700049321255e-06,
      "loss": 0.8828,
      "step": 3613
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.694181517246303,
      "learning_rate": 9.555402314792064e-06,
      "loss": 0.8835,
      "step": 3614
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.690225477390795,
      "learning_rate": 9.555104485178826e-06,
      "loss": 0.8627,
      "step": 3615
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.840025375498851,
      "learning_rate": 9.554806560487758e-06,
      "loss": 0.8871,
      "step": 3616
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.505621260247724,
      "learning_rate": 9.554508540725076e-06,
      "loss": 0.8452,
      "step": 3617
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.76189597737123,
      "learning_rate": 9.554210425897004e-06,
      "loss": 0.9474,
      "step": 3618
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.190951763224067,
      "learning_rate": 9.553912216009762e-06,
      "loss": 0.7859,
      "step": 3619
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.380855987495822,
      "learning_rate": 9.553613911069575e-06,
      "loss": 0.8881,
      "step": 3620
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.0780666386543505,
      "learning_rate": 9.553315511082672e-06,
      "loss": 0.8641,
      "step": 3621
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.357535309015798,
      "learning_rate": 9.553017016055277e-06,
      "loss": 0.8591,
      "step": 3622
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.9332422059054775,
      "learning_rate": 9.552718425993623e-06,
      "loss": 0.852,
      "step": 3623
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.566532311851289,
      "learning_rate": 9.552419740903945e-06,
      "loss": 0.8524,
      "step": 3624
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.242189253068217,
      "learning_rate": 9.55212096079247e-06,
      "loss": 0.8566,
      "step": 3625
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.614162139022515,
      "learning_rate": 9.551822085665442e-06,
      "loss": 0.8595,
      "step": 3626
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.3168672333218625,
      "learning_rate": 9.551523115529095e-06,
      "loss": 0.8494,
      "step": 3627
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.214722240220039,
      "learning_rate": 9.551224050389671e-06,
      "loss": 0.8829,
      "step": 3628
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.39593903947895,
      "learning_rate": 9.550924890253414e-06,
      "loss": 0.8124,
      "step": 3629
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.202938227182617,
      "learning_rate": 9.550625635126564e-06,
      "loss": 0.9282,
      "step": 3630
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.7481690130726095,
      "learning_rate": 9.55032628501537e-06,
      "loss": 0.8321,
      "step": 3631
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.916545475208054,
      "learning_rate": 9.55002683992608e-06,
      "loss": 0.8357,
      "step": 3632
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.613298054409408,
      "learning_rate": 9.549727299864944e-06,
      "loss": 0.8546,
      "step": 3633
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.96381170049042,
      "learning_rate": 9.549427664838213e-06,
      "loss": 0.9277,
      "step": 3634
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.012144668974965,
      "learning_rate": 9.549127934852144e-06,
      "loss": 0.869,
      "step": 3635
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.41290605306973,
      "learning_rate": 9.548828109912993e-06,
      "loss": 0.8657,
      "step": 3636
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.849696741824696,
      "learning_rate": 9.548528190027015e-06,
      "loss": 0.8095,
      "step": 3637
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.918520928842044,
      "learning_rate": 9.548228175200474e-06,
      "loss": 0.9114,
      "step": 3638
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.904793029193729,
      "learning_rate": 9.547928065439628e-06,
      "loss": 0.8066,
      "step": 3639
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.943050747839899,
      "learning_rate": 9.547627860750744e-06,
      "loss": 0.8553,
      "step": 3640
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.536200899930148,
      "learning_rate": 9.547327561140088e-06,
      "loss": 0.8643,
      "step": 3641
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.791893664429239,
      "learning_rate": 9.547027166613928e-06,
      "loss": 0.9466,
      "step": 3642
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.012767063053217,
      "learning_rate": 9.546726677178532e-06,
      "loss": 0.8739,
      "step": 3643
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.877191253181346,
      "learning_rate": 9.546426092840176e-06,
      "loss": 0.7888,
      "step": 3644
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.872271780293323,
      "learning_rate": 9.546125413605131e-06,
      "loss": 0.9118,
      "step": 3645
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.588376972617259,
      "learning_rate": 9.545824639479675e-06,
      "loss": 0.8647,
      "step": 3646
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.931808483366734,
      "learning_rate": 9.545523770470083e-06,
      "loss": 0.8781,
      "step": 3647
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.184685313949505,
      "learning_rate": 9.545222806582636e-06,
      "loss": 0.8876,
      "step": 3648
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.895307968677732,
      "learning_rate": 9.54492174782362e-06,
      "loss": 0.8829,
      "step": 3649
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.968562103586833,
      "learning_rate": 9.544620594199314e-06,
      "loss": 0.8332,
      "step": 3650
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.707689824882312,
      "learning_rate": 9.544319345716006e-06,
      "loss": 0.8467,
      "step": 3651
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.023819707374252,
      "learning_rate": 9.544018002379983e-06,
      "loss": 0.8958,
      "step": 3652
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.007150097353445,
      "learning_rate": 9.543716564197537e-06,
      "loss": 0.8053,
      "step": 3653
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.5800666245311605,
      "learning_rate": 9.543415031174955e-06,
      "loss": 0.8399,
      "step": 3654
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.701517811843559,
      "learning_rate": 9.543113403318535e-06,
      "loss": 0.9074,
      "step": 3655
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.725607151905608,
      "learning_rate": 9.542811680634575e-06,
      "loss": 0.8839,
      "step": 3656
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.344530155934459,
      "learning_rate": 9.542509863129367e-06,
      "loss": 0.8068,
      "step": 3657
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.829513063575374,
      "learning_rate": 9.542207950809214e-06,
      "loss": 0.8008,
      "step": 3658
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.813158808999898,
      "learning_rate": 9.541905943680417e-06,
      "loss": 0.861,
      "step": 3659
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.338683059159795,
      "learning_rate": 9.541603841749281e-06,
      "loss": 0.8966,
      "step": 3660
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.209107833960832,
      "learning_rate": 9.54130164502211e-06,
      "loss": 0.9108,
      "step": 3661
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.904739315771883,
      "learning_rate": 9.540999353505214e-06,
      "loss": 0.8716,
      "step": 3662
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.298362237321713,
      "learning_rate": 9.540696967204899e-06,
      "loss": 0.7907,
      "step": 3663
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.6745802384293285,
      "learning_rate": 9.54039448612748e-06,
      "loss": 0.8326,
      "step": 3664
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.608860018078317,
      "learning_rate": 9.540091910279268e-06,
      "loss": 0.8998,
      "step": 3665
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.010272846414624,
      "learning_rate": 9.539789239666581e-06,
      "loss": 0.9593,
      "step": 3666
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.238395071988777,
      "learning_rate": 9.539486474295735e-06,
      "loss": 0.8193,
      "step": 3667
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.4991890676154975,
      "learning_rate": 9.539183614173053e-06,
      "loss": 0.8596,
      "step": 3668
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.2773830361348315,
      "learning_rate": 9.538880659304852e-06,
      "loss": 0.8184,
      "step": 3669
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.959664184711099,
      "learning_rate": 9.538577609697455e-06,
      "loss": 0.8946,
      "step": 3670
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.554411934066708,
      "learning_rate": 9.538274465357193e-06,
      "loss": 0.847,
      "step": 3671
    },
    {
      "epoch": 0.33,
      "grad_norm": 10.087580267241716,
      "learning_rate": 9.537971226290388e-06,
      "loss": 0.9063,
      "step": 3672
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.175930450932842,
      "learning_rate": 9.537667892503372e-06,
      "loss": 0.8998,
      "step": 3673
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.761344346556277,
      "learning_rate": 9.537364464002476e-06,
      "loss": 0.8898,
      "step": 3674
    },
    {
      "epoch": 0.33,
      "grad_norm": 9.022367507919409,
      "learning_rate": 9.537060940794034e-06,
      "loss": 0.7735,
      "step": 3675
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.071179829433664,
      "learning_rate": 9.53675732288438e-06,
      "loss": 0.8973,
      "step": 3676
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.003399852503,
      "learning_rate": 9.536453610279855e-06,
      "loss": 0.8321,
      "step": 3677
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.27473447232985,
      "learning_rate": 9.536149802986791e-06,
      "loss": 0.8484,
      "step": 3678
    },
    {
      "epoch": 0.33,
      "grad_norm": 10.299095992301412,
      "learning_rate": 9.535845901011537e-06,
      "loss": 0.8339,
      "step": 3679
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.410998409590012,
      "learning_rate": 9.535541904360433e-06,
      "loss": 0.8886,
      "step": 3680
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.617820983313239,
      "learning_rate": 9.535237813039824e-06,
      "loss": 0.8071,
      "step": 3681
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.10709690413607,
      "learning_rate": 9.534933627056058e-06,
      "loss": 0.7827,
      "step": 3682
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.0568007676573155,
      "learning_rate": 9.534629346415482e-06,
      "loss": 0.8318,
      "step": 3683
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.26450949763047,
      "learning_rate": 9.534324971124452e-06,
      "loss": 0.9278,
      "step": 3684
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.243416457858234,
      "learning_rate": 9.534020501189316e-06,
      "loss": 0.8398,
      "step": 3685
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.454239366548421,
      "learning_rate": 9.533715936616432e-06,
      "loss": 0.7901,
      "step": 3686
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.541788799792835,
      "learning_rate": 9.533411277412156e-06,
      "loss": 0.8215,
      "step": 3687
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.814785213599591,
      "learning_rate": 9.533106523582849e-06,
      "loss": 0.8822,
      "step": 3688
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.551371584659849,
      "learning_rate": 9.532801675134869e-06,
      "loss": 0.8078,
      "step": 3689
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.370040477074983,
      "learning_rate": 9.53249673207458e-06,
      "loss": 0.7974,
      "step": 3690
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.129265484975013,
      "learning_rate": 9.532191694408351e-06,
      "loss": 0.8853,
      "step": 3691
    },
    {
      "epoch": 0.33,
      "grad_norm": 10.629380055947278,
      "learning_rate": 9.531886562142542e-06,
      "loss": 0.8701,
      "step": 3692
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.101324212214007,
      "learning_rate": 9.531581335283528e-06,
      "loss": 0.8469,
      "step": 3693
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.032133990008763,
      "learning_rate": 9.531276013837678e-06,
      "loss": 0.829,
      "step": 3694
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.761453891895764,
      "learning_rate": 9.530970597811365e-06,
      "loss": 0.8449,
      "step": 3695
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.210846345511031,
      "learning_rate": 9.530665087210962e-06,
      "loss": 0.8748,
      "step": 3696
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.896821472840307,
      "learning_rate": 9.530359482042846e-06,
      "loss": 0.8706,
      "step": 3697
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.947541877224841,
      "learning_rate": 9.5300537823134e-06,
      "loss": 0.8099,
      "step": 3698
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.278245781553129,
      "learning_rate": 9.529747988029e-06,
      "loss": 0.8913,
      "step": 3699
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.262844261150006,
      "learning_rate": 9.52944209919603e-06,
      "loss": 0.9055,
      "step": 3700
    },
    {
      "epoch": 0.33,
      "grad_norm": 9.792405740057614,
      "learning_rate": 9.529136115820878e-06,
      "loss": 0.8476,
      "step": 3701
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.33032553613948,
      "learning_rate": 9.528830037909927e-06,
      "loss": 0.8378,
      "step": 3702
    },
    {
      "epoch": 0.33,
      "grad_norm": 9.868030506736863,
      "learning_rate": 9.528523865469566e-06,
      "loss": 0.8506,
      "step": 3703
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.425065562257276,
      "learning_rate": 9.528217598506189e-06,
      "loss": 0.7908,
      "step": 3704
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.855087504312761,
      "learning_rate": 9.527911237026184e-06,
      "loss": 0.8505,
      "step": 3705
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.4114711282143695,
      "learning_rate": 9.52760478103595e-06,
      "loss": 0.892,
      "step": 3706
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.934746273380926,
      "learning_rate": 9.527298230541879e-06,
      "loss": 0.8119,
      "step": 3707
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.485316085226549,
      "learning_rate": 9.526991585550372e-06,
      "loss": 0.828,
      "step": 3708
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.047054290539383,
      "learning_rate": 9.52668484606783e-06,
      "loss": 0.8175,
      "step": 3709
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.604450584497912,
      "learning_rate": 9.526378012100657e-06,
      "loss": 0.8588,
      "step": 3710
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.529257932331767,
      "learning_rate": 9.526071083655255e-06,
      "loss": 0.8614,
      "step": 3711
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.183911066760248,
      "learning_rate": 9.52576406073803e-06,
      "loss": 0.8601,
      "step": 3712
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.370669633130773,
      "learning_rate": 9.525456943355393e-06,
      "loss": 0.8721,
      "step": 3713
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.027485464209958,
      "learning_rate": 9.525149731513752e-06,
      "loss": 0.8801,
      "step": 3714
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.84320924935301,
      "learning_rate": 9.524842425219519e-06,
      "loss": 0.9022,
      "step": 3715
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.030887965649005,
      "learning_rate": 9.52453502447911e-06,
      "loss": 0.7968,
      "step": 3716
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.86208440511362,
      "learning_rate": 9.524227529298942e-06,
      "loss": 0.8382,
      "step": 3717
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.380756114451333,
      "learning_rate": 9.523919939685433e-06,
      "loss": 0.9008,
      "step": 3718
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.970636895730147,
      "learning_rate": 9.523612255645e-06,
      "loss": 0.8659,
      "step": 3719
    },
    {
      "epoch": 0.33,
      "grad_norm": 9.035561723833426,
      "learning_rate": 9.52330447718407e-06,
      "loss": 0.8289,
      "step": 3720
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.576031975913975,
      "learning_rate": 9.522996604309065e-06,
      "loss": 0.8906,
      "step": 3721
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.126273131961999,
      "learning_rate": 9.52268863702641e-06,
      "loss": 0.8673,
      "step": 3722
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.989672233155252,
      "learning_rate": 9.522380575342534e-06,
      "loss": 0.8657,
      "step": 3723
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.739046111690475,
      "learning_rate": 9.522072419263867e-06,
      "loss": 0.7641,
      "step": 3724
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.077397089351855,
      "learning_rate": 9.521764168796842e-06,
      "loss": 0.8907,
      "step": 3725
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.106916185813112,
      "learning_rate": 9.521455823947893e-06,
      "loss": 0.8885,
      "step": 3726
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.12483961363923,
      "learning_rate": 9.521147384723455e-06,
      "loss": 0.8842,
      "step": 3727
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.309374569211711,
      "learning_rate": 9.520838851129965e-06,
      "loss": 0.8781,
      "step": 3728
    },
    {
      "epoch": 0.33,
      "grad_norm": 11.05057475917475,
      "learning_rate": 9.520530223173868e-06,
      "loss": 0.8296,
      "step": 3729
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.67937120316523,
      "learning_rate": 9.5202215008616e-06,
      "loss": 0.8343,
      "step": 3730
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.934933139457643,
      "learning_rate": 9.519912684199608e-06,
      "loss": 0.8484,
      "step": 3731
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.875917074405749,
      "learning_rate": 9.519603773194335e-06,
      "loss": 0.8276,
      "step": 3732
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.326086630867792,
      "learning_rate": 9.519294767852234e-06,
      "loss": 0.8375,
      "step": 3733
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.813928287659874,
      "learning_rate": 9.51898566817975e-06,
      "loss": 0.8669,
      "step": 3734
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.6909444791732255,
      "learning_rate": 9.518676474183337e-06,
      "loss": 0.8502,
      "step": 3735
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.506939653731695,
      "learning_rate": 9.518367185869449e-06,
      "loss": 0.8095,
      "step": 3736
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.884874803178075,
      "learning_rate": 9.518057803244541e-06,
      "loss": 0.9271,
      "step": 3737
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.489366685503393,
      "learning_rate": 9.517748326315069e-06,
      "loss": 0.874,
      "step": 3738
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.234189803602368,
      "learning_rate": 9.517438755087496e-06,
      "loss": 0.7762,
      "step": 3739
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.5263246865537,
      "learning_rate": 9.517129089568283e-06,
      "loss": 0.8799,
      "step": 3740
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.03703581512222,
      "learning_rate": 9.516819329763891e-06,
      "loss": 0.8676,
      "step": 3741
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.086825876688838,
      "learning_rate": 9.516509475680787e-06,
      "loss": 0.8225,
      "step": 3742
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.138970682632438,
      "learning_rate": 9.516199527325438e-06,
      "loss": 0.8124,
      "step": 3743
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.41225374783552,
      "learning_rate": 9.515889484704317e-06,
      "loss": 0.899,
      "step": 3744
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.560385359151254,
      "learning_rate": 9.51557934782389e-06,
      "loss": 0.9314,
      "step": 3745
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.247378635443559,
      "learning_rate": 9.515269116690634e-06,
      "loss": 0.8463,
      "step": 3746
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.461447238197859,
      "learning_rate": 9.514958791311023e-06,
      "loss": 0.7991,
      "step": 3747
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.183569608367944,
      "learning_rate": 9.514648371691537e-06,
      "loss": 0.8631,
      "step": 3748
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.8211217104102175,
      "learning_rate": 9.51433785783865e-06,
      "loss": 0.898,
      "step": 3749
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.014793092797788,
      "learning_rate": 9.514027249758848e-06,
      "loss": 0.887,
      "step": 3750
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.481200404598645,
      "learning_rate": 9.513716547458614e-06,
      "loss": 0.8796,
      "step": 3751
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.769288352484387,
      "learning_rate": 9.513405750944428e-06,
      "loss": 0.828,
      "step": 3752
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.358787818373338,
      "learning_rate": 9.513094860222784e-06,
      "loss": 0.8256,
      "step": 3753
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.153201421312751,
      "learning_rate": 9.51278387530017e-06,
      "loss": 0.8475,
      "step": 3754
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.191054295442497,
      "learning_rate": 9.512472796183072e-06,
      "loss": 0.8489,
      "step": 3755
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.615478994081103,
      "learning_rate": 9.512161622877989e-06,
      "loss": 0.8439,
      "step": 3756
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.952598738946193,
      "learning_rate": 9.511850355391412e-06,
      "loss": 0.8245,
      "step": 3757
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.652171037892268,
      "learning_rate": 9.51153899372984e-06,
      "loss": 0.8267,
      "step": 3758
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.846863457111111,
      "learning_rate": 9.511227537899773e-06,
      "loss": 0.8539,
      "step": 3759
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.273774034812369,
      "learning_rate": 9.510915987907708e-06,
      "loss": 0.8751,
      "step": 3760
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.9542678378893585,
      "learning_rate": 9.510604343760152e-06,
      "loss": 0.8645,
      "step": 3761
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.461339285576556,
      "learning_rate": 9.510292605463609e-06,
      "loss": 0.8497,
      "step": 3762
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.050102990424352,
      "learning_rate": 9.509980773024583e-06,
      "loss": 0.8516,
      "step": 3763
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.951712935379132,
      "learning_rate": 9.509668846449586e-06,
      "loss": 0.8364,
      "step": 3764
    },
    {
      "epoch": 0.34,
      "grad_norm": 8.404082498973393,
      "learning_rate": 9.50935682574513e-06,
      "loss": 0.8997,
      "step": 3765
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.514381336972579,
      "learning_rate": 9.509044710917723e-06,
      "loss": 0.8305,
      "step": 3766
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.397477389880263,
      "learning_rate": 9.508732501973883e-06,
      "loss": 0.871,
      "step": 3767
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.895512096232699,
      "learning_rate": 9.508420198920128e-06,
      "loss": 0.7947,
      "step": 3768
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.188782382227938,
      "learning_rate": 9.508107801762972e-06,
      "loss": 0.8238,
      "step": 3769
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.368487492040881,
      "learning_rate": 9.507795310508938e-06,
      "loss": 0.7848,
      "step": 3770
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.368776565094237,
      "learning_rate": 9.50748272516455e-06,
      "loss": 0.9058,
      "step": 3771
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.112407427607459,
      "learning_rate": 9.50717004573633e-06,
      "loss": 0.8901,
      "step": 3772
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.798450863140684,
      "learning_rate": 9.506857272230808e-06,
      "loss": 0.9335,
      "step": 3773
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.792386108033275,
      "learning_rate": 9.50654440465451e-06,
      "loss": 0.9043,
      "step": 3774
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.875447161592311,
      "learning_rate": 9.506231443013966e-06,
      "loss": 0.833,
      "step": 3775
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.70724910875346,
      "learning_rate": 9.50591838731571e-06,
      "loss": 0.9261,
      "step": 3776
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.035179483827681,
      "learning_rate": 9.505605237566273e-06,
      "loss": 0.857,
      "step": 3777
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.681895221381081,
      "learning_rate": 9.505291993772195e-06,
      "loss": 0.9033,
      "step": 3778
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.59832825554872,
      "learning_rate": 9.504978655940013e-06,
      "loss": 0.8348,
      "step": 3779
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.95465567398358,
      "learning_rate": 9.504665224076268e-06,
      "loss": 0.8337,
      "step": 3780
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.780315846969243,
      "learning_rate": 9.504351698187503e-06,
      "loss": 0.8737,
      "step": 3781
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.620895034451064,
      "learning_rate": 9.504038078280257e-06,
      "loss": 0.8567,
      "step": 3782
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.054188871599694,
      "learning_rate": 9.503724364361083e-06,
      "loss": 0.8781,
      "step": 3783
    },
    {
      "epoch": 0.34,
      "grad_norm": 8.477316506862064,
      "learning_rate": 9.503410556436523e-06,
      "loss": 0.7884,
      "step": 3784
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.511885021916745,
      "learning_rate": 9.503096654513132e-06,
      "loss": 0.827,
      "step": 3785
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.898500794400286,
      "learning_rate": 9.50278265859746e-06,
      "loss": 0.8118,
      "step": 3786
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.776420018689311,
      "learning_rate": 9.502468568696059e-06,
      "loss": 0.8617,
      "step": 3787
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.3048458265796015,
      "learning_rate": 9.50215438481549e-06,
      "loss": 0.7822,
      "step": 3788
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.5657949447840185,
      "learning_rate": 9.501840106962306e-06,
      "loss": 0.7977,
      "step": 3789
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.728054767775824,
      "learning_rate": 9.501525735143068e-06,
      "loss": 0.8375,
      "step": 3790
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.457081430010932,
      "learning_rate": 9.50121126936434e-06,
      "loss": 0.8477,
      "step": 3791
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.0782433011071335,
      "learning_rate": 9.500896709632682e-06,
      "loss": 0.8529,
      "step": 3792
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.7552304511393215,
      "learning_rate": 9.500582055954664e-06,
      "loss": 0.8775,
      "step": 3793
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.92949003381526,
      "learning_rate": 9.50026730833685e-06,
      "loss": 0.8454,
      "step": 3794
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.422065188794628,
      "learning_rate": 9.499952466785813e-06,
      "loss": 0.8162,
      "step": 3795
    },
    {
      "epoch": 0.34,
      "grad_norm": 8.169358958917492,
      "learning_rate": 9.49963753130812e-06,
      "loss": 0.8635,
      "step": 3796
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.513608321270964,
      "learning_rate": 9.499322501910351e-06,
      "loss": 0.8637,
      "step": 3797
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.110536145709867,
      "learning_rate": 9.499007378599076e-06,
      "loss": 0.8116,
      "step": 3798
    },
    {
      "epoch": 0.34,
      "grad_norm": 11.534080314436258,
      "learning_rate": 9.498692161380877e-06,
      "loss": 0.8809,
      "step": 3799
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.608406478039386,
      "learning_rate": 9.49837685026233e-06,
      "loss": 0.938,
      "step": 3800
    },
    {
      "epoch": 0.34,
      "grad_norm": 8.037031172188538,
      "learning_rate": 9.498061445250017e-06,
      "loss": 0.8658,
      "step": 3801
    },
    {
      "epoch": 0.34,
      "grad_norm": 8.032756886062943,
      "learning_rate": 9.497745946350524e-06,
      "loss": 0.821,
      "step": 3802
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.360978976709623,
      "learning_rate": 9.497430353570431e-06,
      "loss": 0.8721,
      "step": 3803
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.339875951814523,
      "learning_rate": 9.49711466691633e-06,
      "loss": 0.8387,
      "step": 3804
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.333001569965068,
      "learning_rate": 9.496798886394809e-06,
      "loss": 0.8053,
      "step": 3805
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.210439779830197,
      "learning_rate": 9.496483012012458e-06,
      "loss": 0.766,
      "step": 3806
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.373852368358451,
      "learning_rate": 9.496167043775873e-06,
      "loss": 0.8654,
      "step": 3807
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.889944431966112,
      "learning_rate": 9.495850981691645e-06,
      "loss": 0.8893,
      "step": 3808
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.911679220822672,
      "learning_rate": 9.495534825766376e-06,
      "loss": 0.8482,
      "step": 3809
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.322465434979383,
      "learning_rate": 9.495218576006663e-06,
      "loss": 0.859,
      "step": 3810
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.413139185272793,
      "learning_rate": 9.494902232419103e-06,
      "loss": 0.7437,
      "step": 3811
    },
    {
      "epoch": 0.34,
      "grad_norm": 9.590724933742544,
      "learning_rate": 9.494585795010304e-06,
      "loss": 0.8508,
      "step": 3812
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.965268361289983,
      "learning_rate": 9.494269263786871e-06,
      "loss": 0.8621,
      "step": 3813
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.8286335956164255,
      "learning_rate": 9.493952638755409e-06,
      "loss": 0.896,
      "step": 3814
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.972433183208948,
      "learning_rate": 9.493635919922529e-06,
      "loss": 0.8459,
      "step": 3815
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.904377352205877,
      "learning_rate": 9.493319107294835e-06,
      "loss": 0.8647,
      "step": 3816
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.1348445320086675,
      "learning_rate": 9.49300220087895e-06,
      "loss": 0.8368,
      "step": 3817
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.758166138928909,
      "learning_rate": 9.49268520068148e-06,
      "loss": 0.8088,
      "step": 3818
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.646194225752188,
      "learning_rate": 9.492368106709046e-06,
      "loss": 0.8055,
      "step": 3819
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.034540016930402,
      "learning_rate": 9.492050918968267e-06,
      "loss": 0.8128,
      "step": 3820
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.297212488908784,
      "learning_rate": 9.491733637465764e-06,
      "loss": 0.852,
      "step": 3821
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.368007037607074,
      "learning_rate": 9.491416262208155e-06,
      "loss": 0.8101,
      "step": 3822
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.9641992071244685,
      "learning_rate": 9.491098793202068e-06,
      "loss": 0.8775,
      "step": 3823
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.755214162181776,
      "learning_rate": 9.49078123045413e-06,
      "loss": 0.9137,
      "step": 3824
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.266613587890417,
      "learning_rate": 9.490463573970969e-06,
      "loss": 0.8324,
      "step": 3825
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.430524432709173,
      "learning_rate": 9.490145823759213e-06,
      "loss": 0.8731,
      "step": 3826
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.379278897926894,
      "learning_rate": 9.489827979825496e-06,
      "loss": 0.8396,
      "step": 3827
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.044732234523021,
      "learning_rate": 9.489510042176456e-06,
      "loss": 0.8541,
      "step": 3828
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.209323934333235,
      "learning_rate": 9.489192010818722e-06,
      "loss": 0.8351,
      "step": 3829
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.394154234455542,
      "learning_rate": 9.488873885758937e-06,
      "loss": 0.9217,
      "step": 3830
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.441841477215282,
      "learning_rate": 9.48855566700374e-06,
      "loss": 0.8094,
      "step": 3831
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.442577976455668,
      "learning_rate": 9.488237354559775e-06,
      "loss": 0.8283,
      "step": 3832
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.3672605070953345,
      "learning_rate": 9.487918948433683e-06,
      "loss": 0.8289,
      "step": 3833
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.257439576060358,
      "learning_rate": 9.48760044863211e-06,
      "loss": 0.9499,
      "step": 3834
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.734184023413846,
      "learning_rate": 9.487281855161707e-06,
      "loss": 0.8488,
      "step": 3835
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.130516886262024,
      "learning_rate": 9.486963168029122e-06,
      "loss": 0.8018,
      "step": 3836
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.437399026959671,
      "learning_rate": 9.486644387241006e-06,
      "loss": 0.8625,
      "step": 3837
    },
    {
      "epoch": 0.34,
      "grad_norm": 10.136175968527017,
      "learning_rate": 9.486325512804015e-06,
      "loss": 0.8726,
      "step": 3838
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.3728614915417126,
      "learning_rate": 9.486006544724805e-06,
      "loss": 0.9571,
      "step": 3839
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.081438764102822,
      "learning_rate": 9.48568748301003e-06,
      "loss": 0.8895,
      "step": 3840
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.208663225690133,
      "learning_rate": 9.485368327666354e-06,
      "loss": 0.8454,
      "step": 3841
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.539375474706128,
      "learning_rate": 9.485049078700437e-06,
      "loss": 0.8759,
      "step": 3842
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.8562878039738,
      "learning_rate": 9.484729736118943e-06,
      "loss": 0.8098,
      "step": 3843
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.524471618434453,
      "learning_rate": 9.484410299928535e-06,
      "loss": 0.8821,
      "step": 3844
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.670055824816685,
      "learning_rate": 9.484090770135885e-06,
      "loss": 0.7858,
      "step": 3845
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.684532640585301,
      "learning_rate": 9.483771146747658e-06,
      "loss": 0.8081,
      "step": 3846
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.216452262121781,
      "learning_rate": 9.483451429770529e-06,
      "loss": 0.8153,
      "step": 3847
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.599131626390742,
      "learning_rate": 9.48313161921117e-06,
      "loss": 0.7956,
      "step": 3848
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.409077961262756,
      "learning_rate": 9.482811715076256e-06,
      "loss": 0.8395,
      "step": 3849
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.107259224834636,
      "learning_rate": 9.482491717372464e-06,
      "loss": 0.8102,
      "step": 3850
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.236455514075973,
      "learning_rate": 9.482171626106476e-06,
      "loss": 0.7661,
      "step": 3851
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.133823257787381,
      "learning_rate": 9.481851441284967e-06,
      "loss": 0.8189,
      "step": 3852
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.418793699895365,
      "learning_rate": 9.481531162914628e-06,
      "loss": 0.8624,
      "step": 3853
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.352080205285083,
      "learning_rate": 9.48121079100214e-06,
      "loss": 0.876,
      "step": 3854
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.229177211555696,
      "learning_rate": 9.48089032555419e-06,
      "loss": 0.8822,
      "step": 3855
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.648797124318408,
      "learning_rate": 9.480569766577466e-06,
      "loss": 0.929,
      "step": 3856
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.090931849841327,
      "learning_rate": 9.480249114078662e-06,
      "loss": 0.8021,
      "step": 3857
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.858880982888204,
      "learning_rate": 9.47992836806447e-06,
      "loss": 0.8663,
      "step": 3858
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.321479211496744,
      "learning_rate": 9.479607528541582e-06,
      "loss": 0.8924,
      "step": 3859
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.222967023981757,
      "learning_rate": 9.479286595516698e-06,
      "loss": 0.898,
      "step": 3860
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.285195753771031,
      "learning_rate": 9.478965568996516e-06,
      "loss": 0.9163,
      "step": 3861
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.754916364535729,
      "learning_rate": 9.478644448987737e-06,
      "loss": 0.8194,
      "step": 3862
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.656211267799222,
      "learning_rate": 9.478323235497062e-06,
      "loss": 0.8295,
      "step": 3863
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.502014022138221,
      "learning_rate": 9.478001928531197e-06,
      "loss": 0.8849,
      "step": 3864
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.495767554558469,
      "learning_rate": 9.47768052809685e-06,
      "loss": 0.8495,
      "step": 3865
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.243738926776204,
      "learning_rate": 9.477359034200727e-06,
      "loss": 0.8424,
      "step": 3866
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.6702341436533885,
      "learning_rate": 9.477037446849538e-06,
      "loss": 0.848,
      "step": 3867
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.887446484326033,
      "learning_rate": 9.476715766049997e-06,
      "loss": 0.9434,
      "step": 3868
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.944062244856191,
      "learning_rate": 9.47639399180882e-06,
      "loss": 0.8621,
      "step": 3869
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.306804907096448,
      "learning_rate": 9.47607212413272e-06,
      "loss": 0.8789,
      "step": 3870
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.449146271247235,
      "learning_rate": 9.47575016302842e-06,
      "loss": 0.8102,
      "step": 3871
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.773554480955309,
      "learning_rate": 9.475428108502633e-06,
      "loss": 0.8819,
      "step": 3872
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.612145213437585,
      "learning_rate": 9.475105960562088e-06,
      "loss": 0.8216,
      "step": 3873
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.4768192578436246,
      "learning_rate": 9.474783719213505e-06,
      "loss": 0.8365,
      "step": 3874
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.837452456051207,
      "learning_rate": 9.474461384463612e-06,
      "loss": 0.9091,
      "step": 3875
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.640339434296615,
      "learning_rate": 9.474138956319136e-06,
      "loss": 0.8162,
      "step": 3876
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.486134291606943,
      "learning_rate": 9.473816434786807e-06,
      "loss": 0.8462,
      "step": 3877
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.628469275086714,
      "learning_rate": 9.473493819873358e-06,
      "loss": 0.8338,
      "step": 3878
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.088135000964081,
      "learning_rate": 9.473171111585522e-06,
      "loss": 0.8548,
      "step": 3879
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.168878844813929,
      "learning_rate": 9.472848309930036e-06,
      "loss": 0.8006,
      "step": 3880
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.953254628962278,
      "learning_rate": 9.472525414913637e-06,
      "loss": 0.7236,
      "step": 3881
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.1051727031882175,
      "learning_rate": 9.472202426543064e-06,
      "loss": 0.8983,
      "step": 3882
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.693922117372038,
      "learning_rate": 9.471879344825059e-06,
      "loss": 0.8917,
      "step": 3883
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.8261310338093315,
      "learning_rate": 9.471556169766367e-06,
      "loss": 0.8016,
      "step": 3884
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.847072377642449,
      "learning_rate": 9.471232901373732e-06,
      "loss": 0.834,
      "step": 3885
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.269041846266168,
      "learning_rate": 9.4709095396539e-06,
      "loss": 0.8539,
      "step": 3886
    },
    {
      "epoch": 0.35,
      "grad_norm": 8.407160949736248,
      "learning_rate": 9.470586084613627e-06,
      "loss": 0.9528,
      "step": 3887
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.625429327017089,
      "learning_rate": 9.470262536259655e-06,
      "loss": 0.8272,
      "step": 3888
    },
    {
      "epoch": 0.35,
      "grad_norm": 8.137311813709639,
      "learning_rate": 9.469938894598744e-06,
      "loss": 0.9183,
      "step": 3889
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.106720971149278,
      "learning_rate": 9.46961515963765e-06,
      "loss": 0.8345,
      "step": 3890
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.230130636976884,
      "learning_rate": 9.469291331383126e-06,
      "loss": 0.8055,
      "step": 3891
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.681413279294933,
      "learning_rate": 9.468967409841933e-06,
      "loss": 0.8317,
      "step": 3892
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.987934213115087,
      "learning_rate": 9.46864339502083e-06,
      "loss": 0.8281,
      "step": 3893
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.518269841764804,
      "learning_rate": 9.468319286926586e-06,
      "loss": 0.8164,
      "step": 3894
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.076778144447679,
      "learning_rate": 9.46799508556596e-06,
      "loss": 0.8665,
      "step": 3895
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.287434071373366,
      "learning_rate": 9.467670790945723e-06,
      "loss": 0.8614,
      "step": 3896
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.466698001605577,
      "learning_rate": 9.467346403072641e-06,
      "loss": 0.8689,
      "step": 3897
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.052599027321478,
      "learning_rate": 9.467021921953488e-06,
      "loss": 0.8313,
      "step": 3898
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.099035528307151,
      "learning_rate": 9.466697347595032e-06,
      "loss": 0.8008,
      "step": 3899
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.508056946263018,
      "learning_rate": 9.466372680004052e-06,
      "loss": 0.9037,
      "step": 3900
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.152639120302343,
      "learning_rate": 9.466047919187324e-06,
      "loss": 0.8671,
      "step": 3901
    },
    {
      "epoch": 0.35,
      "grad_norm": 8.185106288406503,
      "learning_rate": 9.465723065151626e-06,
      "loss": 0.8937,
      "step": 3902
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.584100976154333,
      "learning_rate": 9.465398117903739e-06,
      "loss": 0.8267,
      "step": 3903
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.74324567754363,
      "learning_rate": 9.465073077450444e-06,
      "loss": 0.8947,
      "step": 3904
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.440556881118943,
      "learning_rate": 9.46474794379853e-06,
      "loss": 0.8921,
      "step": 3905
    },
    {
      "epoch": 0.35,
      "grad_norm": 8.67447912555166,
      "learning_rate": 9.464422716954777e-06,
      "loss": 0.8318,
      "step": 3906
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.617937706789082,
      "learning_rate": 9.464097396925978e-06,
      "loss": 0.8725,
      "step": 3907
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.557031987121199,
      "learning_rate": 9.463771983718923e-06,
      "loss": 0.8042,
      "step": 3908
    },
    {
      "epoch": 0.35,
      "grad_norm": 8.404633023299303,
      "learning_rate": 9.4634464773404e-06,
      "loss": 0.8732,
      "step": 3909
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.138552905280347,
      "learning_rate": 9.46312087779721e-06,
      "loss": 0.8243,
      "step": 3910
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.836074439865496,
      "learning_rate": 9.462795185096145e-06,
      "loss": 0.8562,
      "step": 3911
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.96404453043018,
      "learning_rate": 9.462469399244002e-06,
      "loss": 0.8596,
      "step": 3912
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.787811332256712,
      "learning_rate": 9.462143520247587e-06,
      "loss": 0.8963,
      "step": 3913
    },
    {
      "epoch": 0.35,
      "grad_norm": 8.384375595573509,
      "learning_rate": 9.461817548113695e-06,
      "loss": 0.8843,
      "step": 3914
    },
    {
      "epoch": 0.35,
      "grad_norm": 8.678717487563077,
      "learning_rate": 9.461491482849132e-06,
      "loss": 0.9014,
      "step": 3915
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.321754818890408,
      "learning_rate": 9.461165324460706e-06,
      "loss": 0.8253,
      "step": 3916
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.6736905745726345,
      "learning_rate": 9.460839072955225e-06,
      "loss": 0.8508,
      "step": 3917
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.73069128183645,
      "learning_rate": 9.460512728339494e-06,
      "loss": 0.7263,
      "step": 3918
    },
    {
      "epoch": 0.35,
      "grad_norm": 8.319258225568124,
      "learning_rate": 9.460186290620333e-06,
      "loss": 0.7863,
      "step": 3919
    },
    {
      "epoch": 0.35,
      "grad_norm": 10.123839665898322,
      "learning_rate": 9.459859759804548e-06,
      "loss": 0.8575,
      "step": 3920
    },
    {
      "epoch": 0.35,
      "grad_norm": 8.056663942780128,
      "learning_rate": 9.459533135898955e-06,
      "loss": 0.8744,
      "step": 3921
    },
    {
      "epoch": 0.35,
      "grad_norm": 8.082430694296736,
      "learning_rate": 9.459206418910379e-06,
      "loss": 0.8321,
      "step": 3922
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.376350513831872,
      "learning_rate": 9.45887960884563e-06,
      "loss": 0.8779,
      "step": 3923
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.290170522772144,
      "learning_rate": 9.458552705711538e-06,
      "loss": 0.9171,
      "step": 3924
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.835131439946,
      "learning_rate": 9.45822570951492e-06,
      "loss": 0.8194,
      "step": 3925
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.287700501614024,
      "learning_rate": 9.457898620262604e-06,
      "loss": 0.8075,
      "step": 3926
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.262028027939857,
      "learning_rate": 9.457571437961417e-06,
      "loss": 0.8757,
      "step": 3927
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.714173363732083,
      "learning_rate": 9.457244162618186e-06,
      "loss": 0.9068,
      "step": 3928
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.989295916553715,
      "learning_rate": 9.456916794239748e-06,
      "loss": 0.8778,
      "step": 3929
    },
    {
      "epoch": 0.35,
      "grad_norm": 8.014936835365715,
      "learning_rate": 9.45658933283293e-06,
      "loss": 0.8642,
      "step": 3930
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.29945266045268,
      "learning_rate": 9.45626177840457e-06,
      "loss": 0.8858,
      "step": 3931
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.513904953646336,
      "learning_rate": 9.455934130961506e-06,
      "loss": 0.8666,
      "step": 3932
    },
    {
      "epoch": 0.35,
      "grad_norm": 8.26905056966033,
      "learning_rate": 9.455606390510574e-06,
      "loss": 0.82,
      "step": 3933
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.328995088873731,
      "learning_rate": 9.455278557058616e-06,
      "loss": 0.7873,
      "step": 3934
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.427441203792534,
      "learning_rate": 9.454950630612473e-06,
      "loss": 0.8561,
      "step": 3935
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.406495305097177,
      "learning_rate": 9.454622611178996e-06,
      "loss": 0.8611,
      "step": 3936
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.815865845149242,
      "learning_rate": 9.454294498765024e-06,
      "loss": 0.8997,
      "step": 3937
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.390519770857665,
      "learning_rate": 9.453966293377412e-06,
      "loss": 0.861,
      "step": 3938
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.614263355867397,
      "learning_rate": 9.453637995023007e-06,
      "loss": 0.8336,
      "step": 3939
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.894388570662001,
      "learning_rate": 9.45330960370866e-06,
      "loss": 0.875,
      "step": 3940
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.3401102991867475,
      "learning_rate": 9.45298111944123e-06,
      "loss": 0.8091,
      "step": 3941
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.8320690873332985,
      "learning_rate": 9.452652542227569e-06,
      "loss": 0.8453,
      "step": 3942
    },
    {
      "epoch": 0.35,
      "grad_norm": 8.21498563699384,
      "learning_rate": 9.452323872074538e-06,
      "loss": 0.8932,
      "step": 3943
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.047918737543299,
      "learning_rate": 9.451995108988996e-06,
      "loss": 0.9228,
      "step": 3944
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.504261842911873,
      "learning_rate": 9.451666252977808e-06,
      "loss": 0.7907,
      "step": 3945
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.9333248734641595,
      "learning_rate": 9.451337304047835e-06,
      "loss": 0.7688,
      "step": 3946
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.676185204890987,
      "learning_rate": 9.451008262205944e-06,
      "loss": 0.92,
      "step": 3947
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.737546669856385,
      "learning_rate": 9.450679127459002e-06,
      "loss": 0.874,
      "step": 3948
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.889043653571835,
      "learning_rate": 9.45034989981388e-06,
      "loss": 0.8584,
      "step": 3949
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.959899715404131,
      "learning_rate": 9.45002057927745e-06,
      "loss": 0.8355,
      "step": 3950
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.180190793483549,
      "learning_rate": 9.449691165856587e-06,
      "loss": 0.9,
      "step": 3951
    },
    {
      "epoch": 0.35,
      "grad_norm": 8.196966054626714,
      "learning_rate": 9.449361659558166e-06,
      "loss": 0.8395,
      "step": 3952
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.0892171164671725,
      "learning_rate": 9.449032060389063e-06,
      "loss": 0.7685,
      "step": 3953
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.997038607116231,
      "learning_rate": 9.448702368356159e-06,
      "loss": 0.9043,
      "step": 3954
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.907717472019158,
      "learning_rate": 9.448372583466336e-06,
      "loss": 0.8638,
      "step": 3955
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.744117114891816,
      "learning_rate": 9.448042705726478e-06,
      "loss": 0.7931,
      "step": 3956
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.635957525537869,
      "learning_rate": 9.447712735143467e-06,
      "loss": 0.7995,
      "step": 3957
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.7305436574088855,
      "learning_rate": 9.447382671724194e-06,
      "loss": 0.8262,
      "step": 3958
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.946360872762408,
      "learning_rate": 9.447052515475548e-06,
      "loss": 0.7889,
      "step": 3959
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.627178413902884,
      "learning_rate": 9.446722266404419e-06,
      "loss": 0.7874,
      "step": 3960
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.272772597654644,
      "learning_rate": 9.446391924517702e-06,
      "loss": 0.8981,
      "step": 3961
    },
    {
      "epoch": 0.35,
      "grad_norm": 8.220749775988565,
      "learning_rate": 9.44606148982229e-06,
      "loss": 0.9034,
      "step": 3962
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.63553930795546,
      "learning_rate": 9.445730962325082e-06,
      "loss": 0.8283,
      "step": 3963
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.188851486320986,
      "learning_rate": 9.445400342032974e-06,
      "loss": 0.8134,
      "step": 3964
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.7029971490227815,
      "learning_rate": 9.44506962895287e-06,
      "loss": 0.7972,
      "step": 3965
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.737148247603911,
      "learning_rate": 9.444738823091673e-06,
      "loss": 0.7919,
      "step": 3966
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.838934210180715,
      "learning_rate": 9.444407924456287e-06,
      "loss": 0.754,
      "step": 3967
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.621725259075146,
      "learning_rate": 9.444076933053618e-06,
      "loss": 0.8708,
      "step": 3968
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.884161656451834,
      "learning_rate": 9.443745848890576e-06,
      "loss": 0.8689,
      "step": 3969
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.542759466017919,
      "learning_rate": 9.44341467197407e-06,
      "loss": 0.7931,
      "step": 3970
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.85715058140632,
      "learning_rate": 9.443083402311015e-06,
      "loss": 0.8654,
      "step": 3971
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.9354989976947055,
      "learning_rate": 9.442752039908322e-06,
      "loss": 0.8734,
      "step": 3972
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.448150035195055,
      "learning_rate": 9.442420584772913e-06,
      "loss": 0.7504,
      "step": 3973
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.916869412370929,
      "learning_rate": 9.4420890369117e-06,
      "loss": 0.7766,
      "step": 3974
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.92938572394969,
      "learning_rate": 9.441757396331607e-06,
      "loss": 0.9213,
      "step": 3975
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.119004313595765,
      "learning_rate": 9.441425663039557e-06,
      "loss": 0.8697,
      "step": 3976
    },
    {
      "epoch": 0.35,
      "grad_norm": 8.179017902798371,
      "learning_rate": 9.441093837042472e-06,
      "loss": 0.8843,
      "step": 3977
    },
    {
      "epoch": 0.35,
      "grad_norm": 8.548589110546544,
      "learning_rate": 9.440761918347278e-06,
      "loss": 0.8762,
      "step": 3978
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.470191206036765,
      "learning_rate": 9.440429906960906e-06,
      "loss": 0.7797,
      "step": 3979
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.519880221685968,
      "learning_rate": 9.440097802890282e-06,
      "loss": 0.8563,
      "step": 3980
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.349072442323493,
      "learning_rate": 9.439765606142342e-06,
      "loss": 0.7744,
      "step": 3981
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.743323057944755,
      "learning_rate": 9.439433316724017e-06,
      "loss": 0.8231,
      "step": 3982
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.57929494929673,
      "learning_rate": 9.439100934642242e-06,
      "loss": 0.8499,
      "step": 3983
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.68304987096151,
      "learning_rate": 9.43876845990396e-06,
      "loss": 0.9037,
      "step": 3984
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.666602331399481,
      "learning_rate": 9.438435892516104e-06,
      "loss": 0.9182,
      "step": 3985
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.606894555627416,
      "learning_rate": 9.438103232485621e-06,
      "loss": 0.8489,
      "step": 3986
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.336517150725684,
      "learning_rate": 9.43777047981945e-06,
      "loss": 0.8058,
      "step": 3987
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.555900508916189,
      "learning_rate": 9.43743763452454e-06,
      "loss": 0.8181,
      "step": 3988
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.5296009184470405,
      "learning_rate": 9.437104696607837e-06,
      "loss": 0.8611,
      "step": 3989
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.498124953347789,
      "learning_rate": 9.436771666076293e-06,
      "loss": 0.9013,
      "step": 3990
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.418509529091909,
      "learning_rate": 9.436438542936853e-06,
      "loss": 0.8241,
      "step": 3991
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.539649129308645,
      "learning_rate": 9.436105327196476e-06,
      "loss": 0.7957,
      "step": 3992
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.391126348787983,
      "learning_rate": 9.435772018862115e-06,
      "loss": 0.8031,
      "step": 3993
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.796072302541556,
      "learning_rate": 9.435438617940727e-06,
      "loss": 0.891,
      "step": 3994
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.561878494604155,
      "learning_rate": 9.435105124439271e-06,
      "loss": 0.8645,
      "step": 3995
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.807892638957509,
      "learning_rate": 9.43477153836471e-06,
      "loss": 0.8148,
      "step": 3996
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.892666251783934,
      "learning_rate": 9.434437859724003e-06,
      "loss": 0.778,
      "step": 3997
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.745435229355464,
      "learning_rate": 9.434104088524119e-06,
      "loss": 0.8569,
      "step": 3998
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.745898461485092,
      "learning_rate": 9.433770224772021e-06,
      "loss": 0.847,
      "step": 3999
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.32879181917969,
      "learning_rate": 9.433436268474679e-06,
      "loss": 0.8067,
      "step": 4000
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.519874437644737,
      "learning_rate": 9.433102219639065e-06,
      "loss": 0.8879,
      "step": 4001
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.67992441824841,
      "learning_rate": 9.43276807827215e-06,
      "loss": 0.8189,
      "step": 4002
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.824068570876983,
      "learning_rate": 9.432433844380909e-06,
      "loss": 0.9759,
      "step": 4003
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.506669151432578,
      "learning_rate": 9.432099517972319e-06,
      "loss": 0.8657,
      "step": 4004
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.191633773161874,
      "learning_rate": 9.431765099053357e-06,
      "loss": 0.8149,
      "step": 4005
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.751026432575095,
      "learning_rate": 9.431430587631006e-06,
      "loss": 0.8386,
      "step": 4006
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.0664615573512215,
      "learning_rate": 9.431095983712243e-06,
      "loss": 0.9023,
      "step": 4007
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.298937982726953,
      "learning_rate": 9.430761287304056e-06,
      "loss": 0.8731,
      "step": 4008
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.456315490546447,
      "learning_rate": 9.430426498413432e-06,
      "loss": 0.8445,
      "step": 4009
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.823697748346717,
      "learning_rate": 9.430091617047356e-06,
      "loss": 0.8736,
      "step": 4010
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.311627957312553,
      "learning_rate": 9.429756643212819e-06,
      "loss": 0.9419,
      "step": 4011
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.228108486166968,
      "learning_rate": 9.429421576916814e-06,
      "loss": 0.876,
      "step": 4012
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.277530322099935,
      "learning_rate": 9.429086418166333e-06,
      "loss": 0.9067,
      "step": 4013
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.259770464309468,
      "learning_rate": 9.428751166968372e-06,
      "loss": 0.8577,
      "step": 4014
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.447599638900323,
      "learning_rate": 9.428415823329932e-06,
      "loss": 0.9179,
      "step": 4015
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.647898893221981,
      "learning_rate": 9.428080387258008e-06,
      "loss": 0.8631,
      "step": 4016
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.188434513448295,
      "learning_rate": 9.427744858759602e-06,
      "loss": 0.8974,
      "step": 4017
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.448741924141476,
      "learning_rate": 9.42740923784172e-06,
      "loss": 0.8599,
      "step": 4018
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.343073688701032,
      "learning_rate": 9.427073524511366e-06,
      "loss": 0.8388,
      "step": 4019
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.589558677677655,
      "learning_rate": 9.426737718775547e-06,
      "loss": 0.9158,
      "step": 4020
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.982896349748364,
      "learning_rate": 9.426401820641274e-06,
      "loss": 0.8615,
      "step": 4021
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.239567514540737,
      "learning_rate": 9.426065830115553e-06,
      "loss": 0.7917,
      "step": 4022
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.7573849156429775,
      "learning_rate": 9.425729747205404e-06,
      "loss": 0.8345,
      "step": 4023
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.801016334765673,
      "learning_rate": 9.425393571917837e-06,
      "loss": 0.8447,
      "step": 4024
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.48706002201493,
      "learning_rate": 9.425057304259872e-06,
      "loss": 0.8647,
      "step": 4025
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.438639425551503,
      "learning_rate": 9.424720944238527e-06,
      "loss": 0.8553,
      "step": 4026
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.309202031210394,
      "learning_rate": 9.424384491860821e-06,
      "loss": 0.8567,
      "step": 4027
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.143033970027757,
      "learning_rate": 9.424047947133778e-06,
      "loss": 0.8256,
      "step": 4028
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.966334445423384,
      "learning_rate": 9.423711310064423e-06,
      "loss": 0.8199,
      "step": 4029
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.412412153084456,
      "learning_rate": 9.42337458065978e-06,
      "loss": 0.8771,
      "step": 4030
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.244566547495396,
      "learning_rate": 9.423037758926883e-06,
      "loss": 0.7986,
      "step": 4031
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.963461234991387,
      "learning_rate": 9.422700844872758e-06,
      "loss": 0.7998,
      "step": 4032
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.978142898110764,
      "learning_rate": 9.422363838504439e-06,
      "loss": 0.7275,
      "step": 4033
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.482714158481245,
      "learning_rate": 9.422026739828958e-06,
      "loss": 0.7616,
      "step": 4034
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.2370892832236615,
      "learning_rate": 9.421689548853352e-06,
      "loss": 0.8748,
      "step": 4035
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.445613961593261,
      "learning_rate": 9.421352265584663e-06,
      "loss": 0.7895,
      "step": 4036
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.249561297585087,
      "learning_rate": 9.421014890029925e-06,
      "loss": 0.9042,
      "step": 4037
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.246486934713767,
      "learning_rate": 9.420677422196185e-06,
      "loss": 0.8928,
      "step": 4038
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.7804784037967245,
      "learning_rate": 9.420339862090485e-06,
      "loss": 0.8648,
      "step": 4039
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.61254421645711,
      "learning_rate": 9.42000220971987e-06,
      "loss": 0.7611,
      "step": 4040
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.504121654781611,
      "learning_rate": 9.419664465091389e-06,
      "loss": 0.8925,
      "step": 4041
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.844011620663812,
      "learning_rate": 9.419326628212091e-06,
      "loss": 0.8382,
      "step": 4042
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.360136452343533,
      "learning_rate": 9.418988699089029e-06,
      "loss": 0.8777,
      "step": 4043
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.808976088887901,
      "learning_rate": 9.418650677729254e-06,
      "loss": 0.7933,
      "step": 4044
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.748157457116361,
      "learning_rate": 9.418312564139822e-06,
      "loss": 0.9138,
      "step": 4045
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.350534290483683,
      "learning_rate": 9.417974358327792e-06,
      "loss": 0.842,
      "step": 4046
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.6169275080348084,
      "learning_rate": 9.417636060300224e-06,
      "loss": 0.8184,
      "step": 4047
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.0848210829061795,
      "learning_rate": 9.417297670064175e-06,
      "loss": 0.8271,
      "step": 4048
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.3900307463238155,
      "learning_rate": 9.416959187626713e-06,
      "loss": 0.8424,
      "step": 4049
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.368493337680685,
      "learning_rate": 9.4166206129949e-06,
      "loss": 0.8527,
      "step": 4050
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.149691808202422,
      "learning_rate": 9.416281946175804e-06,
      "loss": 0.8665,
      "step": 4051
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.025585590090299,
      "learning_rate": 9.415943187176494e-06,
      "loss": 0.8529,
      "step": 4052
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.295905421631316,
      "learning_rate": 9.415604336004041e-06,
      "loss": 0.8373,
      "step": 4053
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.83039467234084,
      "learning_rate": 9.415265392665516e-06,
      "loss": 0.8526,
      "step": 4054
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.1389834791789095,
      "learning_rate": 9.414926357167998e-06,
      "loss": 0.84,
      "step": 4055
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.944072022279057,
      "learning_rate": 9.414587229518557e-06,
      "loss": 0.8312,
      "step": 4056
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.4928633158159546,
      "learning_rate": 9.414248009724279e-06,
      "loss": 0.8684,
      "step": 4057
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.21331563458297,
      "learning_rate": 9.413908697792237e-06,
      "loss": 0.8358,
      "step": 4058
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.346525617381498,
      "learning_rate": 9.413569293729521e-06,
      "loss": 0.8518,
      "step": 4059
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.254646524543499,
      "learning_rate": 9.41322979754321e-06,
      "loss": 0.8294,
      "step": 4060
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.780296557434875,
      "learning_rate": 9.412890209240393e-06,
      "loss": 0.7869,
      "step": 4061
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.078705324756505,
      "learning_rate": 9.412550528828155e-06,
      "loss": 0.8016,
      "step": 4062
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.929555352585771,
      "learning_rate": 9.41221075631359e-06,
      "loss": 0.8271,
      "step": 4063
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.569512272471407,
      "learning_rate": 9.411870891703786e-06,
      "loss": 0.9015,
      "step": 4064
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.639980599077865,
      "learning_rate": 9.41153093500584e-06,
      "loss": 0.7614,
      "step": 4065
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.884095968615489,
      "learning_rate": 9.411190886226847e-06,
      "loss": 0.8509,
      "step": 4066
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.621115716612975,
      "learning_rate": 9.410850745373903e-06,
      "loss": 0.8876,
      "step": 4067
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.925257320880341,
      "learning_rate": 9.410510512454111e-06,
      "loss": 0.8913,
      "step": 4068
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.981394047076977,
      "learning_rate": 9.41017018747457e-06,
      "loss": 0.8078,
      "step": 4069
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.225170007359369,
      "learning_rate": 9.409829770442385e-06,
      "loss": 0.855,
      "step": 4070
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.583389322637002,
      "learning_rate": 9.409489261364662e-06,
      "loss": 0.7814,
      "step": 4071
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.6206346601098165,
      "learning_rate": 9.409148660248505e-06,
      "loss": 0.9044,
      "step": 4072
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.848326359575929,
      "learning_rate": 9.408807967101028e-06,
      "loss": 0.8813,
      "step": 4073
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.23416491812388,
      "learning_rate": 9.408467181929338e-06,
      "loss": 0.8466,
      "step": 4074
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.19406282070563,
      "learning_rate": 9.40812630474055e-06,
      "loss": 0.8192,
      "step": 4075
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.619306814918696,
      "learning_rate": 9.407785335541779e-06,
      "loss": 0.9269,
      "step": 4076
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.048357645297921,
      "learning_rate": 9.407444274340142e-06,
      "loss": 0.794,
      "step": 4077
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.761295287532734,
      "learning_rate": 9.40710312114276e-06,
      "loss": 0.834,
      "step": 4078
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.38922467757638,
      "learning_rate": 9.406761875956748e-06,
      "loss": 0.7574,
      "step": 4079
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.251972770379315,
      "learning_rate": 9.406420538789237e-06,
      "loss": 0.8266,
      "step": 4080
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.671402205661649,
      "learning_rate": 9.406079109647343e-06,
      "loss": 0.8307,
      "step": 4081
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.665099628578525,
      "learning_rate": 9.405737588538199e-06,
      "loss": 0.9054,
      "step": 4082
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.750251520967442,
      "learning_rate": 9.405395975468932e-06,
      "loss": 0.8494,
      "step": 4083
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.748818871388774,
      "learning_rate": 9.40505427044667e-06,
      "loss": 0.834,
      "step": 4084
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.8094259853288985,
      "learning_rate": 9.404712473478547e-06,
      "loss": 0.8103,
      "step": 4085
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.088633839119641,
      "learning_rate": 9.404370584571698e-06,
      "loss": 0.8552,
      "step": 4086
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.900635407596523,
      "learning_rate": 9.404028603733257e-06,
      "loss": 0.8337,
      "step": 4087
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.581933057237395,
      "learning_rate": 9.403686530970365e-06,
      "loss": 0.8087,
      "step": 4088
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.894869788675074,
      "learning_rate": 9.403344366290161e-06,
      "loss": 0.8538,
      "step": 4089
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.177469179533167,
      "learning_rate": 9.403002109699786e-06,
      "loss": 0.8438,
      "step": 4090
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.699913014225009,
      "learning_rate": 9.402659761206384e-06,
      "loss": 0.8601,
      "step": 4091
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.139045172089933,
      "learning_rate": 9.4023173208171e-06,
      "loss": 0.8223,
      "step": 4092
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.527026136808844,
      "learning_rate": 9.401974788539086e-06,
      "loss": 0.8087,
      "step": 4093
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.538049406997626,
      "learning_rate": 9.401632164379485e-06,
      "loss": 0.8609,
      "step": 4094
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.216153341222045,
      "learning_rate": 9.401289448345455e-06,
      "loss": 0.8869,
      "step": 4095
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.49700990414356,
      "learning_rate": 9.400946640444143e-06,
      "loss": 0.8449,
      "step": 4096
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.325892000681776,
      "learning_rate": 9.400603740682712e-06,
      "loss": 0.7727,
      "step": 4097
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.632258695197307,
      "learning_rate": 9.400260749068311e-06,
      "loss": 0.8814,
      "step": 4098
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.0745058348040635,
      "learning_rate": 9.399917665608107e-06,
      "loss": 0.8581,
      "step": 4099
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.171725720041989,
      "learning_rate": 9.399574490309254e-06,
      "loss": 0.9053,
      "step": 4100
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.93356504083452,
      "learning_rate": 9.399231223178919e-06,
      "loss": 0.8308,
      "step": 4101
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.915945308995971,
      "learning_rate": 9.398887864224269e-06,
      "loss": 0.8152,
      "step": 4102
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.553260573458361,
      "learning_rate": 9.398544413452465e-06,
      "loss": 0.8271,
      "step": 4103
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.042196083697854,
      "learning_rate": 9.398200870870681e-06,
      "loss": 0.8612,
      "step": 4104
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.623886831015347,
      "learning_rate": 9.397857236486086e-06,
      "loss": 0.9203,
      "step": 4105
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.770118194982789,
      "learning_rate": 9.397513510305851e-06,
      "loss": 0.8103,
      "step": 4106
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.751010577671964,
      "learning_rate": 9.397169692337154e-06,
      "loss": 0.7862,
      "step": 4107
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.6282650471613,
      "learning_rate": 9.396825782587167e-06,
      "loss": 0.7739,
      "step": 4108
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.663303535640229,
      "learning_rate": 9.396481781063072e-06,
      "loss": 0.8669,
      "step": 4109
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.617495589582204,
      "learning_rate": 9.396137687772047e-06,
      "loss": 0.8446,
      "step": 4110
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.088133707224628,
      "learning_rate": 9.395793502721276e-06,
      "loss": 0.79,
      "step": 4111
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.534911311471697,
      "learning_rate": 9.395449225917941e-06,
      "loss": 0.8703,
      "step": 4112
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.879024837690329,
      "learning_rate": 9.395104857369231e-06,
      "loss": 0.8699,
      "step": 4113
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.872429245706731,
      "learning_rate": 9.39476039708233e-06,
      "loss": 0.8548,
      "step": 4114
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.3200802740680055,
      "learning_rate": 9.394415845064431e-06,
      "loss": 0.8499,
      "step": 4115
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.334078476928942,
      "learning_rate": 9.394071201322726e-06,
      "loss": 0.8258,
      "step": 4116
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.804075622351396,
      "learning_rate": 9.393726465864405e-06,
      "loss": 0.8286,
      "step": 4117
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.226510886778001,
      "learning_rate": 9.39338163869667e-06,
      "loss": 0.9365,
      "step": 4118
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.384944577881193,
      "learning_rate": 9.393036719826711e-06,
      "loss": 0.8641,
      "step": 4119
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.682514794258588,
      "learning_rate": 9.392691709261733e-06,
      "loss": 0.8529,
      "step": 4120
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.400989719999163,
      "learning_rate": 9.392346607008935e-06,
      "loss": 0.8787,
      "step": 4121
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.8691947120610255,
      "learning_rate": 9.39200141307552e-06,
      "loss": 0.8184,
      "step": 4122
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.778652408382469,
      "learning_rate": 9.391656127468696e-06,
      "loss": 0.8352,
      "step": 4123
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.602605217082588,
      "learning_rate": 9.391310750195665e-06,
      "loss": 0.8523,
      "step": 4124
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.186921807978088,
      "learning_rate": 9.39096528126364e-06,
      "loss": 0.8481,
      "step": 4125
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.326505091953193,
      "learning_rate": 9.390619720679833e-06,
      "loss": 0.9144,
      "step": 4126
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.597502591238907,
      "learning_rate": 9.39027406845145e-06,
      "loss": 0.8127,
      "step": 4127
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.692323823014313,
      "learning_rate": 9.389928324585713e-06,
      "loss": 0.8472,
      "step": 4128
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.46676575268346,
      "learning_rate": 9.389582489089837e-06,
      "loss": 0.886,
      "step": 4129
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.6212253638091285,
      "learning_rate": 9.38923656197104e-06,
      "loss": 0.8725,
      "step": 4130
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.6110314410435524,
      "learning_rate": 9.388890543236539e-06,
      "loss": 0.8518,
      "step": 4131
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.54843948671427,
      "learning_rate": 9.388544432893562e-06,
      "loss": 0.8602,
      "step": 4132
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.683441603359437,
      "learning_rate": 9.38819823094933e-06,
      "loss": 0.8373,
      "step": 4133
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.527798564132173,
      "learning_rate": 9.38785193741107e-06,
      "loss": 0.8314,
      "step": 4134
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.43295574119505,
      "learning_rate": 9.387505552286008e-06,
      "loss": 0.8477,
      "step": 4135
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.575464182979885,
      "learning_rate": 9.387159075581378e-06,
      "loss": 0.8621,
      "step": 4136
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.490905909549163,
      "learning_rate": 9.38681250730441e-06,
      "loss": 0.7674,
      "step": 4137
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.69286738294564,
      "learning_rate": 9.386465847462339e-06,
      "loss": 0.8374,
      "step": 4138
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.083088493133638,
      "learning_rate": 9.386119096062396e-06,
      "loss": 0.8673,
      "step": 4139
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.305075502074553,
      "learning_rate": 9.385772253111826e-06,
      "loss": 0.8933,
      "step": 4140
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.746566164748403,
      "learning_rate": 9.385425318617862e-06,
      "loss": 0.8592,
      "step": 4141
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.103635165033205,
      "learning_rate": 9.38507829258775e-06,
      "loss": 0.8032,
      "step": 4142
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.069928405029586,
      "learning_rate": 9.384731175028733e-06,
      "loss": 0.9222,
      "step": 4143
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.3643556074590455,
      "learning_rate": 9.384383965948055e-06,
      "loss": 0.8492,
      "step": 4144
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.3395836253426445,
      "learning_rate": 9.384036665352961e-06,
      "loss": 0.8336,
      "step": 4145
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.763296186313993,
      "learning_rate": 9.383689273250702e-06,
      "loss": 0.8538,
      "step": 4146
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.31832533707095,
      "learning_rate": 9.383341789648533e-06,
      "loss": 0.8369,
      "step": 4147
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.280557918984372,
      "learning_rate": 9.3829942145537e-06,
      "loss": 0.8608,
      "step": 4148
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.063728548077029,
      "learning_rate": 9.382646547973464e-06,
      "loss": 0.8458,
      "step": 4149
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.523396393634446,
      "learning_rate": 9.382298789915079e-06,
      "loss": 0.8528,
      "step": 4150
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.54100705711264,
      "learning_rate": 9.381950940385803e-06,
      "loss": 0.7553,
      "step": 4151
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.2961408456632615,
      "learning_rate": 9.381602999392896e-06,
      "loss": 0.7686,
      "step": 4152
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.749566103104944,
      "learning_rate": 9.381254966943624e-06,
      "loss": 0.8518,
      "step": 4153
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.125573305467597,
      "learning_rate": 9.380906843045248e-06,
      "loss": 0.7937,
      "step": 4154
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.895048481282751,
      "learning_rate": 9.380558627705036e-06,
      "loss": 0.863,
      "step": 4155
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.5717905072356,
      "learning_rate": 9.380210320930256e-06,
      "loss": 0.8986,
      "step": 4156
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.637090765060546,
      "learning_rate": 9.37986192272818e-06,
      "loss": 0.8917,
      "step": 4157
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.993530843801003,
      "learning_rate": 9.379513433106075e-06,
      "loss": 0.8378,
      "step": 4158
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.117855409699876,
      "learning_rate": 9.379164852071221e-06,
      "loss": 0.8441,
      "step": 4159
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.857117890667359,
      "learning_rate": 9.378816179630888e-06,
      "loss": 0.8554,
      "step": 4160
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.9806133568479085,
      "learning_rate": 9.378467415792358e-06,
      "loss": 0.8996,
      "step": 4161
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.727875819859236,
      "learning_rate": 9.37811856056291e-06,
      "loss": 0.8374,
      "step": 4162
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.784603310787402,
      "learning_rate": 9.377769613949825e-06,
      "loss": 0.8768,
      "step": 4163
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.848503059586507,
      "learning_rate": 9.377420575960385e-06,
      "loss": 0.8472,
      "step": 4164
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.68936620822105,
      "learning_rate": 9.377071446601878e-06,
      "loss": 0.858,
      "step": 4165
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.114862406439199,
      "learning_rate": 9.376722225881591e-06,
      "loss": 0.8396,
      "step": 4166
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.875364863901053,
      "learning_rate": 9.37637291380681e-06,
      "loss": 0.827,
      "step": 4167
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.702120842212274,
      "learning_rate": 9.376023510384831e-06,
      "loss": 0.8458,
      "step": 4168
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.518755538146358,
      "learning_rate": 9.375674015622941e-06,
      "loss": 0.8253,
      "step": 4169
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.631253760323815,
      "learning_rate": 9.375324429528443e-06,
      "loss": 0.8653,
      "step": 4170
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.338345343457585,
      "learning_rate": 9.374974752108626e-06,
      "loss": 0.8447,
      "step": 4171
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.257998550546608,
      "learning_rate": 9.374624983370795e-06,
      "loss": 0.8086,
      "step": 4172
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.429756906077807,
      "learning_rate": 9.374275123322245e-06,
      "loss": 0.9058,
      "step": 4173
    },
    {
      "epoch": 0.37,
      "grad_norm": 9.044870236958722,
      "learning_rate": 9.373925171970282e-06,
      "loss": 0.8593,
      "step": 4174
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.63642419578049,
      "learning_rate": 9.37357512932221e-06,
      "loss": 0.8725,
      "step": 4175
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.879325162366704,
      "learning_rate": 9.373224995385336e-06,
      "loss": 0.8629,
      "step": 4176
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.335036251444922,
      "learning_rate": 9.372874770166966e-06,
      "loss": 0.9211,
      "step": 4177
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.044634635740906,
      "learning_rate": 9.372524453674412e-06,
      "loss": 0.8561,
      "step": 4178
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.265686366951105,
      "learning_rate": 9.372174045914987e-06,
      "loss": 0.8514,
      "step": 4179
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.678080930912243,
      "learning_rate": 9.371823546896001e-06,
      "loss": 0.7979,
      "step": 4180
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.751116784504044,
      "learning_rate": 9.371472956624775e-06,
      "loss": 0.8439,
      "step": 4181
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.4584734358888225,
      "learning_rate": 9.371122275108625e-06,
      "loss": 0.8732,
      "step": 4182
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.73503212634114,
      "learning_rate": 9.370771502354867e-06,
      "loss": 0.8559,
      "step": 4183
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.374986659000164,
      "learning_rate": 9.370420638370828e-06,
      "loss": 0.8539,
      "step": 4184
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.418519779362382,
      "learning_rate": 9.37006968316383e-06,
      "loss": 0.8911,
      "step": 4185
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.285369366086445,
      "learning_rate": 9.369718636741196e-06,
      "loss": 0.8948,
      "step": 4186
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.6924506792129534,
      "learning_rate": 9.369367499110254e-06,
      "loss": 0.8408,
      "step": 4187
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.5927187294461005,
      "learning_rate": 9.369016270278335e-06,
      "loss": 0.9031,
      "step": 4188
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.379870611375347,
      "learning_rate": 9.36866495025277e-06,
      "loss": 0.8172,
      "step": 4189
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.06319823238156,
      "learning_rate": 9.368313539040892e-06,
      "loss": 0.848,
      "step": 4190
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.815531155517509,
      "learning_rate": 9.367962036650035e-06,
      "loss": 0.8042,
      "step": 4191
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.597975106976378,
      "learning_rate": 9.367610443087536e-06,
      "loss": 0.8064,
      "step": 4192
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.293436914751377,
      "learning_rate": 9.367258758360733e-06,
      "loss": 0.9103,
      "step": 4193
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.99316343831238,
      "learning_rate": 9.36690698247697e-06,
      "loss": 0.856,
      "step": 4194
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.911543975374293,
      "learning_rate": 9.366555115443584e-06,
      "loss": 0.7736,
      "step": 4195
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.9393088792338515,
      "learning_rate": 9.366203157267923e-06,
      "loss": 0.7731,
      "step": 4196
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.927321662756173,
      "learning_rate": 9.365851107957333e-06,
      "loss": 0.8444,
      "step": 4197
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.885770132108312,
      "learning_rate": 9.365498967519164e-06,
      "loss": 0.9403,
      "step": 4198
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.069652030586314,
      "learning_rate": 9.365146735960763e-06,
      "loss": 0.844,
      "step": 4199
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.60124942388124,
      "learning_rate": 9.364794413289483e-06,
      "loss": 0.7784,
      "step": 4200
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.163774498350081,
      "learning_rate": 9.36444199951268e-06,
      "loss": 0.8196,
      "step": 4201
    },
    {
      "epoch": 0.37,
      "grad_norm": 9.682753938417639,
      "learning_rate": 9.364089494637707e-06,
      "loss": 0.9063,
      "step": 4202
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.540770108217958,
      "learning_rate": 9.363736898671923e-06,
      "loss": 0.8683,
      "step": 4203
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.811097890585324,
      "learning_rate": 9.363384211622688e-06,
      "loss": 0.8357,
      "step": 4204
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.910705659166817,
      "learning_rate": 9.363031433497362e-06,
      "loss": 0.8115,
      "step": 4205
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.365238193088776,
      "learning_rate": 9.362678564303311e-06,
      "loss": 0.7909,
      "step": 4206
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.3510476844499015,
      "learning_rate": 9.362325604047899e-06,
      "loss": 0.801,
      "step": 4207
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.856115252931735,
      "learning_rate": 9.361972552738495e-06,
      "loss": 0.8856,
      "step": 4208
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.416956022113308,
      "learning_rate": 9.361619410382463e-06,
      "loss": 0.8956,
      "step": 4209
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.823312070477005,
      "learning_rate": 9.36126617698718e-06,
      "loss": 0.886,
      "step": 4210
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.367144832749693,
      "learning_rate": 9.360912852560015e-06,
      "loss": 0.8743,
      "step": 4211
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.644403520906778,
      "learning_rate": 9.360559437108346e-06,
      "loss": 0.7568,
      "step": 4212
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.332275888202728,
      "learning_rate": 9.360205930639547e-06,
      "loss": 0.8229,
      "step": 4213
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.886918816473951,
      "learning_rate": 9.359852333161e-06,
      "loss": 0.8362,
      "step": 4214
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.834393949065181,
      "learning_rate": 9.35949864468008e-06,
      "loss": 0.8072,
      "step": 4215
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.330326880041444,
      "learning_rate": 9.359144865204176e-06,
      "loss": 0.8598,
      "step": 4216
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.788126886359555,
      "learning_rate": 9.358790994740669e-06,
      "loss": 0.8957,
      "step": 4217
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.014165574322836,
      "learning_rate": 9.358437033296943e-06,
      "loss": 0.8507,
      "step": 4218
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.049513476981598,
      "learning_rate": 9.358082980880392e-06,
      "loss": 0.8163,
      "step": 4219
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.395693523525062,
      "learning_rate": 9.3577288374984e-06,
      "loss": 0.8007,
      "step": 4220
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.105039626943591,
      "learning_rate": 9.357374603158364e-06,
      "loss": 0.8219,
      "step": 4221
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.223965727975415,
      "learning_rate": 9.357020277867674e-06,
      "loss": 0.8846,
      "step": 4222
    },
    {
      "epoch": 0.38,
      "grad_norm": 9.295629358665192,
      "learning_rate": 9.35666586163373e-06,
      "loss": 0.7818,
      "step": 4223
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.713351597904885,
      "learning_rate": 9.356311354463924e-06,
      "loss": 0.7903,
      "step": 4224
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.617096689184785,
      "learning_rate": 9.35595675636566e-06,
      "loss": 0.8682,
      "step": 4225
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.785333182216809,
      "learning_rate": 9.355602067346337e-06,
      "loss": 0.8409,
      "step": 4226
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.566145019164834,
      "learning_rate": 9.35524728741336e-06,
      "loss": 0.7906,
      "step": 4227
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.136943298837495,
      "learning_rate": 9.354892416574135e-06,
      "loss": 0.7747,
      "step": 4228
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.553953512226192,
      "learning_rate": 9.354537454836067e-06,
      "loss": 0.873,
      "step": 4229
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.311252365644838,
      "learning_rate": 9.354182402206566e-06,
      "loss": 0.8036,
      "step": 4230
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.097597660662817,
      "learning_rate": 9.353827258693043e-06,
      "loss": 0.8583,
      "step": 4231
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.739314986290021,
      "learning_rate": 9.35347202430291e-06,
      "loss": 0.8858,
      "step": 4232
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.313571533153005,
      "learning_rate": 9.353116699043583e-06,
      "loss": 0.8058,
      "step": 4233
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.805854935996429,
      "learning_rate": 9.352761282922479e-06,
      "loss": 0.8168,
      "step": 4234
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.535529316215074,
      "learning_rate": 9.352405775947014e-06,
      "loss": 0.8235,
      "step": 4235
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.476044590396591,
      "learning_rate": 9.35205017812461e-06,
      "loss": 0.9101,
      "step": 4236
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.445937184544161,
      "learning_rate": 9.351694489462689e-06,
      "loss": 0.8733,
      "step": 4237
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.061095670128932,
      "learning_rate": 9.351338709968676e-06,
      "loss": 0.8919,
      "step": 4238
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.906015308536657,
      "learning_rate": 9.350982839649998e-06,
      "loss": 0.8164,
      "step": 4239
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.881305757583553,
      "learning_rate": 9.35062687851408e-06,
      "loss": 0.858,
      "step": 4240
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.185309417685991,
      "learning_rate": 9.350270826568356e-06,
      "loss": 0.8383,
      "step": 4241
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.127411191575992,
      "learning_rate": 9.349914683820252e-06,
      "loss": 0.7821,
      "step": 4242
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.714765347150255,
      "learning_rate": 9.349558450277207e-06,
      "loss": 0.8355,
      "step": 4243
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.391600149573789,
      "learning_rate": 9.349202125946656e-06,
      "loss": 0.8765,
      "step": 4244
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.341236911514885,
      "learning_rate": 9.348845710836034e-06,
      "loss": 0.832,
      "step": 4245
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.162026130439689,
      "learning_rate": 9.348489204952781e-06,
      "loss": 0.8597,
      "step": 4246
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.671547479688158,
      "learning_rate": 9.348132608304338e-06,
      "loss": 0.9192,
      "step": 4247
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.456145263968649,
      "learning_rate": 9.347775920898153e-06,
      "loss": 0.8443,
      "step": 4248
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.5237172892436135,
      "learning_rate": 9.347419142741664e-06,
      "loss": 0.9078,
      "step": 4249
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.847702686183809,
      "learning_rate": 9.347062273842321e-06,
      "loss": 0.8252,
      "step": 4250
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.7965487743060455,
      "learning_rate": 9.346705314207573e-06,
      "loss": 0.8386,
      "step": 4251
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.3759110355259905,
      "learning_rate": 9.346348263844873e-06,
      "loss": 0.783,
      "step": 4252
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.0118384620454135,
      "learning_rate": 9.345991122761668e-06,
      "loss": 0.8703,
      "step": 4253
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.5725085086027235,
      "learning_rate": 9.345633890965418e-06,
      "loss": 0.7744,
      "step": 4254
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.374924460635833,
      "learning_rate": 9.345276568463577e-06,
      "loss": 0.8529,
      "step": 4255
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.572340610802262,
      "learning_rate": 9.344919155263604e-06,
      "loss": 0.8405,
      "step": 4256
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.314437163312313,
      "learning_rate": 9.344561651372958e-06,
      "loss": 0.9246,
      "step": 4257
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.093939518182437,
      "learning_rate": 9.344204056799102e-06,
      "loss": 0.8543,
      "step": 4258
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.179530133294514,
      "learning_rate": 9.3438463715495e-06,
      "loss": 0.7699,
      "step": 4259
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.762505166855348,
      "learning_rate": 9.343488595631619e-06,
      "loss": 0.873,
      "step": 4260
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.963042856942487,
      "learning_rate": 9.343130729052926e-06,
      "loss": 0.8564,
      "step": 4261
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.641218894450112,
      "learning_rate": 9.342772771820887e-06,
      "loss": 0.8399,
      "step": 4262
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.689875145638253,
      "learning_rate": 9.342414723942979e-06,
      "loss": 0.8565,
      "step": 4263
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.056474673237054,
      "learning_rate": 9.342056585426674e-06,
      "loss": 0.82,
      "step": 4264
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.240532086506664,
      "learning_rate": 9.341698356279446e-06,
      "loss": 0.842,
      "step": 4265
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.169032573799889,
      "learning_rate": 9.341340036508773e-06,
      "loss": 0.8565,
      "step": 4266
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.497585143575701,
      "learning_rate": 9.340981626122134e-06,
      "loss": 0.8313,
      "step": 4267
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.77263547684289,
      "learning_rate": 9.34062312512701e-06,
      "loss": 0.9087,
      "step": 4268
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.118237699528283,
      "learning_rate": 9.340264533530886e-06,
      "loss": 0.8626,
      "step": 4269
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.7859350277594075,
      "learning_rate": 9.339905851341242e-06,
      "loss": 0.8797,
      "step": 4270
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.600349599057544,
      "learning_rate": 9.339547078565568e-06,
      "loss": 0.7786,
      "step": 4271
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.449801960611563,
      "learning_rate": 9.339188215211352e-06,
      "loss": 0.6884,
      "step": 4272
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.375935239878161,
      "learning_rate": 9.338829261286087e-06,
      "loss": 0.7891,
      "step": 4273
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.063388020342288,
      "learning_rate": 9.338470216797261e-06,
      "loss": 0.7923,
      "step": 4274
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.40814049296346,
      "learning_rate": 9.33811108175237e-06,
      "loss": 0.8538,
      "step": 4275
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.597402610994749,
      "learning_rate": 9.337751856158912e-06,
      "loss": 0.8646,
      "step": 4276
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.888521865219829,
      "learning_rate": 9.337392540024383e-06,
      "loss": 0.8589,
      "step": 4277
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.206648754962669,
      "learning_rate": 9.337033133356284e-06,
      "loss": 0.7939,
      "step": 4278
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.1704856561570125,
      "learning_rate": 9.336673636162116e-06,
      "loss": 0.8381,
      "step": 4279
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.885888324496254,
      "learning_rate": 9.336314048449382e-06,
      "loss": 0.852,
      "step": 4280
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.086169991644454,
      "learning_rate": 9.33595437022559e-06,
      "loss": 0.8292,
      "step": 4281
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.308688149146375,
      "learning_rate": 9.335594601498246e-06,
      "loss": 0.8078,
      "step": 4282
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.5636153519785125,
      "learning_rate": 9.335234742274858e-06,
      "loss": 0.8634,
      "step": 4283
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.936106150752359,
      "learning_rate": 9.33487479256294e-06,
      "loss": 0.7766,
      "step": 4284
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.848401017570743,
      "learning_rate": 9.334514752370004e-06,
      "loss": 0.9126,
      "step": 4285
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.231825795867669,
      "learning_rate": 9.334154621703564e-06,
      "loss": 0.8721,
      "step": 4286
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.718384398197964,
      "learning_rate": 9.33379440057114e-06,
      "loss": 0.8391,
      "step": 4287
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.0032220738051,
      "learning_rate": 9.333434088980246e-06,
      "loss": 0.8704,
      "step": 4288
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.375546595725357,
      "learning_rate": 9.333073686938406e-06,
      "loss": 0.8605,
      "step": 4289
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.090490395201389,
      "learning_rate": 9.332713194453142e-06,
      "loss": 0.9013,
      "step": 4290
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.389158355548935,
      "learning_rate": 9.332352611531982e-06,
      "loss": 0.7937,
      "step": 4291
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.2912037056374475,
      "learning_rate": 9.331991938182444e-06,
      "loss": 0.8264,
      "step": 4292
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.029326445907776,
      "learning_rate": 9.331631174412063e-06,
      "loss": 0.8329,
      "step": 4293
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.159995059805666,
      "learning_rate": 9.331270320228368e-06,
      "loss": 0.9023,
      "step": 4294
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.964686225193782,
      "learning_rate": 9.33090937563889e-06,
      "loss": 0.793,
      "step": 4295
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.699528193617971,
      "learning_rate": 9.330548340651164e-06,
      "loss": 0.8727,
      "step": 4296
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.169470136001731,
      "learning_rate": 9.330187215272724e-06,
      "loss": 0.7635,
      "step": 4297
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.384229365473055,
      "learning_rate": 9.329825999511109e-06,
      "loss": 0.7864,
      "step": 4298
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.642744545924761,
      "learning_rate": 9.32946469337386e-06,
      "loss": 0.799,
      "step": 4299
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.817878872478266,
      "learning_rate": 9.329103296868514e-06,
      "loss": 0.82,
      "step": 4300
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.036888595957784,
      "learning_rate": 9.32874181000262e-06,
      "loss": 0.9031,
      "step": 4301
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.256451480673778,
      "learning_rate": 9.32838023278372e-06,
      "loss": 0.772,
      "step": 4302
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.384401119957896,
      "learning_rate": 9.32801856521936e-06,
      "loss": 0.8683,
      "step": 4303
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.145436329663048,
      "learning_rate": 9.327656807317092e-06,
      "loss": 0.8643,
      "step": 4304
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.7588488749905205,
      "learning_rate": 9.327294959084466e-06,
      "loss": 0.8561,
      "step": 4305
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.74710087129823,
      "learning_rate": 9.326933020529032e-06,
      "loss": 0.8589,
      "step": 4306
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.576226253192541,
      "learning_rate": 9.326570991658349e-06,
      "loss": 0.8775,
      "step": 4307
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.686530594321312,
      "learning_rate": 9.326208872479971e-06,
      "loss": 0.8109,
      "step": 4308
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.691535447222282,
      "learning_rate": 9.325846663001457e-06,
      "loss": 0.8279,
      "step": 4309
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.436218900351195,
      "learning_rate": 9.325484363230368e-06,
      "loss": 0.7762,
      "step": 4310
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.414517665682313,
      "learning_rate": 9.325121973174265e-06,
      "loss": 0.8566,
      "step": 4311
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.173297352186554,
      "learning_rate": 9.324759492840713e-06,
      "loss": 0.8444,
      "step": 4312
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.97605195714034,
      "learning_rate": 9.324396922237278e-06,
      "loss": 0.8218,
      "step": 4313
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.60891932114542,
      "learning_rate": 9.324034261371527e-06,
      "loss": 0.8226,
      "step": 4314
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.574912900920701,
      "learning_rate": 9.323671510251029e-06,
      "loss": 0.8378,
      "step": 4315
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.385025247785694,
      "learning_rate": 9.32330866888336e-06,
      "loss": 0.8643,
      "step": 4316
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.458643993504901,
      "learning_rate": 9.322945737276088e-06,
      "loss": 0.8851,
      "step": 4317
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.63581409716872,
      "learning_rate": 9.322582715436789e-06,
      "loss": 0.7675,
      "step": 4318
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.529800985273648,
      "learning_rate": 9.322219603373046e-06,
      "loss": 0.9076,
      "step": 4319
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.047900649789184,
      "learning_rate": 9.321856401092433e-06,
      "loss": 0.772,
      "step": 4320
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.308892532142045,
      "learning_rate": 9.321493108602532e-06,
      "loss": 0.8343,
      "step": 4321
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.581800803850134,
      "learning_rate": 9.321129725910926e-06,
      "loss": 0.8932,
      "step": 4322
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.689896762525324,
      "learning_rate": 9.3207662530252e-06,
      "loss": 0.8637,
      "step": 4323
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.77560223405044,
      "learning_rate": 9.320402689952942e-06,
      "loss": 0.8444,
      "step": 4324
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.734211083351347,
      "learning_rate": 9.320039036701738e-06,
      "loss": 0.8892,
      "step": 4325
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.039113570848922,
      "learning_rate": 9.31967529327918e-06,
      "loss": 0.9226,
      "step": 4326
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.542494568331635,
      "learning_rate": 9.319311459692862e-06,
      "loss": 0.8151,
      "step": 4327
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.673367677320942,
      "learning_rate": 9.318947535950375e-06,
      "loss": 0.8323,
      "step": 4328
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.940795820063644,
      "learning_rate": 9.318583522059317e-06,
      "loss": 0.8207,
      "step": 4329
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.505982963422932,
      "learning_rate": 9.318219418027284e-06,
      "loss": 0.8379,
      "step": 4330
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.208320585427727,
      "learning_rate": 9.317855223861878e-06,
      "loss": 0.8008,
      "step": 4331
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.507675363730609,
      "learning_rate": 9.317490939570702e-06,
      "loss": 0.818,
      "step": 4332
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.03228554788471,
      "learning_rate": 9.317126565161357e-06,
      "loss": 0.8385,
      "step": 4333
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.505075197345783,
      "learning_rate": 9.316762100641448e-06,
      "loss": 0.8563,
      "step": 4334
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.991044786583493,
      "learning_rate": 9.316397546018583e-06,
      "loss": 0.8769,
      "step": 4335
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.417279339127916,
      "learning_rate": 9.316032901300374e-06,
      "loss": 0.843,
      "step": 4336
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.122843203492062,
      "learning_rate": 9.31566816649443e-06,
      "loss": 0.8008,
      "step": 4337
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.032643581133637,
      "learning_rate": 9.315303341608364e-06,
      "loss": 0.8663,
      "step": 4338
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.51247204316075,
      "learning_rate": 9.31493842664979e-06,
      "loss": 0.8238,
      "step": 4339
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.936236067160519,
      "learning_rate": 9.314573421626326e-06,
      "loss": 0.7919,
      "step": 4340
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.746648446059553,
      "learning_rate": 9.314208326545592e-06,
      "loss": 0.7591,
      "step": 4341
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.506814486233366,
      "learning_rate": 9.313843141415206e-06,
      "loss": 0.9274,
      "step": 4342
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.436127351951089,
      "learning_rate": 9.313477866242792e-06,
      "loss": 0.8371,
      "step": 4343
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.911529607741707,
      "learning_rate": 9.313112501035975e-06,
      "loss": 0.8038,
      "step": 4344
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.492835974289569,
      "learning_rate": 9.312747045802377e-06,
      "loss": 0.8218,
      "step": 4345
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.7365613150107535,
      "learning_rate": 9.312381500549631e-06,
      "loss": 0.8155,
      "step": 4346
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.2837704003185895,
      "learning_rate": 9.312015865285366e-06,
      "loss": 0.8441,
      "step": 4347
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.129655183365047,
      "learning_rate": 9.311650140017212e-06,
      "loss": 0.8218,
      "step": 4348
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.9815294611422365,
      "learning_rate": 9.311284324752804e-06,
      "loss": 0.847,
      "step": 4349
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.668906164597301,
      "learning_rate": 9.310918419499777e-06,
      "loss": 0.8542,
      "step": 4350
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.9057448491530415,
      "learning_rate": 9.310552424265768e-06,
      "loss": 0.7865,
      "step": 4351
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.351692926996722,
      "learning_rate": 9.31018633905842e-06,
      "loss": 0.9269,
      "step": 4352
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.283725995153035,
      "learning_rate": 9.309820163885368e-06,
      "loss": 0.8742,
      "step": 4353
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.5898157233488535,
      "learning_rate": 9.30945389875426e-06,
      "loss": 0.8158,
      "step": 4354
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.233919442897969,
      "learning_rate": 9.309087543672739e-06,
      "loss": 0.8496,
      "step": 4355
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.752994429182826,
      "learning_rate": 9.308721098648452e-06,
      "loss": 0.847,
      "step": 4356
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.539894713204919,
      "learning_rate": 9.308354563689049e-06,
      "loss": 0.8668,
      "step": 4357
    },
    {
      "epoch": 0.39,
      "grad_norm": 10.909235142424608,
      "learning_rate": 9.307987938802178e-06,
      "loss": 0.8044,
      "step": 4358
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.78464534932656,
      "learning_rate": 9.307621223995496e-06,
      "loss": 0.8215,
      "step": 4359
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.798815352983658,
      "learning_rate": 9.307254419276653e-06,
      "loss": 0.8615,
      "step": 4360
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.527291973213971,
      "learning_rate": 9.306887524653305e-06,
      "loss": 0.9004,
      "step": 4361
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.10565406406722,
      "learning_rate": 9.306520540133113e-06,
      "loss": 0.8408,
      "step": 4362
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.552091232066093,
      "learning_rate": 9.306153465723736e-06,
      "loss": 0.8908,
      "step": 4363
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.830602917196874,
      "learning_rate": 9.305786301432836e-06,
      "loss": 0.7946,
      "step": 4364
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.394372912242018,
      "learning_rate": 9.305419047268077e-06,
      "loss": 0.8586,
      "step": 4365
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.49090933406075,
      "learning_rate": 9.305051703237122e-06,
      "loss": 0.8654,
      "step": 4366
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.622587190197653,
      "learning_rate": 9.304684269347643e-06,
      "loss": 0.8211,
      "step": 4367
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.694005044050268,
      "learning_rate": 9.304316745607307e-06,
      "loss": 0.8051,
      "step": 4368
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.736987060905926,
      "learning_rate": 9.303949132023781e-06,
      "loss": 0.8127,
      "step": 4369
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.711658469118172,
      "learning_rate": 9.303581428604746e-06,
      "loss": 0.8265,
      "step": 4370
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.149537768698715,
      "learning_rate": 9.303213635357874e-06,
      "loss": 0.8095,
      "step": 4371
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.5844938939618745,
      "learning_rate": 9.30284575229084e-06,
      "loss": 0.8464,
      "step": 4372
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.523607577629111,
      "learning_rate": 9.302477779411324e-06,
      "loss": 0.86,
      "step": 4373
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.705816058207198,
      "learning_rate": 9.302109716727006e-06,
      "loss": 0.8674,
      "step": 4374
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.842880076022102,
      "learning_rate": 9.30174156424557e-06,
      "loss": 0.8109,
      "step": 4375
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.0834085074500175,
      "learning_rate": 9.3013733219747e-06,
      "loss": 0.8089,
      "step": 4376
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.610893446462115,
      "learning_rate": 9.30100498992208e-06,
      "loss": 0.8902,
      "step": 4377
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.0217759314467605,
      "learning_rate": 9.300636568095403e-06,
      "loss": 0.8605,
      "step": 4378
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.357175366880541,
      "learning_rate": 9.300268056502353e-06,
      "loss": 0.7841,
      "step": 4379
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.996411991537311,
      "learning_rate": 9.299899455150624e-06,
      "loss": 0.836,
      "step": 4380
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.170984270639248,
      "learning_rate": 9.299530764047913e-06,
      "loss": 0.8013,
      "step": 4381
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.4043245632517465,
      "learning_rate": 9.299161983201912e-06,
      "loss": 0.8962,
      "step": 4382
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.2353953653265854,
      "learning_rate": 9.29879311262032e-06,
      "loss": 0.829,
      "step": 4383
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.634029896596374,
      "learning_rate": 9.298424152310837e-06,
      "loss": 0.8072,
      "step": 4384
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.631814599037949,
      "learning_rate": 9.29805510228116e-06,
      "loss": 0.8519,
      "step": 4385
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.797499165907972,
      "learning_rate": 9.297685962538998e-06,
      "loss": 0.8534,
      "step": 4386
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.791608975422356,
      "learning_rate": 9.297316733092053e-06,
      "loss": 0.7407,
      "step": 4387
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.112398368456486,
      "learning_rate": 9.296947413948031e-06,
      "loss": 0.854,
      "step": 4388
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.212035789700138,
      "learning_rate": 9.296578005114642e-06,
      "loss": 0.8576,
      "step": 4389
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.216902088388244,
      "learning_rate": 9.296208506599597e-06,
      "loss": 0.883,
      "step": 4390
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.602784740199572,
      "learning_rate": 9.295838918410608e-06,
      "loss": 0.7801,
      "step": 4391
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.653722148426496,
      "learning_rate": 9.29546924055539e-06,
      "loss": 0.8486,
      "step": 4392
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.389514508531621,
      "learning_rate": 9.295099473041657e-06,
      "loss": 0.8219,
      "step": 4393
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.823606759527557,
      "learning_rate": 9.294729615877131e-06,
      "loss": 0.9042,
      "step": 4394
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.551974443410783,
      "learning_rate": 9.294359669069527e-06,
      "loss": 0.8711,
      "step": 4395
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.8093333398081795,
      "learning_rate": 9.293989632626571e-06,
      "loss": 0.8165,
      "step": 4396
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.182319160354132,
      "learning_rate": 9.293619506555986e-06,
      "loss": 0.7937,
      "step": 4397
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.51734375109269,
      "learning_rate": 9.293249290865493e-06,
      "loss": 0.865,
      "step": 4398
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.7745689885718345,
      "learning_rate": 9.292878985562826e-06,
      "loss": 0.904,
      "step": 4399
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.087617306345822,
      "learning_rate": 9.292508590655711e-06,
      "loss": 0.842,
      "step": 4400
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.698109916900253,
      "learning_rate": 9.29213810615188e-06,
      "loss": 0.8114,
      "step": 4401
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.105229392625235,
      "learning_rate": 9.291767532059065e-06,
      "loss": 0.8513,
      "step": 4402
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.38839980316784,
      "learning_rate": 9.291396868385e-06,
      "loss": 0.8356,
      "step": 4403
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.210212850235889,
      "learning_rate": 9.291026115137426e-06,
      "loss": 0.8876,
      "step": 4404
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.440264237820538,
      "learning_rate": 9.290655272324078e-06,
      "loss": 0.8485,
      "step": 4405
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.912029853657919,
      "learning_rate": 9.290284339952698e-06,
      "loss": 0.8581,
      "step": 4406
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.028063718428649,
      "learning_rate": 9.289913318031027e-06,
      "loss": 0.8206,
      "step": 4407
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.35255596808982,
      "learning_rate": 9.28954220656681e-06,
      "loss": 0.9145,
      "step": 4408
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.686852935801378,
      "learning_rate": 9.289171005567795e-06,
      "loss": 0.8561,
      "step": 4409
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.957278845109005,
      "learning_rate": 9.288799715041728e-06,
      "loss": 0.8793,
      "step": 4410
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.11654153425826,
      "learning_rate": 9.28842833499636e-06,
      "loss": 0.7913,
      "step": 4411
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.7246318746618865,
      "learning_rate": 9.288056865439439e-06,
      "loss": 0.8224,
      "step": 4412
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.77711808835208,
      "learning_rate": 9.287685306378724e-06,
      "loss": 0.7921,
      "step": 4413
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.667114303770183,
      "learning_rate": 9.287313657821967e-06,
      "loss": 0.8893,
      "step": 4414
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.983479502409955,
      "learning_rate": 9.286941919776928e-06,
      "loss": 0.806,
      "step": 4415
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.7811817169256265,
      "learning_rate": 9.286570092251364e-06,
      "loss": 0.8279,
      "step": 4416
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.416666260184944,
      "learning_rate": 9.286198175253035e-06,
      "loss": 0.8352,
      "step": 4417
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.256330650942187,
      "learning_rate": 9.285826168789707e-06,
      "loss": 0.8542,
      "step": 4418
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.70038586914247,
      "learning_rate": 9.285454072869145e-06,
      "loss": 0.8063,
      "step": 4419
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.852249976771596,
      "learning_rate": 9.285081887499113e-06,
      "loss": 0.8408,
      "step": 4420
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.129210274692351,
      "learning_rate": 9.28470961268738e-06,
      "loss": 0.8103,
      "step": 4421
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.400203200277461,
      "learning_rate": 9.284337248441717e-06,
      "loss": 0.8831,
      "step": 4422
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.326375778532082,
      "learning_rate": 9.283964794769897e-06,
      "loss": 0.8404,
      "step": 4423
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.446147266339482,
      "learning_rate": 9.283592251679694e-06,
      "loss": 0.8391,
      "step": 4424
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.157748236480059,
      "learning_rate": 9.283219619178884e-06,
      "loss": 0.8758,
      "step": 4425
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.583639309983387,
      "learning_rate": 9.282846897275245e-06,
      "loss": 0.7768,
      "step": 4426
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.162724002954905,
      "learning_rate": 9.282474085976553e-06,
      "loss": 0.8339,
      "step": 4427
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.547515039850337,
      "learning_rate": 9.282101185290597e-06,
      "loss": 0.8585,
      "step": 4428
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.731549206484719,
      "learning_rate": 9.281728195225154e-06,
      "loss": 0.8841,
      "step": 4429
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.910253333087144,
      "learning_rate": 9.281355115788014e-06,
      "loss": 0.8583,
      "step": 4430
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.3358170967018275,
      "learning_rate": 9.28098194698696e-06,
      "loss": 0.8144,
      "step": 4431
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.959933946438981,
      "learning_rate": 9.280608688829782e-06,
      "loss": 0.8568,
      "step": 4432
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.926828936236354,
      "learning_rate": 9.280235341324275e-06,
      "loss": 0.9302,
      "step": 4433
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.806216895680405,
      "learning_rate": 9.279861904478226e-06,
      "loss": 0.8305,
      "step": 4434
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.425698991032682,
      "learning_rate": 9.279488378299434e-06,
      "loss": 0.7534,
      "step": 4435
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.219476793574893,
      "learning_rate": 9.279114762795694e-06,
      "loss": 0.7853,
      "step": 4436
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.17955882788759,
      "learning_rate": 9.278741057974804e-06,
      "loss": 0.7896,
      "step": 4437
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.490353714839437,
      "learning_rate": 9.278367263844565e-06,
      "loss": 0.8017,
      "step": 4438
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.894614576984033,
      "learning_rate": 9.277993380412779e-06,
      "loss": 0.8766,
      "step": 4439
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.746740027413282,
      "learning_rate": 9.277619407687248e-06,
      "loss": 0.8512,
      "step": 4440
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.344007958987455,
      "learning_rate": 9.277245345675782e-06,
      "loss": 0.8566,
      "step": 4441
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.367381995463293,
      "learning_rate": 9.276871194386186e-06,
      "loss": 0.8045,
      "step": 4442
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.936519213020619,
      "learning_rate": 9.276496953826269e-06,
      "loss": 0.7777,
      "step": 4443
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.065741517797181,
      "learning_rate": 9.276122624003845e-06,
      "loss": 0.8161,
      "step": 4444
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.400053138991957,
      "learning_rate": 9.275748204926724e-06,
      "loss": 0.83,
      "step": 4445
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.4115173219454125,
      "learning_rate": 9.275373696602724e-06,
      "loss": 0.7389,
      "step": 4446
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.932314848476022,
      "learning_rate": 9.274999099039662e-06,
      "loss": 0.7819,
      "step": 4447
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.924769543374132,
      "learning_rate": 9.274624412245355e-06,
      "loss": 0.8026,
      "step": 4448
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.730134288395782,
      "learning_rate": 9.274249636227623e-06,
      "loss": 0.8706,
      "step": 4449
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.144299548063782,
      "learning_rate": 9.273874770994293e-06,
      "loss": 0.7695,
      "step": 4450
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.9730323602936535,
      "learning_rate": 9.273499816553184e-06,
      "loss": 0.7982,
      "step": 4451
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.294618931645382,
      "learning_rate": 9.273124772912128e-06,
      "loss": 0.8906,
      "step": 4452
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.109562412389788,
      "learning_rate": 9.272749640078948e-06,
      "loss": 0.8686,
      "step": 4453
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.081966968274,
      "learning_rate": 9.272374418061478e-06,
      "loss": 0.8362,
      "step": 4454
    },
    {
      "epoch": 0.4,
      "grad_norm": 9.34008839554988,
      "learning_rate": 9.271999106867546e-06,
      "loss": 0.8251,
      "step": 4455
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.30622389541799,
      "learning_rate": 9.271623706504991e-06,
      "loss": 0.8061,
      "step": 4456
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.617883467896862,
      "learning_rate": 9.271248216981645e-06,
      "loss": 0.8333,
      "step": 4457
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.234373016333952,
      "learning_rate": 9.270872638305347e-06,
      "loss": 0.829,
      "step": 4458
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.21448071901405,
      "learning_rate": 9.270496970483933e-06,
      "loss": 0.8157,
      "step": 4459
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.804136612153014,
      "learning_rate": 9.27012121352525e-06,
      "loss": 0.8365,
      "step": 4460
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.591344943022942,
      "learning_rate": 9.269745367437136e-06,
      "loss": 0.8268,
      "step": 4461
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.119767227870872,
      "learning_rate": 9.26936943222744e-06,
      "loss": 0.8566,
      "step": 4462
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.435060735129838,
      "learning_rate": 9.268993407904004e-06,
      "loss": 0.9066,
      "step": 4463
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.932751072336364,
      "learning_rate": 9.268617294474681e-06,
      "loss": 0.8802,
      "step": 4464
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.405576485531296,
      "learning_rate": 9.268241091947321e-06,
      "loss": 0.8625,
      "step": 4465
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.736086092017031,
      "learning_rate": 9.267864800329776e-06,
      "loss": 0.7884,
      "step": 4466
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.193313356457381,
      "learning_rate": 9.267488419629898e-06,
      "loss": 0.8608,
      "step": 4467
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.2944916780835065,
      "learning_rate": 9.267111949855545e-06,
      "loss": 0.9154,
      "step": 4468
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.238054846844717,
      "learning_rate": 9.266735391014576e-06,
      "loss": 0.8578,
      "step": 4469
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.786347868243233,
      "learning_rate": 9.266358743114849e-06,
      "loss": 0.8121,
      "step": 4470
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.934844210993619,
      "learning_rate": 9.265982006164226e-06,
      "loss": 0.818,
      "step": 4471
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.206032622228986,
      "learning_rate": 9.26560518017057e-06,
      "loss": 0.8797,
      "step": 4472
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.85396257649498,
      "learning_rate": 9.26522826514175e-06,
      "loss": 0.83,
      "step": 4473
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.307579321340161,
      "learning_rate": 9.26485126108563e-06,
      "loss": 0.8419,
      "step": 4474
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.457092279639356,
      "learning_rate": 9.26447416801008e-06,
      "loss": 0.9025,
      "step": 4475
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.790405559107533,
      "learning_rate": 9.26409698592297e-06,
      "loss": 0.8727,
      "step": 4476
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.159003884990671,
      "learning_rate": 9.263719714832173e-06,
      "loss": 0.9158,
      "step": 4477
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.0037296283234705,
      "learning_rate": 9.263342354745565e-06,
      "loss": 0.8316,
      "step": 4478
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.6615278163750125,
      "learning_rate": 9.262964905671021e-06,
      "loss": 0.8481,
      "step": 4479
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.335806548352132,
      "learning_rate": 9.262587367616422e-06,
      "loss": 0.8071,
      "step": 4480
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.756787679015833,
      "learning_rate": 9.262209740589646e-06,
      "loss": 0.8115,
      "step": 4481
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.621887760066865,
      "learning_rate": 9.261832024598576e-06,
      "loss": 0.8272,
      "step": 4482
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.553849919140367,
      "learning_rate": 9.261454219651097e-06,
      "loss": 0.9167,
      "step": 4483
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.928402081631073,
      "learning_rate": 9.26107632575509e-06,
      "loss": 0.8725,
      "step": 4484
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.271218922001157,
      "learning_rate": 9.260698342918448e-06,
      "loss": 0.8599,
      "step": 4485
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.345454270488993,
      "learning_rate": 9.26032027114906e-06,
      "loss": 0.8432,
      "step": 4486
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.357701623476278,
      "learning_rate": 9.259942110454815e-06,
      "loss": 0.9108,
      "step": 4487
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.174531177371189,
      "learning_rate": 9.259563860843608e-06,
      "loss": 0.8065,
      "step": 4488
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.295649704974207,
      "learning_rate": 9.259185522323333e-06,
      "loss": 0.7996,
      "step": 4489
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.421546926449522,
      "learning_rate": 9.258807094901888e-06,
      "loss": 0.8297,
      "step": 4490
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.460814608297221,
      "learning_rate": 9.258428578587174e-06,
      "loss": 0.8173,
      "step": 4491
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.790350480405788,
      "learning_rate": 9.258049973387086e-06,
      "loss": 0.8587,
      "step": 4492
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.921812056973853,
      "learning_rate": 9.25767127930953e-06,
      "loss": 0.8597,
      "step": 4493
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.715286101269676,
      "learning_rate": 9.257292496362412e-06,
      "loss": 0.8034,
      "step": 4494
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.857201392732599,
      "learning_rate": 9.256913624553633e-06,
      "loss": 0.8683,
      "step": 4495
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.495371389209048,
      "learning_rate": 9.256534663891109e-06,
      "loss": 0.8533,
      "step": 4496
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.896327631081806,
      "learning_rate": 9.256155614382744e-06,
      "loss": 0.7857,
      "step": 4497
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.15319544115759,
      "learning_rate": 9.25577647603645e-06,
      "loss": 0.8178,
      "step": 4498
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.30476293926229,
      "learning_rate": 9.255397248860143e-06,
      "loss": 0.8957,
      "step": 4499
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.510767745782046,
      "learning_rate": 9.255017932861739e-06,
      "loss": 0.771,
      "step": 4500
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.22962304386071,
      "learning_rate": 9.254638528049152e-06,
      "loss": 0.8419,
      "step": 4501
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.842237816131411,
      "learning_rate": 9.254259034430304e-06,
      "loss": 0.7841,
      "step": 4502
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.280048032703478,
      "learning_rate": 9.253879452013115e-06,
      "loss": 0.8442,
      "step": 4503
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.2182974561679325,
      "learning_rate": 9.253499780805509e-06,
      "loss": 0.7675,
      "step": 4504
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.187389725325743,
      "learning_rate": 9.253120020815409e-06,
      "loss": 0.8536,
      "step": 4505
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.062254852603474,
      "learning_rate": 9.252740172050743e-06,
      "loss": 0.7873,
      "step": 4506
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.660422862453042,
      "learning_rate": 9.252360234519439e-06,
      "loss": 0.8491,
      "step": 4507
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.5924847606869905,
      "learning_rate": 9.251980208229427e-06,
      "loss": 0.8409,
      "step": 4508
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.48407995674932,
      "learning_rate": 9.251600093188642e-06,
      "loss": 0.8013,
      "step": 4509
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.786509252658318,
      "learning_rate": 9.251219889405014e-06,
      "loss": 0.8487,
      "step": 4510
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.2668582677131806,
      "learning_rate": 9.250839596886482e-06,
      "loss": 0.8405,
      "step": 4511
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.856085928063963,
      "learning_rate": 9.250459215640983e-06,
      "loss": 0.7965,
      "step": 4512
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.181744766791967,
      "learning_rate": 9.250078745676456e-06,
      "loss": 0.8633,
      "step": 4513
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.712586651951771,
      "learning_rate": 9.249698187000841e-06,
      "loss": 0.8348,
      "step": 4514
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.082382429968132,
      "learning_rate": 9.249317539622085e-06,
      "loss": 0.8471,
      "step": 4515
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.939470985014774,
      "learning_rate": 9.24893680354813e-06,
      "loss": 0.8417,
      "step": 4516
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.84107526748278,
      "learning_rate": 9.248555978786923e-06,
      "loss": 0.8463,
      "step": 4517
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.423829657882289,
      "learning_rate": 9.248175065346416e-06,
      "loss": 0.8494,
      "step": 4518
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.4503172952325665,
      "learning_rate": 9.247794063234559e-06,
      "loss": 0.7853,
      "step": 4519
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.407323444107418,
      "learning_rate": 9.247412972459303e-06,
      "loss": 0.92,
      "step": 4520
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.193147010651306,
      "learning_rate": 9.2470317930286e-06,
      "loss": 0.778,
      "step": 4521
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.649815156631681,
      "learning_rate": 9.246650524950411e-06,
      "loss": 0.8484,
      "step": 4522
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.82847858445869,
      "learning_rate": 9.246269168232693e-06,
      "loss": 0.7893,
      "step": 4523
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.68654607910471,
      "learning_rate": 9.245887722883405e-06,
      "loss": 0.8074,
      "step": 4524
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.273545539970788,
      "learning_rate": 9.245506188910509e-06,
      "loss": 0.8886,
      "step": 4525
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.695577564232889,
      "learning_rate": 9.245124566321968e-06,
      "loss": 0.8683,
      "step": 4526
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.221788081239436,
      "learning_rate": 9.244742855125749e-06,
      "loss": 0.8049,
      "step": 4527
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.626461263880344,
      "learning_rate": 9.244361055329818e-06,
      "loss": 0.8965,
      "step": 4528
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.622863026788422,
      "learning_rate": 9.243979166942146e-06,
      "loss": 0.8112,
      "step": 4529
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.849876253183992,
      "learning_rate": 9.243597189970704e-06,
      "loss": 0.7988,
      "step": 4530
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.4470922734163985,
      "learning_rate": 9.24321512442346e-06,
      "loss": 0.8262,
      "step": 4531
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.809774094160614,
      "learning_rate": 9.242832970308398e-06,
      "loss": 0.8593,
      "step": 4532
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.744728442010087,
      "learning_rate": 9.242450727633485e-06,
      "loss": 0.845,
      "step": 4533
    },
    {
      "epoch": 0.4,
      "grad_norm": 10.946559754660253,
      "learning_rate": 9.242068396406706e-06,
      "loss": 0.81,
      "step": 4534
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.375874996676751,
      "learning_rate": 9.24168597663604e-06,
      "loss": 0.7788,
      "step": 4535
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.144671000797504,
      "learning_rate": 9.241303468329467e-06,
      "loss": 0.8297,
      "step": 4536
    },
    {
      "epoch": 0.4,
      "grad_norm": 9.15762997874979,
      "learning_rate": 9.240920871494972e-06,
      "loss": 0.7781,
      "step": 4537
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.24367246562476,
      "learning_rate": 9.240538186140542e-06,
      "loss": 0.7973,
      "step": 4538
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.839971557384537,
      "learning_rate": 9.240155412274163e-06,
      "loss": 0.8287,
      "step": 4539
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.666460818164944,
      "learning_rate": 9.239772549903828e-06,
      "loss": 0.9396,
      "step": 4540
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.286544164530982,
      "learning_rate": 9.239389599037524e-06,
      "loss": 0.8346,
      "step": 4541
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.234382728748434,
      "learning_rate": 9.239006559683246e-06,
      "loss": 0.8207,
      "step": 4542
    },
    {
      "epoch": 0.41,
      "grad_norm": 8.556425511275856,
      "learning_rate": 9.23862343184899e-06,
      "loss": 0.8743,
      "step": 4543
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.960531096786359,
      "learning_rate": 9.238240215542753e-06,
      "loss": 0.8379,
      "step": 4544
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.9028261976243215,
      "learning_rate": 9.237856910772532e-06,
      "loss": 0.8614,
      "step": 4545
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.707959503311916,
      "learning_rate": 9.23747351754633e-06,
      "loss": 0.841,
      "step": 4546
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.836712954200549,
      "learning_rate": 9.237090035872148e-06,
      "loss": 0.8048,
      "step": 4547
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.030422949048965,
      "learning_rate": 9.236706465757987e-06,
      "loss": 0.8285,
      "step": 4548
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.1294575354180525,
      "learning_rate": 9.23632280721186e-06,
      "loss": 0.8369,
      "step": 4549
    },
    {
      "epoch": 0.41,
      "grad_norm": 8.291893863695924,
      "learning_rate": 9.235939060241772e-06,
      "loss": 0.8458,
      "step": 4550
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.405755177790691,
      "learning_rate": 9.235555224855731e-06,
      "loss": 0.8589,
      "step": 4551
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.63150307477105,
      "learning_rate": 9.235171301061752e-06,
      "loss": 0.8356,
      "step": 4552
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.479639959671529,
      "learning_rate": 9.234787288867847e-06,
      "loss": 0.8011,
      "step": 4553
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.189328568776002,
      "learning_rate": 9.23440318828203e-06,
      "loss": 0.8798,
      "step": 4554
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.956024874147626,
      "learning_rate": 9.234018999312321e-06,
      "loss": 0.8664,
      "step": 4555
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.404609216417384,
      "learning_rate": 9.233634721966737e-06,
      "loss": 0.8559,
      "step": 4556
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.826431471290572,
      "learning_rate": 9.233250356253301e-06,
      "loss": 0.7984,
      "step": 4557
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.82008905173775,
      "learning_rate": 9.232865902180032e-06,
      "loss": 0.8184,
      "step": 4558
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.896683001128371,
      "learning_rate": 9.232481359754959e-06,
      "loss": 0.9005,
      "step": 4559
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.010135887965424,
      "learning_rate": 9.232096728986106e-06,
      "loss": 0.7915,
      "step": 4560
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.662167349470587,
      "learning_rate": 9.231712009881501e-06,
      "loss": 0.8354,
      "step": 4561
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.607293285105917,
      "learning_rate": 9.231327202449177e-06,
      "loss": 0.8039,
      "step": 4562
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.329431041390225,
      "learning_rate": 9.230942306697164e-06,
      "loss": 0.8862,
      "step": 4563
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.3813538023988245,
      "learning_rate": 9.230557322633494e-06,
      "loss": 0.9056,
      "step": 4564
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.691697842566199,
      "learning_rate": 9.230172250266206e-06,
      "loss": 0.8901,
      "step": 4565
    },
    {
      "epoch": 0.41,
      "grad_norm": 8.036077988664768,
      "learning_rate": 9.229787089603337e-06,
      "loss": 0.8477,
      "step": 4566
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.58321789079827,
      "learning_rate": 9.229401840652923e-06,
      "loss": 0.817,
      "step": 4567
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.686150347577597,
      "learning_rate": 9.22901650342301e-06,
      "loss": 0.831,
      "step": 4568
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.438881342272416,
      "learning_rate": 9.228631077921637e-06,
      "loss": 0.8572,
      "step": 4569
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.475035050687916,
      "learning_rate": 9.228245564156852e-06,
      "loss": 0.8276,
      "step": 4570
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.90008814898264,
      "learning_rate": 9.2278599621367e-06,
      "loss": 0.7754,
      "step": 4571
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.88885779427092,
      "learning_rate": 9.227474271869233e-06,
      "loss": 0.8844,
      "step": 4572
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.261933266481776,
      "learning_rate": 9.227088493362494e-06,
      "loss": 0.8202,
      "step": 4573
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.402536838503419,
      "learning_rate": 9.226702626624544e-06,
      "loss": 0.8992,
      "step": 4574
    },
    {
      "epoch": 0.41,
      "grad_norm": 8.770798751245627,
      "learning_rate": 9.226316671663431e-06,
      "loss": 0.8266,
      "step": 4575
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.519834236116662,
      "learning_rate": 9.225930628487211e-06,
      "loss": 0.8517,
      "step": 4576
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.0478768363276245,
      "learning_rate": 9.225544497103947e-06,
      "loss": 0.8423,
      "step": 4577
    },
    {
      "epoch": 0.41,
      "grad_norm": 8.353613508116252,
      "learning_rate": 9.225158277521695e-06,
      "loss": 0.8092,
      "step": 4578
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.949163525123579,
      "learning_rate": 9.224771969748517e-06,
      "loss": 0.8947,
      "step": 4579
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.932634832778146,
      "learning_rate": 9.224385573792473e-06,
      "loss": 0.8463,
      "step": 4580
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.7475945243768845,
      "learning_rate": 9.223999089661635e-06,
      "loss": 0.8433,
      "step": 4581
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.884380892089099,
      "learning_rate": 9.223612517364064e-06,
      "loss": 0.8622,
      "step": 4582
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.612144580129278,
      "learning_rate": 9.223225856907834e-06,
      "loss": 0.8212,
      "step": 4583
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.908675815504412,
      "learning_rate": 9.22283910830101e-06,
      "loss": 0.8037,
      "step": 4584
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.512613757981434,
      "learning_rate": 9.222452271551669e-06,
      "loss": 0.8767,
      "step": 4585
    },
    {
      "epoch": 0.41,
      "grad_norm": 8.0025005645414,
      "learning_rate": 9.222065346667883e-06,
      "loss": 0.8208,
      "step": 4586
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.4600158957402,
      "learning_rate": 9.221678333657731e-06,
      "loss": 0.7815,
      "step": 4587
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.318225675733331,
      "learning_rate": 9.221291232529289e-06,
      "loss": 0.8081,
      "step": 4588
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.537378300024944,
      "learning_rate": 9.220904043290635e-06,
      "loss": 0.8256,
      "step": 4589
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.794534035816491,
      "learning_rate": 9.220516765949852e-06,
      "loss": 0.8297,
      "step": 4590
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.3698229396578725,
      "learning_rate": 9.220129400515027e-06,
      "loss": 0.8077,
      "step": 4591
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.711427577827574,
      "learning_rate": 9.21974194699424e-06,
      "loss": 0.7931,
      "step": 4592
    },
    {
      "epoch": 0.41,
      "grad_norm": 8.305680097640726,
      "learning_rate": 9.219354405395584e-06,
      "loss": 0.8309,
      "step": 4593
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.037205886059536,
      "learning_rate": 9.218966775727144e-06,
      "loss": 0.801,
      "step": 4594
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.216392573814221,
      "learning_rate": 9.218579057997011e-06,
      "loss": 0.8539,
      "step": 4595
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.784012827658881,
      "learning_rate": 9.218191252213279e-06,
      "loss": 0.8278,
      "step": 4596
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.256887071164107,
      "learning_rate": 9.217803358384043e-06,
      "loss": 0.8466,
      "step": 4597
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.740580377308551,
      "learning_rate": 9.217415376517397e-06,
      "loss": 0.817,
      "step": 4598
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.713056279429526,
      "learning_rate": 9.217027306621442e-06,
      "loss": 0.8491,
      "step": 4599
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.3650550937095405,
      "learning_rate": 9.216639148704277e-06,
      "loss": 0.9137,
      "step": 4600
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.592982110412086,
      "learning_rate": 9.216250902774004e-06,
      "loss": 0.8564,
      "step": 4601
    },
    {
      "epoch": 0.41,
      "grad_norm": 8.119878921422362,
      "learning_rate": 9.215862568838727e-06,
      "loss": 0.8773,
      "step": 4602
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.263948718499047,
      "learning_rate": 9.215474146906552e-06,
      "loss": 0.7967,
      "step": 4603
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.64974423929634,
      "learning_rate": 9.215085636985586e-06,
      "loss": 0.8212,
      "step": 4604
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.916795611874067,
      "learning_rate": 9.214697039083937e-06,
      "loss": 0.9272,
      "step": 4605
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.068924874527984,
      "learning_rate": 9.214308353209718e-06,
      "loss": 0.8338,
      "step": 4606
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.920273975271022,
      "learning_rate": 9.213919579371043e-06,
      "loss": 0.8347,
      "step": 4607
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.54222852905615,
      "learning_rate": 9.213530717576023e-06,
      "loss": 0.8239,
      "step": 4608
    },
    {
      "epoch": 0.41,
      "grad_norm": 9.20698075647077,
      "learning_rate": 9.213141767832777e-06,
      "loss": 0.8713,
      "step": 4609
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.2198623847524095,
      "learning_rate": 9.212752730149424e-06,
      "loss": 0.838,
      "step": 4610
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.336151561267071,
      "learning_rate": 9.212363604534081e-06,
      "loss": 0.7938,
      "step": 4611
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.541458810217811,
      "learning_rate": 9.211974390994875e-06,
      "loss": 0.8429,
      "step": 4612
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.524534528245318,
      "learning_rate": 9.211585089539928e-06,
      "loss": 0.8542,
      "step": 4613
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.667307202933177,
      "learning_rate": 9.211195700177364e-06,
      "loss": 0.8256,
      "step": 4614
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.523153746195941,
      "learning_rate": 9.210806222915312e-06,
      "loss": 0.8609,
      "step": 4615
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.189639886734872,
      "learning_rate": 9.210416657761903e-06,
      "loss": 0.7957,
      "step": 4616
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.256961265546148,
      "learning_rate": 9.210027004725267e-06,
      "loss": 0.9084,
      "step": 4617
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.29107621214498,
      "learning_rate": 9.209637263813536e-06,
      "loss": 0.8536,
      "step": 4618
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.844041776292312,
      "learning_rate": 9.209247435034848e-06,
      "loss": 0.82,
      "step": 4619
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.172019367627463,
      "learning_rate": 9.208857518397337e-06,
      "loss": 0.755,
      "step": 4620
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.7100666384306535,
      "learning_rate": 9.208467513909141e-06,
      "loss": 0.8701,
      "step": 4621
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.271468745916825,
      "learning_rate": 9.208077421578403e-06,
      "loss": 0.7857,
      "step": 4622
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.28144293408788,
      "learning_rate": 9.207687241413263e-06,
      "loss": 0.8624,
      "step": 4623
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.630682910704307,
      "learning_rate": 9.207296973421869e-06,
      "loss": 0.7886,
      "step": 4624
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.579614243262499,
      "learning_rate": 9.206906617612364e-06,
      "loss": 0.8378,
      "step": 4625
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.717739067057811,
      "learning_rate": 9.206516173992895e-06,
      "loss": 0.8611,
      "step": 4626
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.869163635461895,
      "learning_rate": 9.206125642571613e-06,
      "loss": 0.8275,
      "step": 4627
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.260387305419084,
      "learning_rate": 9.20573502335667e-06,
      "loss": 0.7527,
      "step": 4628
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.11694951709995,
      "learning_rate": 9.20534431635622e-06,
      "loss": 0.8224,
      "step": 4629
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.671453393637832,
      "learning_rate": 9.204953521578415e-06,
      "loss": 0.804,
      "step": 4630
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.084484566426668,
      "learning_rate": 9.204562639031414e-06,
      "loss": 0.8027,
      "step": 4631
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.191963861879142,
      "learning_rate": 9.204171668723377e-06,
      "loss": 0.8724,
      "step": 4632
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.742019131577059,
      "learning_rate": 9.203780610662462e-06,
      "loss": 0.8161,
      "step": 4633
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.517110353378128,
      "learning_rate": 9.203389464856835e-06,
      "loss": 0.8465,
      "step": 4634
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.557451909605078,
      "learning_rate": 9.202998231314656e-06,
      "loss": 0.7583,
      "step": 4635
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.255230740605198,
      "learning_rate": 9.202606910044094e-06,
      "loss": 0.8325,
      "step": 4636
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.9381139425849,
      "learning_rate": 9.202215501053318e-06,
      "loss": 0.7769,
      "step": 4637
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.126971115325527,
      "learning_rate": 9.201824004350495e-06,
      "loss": 0.8265,
      "step": 4638
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.14302129626966,
      "learning_rate": 9.201432419943796e-06,
      "loss": 0.8527,
      "step": 4639
    },
    {
      "epoch": 0.41,
      "grad_norm": 9.047350643744016,
      "learning_rate": 9.2010407478414e-06,
      "loss": 0.7991,
      "step": 4640
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.351892687375672,
      "learning_rate": 9.200648988051474e-06,
      "loss": 0.8875,
      "step": 4641
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.7312209621873595,
      "learning_rate": 9.200257140582202e-06,
      "loss": 0.8918,
      "step": 4642
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.824827510372964,
      "learning_rate": 9.199865205441762e-06,
      "loss": 0.7928,
      "step": 4643
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.992042857263485,
      "learning_rate": 9.199473182638332e-06,
      "loss": 0.8176,
      "step": 4644
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.466490418297384,
      "learning_rate": 9.199081072180097e-06,
      "loss": 0.9162,
      "step": 4645
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.108331579051042,
      "learning_rate": 9.198688874075237e-06,
      "loss": 0.8075,
      "step": 4646
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.440688074380756,
      "learning_rate": 9.198296588331945e-06,
      "loss": 0.89,
      "step": 4647
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.278771069180701,
      "learning_rate": 9.197904214958405e-06,
      "loss": 0.834,
      "step": 4648
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.020776656374454,
      "learning_rate": 9.197511753962806e-06,
      "loss": 0.7754,
      "step": 4649
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.486778318272757,
      "learning_rate": 9.197119205353344e-06,
      "loss": 0.8648,
      "step": 4650
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.163259032246637,
      "learning_rate": 9.196726569138211e-06,
      "loss": 0.7614,
      "step": 4651
    },
    {
      "epoch": 0.42,
      "grad_norm": 8.42909766315542,
      "learning_rate": 9.196333845325597e-06,
      "loss": 0.8476,
      "step": 4652
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.298328093684551,
      "learning_rate": 9.195941033923707e-06,
      "loss": 0.8435,
      "step": 4653
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.732883823542008,
      "learning_rate": 9.195548134940737e-06,
      "loss": 0.8053,
      "step": 4654
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.135783926416529,
      "learning_rate": 9.195155148384888e-06,
      "loss": 0.7672,
      "step": 4655
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.086962787080362,
      "learning_rate": 9.19476207426436e-06,
      "loss": 0.7725,
      "step": 4656
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.24128555485995,
      "learning_rate": 9.194368912587363e-06,
      "loss": 0.7839,
      "step": 4657
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.909360972369439,
      "learning_rate": 9.1939756633621e-06,
      "loss": 0.8338,
      "step": 4658
    },
    {
      "epoch": 0.42,
      "grad_norm": 8.342587624993977,
      "learning_rate": 9.193582326596779e-06,
      "loss": 0.7846,
      "step": 4659
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.600642787529243,
      "learning_rate": 9.19318890229961e-06,
      "loss": 0.8406,
      "step": 4660
    },
    {
      "epoch": 0.42,
      "grad_norm": 8.68669382010525,
      "learning_rate": 9.192795390478806e-06,
      "loss": 0.8319,
      "step": 4661
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.217584273366493,
      "learning_rate": 9.19240179114258e-06,
      "loss": 0.8393,
      "step": 4662
    },
    {
      "epoch": 0.42,
      "grad_norm": 8.019891615918773,
      "learning_rate": 9.19200810429915e-06,
      "loss": 0.8506,
      "step": 4663
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.662247761179312,
      "learning_rate": 9.191614329956728e-06,
      "loss": 0.8422,
      "step": 4664
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.745001954240607,
      "learning_rate": 9.191220468123538e-06,
      "loss": 0.7695,
      "step": 4665
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.467657801754695,
      "learning_rate": 9.1908265188078e-06,
      "loss": 0.8456,
      "step": 4666
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.768773580694558,
      "learning_rate": 9.190432482017734e-06,
      "loss": 0.7899,
      "step": 4667
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.308218653628511,
      "learning_rate": 9.19003835776157e-06,
      "loss": 0.8584,
      "step": 4668
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.924914123839482,
      "learning_rate": 9.18964414604753e-06,
      "loss": 0.8314,
      "step": 4669
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.42287658367356,
      "learning_rate": 9.189249846883841e-06,
      "loss": 0.8423,
      "step": 4670
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.036877929079618,
      "learning_rate": 9.188855460278738e-06,
      "loss": 0.9313,
      "step": 4671
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.22465979459289,
      "learning_rate": 9.188460986240451e-06,
      "loss": 0.8601,
      "step": 4672
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.123062176567409,
      "learning_rate": 9.188066424777214e-06,
      "loss": 0.8594,
      "step": 4673
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.4438097888581725,
      "learning_rate": 9.187671775897261e-06,
      "loss": 0.8169,
      "step": 4674
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.4852635164904555,
      "learning_rate": 9.18727703960883e-06,
      "loss": 0.7808,
      "step": 4675
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.818026606187768,
      "learning_rate": 9.186882215920163e-06,
      "loss": 0.821,
      "step": 4676
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.5844334815031225,
      "learning_rate": 9.186487304839496e-06,
      "loss": 0.8387,
      "step": 4677
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.2162828075218295,
      "learning_rate": 9.186092306375076e-06,
      "loss": 0.8179,
      "step": 4678
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.780998978306586,
      "learning_rate": 9.185697220535147e-06,
      "loss": 0.8427,
      "step": 4679
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.240944431682982,
      "learning_rate": 9.185302047327955e-06,
      "loss": 0.8637,
      "step": 4680
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.140846689883739,
      "learning_rate": 9.184906786761748e-06,
      "loss": 0.8378,
      "step": 4681
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.653230751909554,
      "learning_rate": 9.184511438844777e-06,
      "loss": 0.815,
      "step": 4682
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.132026678794383,
      "learning_rate": 9.184116003585293e-06,
      "loss": 0.8134,
      "step": 4683
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.332322945799071,
      "learning_rate": 9.183720480991552e-06,
      "loss": 0.8287,
      "step": 4684
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.603455083524731,
      "learning_rate": 9.183324871071806e-06,
      "loss": 0.8304,
      "step": 4685
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.778333443927217,
      "learning_rate": 9.182929173834314e-06,
      "loss": 0.8071,
      "step": 4686
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.196576039852963,
      "learning_rate": 9.182533389287338e-06,
      "loss": 0.8096,
      "step": 4687
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.678430904212422,
      "learning_rate": 9.182137517439138e-06,
      "loss": 0.7657,
      "step": 4688
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.30986348863734,
      "learning_rate": 9.181741558297973e-06,
      "loss": 0.8759,
      "step": 4689
    },
    {
      "epoch": 0.42,
      "grad_norm": 10.362392867464221,
      "learning_rate": 9.181345511872113e-06,
      "loss": 0.8088,
      "step": 4690
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.113062511617782,
      "learning_rate": 9.180949378169822e-06,
      "loss": 0.8406,
      "step": 4691
    },
    {
      "epoch": 0.42,
      "grad_norm": 8.001669995935284,
      "learning_rate": 9.18055315719937e-06,
      "loss": 0.8064,
      "step": 4692
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.837290097287818,
      "learning_rate": 9.180156848969024e-06,
      "loss": 0.8201,
      "step": 4693
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.977175091855017,
      "learning_rate": 9.17976045348706e-06,
      "loss": 0.9142,
      "step": 4694
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.265072693362719,
      "learning_rate": 9.179363970761748e-06,
      "loss": 0.8345,
      "step": 4695
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.350564551550255,
      "learning_rate": 9.178967400801367e-06,
      "loss": 0.838,
      "step": 4696
    },
    {
      "epoch": 0.42,
      "grad_norm": 8.319501784920893,
      "learning_rate": 9.178570743614192e-06,
      "loss": 0.8714,
      "step": 4697
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.499125368322284,
      "learning_rate": 9.178173999208506e-06,
      "loss": 0.8316,
      "step": 4698
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.03822431203044,
      "learning_rate": 9.177777167592588e-06,
      "loss": 0.6912,
      "step": 4699
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.743944262018504,
      "learning_rate": 9.177380248774718e-06,
      "loss": 0.817,
      "step": 4700
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.4934930510815665,
      "learning_rate": 9.176983242763184e-06,
      "loss": 0.8252,
      "step": 4701
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.811376148267023,
      "learning_rate": 9.176586149566274e-06,
      "loss": 0.8985,
      "step": 4702
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.04859307179697,
      "learning_rate": 9.176188969192275e-06,
      "loss": 0.8933,
      "step": 4703
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.467906847842432,
      "learning_rate": 9.175791701649476e-06,
      "loss": 0.82,
      "step": 4704
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.999325723413462,
      "learning_rate": 9.17539434694617e-06,
      "loss": 0.792,
      "step": 4705
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.991976808612682,
      "learning_rate": 9.17499690509065e-06,
      "loss": 0.8442,
      "step": 4706
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.520590685550164,
      "learning_rate": 9.174599376091215e-06,
      "loss": 0.851,
      "step": 4707
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.679056358880233,
      "learning_rate": 9.174201759956158e-06,
      "loss": 0.8756,
      "step": 4708
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.911535175533622,
      "learning_rate": 9.173804056693781e-06,
      "loss": 0.8224,
      "step": 4709
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.875786909931423,
      "learning_rate": 9.173406266312386e-06,
      "loss": 0.7981,
      "step": 4710
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.252866175679899,
      "learning_rate": 9.173008388820273e-06,
      "loss": 0.7933,
      "step": 4711
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.525629012749577,
      "learning_rate": 9.17261042422575e-06,
      "loss": 0.8385,
      "step": 4712
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.670774964646604,
      "learning_rate": 9.172212372537123e-06,
      "loss": 0.8983,
      "step": 4713
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.059195413178181,
      "learning_rate": 9.171814233762698e-06,
      "loss": 0.8536,
      "step": 4714
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.714048120285158,
      "learning_rate": 9.171416007910787e-06,
      "loss": 0.8427,
      "step": 4715
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.44401816765405,
      "learning_rate": 9.171017694989704e-06,
      "loss": 0.857,
      "step": 4716
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.70308517100645,
      "learning_rate": 9.170619295007759e-06,
      "loss": 0.8446,
      "step": 4717
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.982646097468416,
      "learning_rate": 9.170220807973272e-06,
      "loss": 0.8179,
      "step": 4718
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.661065177151935,
      "learning_rate": 9.169822233894555e-06,
      "loss": 0.8797,
      "step": 4719
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.5834532028960275,
      "learning_rate": 9.169423572779934e-06,
      "loss": 0.8209,
      "step": 4720
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.396275344176002,
      "learning_rate": 9.169024824637725e-06,
      "loss": 0.8675,
      "step": 4721
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.313810103392968,
      "learning_rate": 9.168625989476253e-06,
      "loss": 0.8381,
      "step": 4722
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.645864530196455,
      "learning_rate": 9.16822706730384e-06,
      "loss": 0.8302,
      "step": 4723
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.453695941460281,
      "learning_rate": 9.167828058128819e-06,
      "loss": 0.8403,
      "step": 4724
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.663758337860535,
      "learning_rate": 9.167428961959514e-06,
      "loss": 0.8575,
      "step": 4725
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.894484204774418,
      "learning_rate": 9.167029778804255e-06,
      "loss": 0.8878,
      "step": 4726
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.2405249285199975,
      "learning_rate": 9.166630508671374e-06,
      "loss": 0.8807,
      "step": 4727
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.9216020963619425,
      "learning_rate": 9.166231151569209e-06,
      "loss": 0.8942,
      "step": 4728
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.539897808644549,
      "learning_rate": 9.16583170750609e-06,
      "loss": 0.9102,
      "step": 4729
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.5400170078339555,
      "learning_rate": 9.16543217649036e-06,
      "loss": 0.8575,
      "step": 4730
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.052055944275131,
      "learning_rate": 9.165032558530353e-06,
      "loss": 0.9064,
      "step": 4731
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.59272018370884,
      "learning_rate": 9.164632853634412e-06,
      "loss": 0.8693,
      "step": 4732
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.908606450948162,
      "learning_rate": 9.16423306181088e-06,
      "loss": 0.7869,
      "step": 4733
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.562565151775326,
      "learning_rate": 9.163833183068105e-06,
      "loss": 0.89,
      "step": 4734
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.912969043096607,
      "learning_rate": 9.163433217414431e-06,
      "loss": 0.8339,
      "step": 4735
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.803424779387007,
      "learning_rate": 9.163033164858204e-06,
      "loss": 0.8829,
      "step": 4736
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.774843240823779,
      "learning_rate": 9.16263302540778e-06,
      "loss": 0.7715,
      "step": 4737
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.981890811756283,
      "learning_rate": 9.162232799071505e-06,
      "loss": 0.8624,
      "step": 4738
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.470433102993778,
      "learning_rate": 9.161832485857736e-06,
      "loss": 0.8495,
      "step": 4739
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.754494412402604,
      "learning_rate": 9.161432085774826e-06,
      "loss": 0.8733,
      "step": 4740
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.149976065165343,
      "learning_rate": 9.161031598831139e-06,
      "loss": 0.8768,
      "step": 4741
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.3280547097008855,
      "learning_rate": 9.160631025035026e-06,
      "loss": 0.7943,
      "step": 4742
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.241206134376276,
      "learning_rate": 9.160230364394852e-06,
      "loss": 0.7718,
      "step": 4743
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.121132152060691,
      "learning_rate": 9.159829616918982e-06,
      "loss": 0.8631,
      "step": 4744
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.337349700946734,
      "learning_rate": 9.159428782615779e-06,
      "loss": 0.7977,
      "step": 4745
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.621838942914706,
      "learning_rate": 9.159027861493608e-06,
      "loss": 0.7948,
      "step": 4746
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.642965770128866,
      "learning_rate": 9.158626853560839e-06,
      "loss": 0.8606,
      "step": 4747
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.6556209701010145,
      "learning_rate": 9.158225758825841e-06,
      "loss": 0.868,
      "step": 4748
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.454390074640307,
      "learning_rate": 9.157824577296987e-06,
      "loss": 0.8533,
      "step": 4749
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.632472262392171,
      "learning_rate": 9.15742330898265e-06,
      "loss": 0.8153,
      "step": 4750
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.485322466204349,
      "learning_rate": 9.157021953891207e-06,
      "loss": 0.8347,
      "step": 4751
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.082983454006973,
      "learning_rate": 9.156620512031033e-06,
      "loss": 0.8395,
      "step": 4752
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.124842004751423,
      "learning_rate": 9.15621898341051e-06,
      "loss": 0.8083,
      "step": 4753
    },
    {
      "epoch": 0.42,
      "grad_norm": 10.345170486231046,
      "learning_rate": 9.155817368038017e-06,
      "loss": 0.9008,
      "step": 4754
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.965529425096823,
      "learning_rate": 9.155415665921938e-06,
      "loss": 0.7945,
      "step": 4755
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.523920055144245,
      "learning_rate": 9.155013877070657e-06,
      "loss": 0.9048,
      "step": 4756
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.150256100001417,
      "learning_rate": 9.15461200149256e-06,
      "loss": 0.8731,
      "step": 4757
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.220728975424585,
      "learning_rate": 9.154210039196037e-06,
      "loss": 0.7604,
      "step": 4758
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.672365428656922,
      "learning_rate": 9.153807990189475e-06,
      "loss": 0.8065,
      "step": 4759
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.800048114974959,
      "learning_rate": 9.15340585448127e-06,
      "loss": 0.8596,
      "step": 4760
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.81196063105043,
      "learning_rate": 9.153003632079812e-06,
      "loss": 0.8403,
      "step": 4761
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.83589784933157,
      "learning_rate": 9.1526013229935e-06,
      "loss": 0.8167,
      "step": 4762
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.51862353525165,
      "learning_rate": 9.15219892723073e-06,
      "loss": 0.7817,
      "step": 4763
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.480691028693473,
      "learning_rate": 9.1517964447999e-06,
      "loss": 0.8806,
      "step": 4764
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.0450123865421,
      "learning_rate": 9.15139387570941e-06,
      "loss": 0.8278,
      "step": 4765
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.296029513174919,
      "learning_rate": 9.150991219967663e-06,
      "loss": 0.8919,
      "step": 4766
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.980978718104485,
      "learning_rate": 9.150588477583067e-06,
      "loss": 0.8677,
      "step": 4767
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.9910759133192055,
      "learning_rate": 9.150185648564025e-06,
      "loss": 0.7479,
      "step": 4768
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.724675384546899,
      "learning_rate": 9.149782732918948e-06,
      "loss": 0.8443,
      "step": 4769
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.598570563913732,
      "learning_rate": 9.149379730656243e-06,
      "loss": 0.8436,
      "step": 4770
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.595044656557839,
      "learning_rate": 9.148976641784324e-06,
      "loss": 0.8133,
      "step": 4771
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.956477102004142,
      "learning_rate": 9.148573466311602e-06,
      "loss": 0.849,
      "step": 4772
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.818363610099549,
      "learning_rate": 9.148170204246496e-06,
      "loss": 0.8941,
      "step": 4773
    },
    {
      "epoch": 0.43,
      "grad_norm": 8.190669552081799,
      "learning_rate": 9.14776685559742e-06,
      "loss": 0.8661,
      "step": 4774
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.345349705564131,
      "learning_rate": 9.147363420372794e-06,
      "loss": 0.8202,
      "step": 4775
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.568846597915994,
      "learning_rate": 9.146959898581042e-06,
      "loss": 0.8467,
      "step": 4776
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.3493734136729225,
      "learning_rate": 9.146556290230582e-06,
      "loss": 0.8407,
      "step": 4777
    },
    {
      "epoch": 0.43,
      "grad_norm": 9.225867706035114,
      "learning_rate": 9.14615259532984e-06,
      "loss": 0.8335,
      "step": 4778
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.083970305487933,
      "learning_rate": 9.145748813887243e-06,
      "loss": 0.8362,
      "step": 4779
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.40676860100397,
      "learning_rate": 9.145344945911217e-06,
      "loss": 0.7882,
      "step": 4780
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.209606567690158,
      "learning_rate": 9.144940991410196e-06,
      "loss": 0.8521,
      "step": 4781
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.206305356943972,
      "learning_rate": 9.144536950392606e-06,
      "loss": 0.7845,
      "step": 4782
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.3792686262589475,
      "learning_rate": 9.144132822866886e-06,
      "loss": 0.8385,
      "step": 4783
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.734815921185136,
      "learning_rate": 9.143728608841469e-06,
      "loss": 0.8237,
      "step": 4784
    },
    {
      "epoch": 0.43,
      "grad_norm": 8.070390396417901,
      "learning_rate": 9.14332430832479e-06,
      "loss": 0.81,
      "step": 4785
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.476656473164878,
      "learning_rate": 9.142919921325291e-06,
      "loss": 0.8128,
      "step": 4786
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.143225017711418,
      "learning_rate": 9.14251544785141e-06,
      "loss": 0.9165,
      "step": 4787
    },
    {
      "epoch": 0.43,
      "grad_norm": 8.370541604439898,
      "learning_rate": 9.142110887911593e-06,
      "loss": 0.8762,
      "step": 4788
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.840805087338073,
      "learning_rate": 9.141706241514281e-06,
      "loss": 0.811,
      "step": 4789
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.552757841728568,
      "learning_rate": 9.141301508667923e-06,
      "loss": 0.8021,
      "step": 4790
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.01865046269057,
      "learning_rate": 9.140896689380964e-06,
      "loss": 0.8641,
      "step": 4791
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.826348439505471,
      "learning_rate": 9.140491783661856e-06,
      "loss": 0.7974,
      "step": 4792
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.359400314798292,
      "learning_rate": 9.140086791519047e-06,
      "loss": 0.7876,
      "step": 4793
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.936475302558658,
      "learning_rate": 9.139681712960994e-06,
      "loss": 0.8361,
      "step": 4794
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.175232159373111,
      "learning_rate": 9.139276547996153e-06,
      "loss": 0.767,
      "step": 4795
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.855227244206391,
      "learning_rate": 9.138871296632977e-06,
      "loss": 0.8444,
      "step": 4796
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.522852548848016,
      "learning_rate": 9.138465958879928e-06,
      "loss": 0.8045,
      "step": 4797
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.0357053652451755,
      "learning_rate": 9.138060534745463e-06,
      "loss": 0.7962,
      "step": 4798
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.581275590192676,
      "learning_rate": 9.137655024238049e-06,
      "loss": 0.8259,
      "step": 4799
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.0547731548834,
      "learning_rate": 9.137249427366145e-06,
      "loss": 0.8104,
      "step": 4800
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.051660682139256,
      "learning_rate": 9.136843744138224e-06,
      "loss": 0.8516,
      "step": 4801
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.4830719191996,
      "learning_rate": 9.136437974562745e-06,
      "loss": 0.8605,
      "step": 4802
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.8539992544952755,
      "learning_rate": 9.136032118648184e-06,
      "loss": 0.815,
      "step": 4803
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.766836564654038,
      "learning_rate": 9.13562617640301e-06,
      "loss": 0.8558,
      "step": 4804
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.96942103059374,
      "learning_rate": 9.135220147835694e-06,
      "loss": 0.8614,
      "step": 4805
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.14138807494017,
      "learning_rate": 9.134814032954717e-06,
      "loss": 0.8396,
      "step": 4806
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.388331877751764,
      "learning_rate": 9.13440783176855e-06,
      "loss": 0.8933,
      "step": 4807
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.918892878758849,
      "learning_rate": 9.134001544285676e-06,
      "loss": 0.8805,
      "step": 4808
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.305146788955883,
      "learning_rate": 9.133595170514571e-06,
      "loss": 0.7844,
      "step": 4809
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.214109285169786,
      "learning_rate": 9.133188710463721e-06,
      "loss": 0.8441,
      "step": 4810
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.706609196675064,
      "learning_rate": 9.132782164141606e-06,
      "loss": 0.7939,
      "step": 4811
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.269334231613863,
      "learning_rate": 9.132375531556714e-06,
      "loss": 0.7106,
      "step": 4812
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.602309220520748,
      "learning_rate": 9.131968812717534e-06,
      "loss": 0.7897,
      "step": 4813
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.717450952703593,
      "learning_rate": 9.131562007632555e-06,
      "loss": 0.8179,
      "step": 4814
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.2638076287388635,
      "learning_rate": 9.131155116310263e-06,
      "loss": 0.9658,
      "step": 4815
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.830889701606198,
      "learning_rate": 9.130748138759157e-06,
      "loss": 0.8031,
      "step": 4816
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.399287363211557,
      "learning_rate": 9.130341074987732e-06,
      "loss": 0.8251,
      "step": 4817
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.768180085559075,
      "learning_rate": 9.12993392500448e-06,
      "loss": 0.8164,
      "step": 4818
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.419118110237655,
      "learning_rate": 9.129526688817902e-06,
      "loss": 0.8303,
      "step": 4819
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.259236668039366,
      "learning_rate": 9.129119366436498e-06,
      "loss": 0.8583,
      "step": 4820
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.2995669792411455,
      "learning_rate": 9.128711957868772e-06,
      "loss": 0.8614,
      "step": 4821
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.854825526177955,
      "learning_rate": 9.128304463123224e-06,
      "loss": 0.8434,
      "step": 4822
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.522533667546708,
      "learning_rate": 9.127896882208359e-06,
      "loss": 0.895,
      "step": 4823
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.279331814772121,
      "learning_rate": 9.127489215132688e-06,
      "loss": 0.8133,
      "step": 4824
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.070670671629508,
      "learning_rate": 9.127081461904719e-06,
      "loss": 0.8714,
      "step": 4825
    },
    {
      "epoch": 0.43,
      "grad_norm": 8.435640450454244,
      "learning_rate": 9.126673622532963e-06,
      "loss": 0.8625,
      "step": 4826
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.9742167599247455,
      "learning_rate": 9.126265697025932e-06,
      "loss": 0.8498,
      "step": 4827
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.896903810621668,
      "learning_rate": 9.125857685392141e-06,
      "loss": 0.8068,
      "step": 4828
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.79387704569665,
      "learning_rate": 9.125449587640107e-06,
      "loss": 0.8017,
      "step": 4829
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.672882399584768,
      "learning_rate": 9.125041403778345e-06,
      "loss": 0.8444,
      "step": 4830
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.562963469245397,
      "learning_rate": 9.12463313381538e-06,
      "loss": 0.7673,
      "step": 4831
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.097294703146544,
      "learning_rate": 9.12422477775973e-06,
      "loss": 0.8075,
      "step": 4832
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.243248426771218,
      "learning_rate": 9.12381633561992e-06,
      "loss": 0.8572,
      "step": 4833
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.325109696860835,
      "learning_rate": 9.123407807404474e-06,
      "loss": 0.8252,
      "step": 4834
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.7054567700402385,
      "learning_rate": 9.122999193121922e-06,
      "loss": 0.8377,
      "step": 4835
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.569833453098052,
      "learning_rate": 9.12259049278079e-06,
      "loss": 0.8157,
      "step": 4836
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.762251706171692,
      "learning_rate": 9.12218170638961e-06,
      "loss": 0.8025,
      "step": 4837
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.610263071413296,
      "learning_rate": 9.121772833956915e-06,
      "loss": 0.8447,
      "step": 4838
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.819459325317752,
      "learning_rate": 9.121363875491236e-06,
      "loss": 0.8955,
      "step": 4839
    },
    {
      "epoch": 0.43,
      "grad_norm": 8.065031663098662,
      "learning_rate": 9.120954831001114e-06,
      "loss": 0.8199,
      "step": 4840
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.128598775592627,
      "learning_rate": 9.120545700495085e-06,
      "loss": 0.8263,
      "step": 4841
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.994503596343405,
      "learning_rate": 9.120136483981688e-06,
      "loss": 0.8711,
      "step": 4842
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.591603739853564,
      "learning_rate": 9.119727181469462e-06,
      "loss": 0.7576,
      "step": 4843
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.422901151681481,
      "learning_rate": 9.119317792966957e-06,
      "loss": 0.8358,
      "step": 4844
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.191764691810066,
      "learning_rate": 9.118908318482711e-06,
      "loss": 0.8594,
      "step": 4845
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.156870919383938,
      "learning_rate": 9.118498758025277e-06,
      "loss": 0.8578,
      "step": 4846
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.660505668626911,
      "learning_rate": 9.1180891116032e-06,
      "loss": 0.7698,
      "step": 4847
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.909224530873166,
      "learning_rate": 9.117679379225031e-06,
      "loss": 0.8256,
      "step": 4848
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.496760772024197,
      "learning_rate": 9.117269560899322e-06,
      "loss": 0.7821,
      "step": 4849
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.982366068952116,
      "learning_rate": 9.116859656634629e-06,
      "loss": 0.8691,
      "step": 4850
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.436438914426638,
      "learning_rate": 9.116449666439504e-06,
      "loss": 0.793,
      "step": 4851
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.933256549510221,
      "learning_rate": 9.116039590322508e-06,
      "loss": 0.8476,
      "step": 4852
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.438844150700897,
      "learning_rate": 9.1156294282922e-06,
      "loss": 0.8428,
      "step": 4853
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.89770910442735,
      "learning_rate": 9.11521918035714e-06,
      "loss": 0.7486,
      "step": 4854
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.496095920576283,
      "learning_rate": 9.114808846525893e-06,
      "loss": 0.766,
      "step": 4855
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.403760885673823,
      "learning_rate": 9.114398426807024e-06,
      "loss": 0.8707,
      "step": 4856
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.0091258254670175,
      "learning_rate": 9.113987921209095e-06,
      "loss": 0.7522,
      "step": 4857
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.853296386558378,
      "learning_rate": 9.11357732974068e-06,
      "loss": 0.8826,
      "step": 4858
    },
    {
      "epoch": 0.43,
      "grad_norm": 9.225204006400393,
      "learning_rate": 9.113166652410344e-06,
      "loss": 0.8522,
      "step": 4859
    },
    {
      "epoch": 0.43,
      "grad_norm": 8.070378011059255,
      "learning_rate": 9.112755889226664e-06,
      "loss": 0.774,
      "step": 4860
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.952803432518575,
      "learning_rate": 9.112345040198212e-06,
      "loss": 0.8162,
      "step": 4861
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.622240886574784,
      "learning_rate": 9.111934105333563e-06,
      "loss": 0.8444,
      "step": 4862
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.392016538822963,
      "learning_rate": 9.111523084641292e-06,
      "loss": 0.7944,
      "step": 4863
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.9825462642169605,
      "learning_rate": 9.111111978129983e-06,
      "loss": 0.8807,
      "step": 4864
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.834063968707749,
      "learning_rate": 9.110700785808216e-06,
      "loss": 0.7289,
      "step": 4865
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.764413796162558,
      "learning_rate": 9.110289507684569e-06,
      "loss": 0.7878,
      "step": 4866
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.790280262182799,
      "learning_rate": 9.109878143767633e-06,
      "loss": 0.821,
      "step": 4867
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.996171763389265,
      "learning_rate": 9.10946669406599e-06,
      "loss": 0.86,
      "step": 4868
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.308244354496188,
      "learning_rate": 9.109055158588226e-06,
      "loss": 0.8747,
      "step": 4869
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.683003077853453,
      "learning_rate": 9.108643537342938e-06,
      "loss": 0.7568,
      "step": 4870
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.169716494751594,
      "learning_rate": 9.108231830338714e-06,
      "loss": 0.8198,
      "step": 4871
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.057289966398745,
      "learning_rate": 9.107820037584147e-06,
      "loss": 0.8373,
      "step": 4872
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.118980415714306,
      "learning_rate": 9.107408159087831e-06,
      "loss": 0.8351,
      "step": 4873
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.668471480557678,
      "learning_rate": 9.106996194858368e-06,
      "loss": 0.8312,
      "step": 4874
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.2501341141926225,
      "learning_rate": 9.106584144904352e-06,
      "loss": 0.821,
      "step": 4875
    },
    {
      "epoch": 0.43,
      "grad_norm": 8.128484954219688,
      "learning_rate": 9.106172009234385e-06,
      "loss": 0.8705,
      "step": 4876
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.421820242883798,
      "learning_rate": 9.105759787857069e-06,
      "loss": 0.8326,
      "step": 4877
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.336248091208248,
      "learning_rate": 9.105347480781008e-06,
      "loss": 0.866,
      "step": 4878
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.6288972372407935,
      "learning_rate": 9.104935088014813e-06,
      "loss": 0.7967,
      "step": 4879
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.649258859580292,
      "learning_rate": 9.104522609567083e-06,
      "loss": 0.8471,
      "step": 4880
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.473163932481095,
      "learning_rate": 9.104110045446435e-06,
      "loss": 0.7515,
      "step": 4881
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.808625417867456,
      "learning_rate": 9.103697395661477e-06,
      "loss": 0.7918,
      "step": 4882
    },
    {
      "epoch": 0.44,
      "grad_norm": 8.41269730698717,
      "learning_rate": 9.103284660220823e-06,
      "loss": 0.7189,
      "step": 4883
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.566026978609095,
      "learning_rate": 9.102871839133087e-06,
      "loss": 0.8262,
      "step": 4884
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.284426809092878,
      "learning_rate": 9.102458932406889e-06,
      "loss": 0.848,
      "step": 4885
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.040023035708307,
      "learning_rate": 9.102045940050843e-06,
      "loss": 0.7947,
      "step": 4886
    },
    {
      "epoch": 0.44,
      "grad_norm": 10.642318195286915,
      "learning_rate": 9.101632862073571e-06,
      "loss": 0.8861,
      "step": 4887
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.296746432152736,
      "learning_rate": 9.101219698483694e-06,
      "loss": 0.8257,
      "step": 4888
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.9750406885018785,
      "learning_rate": 9.10080644928984e-06,
      "loss": 0.8992,
      "step": 4889
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.336430795855631,
      "learning_rate": 9.100393114500632e-06,
      "loss": 0.8137,
      "step": 4890
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.229087324085516,
      "learning_rate": 9.099979694124696e-06,
      "loss": 0.8594,
      "step": 4891
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.121633823963735,
      "learning_rate": 9.099566188170663e-06,
      "loss": 0.8029,
      "step": 4892
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.141951747110666,
      "learning_rate": 9.099152596647165e-06,
      "loss": 0.7847,
      "step": 4893
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.802408005664781,
      "learning_rate": 9.098738919562832e-06,
      "loss": 0.8206,
      "step": 4894
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.968672526431462,
      "learning_rate": 9.0983251569263e-06,
      "loss": 0.8697,
      "step": 4895
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.424975118955648,
      "learning_rate": 9.097911308746206e-06,
      "loss": 0.8517,
      "step": 4896
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.734972061208118,
      "learning_rate": 9.097497375031189e-06,
      "loss": 0.8441,
      "step": 4897
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.288862862494857,
      "learning_rate": 9.097083355789886e-06,
      "loss": 0.9049,
      "step": 4898
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.572106811164453,
      "learning_rate": 9.09666925103094e-06,
      "loss": 0.8221,
      "step": 4899
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.577498237539196,
      "learning_rate": 9.096255060762998e-06,
      "loss": 0.8284,
      "step": 4900
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.501844460247866,
      "learning_rate": 9.095840784994699e-06,
      "loss": 0.869,
      "step": 4901
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.162010420300825,
      "learning_rate": 9.095426423734695e-06,
      "loss": 0.8966,
      "step": 4902
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.919884772221421,
      "learning_rate": 9.095011976991632e-06,
      "loss": 0.867,
      "step": 4903
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.182391125624397,
      "learning_rate": 9.094597444774162e-06,
      "loss": 0.8135,
      "step": 4904
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.644653209275113,
      "learning_rate": 9.09418282709094e-06,
      "loss": 0.8346,
      "step": 4905
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.818937709405059,
      "learning_rate": 9.093768123950616e-06,
      "loss": 0.8098,
      "step": 4906
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.847664273594822,
      "learning_rate": 9.093353335361845e-06,
      "loss": 0.8337,
      "step": 4907
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.396766347544747,
      "learning_rate": 9.09293846133329e-06,
      "loss": 0.895,
      "step": 4908
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.853544707736268,
      "learning_rate": 9.092523501873609e-06,
      "loss": 0.8663,
      "step": 4909
    },
    {
      "epoch": 0.44,
      "grad_norm": 8.794949573888701,
      "learning_rate": 9.09210845699146e-06,
      "loss": 0.885,
      "step": 4910
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.26467628850975,
      "learning_rate": 9.09169332669551e-06,
      "loss": 0.8358,
      "step": 4911
    },
    {
      "epoch": 0.44,
      "grad_norm": 8.63298954072151,
      "learning_rate": 9.091278110994422e-06,
      "loss": 0.9096,
      "step": 4912
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.362068065799047,
      "learning_rate": 9.090862809896864e-06,
      "loss": 0.8052,
      "step": 4913
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.764988674468138,
      "learning_rate": 9.090447423411501e-06,
      "loss": 0.7967,
      "step": 4914
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.5812895241714005,
      "learning_rate": 9.090031951547009e-06,
      "loss": 0.7847,
      "step": 4915
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.135172137192945,
      "learning_rate": 9.089616394312056e-06,
      "loss": 0.8703,
      "step": 4916
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.277565341915346,
      "learning_rate": 9.089200751715318e-06,
      "loss": 0.8592,
      "step": 4917
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.7284509623746045,
      "learning_rate": 9.088785023765467e-06,
      "loss": 0.8267,
      "step": 4918
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.506459956213716,
      "learning_rate": 9.088369210471183e-06,
      "loss": 0.8223,
      "step": 4919
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.3454062294847775,
      "learning_rate": 9.087953311841148e-06,
      "loss": 0.866,
      "step": 4920
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.616842180628503,
      "learning_rate": 9.087537327884037e-06,
      "loss": 0.7944,
      "step": 4921
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.288630382476383,
      "learning_rate": 9.087121258608538e-06,
      "loss": 0.8369,
      "step": 4922
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.499485597169764,
      "learning_rate": 9.086705104023333e-06,
      "loss": 0.8361,
      "step": 4923
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.915982187529836,
      "learning_rate": 9.086288864137108e-06,
      "loss": 0.7949,
      "step": 4924
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.359225796409971,
      "learning_rate": 9.08587253895855e-06,
      "loss": 0.8178,
      "step": 4925
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.366038398522875,
      "learning_rate": 9.085456128496354e-06,
      "loss": 0.8103,
      "step": 4926
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.745203494509066,
      "learning_rate": 9.085039632759208e-06,
      "loss": 0.8787,
      "step": 4927
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.8430864357744206,
      "learning_rate": 9.084623051755803e-06,
      "loss": 0.8436,
      "step": 4928
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.792613389632451,
      "learning_rate": 9.08420638549484e-06,
      "loss": 0.8643,
      "step": 4929
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.9549184682562375,
      "learning_rate": 9.083789633985012e-06,
      "loss": 0.8529,
      "step": 4930
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.180274838788972,
      "learning_rate": 9.083372797235017e-06,
      "loss": 0.868,
      "step": 4931
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.426413842741616,
      "learning_rate": 9.082955875253559e-06,
      "loss": 0.8658,
      "step": 4932
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.452946835775552,
      "learning_rate": 9.08253886804934e-06,
      "loss": 0.8324,
      "step": 4933
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.595017686580911,
      "learning_rate": 9.082121775631058e-06,
      "loss": 0.7809,
      "step": 4934
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.026163602296175,
      "learning_rate": 9.081704598007428e-06,
      "loss": 0.8314,
      "step": 4935
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.260227691118625,
      "learning_rate": 9.081287335187153e-06,
      "loss": 0.8788,
      "step": 4936
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.855432899575218,
      "learning_rate": 9.080869987178941e-06,
      "loss": 0.8118,
      "step": 4937
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.738891803206613,
      "learning_rate": 9.080452553991505e-06,
      "loss": 0.8788,
      "step": 4938
    },
    {
      "epoch": 0.44,
      "grad_norm": 10.565320249373139,
      "learning_rate": 9.080035035633558e-06,
      "loss": 0.8532,
      "step": 4939
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.6631605718541715,
      "learning_rate": 9.079617432113816e-06,
      "loss": 0.7791,
      "step": 4940
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.4116254576292055,
      "learning_rate": 9.079199743440994e-06,
      "loss": 0.8195,
      "step": 4941
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.280108679916767,
      "learning_rate": 9.07878196962381e-06,
      "loss": 0.8276,
      "step": 4942
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.304931483061797,
      "learning_rate": 9.078364110670985e-06,
      "loss": 0.8497,
      "step": 4943
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.767258251264538,
      "learning_rate": 9.077946166591242e-06,
      "loss": 0.8002,
      "step": 4944
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.162269808296781,
      "learning_rate": 9.077528137393302e-06,
      "loss": 0.8459,
      "step": 4945
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.559000757370646,
      "learning_rate": 9.07711002308589e-06,
      "loss": 0.8932,
      "step": 4946
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.484232673893755,
      "learning_rate": 9.07669182367774e-06,
      "loss": 0.7625,
      "step": 4947
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.440623647438592,
      "learning_rate": 9.076273539177572e-06,
      "loss": 0.8138,
      "step": 4948
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.169344608257306,
      "learning_rate": 9.075855169594122e-06,
      "loss": 0.8442,
      "step": 4949
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.7365980069108256,
      "learning_rate": 9.07543671493612e-06,
      "loss": 0.9581,
      "step": 4950
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.147881083545158,
      "learning_rate": 9.075018175212302e-06,
      "loss": 0.8132,
      "step": 4951
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.927444707976578,
      "learning_rate": 9.074599550431405e-06,
      "loss": 0.8052,
      "step": 4952
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.890517908407159,
      "learning_rate": 9.074180840602165e-06,
      "loss": 0.8667,
      "step": 4953
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.885274111604918,
      "learning_rate": 9.073762045733322e-06,
      "loss": 0.8441,
      "step": 4954
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.784214650661166,
      "learning_rate": 9.073343165833617e-06,
      "loss": 0.8358,
      "step": 4955
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.751018323843988,
      "learning_rate": 9.072924200911795e-06,
      "loss": 0.8716,
      "step": 4956
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.428475507912158,
      "learning_rate": 9.072505150976601e-06,
      "loss": 0.7945,
      "step": 4957
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.191817565625338,
      "learning_rate": 9.072086016036779e-06,
      "loss": 0.8498,
      "step": 4958
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.0675998649827925,
      "learning_rate": 9.071666796101078e-06,
      "loss": 0.8176,
      "step": 4959
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.534337219171704,
      "learning_rate": 9.07124749117825e-06,
      "loss": 0.8677,
      "step": 4960
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.485272299104047,
      "learning_rate": 9.070828101277047e-06,
      "loss": 0.7945,
      "step": 4961
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.674192555699343,
      "learning_rate": 9.070408626406224e-06,
      "loss": 0.8332,
      "step": 4962
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.940620218467653,
      "learning_rate": 9.069989066574534e-06,
      "loss": 0.7791,
      "step": 4963
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.964658491085949,
      "learning_rate": 9.069569421790734e-06,
      "loss": 0.7529,
      "step": 4964
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.173534038136184,
      "learning_rate": 9.069149692063585e-06,
      "loss": 0.7701,
      "step": 4965
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.2571631033907185,
      "learning_rate": 9.068729877401848e-06,
      "loss": 0.824,
      "step": 4966
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.215251106126207,
      "learning_rate": 9.068309977814286e-06,
      "loss": 0.815,
      "step": 4967
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.605164420848333,
      "learning_rate": 9.067889993309662e-06,
      "loss": 0.907,
      "step": 4968
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.61463928685527,
      "learning_rate": 9.067469923896743e-06,
      "loss": 0.8169,
      "step": 4969
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.8469789630350375,
      "learning_rate": 9.067049769584295e-06,
      "loss": 0.7721,
      "step": 4970
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.689687597679808,
      "learning_rate": 9.066629530381093e-06,
      "loss": 0.9313,
      "step": 4971
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.497101589072142,
      "learning_rate": 9.066209206295904e-06,
      "loss": 0.8063,
      "step": 4972
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.473484062892589,
      "learning_rate": 9.065788797337502e-06,
      "loss": 0.8157,
      "step": 4973
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.345112737378636,
      "learning_rate": 9.065368303514662e-06,
      "loss": 0.829,
      "step": 4974
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.400023421265176,
      "learning_rate": 9.064947724836163e-06,
      "loss": 0.8148,
      "step": 4975
    },
    {
      "epoch": 0.44,
      "grad_norm": 8.09803200982256,
      "learning_rate": 9.064527061310782e-06,
      "loss": 0.814,
      "step": 4976
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.797807647795938,
      "learning_rate": 9.0641063129473e-06,
      "loss": 0.8752,
      "step": 4977
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.533454338815272,
      "learning_rate": 9.063685479754498e-06,
      "loss": 0.822,
      "step": 4978
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.462620560781152,
      "learning_rate": 9.063264561741163e-06,
      "loss": 0.8112,
      "step": 4979
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.193747367776714,
      "learning_rate": 9.062843558916076e-06,
      "loss": 0.8508,
      "step": 4980
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.4718042542974255,
      "learning_rate": 9.062422471288029e-06,
      "loss": 0.8689,
      "step": 4981
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.554518107171751,
      "learning_rate": 9.062001298865811e-06,
      "loss": 0.8189,
      "step": 4982
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.817209495247513,
      "learning_rate": 9.061580041658209e-06,
      "loss": 0.8424,
      "step": 4983
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.370561119188758,
      "learning_rate": 9.061158699674018e-06,
      "loss": 0.8263,
      "step": 4984
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.989734031404186,
      "learning_rate": 9.060737272922033e-06,
      "loss": 0.7945,
      "step": 4985
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.6375027754978415,
      "learning_rate": 9.060315761411052e-06,
      "loss": 0.8921,
      "step": 4986
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.263250033798762,
      "learning_rate": 9.05989416514987e-06,
      "loss": 0.8436,
      "step": 4987
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.797324841663219,
      "learning_rate": 9.059472484147289e-06,
      "loss": 0.8048,
      "step": 4988
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.561108909718329,
      "learning_rate": 9.05905071841211e-06,
      "loss": 0.8069,
      "step": 4989
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.0231837025332435,
      "learning_rate": 9.058628867953136e-06,
      "loss": 0.8558,
      "step": 4990
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.880013844026016,
      "learning_rate": 9.058206932779174e-06,
      "loss": 0.8393,
      "step": 4991
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.907787108207934,
      "learning_rate": 9.057784912899028e-06,
      "loss": 0.875,
      "step": 4992
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.303137408130783,
      "learning_rate": 9.057362808321509e-06,
      "loss": 0.8022,
      "step": 4993
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.174172975800396,
      "learning_rate": 9.056940619055428e-06,
      "loss": 0.814,
      "step": 4994
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.916546411654465,
      "learning_rate": 9.056518345109595e-06,
      "loss": 0.8254,
      "step": 4995
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.830973464155386,
      "learning_rate": 9.056095986492823e-06,
      "loss": 0.7472,
      "step": 4996
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.61074225800067,
      "learning_rate": 9.055673543213932e-06,
      "loss": 0.8115,
      "step": 4997
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.869034900019781,
      "learning_rate": 9.055251015281739e-06,
      "loss": 0.8475,
      "step": 4998
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.981455221560775,
      "learning_rate": 9.05482840270506e-06,
      "loss": 0.792,
      "step": 4999
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.332255180667185,
      "learning_rate": 9.054405705492718e-06,
      "loss": 0.8712,
      "step": 5000
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.999082067255124,
      "learning_rate": 9.053982923653538e-06,
      "loss": 0.8222,
      "step": 5001
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.118803917373723,
      "learning_rate": 9.05356005719634e-06,
      "loss": 0.8067,
      "step": 5002
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.1201272021944115,
      "learning_rate": 9.053137106129953e-06,
      "loss": 0.7572,
      "step": 5003
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.524797992810213,
      "learning_rate": 9.052714070463206e-06,
      "loss": 0.8331,
      "step": 5004
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.734753551743723,
      "learning_rate": 9.052290950204929e-06,
      "loss": 0.8393,
      "step": 5005
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.696684014156215,
      "learning_rate": 9.051867745363952e-06,
      "loss": 0.8356,
      "step": 5006
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.777284094712152,
      "learning_rate": 9.051444455949107e-06,
      "loss": 0.7918,
      "step": 5007
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.812520270248402,
      "learning_rate": 9.051021081969236e-06,
      "loss": 0.8152,
      "step": 5008
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.255314806640712,
      "learning_rate": 9.050597623433169e-06,
      "loss": 0.789,
      "step": 5009
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.595214791113278,
      "learning_rate": 9.050174080349746e-06,
      "loss": 0.8931,
      "step": 5010
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.6860779585771,
      "learning_rate": 9.04975045272781e-06,
      "loss": 0.807,
      "step": 5011
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.212951510855668,
      "learning_rate": 9.049326740576203e-06,
      "loss": 0.8416,
      "step": 5012
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.857005001403426,
      "learning_rate": 9.048902943903768e-06,
      "loss": 0.7327,
      "step": 5013
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.3887958270933884,
      "learning_rate": 9.04847906271935e-06,
      "loss": 0.8603,
      "step": 5014
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.195653239883354,
      "learning_rate": 9.048055097031797e-06,
      "loss": 0.7875,
      "step": 5015
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.61805331702492,
      "learning_rate": 9.047631046849961e-06,
      "loss": 0.8864,
      "step": 5016
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.4474689879348865,
      "learning_rate": 9.047206912182688e-06,
      "loss": 0.807,
      "step": 5017
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.222143898491759,
      "learning_rate": 9.046782693038837e-06,
      "loss": 0.8142,
      "step": 5018
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.00865213564452,
      "learning_rate": 9.046358389427259e-06,
      "loss": 0.9047,
      "step": 5019
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.504208638795159,
      "learning_rate": 9.04593400135681e-06,
      "loss": 0.8567,
      "step": 5020
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.269079692285917,
      "learning_rate": 9.04550952883635e-06,
      "loss": 0.7735,
      "step": 5021
    },
    {
      "epoch": 0.45,
      "grad_norm": 8.932809846394722,
      "learning_rate": 9.045084971874738e-06,
      "loss": 0.7795,
      "step": 5022
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.8078885121791535,
      "learning_rate": 9.044660330480835e-06,
      "loss": 0.7993,
      "step": 5023
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.573354529655484,
      "learning_rate": 9.044235604663507e-06,
      "loss": 0.7826,
      "step": 5024
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.379444732333957,
      "learning_rate": 9.043810794431615e-06,
      "loss": 0.7929,
      "step": 5025
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.757809206399995,
      "learning_rate": 9.043385899794031e-06,
      "loss": 0.7852,
      "step": 5026
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.155616265141507,
      "learning_rate": 9.04296092075962e-06,
      "loss": 0.835,
      "step": 5027
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.863302810679815,
      "learning_rate": 9.042535857337253e-06,
      "loss": 0.7334,
      "step": 5028
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.860121100378805,
      "learning_rate": 9.042110709535803e-06,
      "loss": 0.8863,
      "step": 5029
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.356797975854452,
      "learning_rate": 9.041685477364146e-06,
      "loss": 0.848,
      "step": 5030
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.210635866219729,
      "learning_rate": 9.041260160831153e-06,
      "loss": 0.7887,
      "step": 5031
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.759612203595883,
      "learning_rate": 9.040834759945707e-06,
      "loss": 0.8679,
      "step": 5032
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.360845218899811,
      "learning_rate": 9.040409274716683e-06,
      "loss": 0.8376,
      "step": 5033
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.778527741462842,
      "learning_rate": 9.039983705152965e-06,
      "loss": 0.8203,
      "step": 5034
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.992529983776203,
      "learning_rate": 9.039558051263433e-06,
      "loss": 0.796,
      "step": 5035
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.700029866465778,
      "learning_rate": 9.039132313056974e-06,
      "loss": 0.807,
      "step": 5036
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.545398275672792,
      "learning_rate": 9.038706490542474e-06,
      "loss": 0.836,
      "step": 5037
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.8768349230037025,
      "learning_rate": 9.03828058372882e-06,
      "loss": 0.8309,
      "step": 5038
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.306846390307188,
      "learning_rate": 9.037854592624902e-06,
      "loss": 0.8378,
      "step": 5039
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.581159556114853,
      "learning_rate": 9.037428517239613e-06,
      "loss": 0.9239,
      "step": 5040
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.95931971618534,
      "learning_rate": 9.037002357581846e-06,
      "loss": 0.8757,
      "step": 5041
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.1490692208678395,
      "learning_rate": 9.036576113660495e-06,
      "loss": 0.8369,
      "step": 5042
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.429267564946473,
      "learning_rate": 9.036149785484457e-06,
      "loss": 0.8338,
      "step": 5043
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.329331772429348,
      "learning_rate": 9.035723373062631e-06,
      "loss": 0.9375,
      "step": 5044
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.625437755033326,
      "learning_rate": 9.035296876403918e-06,
      "loss": 0.7766,
      "step": 5045
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.763544989893001,
      "learning_rate": 9.034870295517222e-06,
      "loss": 0.823,
      "step": 5046
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.3032926454762315,
      "learning_rate": 9.034443630411442e-06,
      "loss": 0.8104,
      "step": 5047
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.404778360169698,
      "learning_rate": 9.034016881095489e-06,
      "loss": 0.8287,
      "step": 5048
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.553873155547471,
      "learning_rate": 9.033590047578266e-06,
      "loss": 0.8656,
      "step": 5049
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.413849168099422,
      "learning_rate": 9.033163129868685e-06,
      "loss": 0.8878,
      "step": 5050
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.188355117046866,
      "learning_rate": 9.032736127975654e-06,
      "loss": 0.8031,
      "step": 5051
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.116859864046155,
      "learning_rate": 9.032309041908091e-06,
      "loss": 0.8395,
      "step": 5052
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.06367771971315,
      "learning_rate": 9.031881871674906e-06,
      "loss": 0.7694,
      "step": 5053
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.3364632928642255,
      "learning_rate": 9.031454617285015e-06,
      "loss": 0.7693,
      "step": 5054
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.897152427963591,
      "learning_rate": 9.031027278747341e-06,
      "loss": 0.8419,
      "step": 5055
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.109566328941059,
      "learning_rate": 9.030599856070799e-06,
      "loss": 0.8615,
      "step": 5056
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.777331246592382,
      "learning_rate": 9.03017234926431e-06,
      "loss": 0.8316,
      "step": 5057
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.226816638488124,
      "learning_rate": 9.029744758336802e-06,
      "loss": 0.8193,
      "step": 5058
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.419800045022271,
      "learning_rate": 9.029317083297196e-06,
      "loss": 0.8621,
      "step": 5059
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.150291377237508,
      "learning_rate": 9.028889324154419e-06,
      "loss": 0.8712,
      "step": 5060
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.336730023253211,
      "learning_rate": 9.028461480917401e-06,
      "loss": 0.8897,
      "step": 5061
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.114759163251798,
      "learning_rate": 9.028033553595072e-06,
      "loss": 0.8256,
      "step": 5062
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.363603961033879,
      "learning_rate": 9.027605542196364e-06,
      "loss": 0.7794,
      "step": 5063
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.350532044329956,
      "learning_rate": 9.02717744673021e-06,
      "loss": 0.8246,
      "step": 5064
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.346170210083831,
      "learning_rate": 9.026749267205547e-06,
      "loss": 0.8799,
      "step": 5065
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.516150684106209,
      "learning_rate": 9.026321003631311e-06,
      "loss": 0.7709,
      "step": 5066
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.569509963222773,
      "learning_rate": 9.025892656016442e-06,
      "loss": 0.903,
      "step": 5067
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.349875774489,
      "learning_rate": 9.02546422436988e-06,
      "loss": 0.8563,
      "step": 5068
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.997665960489791,
      "learning_rate": 9.025035708700566e-06,
      "loss": 0.8935,
      "step": 5069
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.793268851754902,
      "learning_rate": 9.024607109017449e-06,
      "loss": 0.8421,
      "step": 5070
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.027803008821304,
      "learning_rate": 9.024178425329472e-06,
      "loss": 0.8669,
      "step": 5071
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.787558677658023,
      "learning_rate": 9.023749657645581e-06,
      "loss": 0.8416,
      "step": 5072
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.482870303295683,
      "learning_rate": 9.023320805974731e-06,
      "loss": 0.8029,
      "step": 5073
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.702986097889686,
      "learning_rate": 9.022891870325869e-06,
      "loss": 0.8299,
      "step": 5074
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.3645706328934475,
      "learning_rate": 9.022462850707949e-06,
      "loss": 0.8023,
      "step": 5075
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.689310822044897,
      "learning_rate": 9.022033747129925e-06,
      "loss": 0.7945,
      "step": 5076
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.675860918625916,
      "learning_rate": 9.021604559600756e-06,
      "loss": 0.7843,
      "step": 5077
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.302481394923712,
      "learning_rate": 9.021175288129398e-06,
      "loss": 0.8404,
      "step": 5078
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.976142110182973,
      "learning_rate": 9.020745932724812e-06,
      "loss": 0.7328,
      "step": 5079
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.8601038683117,
      "learning_rate": 9.02031649339596e-06,
      "loss": 0.7378,
      "step": 5080
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.881645632757235,
      "learning_rate": 9.019886970151805e-06,
      "loss": 0.7992,
      "step": 5081
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.135894137543992,
      "learning_rate": 9.019457363001316e-06,
      "loss": 0.8454,
      "step": 5082
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.053055355667253,
      "learning_rate": 9.019027671953453e-06,
      "loss": 0.824,
      "step": 5083
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.4202877647627545,
      "learning_rate": 9.018597897017192e-06,
      "loss": 0.8307,
      "step": 5084
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.508710790819579,
      "learning_rate": 9.018168038201498e-06,
      "loss": 0.8101,
      "step": 5085
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.898263908006392,
      "learning_rate": 9.017738095515347e-06,
      "loss": 0.8283,
      "step": 5086
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.0921734551170585,
      "learning_rate": 9.017308068967712e-06,
      "loss": 0.7883,
      "step": 5087
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.076802046205431,
      "learning_rate": 9.01687795856757e-06,
      "loss": 0.7893,
      "step": 5088
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.915784806173987,
      "learning_rate": 9.016447764323894e-06,
      "loss": 0.7996,
      "step": 5089
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.480522973906189,
      "learning_rate": 9.01601748624567e-06,
      "loss": 0.8089,
      "step": 5090
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.282058408406082,
      "learning_rate": 9.015587124341875e-06,
      "loss": 0.9167,
      "step": 5091
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.669236243660897,
      "learning_rate": 9.015156678621492e-06,
      "loss": 0.8799,
      "step": 5092
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.167515446005485,
      "learning_rate": 9.014726149093508e-06,
      "loss": 0.8505,
      "step": 5093
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.287330451320698,
      "learning_rate": 9.014295535766905e-06,
      "loss": 0.8295,
      "step": 5094
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.136122110704356,
      "learning_rate": 9.013864838650677e-06,
      "loss": 0.8993,
      "step": 5095
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.1019736515795415,
      "learning_rate": 9.013434057753809e-06,
      "loss": 0.8808,
      "step": 5096
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.686934123663191,
      "learning_rate": 9.013003193085295e-06,
      "loss": 0.8359,
      "step": 5097
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.869360315873249,
      "learning_rate": 9.012572244654128e-06,
      "loss": 0.846,
      "step": 5098
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.712785425012518,
      "learning_rate": 9.0121412124693e-06,
      "loss": 0.8592,
      "step": 5099
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.925459604106917,
      "learning_rate": 9.011710096539814e-06,
      "loss": 0.7939,
      "step": 5100
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.9944928076431925,
      "learning_rate": 9.011278896874664e-06,
      "loss": 0.8047,
      "step": 5101
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.783928218196069,
      "learning_rate": 9.010847613482852e-06,
      "loss": 0.8844,
      "step": 5102
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.810469391310199,
      "learning_rate": 9.01041624637338e-06,
      "loss": 0.7352,
      "step": 5103
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.63486633565103,
      "learning_rate": 9.009984795555248e-06,
      "loss": 0.8558,
      "step": 5104
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.665924497846773,
      "learning_rate": 9.00955326103747e-06,
      "loss": 0.9132,
      "step": 5105
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.0348920471682534,
      "learning_rate": 9.009121642829046e-06,
      "loss": 0.8375,
      "step": 5106
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.053244324521354,
      "learning_rate": 9.008689940938988e-06,
      "loss": 0.8438,
      "step": 5107
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.318383359651649,
      "learning_rate": 9.008258155376307e-06,
      "loss": 0.808,
      "step": 5108
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.658893032040583,
      "learning_rate": 9.007826286150011e-06,
      "loss": 0.8505,
      "step": 5109
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.855846196599967,
      "learning_rate": 9.007394333269124e-06,
      "loss": 0.8196,
      "step": 5110
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.130684104016678,
      "learning_rate": 9.006962296742653e-06,
      "loss": 0.9047,
      "step": 5111
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.553594675646942,
      "learning_rate": 9.00653017657962e-06,
      "loss": 0.8776,
      "step": 5112
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.724512367550496,
      "learning_rate": 9.006097972789042e-06,
      "loss": 0.7866,
      "step": 5113
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.244919113574722,
      "learning_rate": 9.005665685379945e-06,
      "loss": 0.8236,
      "step": 5114
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.66106884038679,
      "learning_rate": 9.005233314361349e-06,
      "loss": 0.858,
      "step": 5115
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.37173292566724,
      "learning_rate": 9.004800859742276e-06,
      "loss": 0.8256,
      "step": 5116
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.537981204527235,
      "learning_rate": 9.004368321531757e-06,
      "loss": 0.8764,
      "step": 5117
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.308820411411482,
      "learning_rate": 9.00393569973882e-06,
      "loss": 0.8887,
      "step": 5118
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.612723373572323,
      "learning_rate": 9.00350299437249e-06,
      "loss": 0.8673,
      "step": 5119
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.599560516400066,
      "learning_rate": 9.003070205441807e-06,
      "loss": 0.7947,
      "step": 5120
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.742870089713064,
      "learning_rate": 9.002637332955797e-06,
      "loss": 0.7917,
      "step": 5121
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.248537408982362,
      "learning_rate": 9.0022043769235e-06,
      "loss": 0.8272,
      "step": 5122
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.9122783248461594,
      "learning_rate": 9.00177133735395e-06,
      "loss": 0.8297,
      "step": 5123
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.385495752678054,
      "learning_rate": 9.001338214256188e-06,
      "loss": 0.808,
      "step": 5124
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.334458326584334,
      "learning_rate": 9.000905007639252e-06,
      "loss": 0.8167,
      "step": 5125
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.6127378255930305,
      "learning_rate": 9.000471717512188e-06,
      "loss": 0.8733,
      "step": 5126
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.429854361101258,
      "learning_rate": 9.000038343884037e-06,
      "loss": 0.7266,
      "step": 5127
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.6180991483049985,
      "learning_rate": 8.999604886763844e-06,
      "loss": 0.8369,
      "step": 5128
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.057339578458408,
      "learning_rate": 8.99917134616066e-06,
      "loss": 0.8002,
      "step": 5129
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.068384668724896,
      "learning_rate": 8.998737722083531e-06,
      "loss": 0.7729,
      "step": 5130
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.11828404002208,
      "learning_rate": 8.99830401454151e-06,
      "loss": 0.8108,
      "step": 5131
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.239775693636682,
      "learning_rate": 8.997870223543647e-06,
      "loss": 0.8175,
      "step": 5132
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.828374962799092,
      "learning_rate": 8.997436349099e-06,
      "loss": 0.8217,
      "step": 5133
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.221169000982274,
      "learning_rate": 8.997002391216623e-06,
      "loss": 0.8438,
      "step": 5134
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.812759791923963,
      "learning_rate": 8.996568349905577e-06,
      "loss": 0.8034,
      "step": 5135
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.945024855185755,
      "learning_rate": 8.996134225174915e-06,
      "loss": 0.834,
      "step": 5136
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.334795585499347,
      "learning_rate": 8.995700017033703e-06,
      "loss": 0.8283,
      "step": 5137
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.746877129333405,
      "learning_rate": 8.995265725491007e-06,
      "loss": 0.8149,
      "step": 5138
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.920857454789024,
      "learning_rate": 8.994831350555885e-06,
      "loss": 0.8442,
      "step": 5139
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.530149700830683,
      "learning_rate": 8.994396892237409e-06,
      "loss": 0.8359,
      "step": 5140
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.9796731115184185,
      "learning_rate": 8.993962350544643e-06,
      "loss": 0.8328,
      "step": 5141
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.387867982033721,
      "learning_rate": 8.993527725486662e-06,
      "loss": 0.8447,
      "step": 5142
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.098297819818132,
      "learning_rate": 8.993093017072535e-06,
      "loss": 0.8213,
      "step": 5143
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.846050224792815,
      "learning_rate": 8.992658225311334e-06,
      "loss": 0.8553,
      "step": 5144
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.265988385052772,
      "learning_rate": 8.99222335021214e-06,
      "loss": 0.8149,
      "step": 5145
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.2900752130342275,
      "learning_rate": 8.991788391784022e-06,
      "loss": 0.7816,
      "step": 5146
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.465218725048546,
      "learning_rate": 8.991353350036065e-06,
      "loss": 0.7875,
      "step": 5147
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.391564226014759,
      "learning_rate": 8.99091822497735e-06,
      "loss": 0.8115,
      "step": 5148
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.552076257155046,
      "learning_rate": 8.990483016616955e-06,
      "loss": 0.8448,
      "step": 5149
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.13387548653464,
      "learning_rate": 8.990047724963967e-06,
      "loss": 0.8243,
      "step": 5150
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.071567307182445,
      "learning_rate": 8.989612350027467e-06,
      "loss": 0.8358,
      "step": 5151
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.114891435105107,
      "learning_rate": 8.989176891816551e-06,
      "loss": 0.8423,
      "step": 5152
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.07647478414593,
      "learning_rate": 8.988741350340303e-06,
      "loss": 0.8246,
      "step": 5153
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.344994326234233,
      "learning_rate": 8.988305725607812e-06,
      "loss": 0.8405,
      "step": 5154
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.779992611325056,
      "learning_rate": 8.987870017628174e-06,
      "loss": 0.7886,
      "step": 5155
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.015998749017503,
      "learning_rate": 8.987434226410483e-06,
      "loss": 0.8487,
      "step": 5156
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.883043034320816,
      "learning_rate": 8.986998351963836e-06,
      "loss": 0.9356,
      "step": 5157
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.202879932313559,
      "learning_rate": 8.986562394297329e-06,
      "loss": 0.8076,
      "step": 5158
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.714095780884199,
      "learning_rate": 8.986126353420062e-06,
      "loss": 0.8207,
      "step": 5159
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.809141119482261,
      "learning_rate": 8.985690229341137e-06,
      "loss": 0.8242,
      "step": 5160
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.996542759902648,
      "learning_rate": 8.985254022069658e-06,
      "loss": 0.8153,
      "step": 5161
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.369077671655855,
      "learning_rate": 8.984817731614727e-06,
      "loss": 0.8097,
      "step": 5162
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.565612826226608,
      "learning_rate": 8.984381357985452e-06,
      "loss": 0.8353,
      "step": 5163
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.06314637500562,
      "learning_rate": 8.983944901190944e-06,
      "loss": 0.8857,
      "step": 5164
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.57835559306156,
      "learning_rate": 8.983508361240309e-06,
      "loss": 0.9373,
      "step": 5165
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.843295978243513,
      "learning_rate": 8.983071738142663e-06,
      "loss": 0.8606,
      "step": 5166
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.206879582172523,
      "learning_rate": 8.982635031907116e-06,
      "loss": 0.7973,
      "step": 5167
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.235041055742097,
      "learning_rate": 8.982198242542785e-06,
      "loss": 0.8387,
      "step": 5168
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.7543881384320805,
      "learning_rate": 8.981761370058785e-06,
      "loss": 0.814,
      "step": 5169
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.401243623358166,
      "learning_rate": 8.981324414464238e-06,
      "loss": 0.8338,
      "step": 5170
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.339068297047841,
      "learning_rate": 8.980887375768263e-06,
      "loss": 0.8361,
      "step": 5171
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.5602655655487725,
      "learning_rate": 8.980450253979981e-06,
      "loss": 0.8295,
      "step": 5172
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.101326185881579,
      "learning_rate": 8.980013049108517e-06,
      "loss": 0.7551,
      "step": 5173
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.296463563427267,
      "learning_rate": 8.979575761162998e-06,
      "loss": 0.7611,
      "step": 5174
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.09585591135136,
      "learning_rate": 8.979138390152548e-06,
      "loss": 0.7654,
      "step": 5175
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.1088611974847105,
      "learning_rate": 8.9787009360863e-06,
      "loss": 0.7723,
      "step": 5176
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.158167316183559,
      "learning_rate": 8.978263398973381e-06,
      "loss": 0.7739,
      "step": 5177
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.2488166649922485,
      "learning_rate": 8.977825778822928e-06,
      "loss": 0.8288,
      "step": 5178
    },
    {
      "epoch": 0.46,
      "grad_norm": 9.61263395462851,
      "learning_rate": 8.977388075644073e-06,
      "loss": 0.8904,
      "step": 5179
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.039188953313096,
      "learning_rate": 8.976950289445951e-06,
      "loss": 0.8214,
      "step": 5180
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.4035257095455185,
      "learning_rate": 8.976512420237701e-06,
      "loss": 0.8545,
      "step": 5181
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.951074351113605,
      "learning_rate": 8.976074468028464e-06,
      "loss": 0.712,
      "step": 5182
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.456455065554696,
      "learning_rate": 8.97563643282738e-06,
      "loss": 0.7781,
      "step": 5183
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.88355198962257,
      "learning_rate": 8.975198314643591e-06,
      "loss": 0.8499,
      "step": 5184
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.28667464256252,
      "learning_rate": 8.974760113486243e-06,
      "loss": 0.8351,
      "step": 5185
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.699102688562684,
      "learning_rate": 8.974321829364483e-06,
      "loss": 0.8014,
      "step": 5186
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.488250766789452,
      "learning_rate": 8.973883462287456e-06,
      "loss": 0.8023,
      "step": 5187
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.403220260624605,
      "learning_rate": 8.973445012264316e-06,
      "loss": 0.8028,
      "step": 5188
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.322901228840251,
      "learning_rate": 8.973006479304213e-06,
      "loss": 0.7801,
      "step": 5189
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.479593301071975,
      "learning_rate": 8.972567863416302e-06,
      "loss": 0.8203,
      "step": 5190
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.833763835012731,
      "learning_rate": 8.972129164609735e-06,
      "loss": 0.8088,
      "step": 5191
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.977470525889851,
      "learning_rate": 8.97169038289367e-06,
      "loss": 0.7762,
      "step": 5192
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.152652409203137,
      "learning_rate": 8.971251518277266e-06,
      "loss": 0.8257,
      "step": 5193
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.027536949951563,
      "learning_rate": 8.970812570769686e-06,
      "loss": 0.7388,
      "step": 5194
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.142297227587591,
      "learning_rate": 8.970373540380086e-06,
      "loss": 0.8232,
      "step": 5195
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.434531033459896,
      "learning_rate": 8.969934427117636e-06,
      "loss": 0.9112,
      "step": 5196
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.097043335564176,
      "learning_rate": 8.969495230991498e-06,
      "loss": 0.8114,
      "step": 5197
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.045995590797559,
      "learning_rate": 8.969055952010838e-06,
      "loss": 0.8533,
      "step": 5198
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.443426474926982,
      "learning_rate": 8.96861659018483e-06,
      "loss": 0.8187,
      "step": 5199
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.285091940770798,
      "learning_rate": 8.968177145522641e-06,
      "loss": 0.8051,
      "step": 5200
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.412123384082406,
      "learning_rate": 8.967737618033445e-06,
      "loss": 0.8381,
      "step": 5201
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.763880256149526,
      "learning_rate": 8.967298007726415e-06,
      "loss": 0.8504,
      "step": 5202
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.340965938882367,
      "learning_rate": 8.966858314610728e-06,
      "loss": 0.8887,
      "step": 5203
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.506980314028418,
      "learning_rate": 8.966418538695558e-06,
      "loss": 0.8559,
      "step": 5204
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.553849548043362,
      "learning_rate": 8.965978679990088e-06,
      "loss": 0.847,
      "step": 5205
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.540418156651894,
      "learning_rate": 8.965538738503501e-06,
      "loss": 0.891,
      "step": 5206
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.285717176122889,
      "learning_rate": 8.965098714244976e-06,
      "loss": 0.8535,
      "step": 5207
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.266638916781551,
      "learning_rate": 8.964658607223697e-06,
      "loss": 0.8351,
      "step": 5208
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.75917253528263,
      "learning_rate": 8.964218417448854e-06,
      "loss": 0.8269,
      "step": 5209
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.971151463381407,
      "learning_rate": 8.963778144929633e-06,
      "loss": 0.9389,
      "step": 5210
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.4457510589096065,
      "learning_rate": 8.963337789675224e-06,
      "loss": 0.7858,
      "step": 5211
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.716159038134252,
      "learning_rate": 8.962897351694817e-06,
      "loss": 0.8264,
      "step": 5212
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.468285415744158,
      "learning_rate": 8.962456830997606e-06,
      "loss": 0.8503,
      "step": 5213
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.4908268860858955,
      "learning_rate": 8.962016227592788e-06,
      "loss": 0.8599,
      "step": 5214
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.878752086515638,
      "learning_rate": 8.961575541489558e-06,
      "loss": 0.8394,
      "step": 5215
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.442373981101603,
      "learning_rate": 8.961134772697113e-06,
      "loss": 0.8529,
      "step": 5216
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.196767770181244,
      "learning_rate": 8.960693921224654e-06,
      "loss": 0.8085,
      "step": 5217
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.122279807007636,
      "learning_rate": 8.960252987081383e-06,
      "loss": 0.8182,
      "step": 5218
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.998241515219926,
      "learning_rate": 8.959811970276506e-06,
      "loss": 0.8165,
      "step": 5219
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.868976616064946,
      "learning_rate": 8.959370870819224e-06,
      "loss": 0.7728,
      "step": 5220
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.799771681075606,
      "learning_rate": 8.958929688718746e-06,
      "loss": 0.7742,
      "step": 5221
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.504477405125737,
      "learning_rate": 8.958488423984281e-06,
      "loss": 0.8114,
      "step": 5222
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.589764665500126,
      "learning_rate": 8.95804707662504e-06,
      "loss": 0.7968,
      "step": 5223
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.189852195532309,
      "learning_rate": 8.957605646650233e-06,
      "loss": 0.8856,
      "step": 5224
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.38207722812808,
      "learning_rate": 8.957164134069076e-06,
      "loss": 0.7867,
      "step": 5225
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.7203235102836105,
      "learning_rate": 8.956722538890784e-06,
      "loss": 0.8228,
      "step": 5226
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.7841319955173125,
      "learning_rate": 8.956280861124573e-06,
      "loss": 0.9064,
      "step": 5227
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.58204092359681,
      "learning_rate": 8.955839100779665e-06,
      "loss": 0.8261,
      "step": 5228
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.521866671998887,
      "learning_rate": 8.955397257865277e-06,
      "loss": 0.8659,
      "step": 5229
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.22538829217875,
      "learning_rate": 8.954955332390633e-06,
      "loss": 0.8521,
      "step": 5230
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.422823145618653,
      "learning_rate": 8.95451332436496e-06,
      "loss": 0.8191,
      "step": 5231
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.23407617471784,
      "learning_rate": 8.954071233797479e-06,
      "loss": 0.8164,
      "step": 5232
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.938111105481462,
      "learning_rate": 8.953629060697422e-06,
      "loss": 0.838,
      "step": 5233
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.846641478438352,
      "learning_rate": 8.953186805074016e-06,
      "loss": 0.8513,
      "step": 5234
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.055895661410474,
      "learning_rate": 8.952744466936492e-06,
      "loss": 0.8326,
      "step": 5235
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.512577905217255,
      "learning_rate": 8.952302046294085e-06,
      "loss": 0.9053,
      "step": 5236
    },
    {
      "epoch": 0.47,
      "grad_norm": 9.078875239990236,
      "learning_rate": 8.951859543156028e-06,
      "loss": 0.7941,
      "step": 5237
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.0505724141337875,
      "learning_rate": 8.951416957531555e-06,
      "loss": 0.8385,
      "step": 5238
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.31826551146731,
      "learning_rate": 8.95097428942991e-06,
      "loss": 0.8402,
      "step": 5239
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.306700356111773,
      "learning_rate": 8.950531538860328e-06,
      "loss": 0.8285,
      "step": 5240
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.105484277795808,
      "learning_rate": 8.950088705832052e-06,
      "loss": 0.8023,
      "step": 5241
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.581645647444966,
      "learning_rate": 8.949645790354325e-06,
      "loss": 0.8123,
      "step": 5242
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.418046318691268,
      "learning_rate": 8.94920279243639e-06,
      "loss": 0.8593,
      "step": 5243
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.592455134514821,
      "learning_rate": 8.948759712087496e-06,
      "loss": 0.8518,
      "step": 5244
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.640358176454387,
      "learning_rate": 8.948316549316891e-06,
      "loss": 0.7954,
      "step": 5245
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.338221490553725,
      "learning_rate": 8.947873304133824e-06,
      "loss": 0.8406,
      "step": 5246
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.082911742259069,
      "learning_rate": 8.947429976547549e-06,
      "loss": 0.7942,
      "step": 5247
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.027675693824544,
      "learning_rate": 8.946986566567317e-06,
      "loss": 0.8385,
      "step": 5248
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.633735265626218,
      "learning_rate": 8.946543074202382e-06,
      "loss": 0.9167,
      "step": 5249
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.8870293724140135,
      "learning_rate": 8.946099499462006e-06,
      "loss": 0.762,
      "step": 5250
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.342132237880394,
      "learning_rate": 8.945655842355445e-06,
      "loss": 0.8695,
      "step": 5251
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.84119282972036,
      "learning_rate": 8.945212102891958e-06,
      "loss": 0.8116,
      "step": 5252
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.835468493849038,
      "learning_rate": 8.94476828108081e-06,
      "loss": 0.7935,
      "step": 5253
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.184435876494331,
      "learning_rate": 8.94432437693126e-06,
      "loss": 0.8297,
      "step": 5254
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.153134071920158,
      "learning_rate": 8.943880390452578e-06,
      "loss": 0.7927,
      "step": 5255
    },
    {
      "epoch": 0.47,
      "grad_norm": 9.821283638407671,
      "learning_rate": 8.94343632165403e-06,
      "loss": 0.8571,
      "step": 5256
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.214367138651282,
      "learning_rate": 8.942992170544884e-06,
      "loss": 0.8542,
      "step": 5257
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.809967284514573,
      "learning_rate": 8.942547937134413e-06,
      "loss": 0.8081,
      "step": 5258
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.1562504931129185,
      "learning_rate": 8.942103621431885e-06,
      "loss": 0.8334,
      "step": 5259
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.216617411811801,
      "learning_rate": 8.941659223446579e-06,
      "loss": 0.8112,
      "step": 5260
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.8603164273013615,
      "learning_rate": 8.94121474318777e-06,
      "loss": 0.7462,
      "step": 5261
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.78220445738919,
      "learning_rate": 8.940770180664734e-06,
      "loss": 0.7742,
      "step": 5262
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.769294610984565,
      "learning_rate": 8.940325535886748e-06,
      "loss": 0.8383,
      "step": 5263
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.017907005765718,
      "learning_rate": 8.939880808863097e-06,
      "loss": 0.8324,
      "step": 5264
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.986055207079286,
      "learning_rate": 8.939435999603063e-06,
      "loss": 0.7738,
      "step": 5265
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.041452994689371,
      "learning_rate": 8.938991108115928e-06,
      "loss": 0.8689,
      "step": 5266
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.846724443101808,
      "learning_rate": 8.938546134410983e-06,
      "loss": 0.7621,
      "step": 5267
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.192495936843143,
      "learning_rate": 8.938101078497509e-06,
      "loss": 0.7984,
      "step": 5268
    },
    {
      "epoch": 0.47,
      "grad_norm": 8.077621494744491,
      "learning_rate": 8.937655940384802e-06,
      "loss": 0.8313,
      "step": 5269
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.332898912414202,
      "learning_rate": 8.93721072008215e-06,
      "loss": 0.7399,
      "step": 5270
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.020884754018453,
      "learning_rate": 8.936765417598845e-06,
      "loss": 0.7243,
      "step": 5271
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.667624981195776,
      "learning_rate": 8.936320032944185e-06,
      "loss": 0.8517,
      "step": 5272
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.343391834524811,
      "learning_rate": 8.935874566127465e-06,
      "loss": 0.7575,
      "step": 5273
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.040696890070989,
      "learning_rate": 8.935429017157982e-06,
      "loss": 0.8013,
      "step": 5274
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.135464145871769,
      "learning_rate": 8.934983386045037e-06,
      "loss": 0.7702,
      "step": 5275
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.4290769900003015,
      "learning_rate": 8.934537672797931e-06,
      "loss": 0.7912,
      "step": 5276
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.285617158832652,
      "learning_rate": 8.934091877425968e-06,
      "loss": 0.8765,
      "step": 5277
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.429904374260959,
      "learning_rate": 8.933645999938452e-06,
      "loss": 0.8637,
      "step": 5278
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.924515122887034,
      "learning_rate": 8.933200040344689e-06,
      "loss": 0.7432,
      "step": 5279
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.141559201370439,
      "learning_rate": 8.932753998653991e-06,
      "loss": 0.8812,
      "step": 5280
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.499340881505883,
      "learning_rate": 8.932307874875666e-06,
      "loss": 0.7377,
      "step": 5281
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.557013221993514,
      "learning_rate": 8.931861669019024e-06,
      "loss": 0.8109,
      "step": 5282
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.277451564421584,
      "learning_rate": 8.93141538109338e-06,
      "loss": 0.8543,
      "step": 5283
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.99043430344877,
      "learning_rate": 8.930969011108051e-06,
      "loss": 0.8657,
      "step": 5284
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.341202914576589,
      "learning_rate": 8.930522559072352e-06,
      "loss": 0.8933,
      "step": 5285
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.183970338841475,
      "learning_rate": 8.930076024995604e-06,
      "loss": 0.7764,
      "step": 5286
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.906125349140781,
      "learning_rate": 8.929629408887123e-06,
      "loss": 0.8576,
      "step": 5287
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.659708402447854,
      "learning_rate": 8.929182710756234e-06,
      "loss": 0.792,
      "step": 5288
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.738814870421521,
      "learning_rate": 8.928735930612262e-06,
      "loss": 0.7916,
      "step": 5289
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.479468076682971,
      "learning_rate": 8.928289068464532e-06,
      "loss": 0.8198,
      "step": 5290
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.7531356612383595,
      "learning_rate": 8.92784212432237e-06,
      "loss": 0.831,
      "step": 5291
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.665324903959377,
      "learning_rate": 8.927395098195105e-06,
      "loss": 0.8419,
      "step": 5292
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.226943784445617,
      "learning_rate": 8.926947990092068e-06,
      "loss": 0.8024,
      "step": 5293
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.210213377288032,
      "learning_rate": 8.92650080002259e-06,
      "loss": 0.8168,
      "step": 5294
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.225332433045818,
      "learning_rate": 8.92605352799601e-06,
      "loss": 0.8271,
      "step": 5295
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.660926729972117,
      "learning_rate": 8.925606174021656e-06,
      "loss": 0.7718,
      "step": 5296
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.843594189518743,
      "learning_rate": 8.925158738108875e-06,
      "loss": 0.7625,
      "step": 5297
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.992034921635682,
      "learning_rate": 8.924711220266999e-06,
      "loss": 0.8382,
      "step": 5298
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.450297925154189,
      "learning_rate": 8.924263620505372e-06,
      "loss": 0.7827,
      "step": 5299
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.837132163652643,
      "learning_rate": 8.923815938833336e-06,
      "loss": 0.8834,
      "step": 5300
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.444131392192165,
      "learning_rate": 8.923368175260233e-06,
      "loss": 0.7511,
      "step": 5301
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.605876668038056,
      "learning_rate": 8.922920329795414e-06,
      "loss": 0.8958,
      "step": 5302
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.16310070569928,
      "learning_rate": 8.922472402448223e-06,
      "loss": 0.8405,
      "step": 5303
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.855634288380377,
      "learning_rate": 8.92202439322801e-06,
      "loss": 0.7465,
      "step": 5304
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.037764649450968,
      "learning_rate": 8.921576302144129e-06,
      "loss": 0.8274,
      "step": 5305
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.002688099143435,
      "learning_rate": 8.92112812920593e-06,
      "loss": 0.8229,
      "step": 5306
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.552781726659676,
      "learning_rate": 8.920679874422766e-06,
      "loss": 0.8091,
      "step": 5307
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.905196978528415,
      "learning_rate": 8.920231537803998e-06,
      "loss": 0.9041,
      "step": 5308
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.858175567241367,
      "learning_rate": 8.919783119358983e-06,
      "loss": 0.7927,
      "step": 5309
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.7571741968332155,
      "learning_rate": 8.919334619097078e-06,
      "loss": 0.8503,
      "step": 5310
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.2673708420786705,
      "learning_rate": 8.918886037027646e-06,
      "loss": 0.838,
      "step": 5311
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.249237502490381,
      "learning_rate": 8.918437373160049e-06,
      "loss": 0.8436,
      "step": 5312
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.880788704380828,
      "learning_rate": 8.917988627503655e-06,
      "loss": 0.8229,
      "step": 5313
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.858748326674754,
      "learning_rate": 8.917539800067827e-06,
      "loss": 0.8202,
      "step": 5314
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.478832612369946,
      "learning_rate": 8.917090890861938e-06,
      "loss": 0.7812,
      "step": 5315
    },
    {
      "epoch": 0.47,
      "grad_norm": 8.218878147570763,
      "learning_rate": 8.916641899895351e-06,
      "loss": 0.8551,
      "step": 5316
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.2641647656701585,
      "learning_rate": 8.916192827177444e-06,
      "loss": 0.8206,
      "step": 5317
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.675589177879057,
      "learning_rate": 8.915743672717588e-06,
      "loss": 0.8406,
      "step": 5318
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.719289326357555,
      "learning_rate": 8.915294436525158e-06,
      "loss": 0.9279,
      "step": 5319
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.843101217239931,
      "learning_rate": 8.91484511860953e-06,
      "loss": 0.8037,
      "step": 5320
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.010792712561126,
      "learning_rate": 8.914395718980084e-06,
      "loss": 0.8942,
      "step": 5321
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.32345497024361,
      "learning_rate": 8.9139462376462e-06,
      "loss": 0.8201,
      "step": 5322
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.585061450573544,
      "learning_rate": 8.91349667461726e-06,
      "loss": 0.7794,
      "step": 5323
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.142161132611303,
      "learning_rate": 8.913047029902646e-06,
      "loss": 0.8305,
      "step": 5324
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.779243141921022,
      "learning_rate": 8.912597303511746e-06,
      "loss": 0.8847,
      "step": 5325
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.327577775346467,
      "learning_rate": 8.912147495453946e-06,
      "loss": 0.7883,
      "step": 5326
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.967547649450736,
      "learning_rate": 8.911697605738634e-06,
      "loss": 0.843,
      "step": 5327
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.859851833829306,
      "learning_rate": 8.911247634375201e-06,
      "loss": 0.8525,
      "step": 5328
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.784127138607858,
      "learning_rate": 8.910797581373038e-06,
      "loss": 0.7835,
      "step": 5329
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.374779572002459,
      "learning_rate": 8.910347446741541e-06,
      "loss": 0.8095,
      "step": 5330
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.969599572463945,
      "learning_rate": 8.909897230490105e-06,
      "loss": 0.8144,
      "step": 5331
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.88542768037812,
      "learning_rate": 8.909446932628128e-06,
      "loss": 0.7873,
      "step": 5332
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.7920789211157375,
      "learning_rate": 8.908996553165005e-06,
      "loss": 0.8235,
      "step": 5333
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.984841442525056,
      "learning_rate": 8.90854609211014e-06,
      "loss": 0.7938,
      "step": 5334
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.221739521481055,
      "learning_rate": 8.908095549472936e-06,
      "loss": 0.8547,
      "step": 5335
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.650276747396451,
      "learning_rate": 8.907644925262796e-06,
      "loss": 0.7847,
      "step": 5336
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.81315313095072,
      "learning_rate": 8.907194219489125e-06,
      "loss": 0.7794,
      "step": 5337
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.664181380089211,
      "learning_rate": 8.90674343216133e-06,
      "loss": 0.803,
      "step": 5338
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.22548318381688,
      "learning_rate": 8.906292563288826e-06,
      "loss": 0.7532,
      "step": 5339
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.95294562980093,
      "learning_rate": 8.905841612881015e-06,
      "loss": 0.9053,
      "step": 5340
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.485863873383317,
      "learning_rate": 8.905390580947315e-06,
      "loss": 0.8524,
      "step": 5341
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.053201839413083,
      "learning_rate": 8.90493946749714e-06,
      "loss": 0.884,
      "step": 5342
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.386235207508995,
      "learning_rate": 8.904488272539905e-06,
      "loss": 0.8453,
      "step": 5343
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.493222314721316,
      "learning_rate": 8.904036996085028e-06,
      "loss": 0.8647,
      "step": 5344
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.334128233296175,
      "learning_rate": 8.90358563814193e-06,
      "loss": 0.8377,
      "step": 5345
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.882368424034483,
      "learning_rate": 8.903134198720027e-06,
      "loss": 0.8149,
      "step": 5346
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.755851540560259,
      "learning_rate": 8.902682677828749e-06,
      "loss": 0.8266,
      "step": 5347
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.914346777970283,
      "learning_rate": 8.902231075477514e-06,
      "loss": 0.9492,
      "step": 5348
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.585028324860895,
      "learning_rate": 8.901779391675752e-06,
      "loss": 0.7595,
      "step": 5349
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.612288952961467,
      "learning_rate": 8.90132762643289e-06,
      "loss": 0.7445,
      "step": 5350
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.172014577916074,
      "learning_rate": 8.900875779758358e-06,
      "loss": 0.804,
      "step": 5351
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.898302126593629,
      "learning_rate": 8.900423851661586e-06,
      "loss": 0.7638,
      "step": 5352
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.2239629391039015,
      "learning_rate": 8.899971842152008e-06,
      "loss": 0.8124,
      "step": 5353
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.197159674990731,
      "learning_rate": 8.89951975123906e-06,
      "loss": 0.8412,
      "step": 5354
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.265370229346946,
      "learning_rate": 8.899067578932177e-06,
      "loss": 0.813,
      "step": 5355
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.535559382986295,
      "learning_rate": 8.898615325240796e-06,
      "loss": 0.7463,
      "step": 5356
    },
    {
      "epoch": 0.48,
      "grad_norm": 8.306020248533402,
      "learning_rate": 8.898162990174358e-06,
      "loss": 0.8317,
      "step": 5357
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.938115533948314,
      "learning_rate": 8.897710573742304e-06,
      "loss": 0.8241,
      "step": 5358
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.786529313239252,
      "learning_rate": 8.897258075954078e-06,
      "loss": 0.8824,
      "step": 5359
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.068772430843752,
      "learning_rate": 8.896805496819125e-06,
      "loss": 0.8877,
      "step": 5360
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.965898232876674,
      "learning_rate": 8.89635283634689e-06,
      "loss": 0.9091,
      "step": 5361
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.580070472307052,
      "learning_rate": 8.895900094546823e-06,
      "loss": 0.8242,
      "step": 5362
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.941670926798362,
      "learning_rate": 8.895447271428374e-06,
      "loss": 0.8729,
      "step": 5363
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.4721469470497075,
      "learning_rate": 8.894994367000993e-06,
      "loss": 0.8346,
      "step": 5364
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.909836003334355,
      "learning_rate": 8.894541381274136e-06,
      "loss": 0.7766,
      "step": 5365
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.298420437768125,
      "learning_rate": 8.894088314257254e-06,
      "loss": 0.8287,
      "step": 5366
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.609417397752601,
      "learning_rate": 8.893635165959809e-06,
      "loss": 0.8252,
      "step": 5367
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.331677654913266,
      "learning_rate": 8.893181936391255e-06,
      "loss": 0.7883,
      "step": 5368
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.559048133082869,
      "learning_rate": 8.892728625561056e-06,
      "loss": 0.8719,
      "step": 5369
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.934668766060812,
      "learning_rate": 8.892275233478668e-06,
      "loss": 0.8342,
      "step": 5370
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.9023831122771515,
      "learning_rate": 8.891821760153563e-06,
      "loss": 0.8237,
      "step": 5371
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.863793100930706,
      "learning_rate": 8.891368205595198e-06,
      "loss": 0.8008,
      "step": 5372
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.6049163018328345,
      "learning_rate": 8.890914569813048e-06,
      "loss": 0.9064,
      "step": 5373
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.048074001846459,
      "learning_rate": 8.890460852816574e-06,
      "loss": 0.7912,
      "step": 5374
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.8131741318618495,
      "learning_rate": 8.89000705461525e-06,
      "loss": 0.7957,
      "step": 5375
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.82703367553774,
      "learning_rate": 8.889553175218549e-06,
      "loss": 0.8743,
      "step": 5376
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.654266462041014,
      "learning_rate": 8.889099214635941e-06,
      "loss": 0.8532,
      "step": 5377
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.446130627651068,
      "learning_rate": 8.888645172876907e-06,
      "loss": 0.8111,
      "step": 5378
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.795044736553056,
      "learning_rate": 8.888191049950918e-06,
      "loss": 0.8268,
      "step": 5379
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.0516035599076465,
      "learning_rate": 8.887736845867458e-06,
      "loss": 0.8623,
      "step": 5380
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.268227876017758,
      "learning_rate": 8.887282560636003e-06,
      "loss": 0.8186,
      "step": 5381
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.43316159516123,
      "learning_rate": 8.88682819426604e-06,
      "loss": 0.8161,
      "step": 5382
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.779326256832511,
      "learning_rate": 8.88637374676705e-06,
      "loss": 0.8285,
      "step": 5383
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.038199355160814,
      "learning_rate": 8.885919218148519e-06,
      "loss": 0.9075,
      "step": 5384
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.363384528890565,
      "learning_rate": 8.885464608419934e-06,
      "loss": 0.8519,
      "step": 5385
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.8592881427923444,
      "learning_rate": 8.885009917590783e-06,
      "loss": 0.8579,
      "step": 5386
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.61344661247994,
      "learning_rate": 8.884555145670561e-06,
      "loss": 0.8018,
      "step": 5387
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.517602282236666,
      "learning_rate": 8.884100292668755e-06,
      "loss": 0.8161,
      "step": 5388
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.934780603085137,
      "learning_rate": 8.883645358594863e-06,
      "loss": 0.8435,
      "step": 5389
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.372529534697981,
      "learning_rate": 8.883190343458378e-06,
      "loss": 0.7458,
      "step": 5390
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.482037933168966,
      "learning_rate": 8.882735247268798e-06,
      "loss": 0.8057,
      "step": 5391
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.360708653475338,
      "learning_rate": 8.882280070035625e-06,
      "loss": 0.8733,
      "step": 5392
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.36592359457441,
      "learning_rate": 8.881824811768356e-06,
      "loss": 0.7961,
      "step": 5393
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.535807727213531,
      "learning_rate": 8.881369472476495e-06,
      "loss": 0.8504,
      "step": 5394
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.838749433334302,
      "learning_rate": 8.880914052169547e-06,
      "loss": 0.7549,
      "step": 5395
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.024264638015091,
      "learning_rate": 8.880458550857016e-06,
      "loss": 0.8235,
      "step": 5396
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.545301739993742,
      "learning_rate": 8.880002968548411e-06,
      "loss": 0.8193,
      "step": 5397
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.734990158845536,
      "learning_rate": 8.879547305253243e-06,
      "loss": 0.8046,
      "step": 5398
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.149357074085558,
      "learning_rate": 8.879091560981018e-06,
      "loss": 0.8295,
      "step": 5399
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.398068584664788,
      "learning_rate": 8.878635735741255e-06,
      "loss": 0.8071,
      "step": 5400
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.94047688254384,
      "learning_rate": 8.878179829543464e-06,
      "loss": 0.8302,
      "step": 5401
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.279321145065282,
      "learning_rate": 8.87772384239716e-06,
      "loss": 0.8166,
      "step": 5402
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.591775707857859,
      "learning_rate": 8.877267774311866e-06,
      "loss": 0.8167,
      "step": 5403
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.865245089863423,
      "learning_rate": 8.876811625297097e-06,
      "loss": 0.8332,
      "step": 5404
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.892108104350891,
      "learning_rate": 8.876355395362376e-06,
      "loss": 0.7848,
      "step": 5405
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.364838412407481,
      "learning_rate": 8.875899084517225e-06,
      "loss": 0.8589,
      "step": 5406
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.407991646383552,
      "learning_rate": 8.875442692771168e-06,
      "loss": 0.8949,
      "step": 5407
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.205000380009722,
      "learning_rate": 8.874986220133731e-06,
      "loss": 0.8754,
      "step": 5408
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.411793670492893,
      "learning_rate": 8.874529666614444e-06,
      "loss": 0.923,
      "step": 5409
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.418524750629432,
      "learning_rate": 8.874073032222836e-06,
      "loss": 0.8153,
      "step": 5410
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.902290378716985,
      "learning_rate": 8.873616316968436e-06,
      "loss": 0.8398,
      "step": 5411
    },
    {
      "epoch": 0.48,
      "grad_norm": 8.365000742655269,
      "learning_rate": 8.87315952086078e-06,
      "loss": 0.8416,
      "step": 5412
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.4099371610854465,
      "learning_rate": 8.8727026439094e-06,
      "loss": 0.8034,
      "step": 5413
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.412474507641492,
      "learning_rate": 8.872245686123832e-06,
      "loss": 0.8056,
      "step": 5414
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.502540192278813,
      "learning_rate": 8.871788647513617e-06,
      "loss": 0.7985,
      "step": 5415
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.05957808188602,
      "learning_rate": 8.871331528088294e-06,
      "loss": 0.7476,
      "step": 5416
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.791314348128598,
      "learning_rate": 8.870874327857401e-06,
      "loss": 0.8009,
      "step": 5417
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.486625955336836,
      "learning_rate": 8.870417046830487e-06,
      "loss": 0.8294,
      "step": 5418
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.229304740321843,
      "learning_rate": 8.86995968501709e-06,
      "loss": 0.8524,
      "step": 5419
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.610055167633258,
      "learning_rate": 8.869502242426761e-06,
      "loss": 0.8392,
      "step": 5420
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.2085571817770235,
      "learning_rate": 8.869044719069047e-06,
      "loss": 0.7903,
      "step": 5421
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.77853831865441,
      "learning_rate": 8.868587114953497e-06,
      "loss": 0.7917,
      "step": 5422
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.237914822932454,
      "learning_rate": 8.868129430089663e-06,
      "loss": 0.8811,
      "step": 5423
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.766112083211153,
      "learning_rate": 8.867671664487099e-06,
      "loss": 0.8341,
      "step": 5424
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.790786625719171,
      "learning_rate": 8.867213818155357e-06,
      "loss": 0.8623,
      "step": 5425
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.782346363002075,
      "learning_rate": 8.866755891103997e-06,
      "loss": 0.758,
      "step": 5426
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.702478347269997,
      "learning_rate": 8.866297883342577e-06,
      "loss": 0.8115,
      "step": 5427
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.184316421012745,
      "learning_rate": 8.865839794880653e-06,
      "loss": 0.7868,
      "step": 5428
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.211516453327603,
      "learning_rate": 8.865381625727791e-06,
      "loss": 0.8049,
      "step": 5429
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.3510942449992305,
      "learning_rate": 8.864923375893552e-06,
      "loss": 0.957,
      "step": 5430
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.128167190911886,
      "learning_rate": 8.864465045387502e-06,
      "loss": 0.834,
      "step": 5431
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.627019375161246,
      "learning_rate": 8.864006634219207e-06,
      "loss": 0.8569,
      "step": 5432
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.620272146979533,
      "learning_rate": 8.863548142398238e-06,
      "loss": 0.8455,
      "step": 5433
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.816628448398909,
      "learning_rate": 8.863089569934158e-06,
      "loss": 0.8014,
      "step": 5434
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.650120162193779,
      "learning_rate": 8.862630916836547e-06,
      "loss": 0.8295,
      "step": 5435
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.334862531189611,
      "learning_rate": 8.862172183114972e-06,
      "loss": 0.8389,
      "step": 5436
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.167755882044029,
      "learning_rate": 8.861713368779013e-06,
      "loss": 0.8195,
      "step": 5437
    },
    {
      "epoch": 0.49,
      "grad_norm": 11.633534148378313,
      "learning_rate": 8.861254473838244e-06,
      "loss": 0.9308,
      "step": 5438
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.814310067894542,
      "learning_rate": 8.860795498302242e-06,
      "loss": 0.8424,
      "step": 5439
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.631489406704529,
      "learning_rate": 8.860336442180591e-06,
      "loss": 0.8393,
      "step": 5440
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.7961697150847025,
      "learning_rate": 8.85987730548287e-06,
      "loss": 0.7876,
      "step": 5441
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.301040409056121,
      "learning_rate": 8.859418088218664e-06,
      "loss": 0.8815,
      "step": 5442
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.010141640063692,
      "learning_rate": 8.858958790397554e-06,
      "loss": 0.8294,
      "step": 5443
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.7657144918935,
      "learning_rate": 8.858499412029134e-06,
      "loss": 0.7738,
      "step": 5444
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.140036710698247,
      "learning_rate": 8.858039953122988e-06,
      "loss": 0.7698,
      "step": 5445
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.755576234117862,
      "learning_rate": 8.857580413688706e-06,
      "loss": 0.822,
      "step": 5446
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.547285611644854,
      "learning_rate": 8.857120793735881e-06,
      "loss": 0.823,
      "step": 5447
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.871221980081119,
      "learning_rate": 8.856661093274105e-06,
      "loss": 0.8055,
      "step": 5448
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.609805085422755,
      "learning_rate": 8.856201312312978e-06,
      "loss": 0.8872,
      "step": 5449
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.887767840571917,
      "learning_rate": 8.855741450862091e-06,
      "loss": 0.829,
      "step": 5450
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.080567377623554,
      "learning_rate": 8.855281508931047e-06,
      "loss": 0.8142,
      "step": 5451
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.491634993569011,
      "learning_rate": 8.854821486529443e-06,
      "loss": 0.7731,
      "step": 5452
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.7692879060481435,
      "learning_rate": 8.854361383666883e-06,
      "loss": 0.8305,
      "step": 5453
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.409847738114942,
      "learning_rate": 8.85390120035297e-06,
      "loss": 0.8526,
      "step": 5454
    },
    {
      "epoch": 0.49,
      "grad_norm": 9.010385641662076,
      "learning_rate": 8.853440936597308e-06,
      "loss": 0.7644,
      "step": 5455
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.359258545695406,
      "learning_rate": 8.852980592409507e-06,
      "loss": 0.7861,
      "step": 5456
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.8658166230734015,
      "learning_rate": 8.852520167799173e-06,
      "loss": 0.8586,
      "step": 5457
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.161745509601713,
      "learning_rate": 8.852059662775918e-06,
      "loss": 0.8605,
      "step": 5458
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.120811378207375,
      "learning_rate": 8.851599077349354e-06,
      "loss": 0.7828,
      "step": 5459
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.672714923819332,
      "learning_rate": 8.851138411529094e-06,
      "loss": 0.8461,
      "step": 5460
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.680910291214889,
      "learning_rate": 8.850677665324751e-06,
      "loss": 0.8463,
      "step": 5461
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.695871571161157,
      "learning_rate": 8.850216838745947e-06,
      "loss": 0.8411,
      "step": 5462
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.522777129331663,
      "learning_rate": 8.849755931802298e-06,
      "loss": 0.8357,
      "step": 5463
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.327416702450108,
      "learning_rate": 8.849294944503424e-06,
      "loss": 0.7882,
      "step": 5464
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.546660644137393,
      "learning_rate": 8.848833876858949e-06,
      "loss": 0.8092,
      "step": 5465
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.767951619055209,
      "learning_rate": 8.848372728878495e-06,
      "loss": 0.8273,
      "step": 5466
    },
    {
      "epoch": 0.49,
      "grad_norm": 8.358040895830687,
      "learning_rate": 8.847911500571688e-06,
      "loss": 0.8264,
      "step": 5467
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.772754493476175,
      "learning_rate": 8.847450191948155e-06,
      "loss": 0.8215,
      "step": 5468
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.007811361640313,
      "learning_rate": 8.846988803017526e-06,
      "loss": 0.8833,
      "step": 5469
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.824473978082111,
      "learning_rate": 8.84652733378943e-06,
      "loss": 0.8517,
      "step": 5470
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.460611978482936,
      "learning_rate": 8.846065784273499e-06,
      "loss": 0.8049,
      "step": 5471
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.280705751284739,
      "learning_rate": 8.845604154479368e-06,
      "loss": 0.7656,
      "step": 5472
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.745439486371974,
      "learning_rate": 8.845142444416672e-06,
      "loss": 0.801,
      "step": 5473
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.969438824931848,
      "learning_rate": 8.844680654095048e-06,
      "loss": 0.8482,
      "step": 5474
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.815412910498652,
      "learning_rate": 8.844218783524135e-06,
      "loss": 0.8338,
      "step": 5475
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.611102785561891,
      "learning_rate": 8.843756832713575e-06,
      "loss": 0.8473,
      "step": 5476
    },
    {
      "epoch": 0.49,
      "grad_norm": 8.985500794668559,
      "learning_rate": 8.843294801673007e-06,
      "loss": 0.8337,
      "step": 5477
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.319099780359786,
      "learning_rate": 8.842832690412078e-06,
      "loss": 0.8316,
      "step": 5478
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.280261643715185,
      "learning_rate": 8.84237049894043e-06,
      "loss": 0.7898,
      "step": 5479
    },
    {
      "epoch": 0.49,
      "grad_norm": 8.636365096031428,
      "learning_rate": 8.841908227267714e-06,
      "loss": 0.7836,
      "step": 5480
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.349142595841628,
      "learning_rate": 8.841445875403578e-06,
      "loss": 0.7616,
      "step": 5481
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.2571466047928865,
      "learning_rate": 8.84098344335767e-06,
      "loss": 0.804,
      "step": 5482
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.700031643782717,
      "learning_rate": 8.840520931139647e-06,
      "loss": 0.8332,
      "step": 5483
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.398499009596309,
      "learning_rate": 8.840058338759159e-06,
      "loss": 0.7936,
      "step": 5484
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.16295571937463,
      "learning_rate": 8.839595666225862e-06,
      "loss": 0.8691,
      "step": 5485
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.466730670706551,
      "learning_rate": 8.839132913549415e-06,
      "loss": 0.861,
      "step": 5486
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.701858505388962,
      "learning_rate": 8.838670080739475e-06,
      "loss": 0.8563,
      "step": 5487
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.894613893901296,
      "learning_rate": 8.838207167805704e-06,
      "loss": 0.8764,
      "step": 5488
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.007647514809908,
      "learning_rate": 8.837744174757765e-06,
      "loss": 0.8085,
      "step": 5489
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.281215170492915,
      "learning_rate": 8.83728110160532e-06,
      "loss": 0.8811,
      "step": 5490
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.63612072243749,
      "learning_rate": 8.836817948358035e-06,
      "loss": 0.8269,
      "step": 5491
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.53196969016628,
      "learning_rate": 8.83635471502558e-06,
      "loss": 0.8707,
      "step": 5492
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.415027663847866,
      "learning_rate": 8.835891401617617e-06,
      "loss": 0.7876,
      "step": 5493
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.772663549058861,
      "learning_rate": 8.835428008143825e-06,
      "loss": 0.8728,
      "step": 5494
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.481638207990558,
      "learning_rate": 8.834964534613873e-06,
      "loss": 0.818,
      "step": 5495
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.093130864434619,
      "learning_rate": 8.834500981037435e-06,
      "loss": 0.8075,
      "step": 5496
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.259009265893882,
      "learning_rate": 8.834037347424185e-06,
      "loss": 0.8193,
      "step": 5497
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.079136867732288,
      "learning_rate": 8.833573633783801e-06,
      "loss": 0.8508,
      "step": 5498
    },
    {
      "epoch": 0.49,
      "grad_norm": 8.087602471488859,
      "learning_rate": 8.833109840125964e-06,
      "loss": 0.8533,
      "step": 5499
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.815014187441658,
      "learning_rate": 8.832645966460353e-06,
      "loss": 0.7872,
      "step": 5500
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.038548263659589,
      "learning_rate": 8.832182012796653e-06,
      "loss": 0.804,
      "step": 5501
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.051079691388929,
      "learning_rate": 8.831717979144544e-06,
      "loss": 0.8355,
      "step": 5502
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.33231088081368,
      "learning_rate": 8.831253865513713e-06,
      "loss": 0.7953,
      "step": 5503
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.292162416451263,
      "learning_rate": 8.830789671913849e-06,
      "loss": 0.756,
      "step": 5504
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.359062289872673,
      "learning_rate": 8.830325398354639e-06,
      "loss": 0.7167,
      "step": 5505
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.989652288671198,
      "learning_rate": 8.829861044845775e-06,
      "loss": 0.8359,
      "step": 5506
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.237497392495298,
      "learning_rate": 8.82939661139695e-06,
      "loss": 0.8418,
      "step": 5507
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.89708717575345,
      "learning_rate": 8.828932098017855e-06,
      "loss": 0.8272,
      "step": 5508
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.928310920873493,
      "learning_rate": 8.828467504718188e-06,
      "loss": 0.8079,
      "step": 5509
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.879394927845364,
      "learning_rate": 8.828002831507648e-06,
      "loss": 0.7699,
      "step": 5510
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.547101142463651,
      "learning_rate": 8.82753807839593e-06,
      "loss": 0.7715,
      "step": 5511
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.610563795519481,
      "learning_rate": 8.827073245392736e-06,
      "loss": 0.8862,
      "step": 5512
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.378056086251798,
      "learning_rate": 8.82660833250777e-06,
      "loss": 0.832,
      "step": 5513
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.692397991489804,
      "learning_rate": 8.826143339750737e-06,
      "loss": 0.8358,
      "step": 5514
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.575583704582548,
      "learning_rate": 8.82567826713134e-06,
      "loss": 0.8118,
      "step": 5515
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.803273868418658,
      "learning_rate": 8.825213114659286e-06,
      "loss": 0.7884,
      "step": 5516
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.275578057843489,
      "learning_rate": 8.824747882344286e-06,
      "loss": 0.8129,
      "step": 5517
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.3620056644674525,
      "learning_rate": 8.824282570196051e-06,
      "loss": 0.8162,
      "step": 5518
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.935843539546981,
      "learning_rate": 8.823817178224292e-06,
      "loss": 0.7283,
      "step": 5519
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.992701268700711,
      "learning_rate": 8.823351706438722e-06,
      "loss": 0.8933,
      "step": 5520
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.4977656823201375,
      "learning_rate": 8.82288615484906e-06,
      "loss": 0.7773,
      "step": 5521
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.821219411381136,
      "learning_rate": 8.82242052346502e-06,
      "loss": 0.9303,
      "step": 5522
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.038045182833607,
      "learning_rate": 8.821954812296325e-06,
      "loss": 0.8098,
      "step": 5523
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.642245397080293,
      "learning_rate": 8.821489021352691e-06,
      "loss": 0.8493,
      "step": 5524
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.678296364222998,
      "learning_rate": 8.821023150643843e-06,
      "loss": 0.8182,
      "step": 5525
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.744838168195922,
      "learning_rate": 8.820557200179505e-06,
      "loss": 0.8099,
      "step": 5526
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.101604287224704,
      "learning_rate": 8.8200911699694e-06,
      "loss": 0.8024,
      "step": 5527
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.205420755092225,
      "learning_rate": 8.81962506002326e-06,
      "loss": 0.8363,
      "step": 5528
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.395375879570854,
      "learning_rate": 8.819158870350813e-06,
      "loss": 0.8287,
      "step": 5529
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.870131073485972,
      "learning_rate": 8.818692600961786e-06,
      "loss": 0.8635,
      "step": 5530
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.509271488746129,
      "learning_rate": 8.818226251865915e-06,
      "loss": 0.77,
      "step": 5531
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.787917101235761,
      "learning_rate": 8.817759823072931e-06,
      "loss": 0.8023,
      "step": 5532
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.862523367151897,
      "learning_rate": 8.817293314592574e-06,
      "loss": 0.7562,
      "step": 5533
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.0162477258097455,
      "learning_rate": 8.816826726434578e-06,
      "loss": 0.7948,
      "step": 5534
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.772189120286938,
      "learning_rate": 8.816360058608681e-06,
      "loss": 0.8797,
      "step": 5535
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.92605402421492,
      "learning_rate": 8.815893311124626e-06,
      "loss": 0.7921,
      "step": 5536
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.2204181436127515,
      "learning_rate": 8.815426483992156e-06,
      "loss": 0.805,
      "step": 5537
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.950570599642973,
      "learning_rate": 8.814959577221014e-06,
      "loss": 0.8243,
      "step": 5538
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.637243572562901,
      "learning_rate": 8.814492590820944e-06,
      "loss": 0.7641,
      "step": 5539
    },
    {
      "epoch": 0.49,
      "grad_norm": 9.948119929150002,
      "learning_rate": 8.814025524801698e-06,
      "loss": 0.8242,
      "step": 5540
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.270505237682784,
      "learning_rate": 8.813558379173018e-06,
      "loss": 0.8194,
      "step": 5541
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.350967312599104,
      "learning_rate": 8.81309115394466e-06,
      "loss": 0.7967,
      "step": 5542
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.0784511337081195,
      "learning_rate": 8.812623849126374e-06,
      "loss": 0.8551,
      "step": 5543
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.423206428464275,
      "learning_rate": 8.812156464727914e-06,
      "loss": 0.8839,
      "step": 5544
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.048261481207874,
      "learning_rate": 8.811689000759039e-06,
      "loss": 0.8444,
      "step": 5545
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.82661507029909,
      "learning_rate": 8.8112214572295e-06,
      "loss": 0.8767,
      "step": 5546
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.830945119570509,
      "learning_rate": 8.810753834149061e-06,
      "loss": 0.8249,
      "step": 5547
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.494961929740061,
      "learning_rate": 8.810286131527482e-06,
      "loss": 0.7788,
      "step": 5548
    },
    {
      "epoch": 0.5,
      "grad_norm": 8.4813436457888,
      "learning_rate": 8.809818349374525e-06,
      "loss": 0.8084,
      "step": 5549
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.7445059558716425,
      "learning_rate": 8.809350487699952e-06,
      "loss": 0.8264,
      "step": 5550
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.760521975681675,
      "learning_rate": 8.80888254651353e-06,
      "loss": 0.7546,
      "step": 5551
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.905108546444297,
      "learning_rate": 8.808414525825026e-06,
      "loss": 0.8335,
      "step": 5552
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.3083934368827554,
      "learning_rate": 8.807946425644209e-06,
      "loss": 0.8347,
      "step": 5553
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.452110977896546,
      "learning_rate": 8.807478245980851e-06,
      "loss": 0.8396,
      "step": 5554
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.723570500052544,
      "learning_rate": 8.80700998684472e-06,
      "loss": 0.8131,
      "step": 5555
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.765869890188234,
      "learning_rate": 8.806541648245595e-06,
      "loss": 0.8214,
      "step": 5556
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.845920796871105,
      "learning_rate": 8.806073230193251e-06,
      "loss": 0.876,
      "step": 5557
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.526438446816304,
      "learning_rate": 8.80560473269746e-06,
      "loss": 0.905,
      "step": 5558
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.212430867488916,
      "learning_rate": 8.805136155768005e-06,
      "loss": 0.8247,
      "step": 5559
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.162608944181552,
      "learning_rate": 8.804667499414668e-06,
      "loss": 0.8491,
      "step": 5560
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.846372029558468,
      "learning_rate": 8.804198763647227e-06,
      "loss": 0.8622,
      "step": 5561
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.124835685312793,
      "learning_rate": 8.803729948475467e-06,
      "loss": 0.8229,
      "step": 5562
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.813329825246744,
      "learning_rate": 8.803261053909177e-06,
      "loss": 0.8313,
      "step": 5563
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.1665416802696535,
      "learning_rate": 8.80279207995814e-06,
      "loss": 0.7806,
      "step": 5564
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.4733843362881105,
      "learning_rate": 8.802323026632147e-06,
      "loss": 0.8733,
      "step": 5565
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.289806814233937,
      "learning_rate": 8.801853893940987e-06,
      "loss": 0.8078,
      "step": 5566
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.166284056659091,
      "learning_rate": 8.801384681894453e-06,
      "loss": 0.8514,
      "step": 5567
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.182258535535486,
      "learning_rate": 8.800915390502338e-06,
      "loss": 0.8115,
      "step": 5568
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.661992984346423,
      "learning_rate": 8.80044601977444e-06,
      "loss": 0.7609,
      "step": 5569
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.922986478312525,
      "learning_rate": 8.799976569720552e-06,
      "loss": 0.7995,
      "step": 5570
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.4527155106389,
      "learning_rate": 8.799507040350476e-06,
      "loss": 0.8414,
      "step": 5571
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.425482336298079,
      "learning_rate": 8.799037431674012e-06,
      "loss": 0.8619,
      "step": 5572
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.909608891076864,
      "learning_rate": 8.79856774370096e-06,
      "loss": 0.8441,
      "step": 5573
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.140563043841346,
      "learning_rate": 8.798097976441126e-06,
      "loss": 0.7949,
      "step": 5574
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.462764458378051,
      "learning_rate": 8.797628129904314e-06,
      "loss": 0.8478,
      "step": 5575
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.864667849553765,
      "learning_rate": 8.797158204100332e-06,
      "loss": 0.839,
      "step": 5576
    },
    {
      "epoch": 0.5,
      "grad_norm": 10.475613685631325,
      "learning_rate": 8.796688199038987e-06,
      "loss": 0.9458,
      "step": 5577
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.0872039393537385,
      "learning_rate": 8.796218114730092e-06,
      "loss": 0.7936,
      "step": 5578
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.260676665586891,
      "learning_rate": 8.795747951183457e-06,
      "loss": 0.7494,
      "step": 5579
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.097795977628091,
      "learning_rate": 8.795277708408896e-06,
      "loss": 0.8645,
      "step": 5580
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.432444455672716,
      "learning_rate": 8.794807386416224e-06,
      "loss": 0.9038,
      "step": 5581
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.704478679427868,
      "learning_rate": 8.794336985215258e-06,
      "loss": 0.8547,
      "step": 5582
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.533703270259658,
      "learning_rate": 8.793866504815818e-06,
      "loss": 0.8349,
      "step": 5583
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.169590163062241,
      "learning_rate": 8.793395945227721e-06,
      "loss": 0.914,
      "step": 5584
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.116209001314709,
      "learning_rate": 8.792925306460793e-06,
      "loss": 0.8366,
      "step": 5585
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.196099869791053,
      "learning_rate": 8.792454588524855e-06,
      "loss": 0.8674,
      "step": 5586
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.864568133713459,
      "learning_rate": 8.791983791429732e-06,
      "loss": 0.8403,
      "step": 5587
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.555023958616594,
      "learning_rate": 8.791512915185254e-06,
      "loss": 0.8504,
      "step": 5588
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.967745313050185,
      "learning_rate": 8.791041959801245e-06,
      "loss": 0.7898,
      "step": 5589
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.408646893938039,
      "learning_rate": 8.790570925287538e-06,
      "loss": 0.7598,
      "step": 5590
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.948070460809465,
      "learning_rate": 8.790099811653963e-06,
      "loss": 0.8019,
      "step": 5591
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.437654839604383,
      "learning_rate": 8.789628618910357e-06,
      "loss": 0.7745,
      "step": 5592
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.868528546766006,
      "learning_rate": 8.789157347066552e-06,
      "loss": 0.8234,
      "step": 5593
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.48313342678063,
      "learning_rate": 8.788685996132384e-06,
      "loss": 0.8387,
      "step": 5594
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.803805977904739,
      "learning_rate": 8.788214566117694e-06,
      "loss": 0.7895,
      "step": 5595
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.69988015977202,
      "learning_rate": 8.78774305703232e-06,
      "loss": 0.8981,
      "step": 5596
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.746695794606797,
      "learning_rate": 8.787271468886107e-06,
      "loss": 0.8371,
      "step": 5597
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.136166592481909,
      "learning_rate": 8.786799801688894e-06,
      "loss": 0.8289,
      "step": 5598
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.1258418236165095,
      "learning_rate": 8.786328055450528e-06,
      "loss": 0.8845,
      "step": 5599
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.855236411487828,
      "learning_rate": 8.785856230180857e-06,
      "loss": 0.8081,
      "step": 5600
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.86665439317705,
      "learning_rate": 8.785384325889728e-06,
      "loss": 0.8034,
      "step": 5601
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.404959450403622,
      "learning_rate": 8.784912342586991e-06,
      "loss": 0.8229,
      "step": 5602
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.489964474685759,
      "learning_rate": 8.784440280282497e-06,
      "loss": 0.7587,
      "step": 5603
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.9481701347505815,
      "learning_rate": 8.7839681389861e-06,
      "loss": 0.8232,
      "step": 5604
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.593131991702785,
      "learning_rate": 8.783495918707653e-06,
      "loss": 0.8456,
      "step": 5605
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.63372069218948,
      "learning_rate": 8.783023619457019e-06,
      "loss": 0.8471,
      "step": 5606
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.150534282242147,
      "learning_rate": 8.782551241244049e-06,
      "loss": 0.831,
      "step": 5607
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.778367068231051,
      "learning_rate": 8.782078784078604e-06,
      "loss": 0.7912,
      "step": 5608
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.932322018378101,
      "learning_rate": 8.78160624797055e-06,
      "loss": 0.8051,
      "step": 5609
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.233044214794863,
      "learning_rate": 8.781133632929745e-06,
      "loss": 0.8054,
      "step": 5610
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.251248701755817,
      "learning_rate": 8.780660938966057e-06,
      "loss": 0.8654,
      "step": 5611
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.553410706448754,
      "learning_rate": 8.78018816608935e-06,
      "loss": 0.8432,
      "step": 5612
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.935603325677223,
      "learning_rate": 8.779715314309494e-06,
      "loss": 0.8284,
      "step": 5613
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.26202972525311,
      "learning_rate": 8.77924238363636e-06,
      "loss": 0.7723,
      "step": 5614
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.309065282758038,
      "learning_rate": 8.778769374079816e-06,
      "loss": 0.8525,
      "step": 5615
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.6085670289149085,
      "learning_rate": 8.778296285649738e-06,
      "loss": 0.8518,
      "step": 5616
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.3081597782879255,
      "learning_rate": 8.777823118355999e-06,
      "loss": 0.8348,
      "step": 5617
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.5286207106142315,
      "learning_rate": 8.777349872208475e-06,
      "loss": 0.851,
      "step": 5618
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.71870810454012,
      "learning_rate": 8.776876547217043e-06,
      "loss": 0.799,
      "step": 5619
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.162888191748796,
      "learning_rate": 8.776403143391588e-06,
      "loss": 0.8324,
      "step": 5620
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.060984465007856,
      "learning_rate": 8.775929660741985e-06,
      "loss": 0.842,
      "step": 5621
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.9893998801331865,
      "learning_rate": 8.77545609927812e-06,
      "loss": 0.831,
      "step": 5622
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.33976056431188,
      "learning_rate": 8.774982459009877e-06,
      "loss": 0.836,
      "step": 5623
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.813902217177362,
      "learning_rate": 8.774508739947143e-06,
      "loss": 0.8505,
      "step": 5624
    },
    {
      "epoch": 0.5,
      "grad_norm": 9.865401065238201,
      "learning_rate": 8.774034942099804e-06,
      "loss": 0.8582,
      "step": 5625
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.262516745457644,
      "learning_rate": 8.773561065477751e-06,
      "loss": 0.7963,
      "step": 5626
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.642995469597107,
      "learning_rate": 8.773087110090875e-06,
      "loss": 0.821,
      "step": 5627
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.328377505500768,
      "learning_rate": 8.772613075949067e-06,
      "loss": 0.7877,
      "step": 5628
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.927730861143127,
      "learning_rate": 8.772138963062227e-06,
      "loss": 0.8596,
      "step": 5629
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.439247217507778,
      "learning_rate": 8.771664771440243e-06,
      "loss": 0.8435,
      "step": 5630
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.931586926104177,
      "learning_rate": 8.771190501093019e-06,
      "loss": 0.8365,
      "step": 5631
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.787152949572916,
      "learning_rate": 8.77071615203045e-06,
      "loss": 0.8594,
      "step": 5632
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.195628472568788,
      "learning_rate": 8.77024172426244e-06,
      "loss": 0.8154,
      "step": 5633
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.254848165382624,
      "learning_rate": 8.76976721779889e-06,
      "loss": 0.7498,
      "step": 5634
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.449395740567542,
      "learning_rate": 8.769292632649707e-06,
      "loss": 0.8592,
      "step": 5635
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.689443388093119,
      "learning_rate": 8.768817968824794e-06,
      "loss": 0.7696,
      "step": 5636
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.680673409801719,
      "learning_rate": 8.768343226334059e-06,
      "loss": 0.8095,
      "step": 5637
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.262625844537167,
      "learning_rate": 8.767868405187411e-06,
      "loss": 0.8097,
      "step": 5638
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.216956288986688,
      "learning_rate": 8.767393505394764e-06,
      "loss": 0.7933,
      "step": 5639
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.628314001969875,
      "learning_rate": 8.766918526966025e-06,
      "loss": 0.8447,
      "step": 5640
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.75616978790288,
      "learning_rate": 8.766443469911111e-06,
      "loss": 0.785,
      "step": 5641
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.701206197121069,
      "learning_rate": 8.76596833423994e-06,
      "loss": 0.8606,
      "step": 5642
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.56729942775858,
      "learning_rate": 8.765493119962429e-06,
      "loss": 0.8201,
      "step": 5643
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.106175028870563,
      "learning_rate": 8.765017827088492e-06,
      "loss": 0.8194,
      "step": 5644
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.21983949232419,
      "learning_rate": 8.764542455628054e-06,
      "loss": 0.882,
      "step": 5645
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.400398669053492,
      "learning_rate": 8.764067005591037e-06,
      "loss": 0.7719,
      "step": 5646
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.254060418030271,
      "learning_rate": 8.763591476987364e-06,
      "loss": 0.8742,
      "step": 5647
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.867158231715006,
      "learning_rate": 8.763115869826962e-06,
      "loss": 0.8092,
      "step": 5648
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.46070155117413,
      "learning_rate": 8.762640184119756e-06,
      "loss": 0.8162,
      "step": 5649
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.032526274530505,
      "learning_rate": 8.762164419875676e-06,
      "loss": 0.8577,
      "step": 5650
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.923725302659771,
      "learning_rate": 8.761688577104655e-06,
      "loss": 0.8229,
      "step": 5651
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.941836289431967,
      "learning_rate": 8.76121265581662e-06,
      "loss": 0.8211,
      "step": 5652
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.0512104105025575,
      "learning_rate": 8.76073665602151e-06,
      "loss": 0.8043,
      "step": 5653
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.242077460332891,
      "learning_rate": 8.760260577729258e-06,
      "loss": 0.8463,
      "step": 5654
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.284501072106909,
      "learning_rate": 8.7597844209498e-06,
      "loss": 0.7865,
      "step": 5655
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.222491679829832,
      "learning_rate": 8.759308185693076e-06,
      "loss": 0.7859,
      "step": 5656
    },
    {
      "epoch": 0.5,
      "grad_norm": 9.042084352625805,
      "learning_rate": 8.758831871969026e-06,
      "loss": 0.8438,
      "step": 5657
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.869249129898664,
      "learning_rate": 8.758355479787594e-06,
      "loss": 0.8273,
      "step": 5658
    },
    {
      "epoch": 0.5,
      "grad_norm": 8.533630442065972,
      "learning_rate": 8.757879009158722e-06,
      "loss": 0.8466,
      "step": 5659
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.188883191239206,
      "learning_rate": 8.757402460092355e-06,
      "loss": 0.8547,
      "step": 5660
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.897186848015153,
      "learning_rate": 8.75692583259844e-06,
      "loss": 0.8594,
      "step": 5661
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.406586845676556,
      "learning_rate": 8.756449126686925e-06,
      "loss": 0.8323,
      "step": 5662
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.329968324531108,
      "learning_rate": 8.755972342367762e-06,
      "loss": 0.8307,
      "step": 5663
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.979975380275228,
      "learning_rate": 8.755495479650901e-06,
      "loss": 0.8628,
      "step": 5664
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.29714989632913,
      "learning_rate": 8.755018538546297e-06,
      "loss": 0.7624,
      "step": 5665
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.694647620445541,
      "learning_rate": 8.754541519063904e-06,
      "loss": 0.7829,
      "step": 5666
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.869824930763273,
      "learning_rate": 8.754064421213679e-06,
      "loss": 0.8452,
      "step": 5667
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.264250254116545,
      "learning_rate": 8.75358724500558e-06,
      "loss": 0.8158,
      "step": 5668
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.05238796507581,
      "learning_rate": 8.753109990449568e-06,
      "loss": 0.8297,
      "step": 5669
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.597453912168418,
      "learning_rate": 8.752632657555605e-06,
      "loss": 0.7604,
      "step": 5670
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.743382954307651,
      "learning_rate": 8.752155246333652e-06,
      "loss": 0.7958,
      "step": 5671
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.091310088065734,
      "learning_rate": 8.751677756793676e-06,
      "loss": 0.7707,
      "step": 5672
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.810017589462644,
      "learning_rate": 8.751200188945642e-06,
      "loss": 0.7633,
      "step": 5673
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.9523604277509214,
      "learning_rate": 8.750722542799518e-06,
      "loss": 0.8334,
      "step": 5674
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.411044113048015,
      "learning_rate": 8.750244818365278e-06,
      "loss": 0.8564,
      "step": 5675
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.731058648822523,
      "learning_rate": 8.74976701565289e-06,
      "loss": 0.7977,
      "step": 5676
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.91905231375966,
      "learning_rate": 8.749289134672326e-06,
      "loss": 0.8449,
      "step": 5677
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.0551932984726955,
      "learning_rate": 8.748811175433563e-06,
      "loss": 0.7952,
      "step": 5678
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.356728392784034,
      "learning_rate": 8.748333137946574e-06,
      "loss": 0.794,
      "step": 5679
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.3842715881446965,
      "learning_rate": 8.747855022221342e-06,
      "loss": 0.7705,
      "step": 5680
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.389805113520474,
      "learning_rate": 8.747376828267843e-06,
      "loss": 0.8228,
      "step": 5681
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.661018478830402,
      "learning_rate": 8.746898556096061e-06,
      "loss": 0.8355,
      "step": 5682
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.040941038249002,
      "learning_rate": 8.746420205715976e-06,
      "loss": 0.7814,
      "step": 5683
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.217049567726062,
      "learning_rate": 8.745941777137573e-06,
      "loss": 0.7889,
      "step": 5684
    },
    {
      "epoch": 0.51,
      "grad_norm": 8.026812855837257,
      "learning_rate": 8.74546327037084e-06,
      "loss": 0.7786,
      "step": 5685
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.2085834249706355,
      "learning_rate": 8.744984685425764e-06,
      "loss": 0.7807,
      "step": 5686
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.8884887152527625,
      "learning_rate": 8.744506022312332e-06,
      "loss": 0.8734,
      "step": 5687
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.169234871145174,
      "learning_rate": 8.74402728104054e-06,
      "loss": 0.8219,
      "step": 5688
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.858647940584735,
      "learning_rate": 8.743548461620374e-06,
      "loss": 0.8056,
      "step": 5689
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.759621969499233,
      "learning_rate": 8.743069564061835e-06,
      "loss": 0.7411,
      "step": 5690
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.208776677097714,
      "learning_rate": 8.742590588374916e-06,
      "loss": 0.7872,
      "step": 5691
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.5817215076216,
      "learning_rate": 8.742111534569612e-06,
      "loss": 0.8095,
      "step": 5692
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.875875208647669,
      "learning_rate": 8.741632402655928e-06,
      "loss": 0.8706,
      "step": 5693
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.87200717368755,
      "learning_rate": 8.741153192643861e-06,
      "loss": 0.7908,
      "step": 5694
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.931799506060896,
      "learning_rate": 8.740673904543413e-06,
      "loss": 0.8056,
      "step": 5695
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.512019445236444,
      "learning_rate": 8.740194538364588e-06,
      "loss": 0.8469,
      "step": 5696
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.645883466274443,
      "learning_rate": 8.739715094117396e-06,
      "loss": 0.9058,
      "step": 5697
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.447206364154061,
      "learning_rate": 8.73923557181184e-06,
      "loss": 0.7739,
      "step": 5698
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.754737226520066,
      "learning_rate": 8.73875597145793e-06,
      "loss": 0.7503,
      "step": 5699
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.410728986892296,
      "learning_rate": 8.738276293065677e-06,
      "loss": 0.8288,
      "step": 5700
    },
    {
      "epoch": 0.51,
      "grad_norm": 8.348749802497387,
      "learning_rate": 8.737796536645093e-06,
      "loss": 0.9034,
      "step": 5701
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.35455158614091,
      "learning_rate": 8.73731670220619e-06,
      "loss": 0.8243,
      "step": 5702
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.12106015111757,
      "learning_rate": 8.73683678975899e-06,
      "loss": 0.8184,
      "step": 5703
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.660868934674814,
      "learning_rate": 8.736356799313501e-06,
      "loss": 0.7982,
      "step": 5704
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.0927921577529975,
      "learning_rate": 8.73587673087975e-06,
      "loss": 0.8281,
      "step": 5705
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.359008475168115,
      "learning_rate": 8.73539658446775e-06,
      "loss": 0.8467,
      "step": 5706
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.159921838378046,
      "learning_rate": 8.734916360087528e-06,
      "loss": 0.8366,
      "step": 5707
    },
    {
      "epoch": 0.51,
      "grad_norm": 8.008967987703356,
      "learning_rate": 8.734436057749108e-06,
      "loss": 0.8166,
      "step": 5708
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.071136274209243,
      "learning_rate": 8.733955677462511e-06,
      "loss": 0.8124,
      "step": 5709
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.044791193065917,
      "learning_rate": 8.733475219237768e-06,
      "loss": 0.8317,
      "step": 5710
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.905303080640748,
      "learning_rate": 8.732994683084907e-06,
      "loss": 0.7873,
      "step": 5711
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.852016146099419,
      "learning_rate": 8.732514069013956e-06,
      "loss": 0.8052,
      "step": 5712
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.666027579510247,
      "learning_rate": 8.732033377034947e-06,
      "loss": 0.8402,
      "step": 5713
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.786548116080704,
      "learning_rate": 8.731552607157914e-06,
      "loss": 0.7912,
      "step": 5714
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.38621025409749,
      "learning_rate": 8.731071759392895e-06,
      "loss": 0.8353,
      "step": 5715
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.967842235591543,
      "learning_rate": 8.73059083374992e-06,
      "loss": 0.7529,
      "step": 5716
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.2481751101313465,
      "learning_rate": 8.730109830239036e-06,
      "loss": 0.7862,
      "step": 5717
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.469022907867675,
      "learning_rate": 8.729628748870274e-06,
      "loss": 0.8234,
      "step": 5718
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.235253129413111,
      "learning_rate": 8.729147589653681e-06,
      "loss": 0.8412,
      "step": 5719
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.544024395441532,
      "learning_rate": 8.7286663525993e-06,
      "loss": 0.7904,
      "step": 5720
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.133812139757403,
      "learning_rate": 8.728185037717172e-06,
      "loss": 0.7928,
      "step": 5721
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.643997472334568,
      "learning_rate": 8.727703645017349e-06,
      "loss": 0.804,
      "step": 5722
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.640185678733757,
      "learning_rate": 8.727222174509873e-06,
      "loss": 0.8106,
      "step": 5723
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.117011654036177,
      "learning_rate": 8.726740626204799e-06,
      "loss": 0.8205,
      "step": 5724
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.128560760410511,
      "learning_rate": 8.726259000112174e-06,
      "loss": 0.8723,
      "step": 5725
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.71083419413605,
      "learning_rate": 8.725777296242054e-06,
      "loss": 0.8741,
      "step": 5726
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.821441086629458,
      "learning_rate": 8.725295514604492e-06,
      "loss": 0.7476,
      "step": 5727
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.374982813778905,
      "learning_rate": 8.724813655209543e-06,
      "loss": 0.7948,
      "step": 5728
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.830079196366561,
      "learning_rate": 8.724331718067269e-06,
      "loss": 0.8634,
      "step": 5729
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.131750921932709,
      "learning_rate": 8.723849703187723e-06,
      "loss": 0.798,
      "step": 5730
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.752241371375161,
      "learning_rate": 8.723367610580973e-06,
      "loss": 0.8067,
      "step": 5731
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.663078251562671,
      "learning_rate": 8.722885440257075e-06,
      "loss": 0.7794,
      "step": 5732
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.053920294127615,
      "learning_rate": 8.722403192226099e-06,
      "loss": 0.8893,
      "step": 5733
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.805347780119679,
      "learning_rate": 8.721920866498107e-06,
      "loss": 0.8055,
      "step": 5734
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.675019286880624,
      "learning_rate": 8.721438463083168e-06,
      "loss": 0.8166,
      "step": 5735
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.006866503667122,
      "learning_rate": 8.72095598199135e-06,
      "loss": 0.8193,
      "step": 5736
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.632571902047563,
      "learning_rate": 8.720473423232726e-06,
      "loss": 0.7751,
      "step": 5737
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.905215083505556,
      "learning_rate": 8.719990786817367e-06,
      "loss": 0.8008,
      "step": 5738
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.5411357539883195,
      "learning_rate": 8.719508072755345e-06,
      "loss": 0.8538,
      "step": 5739
    },
    {
      "epoch": 0.51,
      "grad_norm": 9.278517256201297,
      "learning_rate": 8.71902528105674e-06,
      "loss": 0.9175,
      "step": 5740
    },
    {
      "epoch": 0.51,
      "grad_norm": 8.535211358742204,
      "learning_rate": 8.718542411731626e-06,
      "loss": 0.8957,
      "step": 5741
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.623199942488983,
      "learning_rate": 8.718059464790082e-06,
      "loss": 0.8396,
      "step": 5742
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.059933438873013,
      "learning_rate": 8.71757644024219e-06,
      "loss": 0.8001,
      "step": 5743
    },
    {
      "epoch": 0.51,
      "grad_norm": 9.957385591870104,
      "learning_rate": 8.71709333809803e-06,
      "loss": 0.8119,
      "step": 5744
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.7677950714420145,
      "learning_rate": 8.716610158367688e-06,
      "loss": 0.8875,
      "step": 5745
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.446171842213826,
      "learning_rate": 8.716126901061248e-06,
      "loss": 0.8732,
      "step": 5746
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.381789993071132,
      "learning_rate": 8.715643566188797e-06,
      "loss": 0.8128,
      "step": 5747
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.464851828255418,
      "learning_rate": 8.715160153760422e-06,
      "loss": 0.876,
      "step": 5748
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.647085603101719,
      "learning_rate": 8.714676663786216e-06,
      "loss": 0.7801,
      "step": 5749
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.185902678256492,
      "learning_rate": 8.714193096276272e-06,
      "loss": 0.812,
      "step": 5750
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.398991162276467,
      "learning_rate": 8.713709451240679e-06,
      "loss": 0.957,
      "step": 5751
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.572346398902644,
      "learning_rate": 8.713225728689534e-06,
      "loss": 0.7716,
      "step": 5752
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.563499805892573,
      "learning_rate": 8.712741928632933e-06,
      "loss": 1.0096,
      "step": 5753
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.422860872744204,
      "learning_rate": 8.712258051080975e-06,
      "loss": 0.8319,
      "step": 5754
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.13073235937632,
      "learning_rate": 8.711774096043762e-06,
      "loss": 0.8576,
      "step": 5755
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.315718873527776,
      "learning_rate": 8.711290063531393e-06,
      "loss": 0.7578,
      "step": 5756
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.649425923569732,
      "learning_rate": 8.710805953553971e-06,
      "loss": 0.8205,
      "step": 5757
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.828016439689453,
      "learning_rate": 8.7103217661216e-06,
      "loss": 0.7856,
      "step": 5758
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.52548865112177,
      "learning_rate": 8.70983750124439e-06,
      "loss": 0.8168,
      "step": 5759
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.154187500715567,
      "learning_rate": 8.709353158932442e-06,
      "loss": 0.8278,
      "step": 5760
    },
    {
      "epoch": 0.51,
      "grad_norm": 22.781189933084534,
      "learning_rate": 8.708868739195875e-06,
      "loss": 0.8343,
      "step": 5761
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.773865457858816,
      "learning_rate": 8.708384242044793e-06,
      "loss": 0.7413,
      "step": 5762
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.100039627454103,
      "learning_rate": 8.707899667489312e-06,
      "loss": 0.8376,
      "step": 5763
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.844058003056372,
      "learning_rate": 8.707415015539546e-06,
      "loss": 0.8395,
      "step": 5764
    },
    {
      "epoch": 0.51,
      "grad_norm": 8.261438581169475,
      "learning_rate": 8.706930286205611e-06,
      "loss": 0.8271,
      "step": 5765
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.4421742541436116,
      "learning_rate": 8.706445479497623e-06,
      "loss": 0.8262,
      "step": 5766
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.987426583221405,
      "learning_rate": 8.705960595425701e-06,
      "loss": 0.8273,
      "step": 5767
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.384476868382562,
      "learning_rate": 8.705475633999971e-06,
      "loss": 0.7961,
      "step": 5768
    },
    {
      "epoch": 0.51,
      "grad_norm": 8.109813634613358,
      "learning_rate": 8.70499059523055e-06,
      "loss": 0.8307,
      "step": 5769
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.381037406673442,
      "learning_rate": 8.704505479127564e-06,
      "loss": 0.8198,
      "step": 5770
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.956425373451172,
      "learning_rate": 8.70402028570114e-06,
      "loss": 0.8263,
      "step": 5771
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.094193535875792,
      "learning_rate": 8.703535014961404e-06,
      "loss": 0.7822,
      "step": 5772
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.0756566071039355,
      "learning_rate": 8.703049666918484e-06,
      "loss": 0.8449,
      "step": 5773
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.862271692051127,
      "learning_rate": 8.702564241582513e-06,
      "loss": 0.7016,
      "step": 5774
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.775134299311187,
      "learning_rate": 8.702078738963623e-06,
      "loss": 0.8105,
      "step": 5775
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.187425287393864,
      "learning_rate": 8.701593159071945e-06,
      "loss": 0.8674,
      "step": 5776
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.8075496066449315,
      "learning_rate": 8.701107501917616e-06,
      "loss": 0.828,
      "step": 5777
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.959180396726149,
      "learning_rate": 8.700621767510774e-06,
      "loss": 0.7748,
      "step": 5778
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.709156774966415,
      "learning_rate": 8.700135955861555e-06,
      "loss": 0.7852,
      "step": 5779
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.993845937908161,
      "learning_rate": 8.699650066980102e-06,
      "loss": 0.834,
      "step": 5780
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.0874468829899735,
      "learning_rate": 8.699164100876557e-06,
      "loss": 0.8023,
      "step": 5781
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.42674979256416,
      "learning_rate": 8.69867805756106e-06,
      "loss": 0.898,
      "step": 5782
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.401841924411248,
      "learning_rate": 8.69819193704376e-06,
      "loss": 0.824,
      "step": 5783
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.087579637160131,
      "learning_rate": 8.697705739334805e-06,
      "loss": 0.8709,
      "step": 5784
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.523768588211032,
      "learning_rate": 8.697219464444337e-06,
      "loss": 0.8042,
      "step": 5785
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.8360967103691,
      "learning_rate": 8.69673311238251e-06,
      "loss": 0.8115,
      "step": 5786
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.856844644695396,
      "learning_rate": 8.696246683159476e-06,
      "loss": 0.8215,
      "step": 5787
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.788478296977279,
      "learning_rate": 8.695760176785386e-06,
      "loss": 0.8629,
      "step": 5788
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.216487547235108,
      "learning_rate": 8.695273593270396e-06,
      "loss": 0.7657,
      "step": 5789
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.181336419466348,
      "learning_rate": 8.694786932624663e-06,
      "loss": 0.7817,
      "step": 5790
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.740612031477996,
      "learning_rate": 8.694300194858345e-06,
      "loss": 0.7367,
      "step": 5791
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.9782173761304325,
      "learning_rate": 8.6938133799816e-06,
      "loss": 0.7672,
      "step": 5792
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.025564952346374,
      "learning_rate": 8.69332648800459e-06,
      "loss": 0.8586,
      "step": 5793
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.195319012424117,
      "learning_rate": 8.692839518937477e-06,
      "loss": 0.8284,
      "step": 5794
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.18304070463878,
      "learning_rate": 8.692352472790427e-06,
      "loss": 0.8076,
      "step": 5795
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.6113535148407685,
      "learning_rate": 8.691865349573606e-06,
      "loss": 0.7873,
      "step": 5796
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.929410780940768,
      "learning_rate": 8.69137814929718e-06,
      "loss": 0.7224,
      "step": 5797
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.58702630737882,
      "learning_rate": 8.690890871971318e-06,
      "loss": 0.8173,
      "step": 5798
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.128813248183907,
      "learning_rate": 8.690403517606193e-06,
      "loss": 0.7042,
      "step": 5799
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.399927092925807,
      "learning_rate": 8.689916086211977e-06,
      "loss": 0.8164,
      "step": 5800
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.1710019912427825,
      "learning_rate": 8.689428577798843e-06,
      "loss": 0.7981,
      "step": 5801
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.604813997565266,
      "learning_rate": 8.688940992376966e-06,
      "loss": 0.8086,
      "step": 5802
    },
    {
      "epoch": 0.52,
      "grad_norm": 9.760176906734495,
      "learning_rate": 8.688453329956526e-06,
      "loss": 0.9153,
      "step": 5803
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.854059398888365,
      "learning_rate": 8.6879655905477e-06,
      "loss": 0.8553,
      "step": 5804
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.416887246472852,
      "learning_rate": 8.687477774160668e-06,
      "loss": 0.8488,
      "step": 5805
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.1784643533734265,
      "learning_rate": 8.686989880805614e-06,
      "loss": 0.753,
      "step": 5806
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.344982546634153,
      "learning_rate": 8.68650191049272e-06,
      "loss": 0.7595,
      "step": 5807
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.5652850099482025,
      "learning_rate": 8.686013863232173e-06,
      "loss": 0.8452,
      "step": 5808
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.751502329410952,
      "learning_rate": 8.685525739034158e-06,
      "loss": 0.847,
      "step": 5809
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.834847388684795,
      "learning_rate": 8.685037537908865e-06,
      "loss": 0.844,
      "step": 5810
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.7146583685829375,
      "learning_rate": 8.684549259866481e-06,
      "loss": 0.8672,
      "step": 5811
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.034218441622979,
      "learning_rate": 8.684060904917201e-06,
      "loss": 0.8741,
      "step": 5812
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.032366003820098,
      "learning_rate": 8.68357247307122e-06,
      "loss": 0.8237,
      "step": 5813
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.088184677539254,
      "learning_rate": 8.683083964338728e-06,
      "loss": 0.9411,
      "step": 5814
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.937311078827777,
      "learning_rate": 8.682595378729925e-06,
      "loss": 0.7493,
      "step": 5815
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.759193239561655,
      "learning_rate": 8.68210671625501e-06,
      "loss": 0.8668,
      "step": 5816
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.970133383131887,
      "learning_rate": 8.681617976924179e-06,
      "loss": 0.8696,
      "step": 5817
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.1452477626782915,
      "learning_rate": 8.681129160747636e-06,
      "loss": 0.8789,
      "step": 5818
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.337485276630007,
      "learning_rate": 8.680640267735582e-06,
      "loss": 0.8619,
      "step": 5819
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.620873318200385,
      "learning_rate": 8.680151297898224e-06,
      "loss": 0.7877,
      "step": 5820
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.452621604022749,
      "learning_rate": 8.679662251245768e-06,
      "loss": 0.8181,
      "step": 5821
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.446134143378865,
      "learning_rate": 8.679173127788418e-06,
      "loss": 0.8256,
      "step": 5822
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.390551931565307,
      "learning_rate": 8.678683927536389e-06,
      "loss": 0.8467,
      "step": 5823
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.001840565952541,
      "learning_rate": 8.678194650499888e-06,
      "loss": 0.7517,
      "step": 5824
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.725662748993245,
      "learning_rate": 8.67770529668913e-06,
      "loss": 0.8347,
      "step": 5825
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.937090576892703,
      "learning_rate": 8.677215866114326e-06,
      "loss": 0.7682,
      "step": 5826
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.664380607957075,
      "learning_rate": 8.676726358785695e-06,
      "loss": 0.8393,
      "step": 5827
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.21904737552294,
      "learning_rate": 8.67623677471345e-06,
      "loss": 0.8114,
      "step": 5828
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.1324640590837785,
      "learning_rate": 8.675747113907818e-06,
      "loss": 0.8728,
      "step": 5829
    },
    {
      "epoch": 0.52,
      "grad_norm": 8.554614584810185,
      "learning_rate": 8.675257376379013e-06,
      "loss": 0.8018,
      "step": 5830
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.421788661747229,
      "learning_rate": 8.674767562137258e-06,
      "loss": 0.798,
      "step": 5831
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.452455108181789,
      "learning_rate": 8.67427767119278e-06,
      "loss": 0.7945,
      "step": 5832
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.776367611496056,
      "learning_rate": 8.673787703555797e-06,
      "loss": 0.8481,
      "step": 5833
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.154290080262535,
      "learning_rate": 8.673297659236546e-06,
      "loss": 0.8308,
      "step": 5834
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.881841263429405,
      "learning_rate": 8.67280753824525e-06,
      "loss": 0.7705,
      "step": 5835
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.284639909928041,
      "learning_rate": 8.67231734059214e-06,
      "loss": 0.8564,
      "step": 5836
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.907975559718964,
      "learning_rate": 8.671827066287446e-06,
      "loss": 0.7754,
      "step": 5837
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.116066419596677,
      "learning_rate": 8.671336715341406e-06,
      "loss": 0.8135,
      "step": 5838
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.556805765809384,
      "learning_rate": 8.670846287764249e-06,
      "loss": 0.7778,
      "step": 5839
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.4224575446521674,
      "learning_rate": 8.670355783566216e-06,
      "loss": 0.8272,
      "step": 5840
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.525716811210102,
      "learning_rate": 8.669865202757545e-06,
      "loss": 0.7752,
      "step": 5841
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.038632069834694,
      "learning_rate": 8.669374545348474e-06,
      "loss": 0.8002,
      "step": 5842
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.785049194308323,
      "learning_rate": 8.668883811349244e-06,
      "loss": 0.7265,
      "step": 5843
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.859634297291458,
      "learning_rate": 8.668393000770101e-06,
      "loss": 0.7786,
      "step": 5844
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.1976172305602315,
      "learning_rate": 8.667902113621287e-06,
      "loss": 0.8001,
      "step": 5845
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.6828375295949565,
      "learning_rate": 8.667411149913049e-06,
      "loss": 0.8118,
      "step": 5846
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.278795048680833,
      "learning_rate": 8.666920109655635e-06,
      "loss": 0.8459,
      "step": 5847
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.209184965448266,
      "learning_rate": 8.666428992859292e-06,
      "loss": 0.845,
      "step": 5848
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.243022643755357,
      "learning_rate": 8.665937799534276e-06,
      "loss": 0.8816,
      "step": 5849
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.189025727739834,
      "learning_rate": 8.665446529690835e-06,
      "loss": 0.8617,
      "step": 5850
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.725799718597585,
      "learning_rate": 8.664955183339225e-06,
      "loss": 0.7671,
      "step": 5851
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.26458728182319,
      "learning_rate": 8.664463760489702e-06,
      "loss": 0.8669,
      "step": 5852
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.009630389915014,
      "learning_rate": 8.663972261152525e-06,
      "loss": 0.8506,
      "step": 5853
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.590625896228143,
      "learning_rate": 8.663480685337949e-06,
      "loss": 0.807,
      "step": 5854
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.442163173205468,
      "learning_rate": 8.662989033056236e-06,
      "loss": 0.8555,
      "step": 5855
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.439306543111314,
      "learning_rate": 8.66249730431765e-06,
      "loss": 0.8224,
      "step": 5856
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.142572838905145,
      "learning_rate": 8.662005499132453e-06,
      "loss": 0.8121,
      "step": 5857
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.548074491232959,
      "learning_rate": 8.66151361751091e-06,
      "loss": 0.7792,
      "step": 5858
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.01415768658376,
      "learning_rate": 8.661021659463291e-06,
      "loss": 0.8657,
      "step": 5859
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.353882542965322,
      "learning_rate": 8.66052962499986e-06,
      "loss": 0.87,
      "step": 5860
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.7456530099919805,
      "learning_rate": 8.66003751413089e-06,
      "loss": 0.7978,
      "step": 5861
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.555662987495287,
      "learning_rate": 8.659545326866654e-06,
      "loss": 0.804,
      "step": 5862
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.78475662611469,
      "learning_rate": 8.659053063217423e-06,
      "loss": 0.8775,
      "step": 5863
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.701464067362781,
      "learning_rate": 8.65856072319347e-06,
      "loss": 0.836,
      "step": 5864
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.496276721066393,
      "learning_rate": 8.658068306805076e-06,
      "loss": 0.8648,
      "step": 5865
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.048691600795879,
      "learning_rate": 8.657575814062518e-06,
      "loss": 0.8248,
      "step": 5866
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.0588862343307825,
      "learning_rate": 8.657083244976074e-06,
      "loss": 0.8256,
      "step": 5867
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.902783416051479,
      "learning_rate": 8.656590599556024e-06,
      "loss": 0.9138,
      "step": 5868
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.461219340601897,
      "learning_rate": 8.656097877812658e-06,
      "loss": 0.8361,
      "step": 5869
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.461641741271912,
      "learning_rate": 8.65560507975625e-06,
      "loss": 0.8137,
      "step": 5870
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.62478958979562,
      "learning_rate": 8.655112205397096e-06,
      "loss": 0.7629,
      "step": 5871
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.830251388527715,
      "learning_rate": 8.654619254745476e-06,
      "loss": 0.8386,
      "step": 5872
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.173822926207818,
      "learning_rate": 8.654126227811685e-06,
      "loss": 0.8908,
      "step": 5873
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.192119037646925,
      "learning_rate": 8.65363312460601e-06,
      "loss": 0.8485,
      "step": 5874
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.286649318842872,
      "learning_rate": 8.653139945138746e-06,
      "loss": 0.9107,
      "step": 5875
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.802309909232628,
      "learning_rate": 8.652646689420184e-06,
      "loss": 0.8234,
      "step": 5876
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.9387858974413295,
      "learning_rate": 8.652153357460622e-06,
      "loss": 0.7789,
      "step": 5877
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.9985928919903655,
      "learning_rate": 8.65165994927036e-06,
      "loss": 0.7443,
      "step": 5878
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.511388774308987,
      "learning_rate": 8.651166464859689e-06,
      "loss": 0.835,
      "step": 5879
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.379462188182881,
      "learning_rate": 8.650672904238917e-06,
      "loss": 0.7797,
      "step": 5880
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.965018981372429,
      "learning_rate": 8.65017926741834e-06,
      "loss": 0.8612,
      "step": 5881
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.4198053829118615,
      "learning_rate": 8.649685554408267e-06,
      "loss": 0.7675,
      "step": 5882
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.255484195011597,
      "learning_rate": 8.649191765219001e-06,
      "loss": 0.7866,
      "step": 5883
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.457003922008676,
      "learning_rate": 8.648697899860848e-06,
      "loss": 0.7731,
      "step": 5884
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.716223652204424,
      "learning_rate": 8.648203958344115e-06,
      "loss": 0.7296,
      "step": 5885
    },
    {
      "epoch": 0.53,
      "grad_norm": 10.834241749051515,
      "learning_rate": 8.647709940679116e-06,
      "loss": 0.8541,
      "step": 5886
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.944107135815664,
      "learning_rate": 8.647215846876158e-06,
      "loss": 0.8406,
      "step": 5887
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.176779518193242,
      "learning_rate": 8.646721676945559e-06,
      "loss": 0.8582,
      "step": 5888
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.801428213908792,
      "learning_rate": 8.64622743089763e-06,
      "loss": 0.8694,
      "step": 5889
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.229578357317486,
      "learning_rate": 8.645733108742689e-06,
      "loss": 0.7349,
      "step": 5890
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.906458347544252,
      "learning_rate": 8.645238710491054e-06,
      "loss": 0.824,
      "step": 5891
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.581578044411557,
      "learning_rate": 8.644744236153042e-06,
      "loss": 0.7604,
      "step": 5892
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.676728896862765,
      "learning_rate": 8.644249685738977e-06,
      "loss": 0.8697,
      "step": 5893
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.945961269684335,
      "learning_rate": 8.643755059259181e-06,
      "loss": 0.8648,
      "step": 5894
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.7761591165840835,
      "learning_rate": 8.643260356723977e-06,
      "loss": 0.8491,
      "step": 5895
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.963085768233817,
      "learning_rate": 8.64276557814369e-06,
      "loss": 0.8552,
      "step": 5896
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.017270489091205,
      "learning_rate": 8.642270723528652e-06,
      "loss": 0.8453,
      "step": 5897
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.176265424313485,
      "learning_rate": 8.641775792889188e-06,
      "loss": 0.8161,
      "step": 5898
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.518456062944514,
      "learning_rate": 8.641280786235631e-06,
      "loss": 0.7776,
      "step": 5899
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.827923844505221,
      "learning_rate": 8.64078570357831e-06,
      "loss": 0.849,
      "step": 5900
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.405540099988824,
      "learning_rate": 8.640290544927561e-06,
      "loss": 0.8377,
      "step": 5901
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.482105532833051,
      "learning_rate": 8.63979531029372e-06,
      "loss": 0.8671,
      "step": 5902
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.180400037588014,
      "learning_rate": 8.63929999968712e-06,
      "loss": 0.8458,
      "step": 5903
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.708669658746596,
      "learning_rate": 8.638804613118106e-06,
      "loss": 0.8311,
      "step": 5904
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.643412347997996,
      "learning_rate": 8.638309150597013e-06,
      "loss": 0.9175,
      "step": 5905
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.642704343252684,
      "learning_rate": 8.637813612134183e-06,
      "loss": 0.8153,
      "step": 5906
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.690201966400359,
      "learning_rate": 8.637317997739962e-06,
      "loss": 0.8545,
      "step": 5907
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.197513503545189,
      "learning_rate": 8.636822307424694e-06,
      "loss": 0.8307,
      "step": 5908
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.624004293318079,
      "learning_rate": 8.636326541198724e-06,
      "loss": 0.7764,
      "step": 5909
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.421440928363879,
      "learning_rate": 8.6358306990724e-06,
      "loss": 0.8429,
      "step": 5910
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.732823558050309,
      "learning_rate": 8.635334781056074e-06,
      "loss": 0.861,
      "step": 5911
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.921574147408365,
      "learning_rate": 8.634838787160096e-06,
      "loss": 0.8272,
      "step": 5912
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.32225569071901,
      "learning_rate": 8.634342717394817e-06,
      "loss": 0.8917,
      "step": 5913
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.802714798385363,
      "learning_rate": 8.633846571770593e-06,
      "loss": 0.8183,
      "step": 5914
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.929354104310692,
      "learning_rate": 8.63335035029778e-06,
      "loss": 0.8429,
      "step": 5915
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.2772776954939635,
      "learning_rate": 8.632854052986736e-06,
      "loss": 0.8284,
      "step": 5916
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.662468199443719,
      "learning_rate": 8.63235767984782e-06,
      "loss": 0.8029,
      "step": 5917
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.890829798335322,
      "learning_rate": 8.631861230891391e-06,
      "loss": 0.7785,
      "step": 5918
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.520583680228424,
      "learning_rate": 8.631364706127812e-06,
      "loss": 0.808,
      "step": 5919
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.857123804299668,
      "learning_rate": 8.630868105567449e-06,
      "loss": 0.8411,
      "step": 5920
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.2795063651370135,
      "learning_rate": 8.630371429220665e-06,
      "loss": 0.7814,
      "step": 5921
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.40924820880883,
      "learning_rate": 8.629874677097827e-06,
      "loss": 0.873,
      "step": 5922
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.769645820266975,
      "learning_rate": 8.629377849209303e-06,
      "loss": 0.8272,
      "step": 5923
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.762479328512109,
      "learning_rate": 8.628880945565468e-06,
      "loss": 0.8113,
      "step": 5924
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.69112130342404,
      "learning_rate": 8.628383966176688e-06,
      "loss": 0.8505,
      "step": 5925
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.310612958416333,
      "learning_rate": 8.62788691105334e-06,
      "loss": 0.8108,
      "step": 5926
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.588636258431608,
      "learning_rate": 8.627389780205798e-06,
      "loss": 0.8097,
      "step": 5927
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.459016232186524,
      "learning_rate": 8.626892573644437e-06,
      "loss": 0.8005,
      "step": 5928
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.429122217944679,
      "learning_rate": 8.626395291379636e-06,
      "loss": 0.8408,
      "step": 5929
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.2643040375769194,
      "learning_rate": 8.625897933421777e-06,
      "loss": 0.7563,
      "step": 5930
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.513483902972747,
      "learning_rate": 8.625400499781239e-06,
      "loss": 0.7732,
      "step": 5931
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.115078516425911,
      "learning_rate": 8.624902990468404e-06,
      "loss": 0.7783,
      "step": 5932
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.311436106180336,
      "learning_rate": 8.624405405493656e-06,
      "loss": 0.8585,
      "step": 5933
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.331074276461183,
      "learning_rate": 8.623907744867384e-06,
      "loss": 0.8107,
      "step": 5934
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.799763763312336,
      "learning_rate": 8.623410008599975e-06,
      "loss": 0.8502,
      "step": 5935
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.900972926138916,
      "learning_rate": 8.622912196701816e-06,
      "loss": 0.8297,
      "step": 5936
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.029524088305606,
      "learning_rate": 8.6224143091833e-06,
      "loss": 0.871,
      "step": 5937
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.438269374370796,
      "learning_rate": 8.621916346054816e-06,
      "loss": 0.8998,
      "step": 5938
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.750549003815412,
      "learning_rate": 8.621418307326761e-06,
      "loss": 0.7803,
      "step": 5939
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.106996777166243,
      "learning_rate": 8.62092019300953e-06,
      "loss": 0.9078,
      "step": 5940
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.83939390845059,
      "learning_rate": 8.62042200311352e-06,
      "loss": 0.8817,
      "step": 5941
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.9092518171983945,
      "learning_rate": 8.61992373764913e-06,
      "loss": 0.7392,
      "step": 5942
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.315744593338969,
      "learning_rate": 8.619425396626758e-06,
      "loss": 0.8435,
      "step": 5943
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.721528522038859,
      "learning_rate": 8.618926980056808e-06,
      "loss": 0.8222,
      "step": 5944
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.718669231105705,
      "learning_rate": 8.618428487949682e-06,
      "loss": 0.8099,
      "step": 5945
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.554078076711697,
      "learning_rate": 8.617929920315787e-06,
      "loss": 0.708,
      "step": 5946
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.275317540329718,
      "learning_rate": 8.617431277165526e-06,
      "loss": 0.7724,
      "step": 5947
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.034451063293915,
      "learning_rate": 8.61693255850931e-06,
      "loss": 0.8317,
      "step": 5948
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.3283445802996185,
      "learning_rate": 8.616433764357547e-06,
      "loss": 0.8071,
      "step": 5949
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.41451294868157,
      "learning_rate": 8.615934894720653e-06,
      "loss": 0.8824,
      "step": 5950
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.7283094830664165,
      "learning_rate": 8.615435949609032e-06,
      "loss": 0.8068,
      "step": 5951
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.864314246203908,
      "learning_rate": 8.614936929033105e-06,
      "loss": 0.8084,
      "step": 5952
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.866311440323056,
      "learning_rate": 8.614437833003286e-06,
      "loss": 0.8776,
      "step": 5953
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.689370667413159,
      "learning_rate": 8.613938661529992e-06,
      "loss": 0.7869,
      "step": 5954
    },
    {
      "epoch": 0.53,
      "grad_norm": 8.425490894399791,
      "learning_rate": 8.613439414623645e-06,
      "loss": 0.9229,
      "step": 5955
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.093928133776955,
      "learning_rate": 8.612940092294661e-06,
      "loss": 0.7636,
      "step": 5956
    },
    {
      "epoch": 0.53,
      "grad_norm": 8.10916176993494,
      "learning_rate": 8.612440694553468e-06,
      "loss": 0.857,
      "step": 5957
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.705470878358799,
      "learning_rate": 8.611941221410484e-06,
      "loss": 0.7981,
      "step": 5958
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.96723794157607,
      "learning_rate": 8.611441672876137e-06,
      "loss": 0.7793,
      "step": 5959
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.940457766662916,
      "learning_rate": 8.610942048960855e-06,
      "loss": 0.7517,
      "step": 5960
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.694537577750331,
      "learning_rate": 8.610442349675065e-06,
      "loss": 0.8323,
      "step": 5961
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.5777513851493286,
      "learning_rate": 8.609942575029198e-06,
      "loss": 0.762,
      "step": 5962
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.003939279995617,
      "learning_rate": 8.609442725033685e-06,
      "loss": 0.8118,
      "step": 5963
    },
    {
      "epoch": 0.53,
      "grad_norm": 14.934523793615586,
      "learning_rate": 8.60894279969896e-06,
      "loss": 0.8522,
      "step": 5964
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.643278177559672,
      "learning_rate": 8.608442799035458e-06,
      "loss": 0.8485,
      "step": 5965
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.1281810442417655,
      "learning_rate": 8.607942723053615e-06,
      "loss": 0.8854,
      "step": 5966
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.792865586905612,
      "learning_rate": 8.607442571763868e-06,
      "loss": 0.7569,
      "step": 5967
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.025156682112292,
      "learning_rate": 8.606942345176658e-06,
      "loss": 0.8217,
      "step": 5968
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.161841221024935,
      "learning_rate": 8.606442043302426e-06,
      "loss": 0.7964,
      "step": 5969
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.331667915014399,
      "learning_rate": 8.605941666151613e-06,
      "loss": 0.8385,
      "step": 5970
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.673320235373131,
      "learning_rate": 8.605441213734664e-06,
      "loss": 0.8292,
      "step": 5971
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.9204880133349675,
      "learning_rate": 8.604940686062027e-06,
      "loss": 0.8063,
      "step": 5972
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.079683749632774,
      "learning_rate": 8.604440083144147e-06,
      "loss": 0.8523,
      "step": 5973
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.414706538958956,
      "learning_rate": 8.603939404991474e-06,
      "loss": 0.9264,
      "step": 5974
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.315688749420987,
      "learning_rate": 8.60343865161446e-06,
      "loss": 0.8248,
      "step": 5975
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.040535569565349,
      "learning_rate": 8.602937823023554e-06,
      "loss": 0.8514,
      "step": 5976
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.275477441647361,
      "learning_rate": 8.602436919229214e-06,
      "loss": 0.8042,
      "step": 5977
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.838148420512196,
      "learning_rate": 8.601935940241888e-06,
      "loss": 0.8622,
      "step": 5978
    },
    {
      "epoch": 0.53,
      "grad_norm": 8.127351709405445,
      "learning_rate": 8.60143488607204e-06,
      "loss": 0.8294,
      "step": 5979
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.229731831395928,
      "learning_rate": 8.600933756730126e-06,
      "loss": 0.8372,
      "step": 5980
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.687414413745216,
      "learning_rate": 8.600432552226604e-06,
      "loss": 0.7938,
      "step": 5981
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.1706882164397365,
      "learning_rate": 8.59993127257194e-06,
      "loss": 0.8087,
      "step": 5982
    },
    {
      "epoch": 0.53,
      "grad_norm": 8.378666956741002,
      "learning_rate": 8.599429917776592e-06,
      "loss": 0.9085,
      "step": 5983
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.427278709788023,
      "learning_rate": 8.598928487851026e-06,
      "loss": 0.8498,
      "step": 5984
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.083798654508727,
      "learning_rate": 8.598426982805712e-06,
      "loss": 0.8324,
      "step": 5985
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.520173582664215,
      "learning_rate": 8.597925402651114e-06,
      "loss": 0.8136,
      "step": 5986
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.295222407333342,
      "learning_rate": 8.597423747397705e-06,
      "loss": 0.8119,
      "step": 5987
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.023505716786854,
      "learning_rate": 8.59692201705595e-06,
      "loss": 0.8329,
      "step": 5988
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.784247821358579,
      "learning_rate": 8.596420211636328e-06,
      "loss": 0.7894,
      "step": 5989
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.596441308285133,
      "learning_rate": 8.595918331149307e-06,
      "loss": 0.8635,
      "step": 5990
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.108078004362753,
      "learning_rate": 8.595416375605367e-06,
      "loss": 0.799,
      "step": 5991
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.429759819793662,
      "learning_rate": 8.594914345014984e-06,
      "loss": 0.8011,
      "step": 5992
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.4086308965482015,
      "learning_rate": 8.594412239388638e-06,
      "loss": 0.8214,
      "step": 5993
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.466529788010683,
      "learning_rate": 8.593910058736807e-06,
      "loss": 0.8241,
      "step": 5994
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.799862594530469,
      "learning_rate": 8.593407803069973e-06,
      "loss": 0.8075,
      "step": 5995
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.414700947156203,
      "learning_rate": 8.592905472398622e-06,
      "loss": 0.8647,
      "step": 5996
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.617586864207485,
      "learning_rate": 8.592403066733237e-06,
      "loss": 0.8335,
      "step": 5997
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.370469646604213,
      "learning_rate": 8.591900586084305e-06,
      "loss": 0.806,
      "step": 5998
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.142305241946222,
      "learning_rate": 8.591398030462315e-06,
      "loss": 0.8813,
      "step": 5999
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.988350473328172,
      "learning_rate": 8.590895399877757e-06,
      "loss": 0.7867,
      "step": 6000
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.294499847417179,
      "learning_rate": 8.590392694341121e-06,
      "loss": 0.7925,
      "step": 6001
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.517592930473238,
      "learning_rate": 8.589889913862899e-06,
      "loss": 0.8611,
      "step": 6002
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.750257312459497,
      "learning_rate": 8.589387058453588e-06,
      "loss": 0.8125,
      "step": 6003
    },
    {
      "epoch": 0.54,
      "grad_norm": 8.33561830592563,
      "learning_rate": 8.588884128123682e-06,
      "loss": 0.8147,
      "step": 6004
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.21561781348251,
      "learning_rate": 8.58838112288368e-06,
      "loss": 0.8001,
      "step": 6005
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.318249581673096,
      "learning_rate": 8.58787804274408e-06,
      "loss": 0.813,
      "step": 6006
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.123830957447272,
      "learning_rate": 8.587374887715385e-06,
      "loss": 0.7966,
      "step": 6007
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.957086316518475,
      "learning_rate": 8.586871657808092e-06,
      "loss": 0.7476,
      "step": 6008
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.728084802369861,
      "learning_rate": 8.58636835303271e-06,
      "loss": 0.9083,
      "step": 6009
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.137105485595029,
      "learning_rate": 8.585864973399743e-06,
      "loss": 0.8642,
      "step": 6010
    },
    {
      "epoch": 0.54,
      "grad_norm": 12.71945085109548,
      "learning_rate": 8.585361518919698e-06,
      "loss": 0.8418,
      "step": 6011
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.993426007414448,
      "learning_rate": 8.584857989603082e-06,
      "loss": 0.8366,
      "step": 6012
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.948656286437939,
      "learning_rate": 8.584354385460408e-06,
      "loss": 0.8575,
      "step": 6013
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.6570305329388075,
      "learning_rate": 8.583850706502184e-06,
      "loss": 0.7891,
      "step": 6014
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.882073310637497,
      "learning_rate": 8.583346952738924e-06,
      "loss": 0.7639,
      "step": 6015
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.329641906290548,
      "learning_rate": 8.582843124181147e-06,
      "loss": 0.8096,
      "step": 6016
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.684171315254989,
      "learning_rate": 8.582339220839363e-06,
      "loss": 0.8158,
      "step": 6017
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.158475102298465,
      "learning_rate": 8.581835242724095e-06,
      "loss": 0.7874,
      "step": 6018
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.610236124576656,
      "learning_rate": 8.581331189845859e-06,
      "loss": 0.8435,
      "step": 6019
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.536418122890589,
      "learning_rate": 8.580827062215178e-06,
      "loss": 0.8213,
      "step": 6020
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.9736651117460235,
      "learning_rate": 8.580322859842574e-06,
      "loss": 0.8352,
      "step": 6021
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.594432355310355,
      "learning_rate": 8.57981858273857e-06,
      "loss": 0.8288,
      "step": 6022
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.781530037782979,
      "learning_rate": 8.579314230913693e-06,
      "loss": 0.7503,
      "step": 6023
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.577182540948343,
      "learning_rate": 8.578809804378472e-06,
      "loss": 0.7775,
      "step": 6024
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.235089924482543,
      "learning_rate": 8.578305303143433e-06,
      "loss": 0.8512,
      "step": 6025
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.913389997235769,
      "learning_rate": 8.577800727219106e-06,
      "loss": 0.8416,
      "step": 6026
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.215529289311893,
      "learning_rate": 8.577296076616025e-06,
      "loss": 0.8449,
      "step": 6027
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.343390857706067,
      "learning_rate": 8.576791351344722e-06,
      "loss": 0.7862,
      "step": 6028
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.7929412914168115,
      "learning_rate": 8.576286551415733e-06,
      "loss": 0.7394,
      "step": 6029
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.158746094612381,
      "learning_rate": 8.575781676839594e-06,
      "loss": 0.7281,
      "step": 6030
    },
    {
      "epoch": 0.54,
      "grad_norm": 8.797987756338955,
      "learning_rate": 8.575276727626843e-06,
      "loss": 0.8129,
      "step": 6031
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.33901101128853,
      "learning_rate": 8.574771703788021e-06,
      "loss": 0.8704,
      "step": 6032
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.311812695714598,
      "learning_rate": 8.574266605333668e-06,
      "loss": 0.8472,
      "step": 6033
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.108686108660076,
      "learning_rate": 8.573761432274326e-06,
      "loss": 0.8005,
      "step": 6034
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.869878440707376,
      "learning_rate": 8.573256184620542e-06,
      "loss": 0.8498,
      "step": 6035
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.172123977414717,
      "learning_rate": 8.57275086238286e-06,
      "loss": 0.838,
      "step": 6036
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.355445572149422,
      "learning_rate": 8.572245465571828e-06,
      "loss": 0.8196,
      "step": 6037
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.789499685135674,
      "learning_rate": 8.571739994197994e-06,
      "loss": 0.8848,
      "step": 6038
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.853447641193016,
      "learning_rate": 8.571234448271911e-06,
      "loss": 0.8399,
      "step": 6039
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.165967187492139,
      "learning_rate": 8.57072882780413e-06,
      "loss": 0.7947,
      "step": 6040
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.608968489966911,
      "learning_rate": 8.570223132805202e-06,
      "loss": 0.777,
      "step": 6041
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.243848141933354,
      "learning_rate": 8.569717363285685e-06,
      "loss": 0.7717,
      "step": 6042
    },
    {
      "epoch": 0.54,
      "grad_norm": 10.477802172375956,
      "learning_rate": 8.569211519256138e-06,
      "loss": 0.8604,
      "step": 6043
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.981491507194219,
      "learning_rate": 8.568705600727117e-06,
      "loss": 0.7751,
      "step": 6044
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.0447522740171,
      "learning_rate": 8.56819960770918e-06,
      "loss": 0.8447,
      "step": 6045
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.381274889033219,
      "learning_rate": 8.56769354021289e-06,
      "loss": 0.7847,
      "step": 6046
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.584585391135997,
      "learning_rate": 8.567187398248813e-06,
      "loss": 0.8455,
      "step": 6047
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.746923541961522,
      "learning_rate": 8.56668118182751e-06,
      "loss": 0.7936,
      "step": 6048
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.524721831482064,
      "learning_rate": 8.566174890959547e-06,
      "loss": 0.795,
      "step": 6049
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.1344982758962265,
      "learning_rate": 8.565668525655495e-06,
      "loss": 0.8247,
      "step": 6050
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.383353942600456,
      "learning_rate": 8.56516208592592e-06,
      "loss": 0.7965,
      "step": 6051
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.039456429482954,
      "learning_rate": 8.564655571781393e-06,
      "loss": 0.7687,
      "step": 6052
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.312853261598657,
      "learning_rate": 8.56414898323249e-06,
      "loss": 0.7442,
      "step": 6053
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.911361852922648,
      "learning_rate": 8.56364232028978e-06,
      "loss": 0.8076,
      "step": 6054
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.959570682076482,
      "learning_rate": 8.563135582963841e-06,
      "loss": 0.7801,
      "step": 6055
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.609395870367582,
      "learning_rate": 8.56262877126525e-06,
      "loss": 0.7987,
      "step": 6056
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.06570991456562,
      "learning_rate": 8.562121885204587e-06,
      "loss": 0.8131,
      "step": 6057
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.045374404102717,
      "learning_rate": 8.56161492479243e-06,
      "loss": 0.7604,
      "step": 6058
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.725501047047067,
      "learning_rate": 8.561107890039359e-06,
      "loss": 0.7825,
      "step": 6059
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.225849473414879,
      "learning_rate": 8.560600780955964e-06,
      "loss": 0.7995,
      "step": 6060
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.614255153649072,
      "learning_rate": 8.56009359755282e-06,
      "loss": 0.8815,
      "step": 6061
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.682355715574637,
      "learning_rate": 8.559586339840522e-06,
      "loss": 0.8345,
      "step": 6062
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.972625146098602,
      "learning_rate": 8.559079007829654e-06,
      "loss": 0.8417,
      "step": 6063
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.94103059165331,
      "learning_rate": 8.558571601530805e-06,
      "loss": 0.8761,
      "step": 6064
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.406715050488156,
      "learning_rate": 8.558064120954569e-06,
      "loss": 0.8264,
      "step": 6065
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.903030652934619,
      "learning_rate": 8.557556566111535e-06,
      "loss": 0.8228,
      "step": 6066
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.3949183498334,
      "learning_rate": 8.557048937012298e-06,
      "loss": 0.8407,
      "step": 6067
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.52738558684205,
      "learning_rate": 8.556541233667454e-06,
      "loss": 0.7759,
      "step": 6068
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.000199939478609,
      "learning_rate": 8.556033456087602e-06,
      "loss": 0.8439,
      "step": 6069
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.942933993521471,
      "learning_rate": 8.555525604283338e-06,
      "loss": 0.743,
      "step": 6070
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.018904782405857,
      "learning_rate": 8.555017678265265e-06,
      "loss": 0.8322,
      "step": 6071
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.151466641347401,
      "learning_rate": 8.554509678043983e-06,
      "loss": 0.84,
      "step": 6072
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.668336978515194,
      "learning_rate": 8.554001603630094e-06,
      "loss": 0.7679,
      "step": 6073
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.773401107237038,
      "learning_rate": 8.553493455034205e-06,
      "loss": 0.8802,
      "step": 6074
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.921911830248139,
      "learning_rate": 8.552985232266924e-06,
      "loss": 0.7559,
      "step": 6075
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.5261787581851305,
      "learning_rate": 8.552476935338855e-06,
      "loss": 0.8851,
      "step": 6076
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.323435997468076,
      "learning_rate": 8.551968564260613e-06,
      "loss": 0.8498,
      "step": 6077
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.031032940954554,
      "learning_rate": 8.551460119042803e-06,
      "loss": 0.7812,
      "step": 6078
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.8417264523951715,
      "learning_rate": 8.550951599696041e-06,
      "loss": 0.8224,
      "step": 6079
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.039826768181223,
      "learning_rate": 8.550443006230942e-06,
      "loss": 0.847,
      "step": 6080
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.15145924041698,
      "learning_rate": 8.549934338658118e-06,
      "loss": 0.8551,
      "step": 6081
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.287392523044316,
      "learning_rate": 8.549425596988191e-06,
      "loss": 0.824,
      "step": 6082
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.654812395923941,
      "learning_rate": 8.548916781231778e-06,
      "loss": 0.7923,
      "step": 6083
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.315248697287516,
      "learning_rate": 8.548407891399498e-06,
      "loss": 0.8943,
      "step": 6084
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.733155089508796,
      "learning_rate": 8.547898927501975e-06,
      "loss": 0.7794,
      "step": 6085
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.429365455833748,
      "learning_rate": 8.547389889549831e-06,
      "loss": 0.7851,
      "step": 6086
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.892537681221298,
      "learning_rate": 8.546880777553692e-06,
      "loss": 0.8742,
      "step": 6087
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.295660598221255,
      "learning_rate": 8.546371591524185e-06,
      "loss": 0.8295,
      "step": 6088
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.673502318616251,
      "learning_rate": 8.545862331471938e-06,
      "loss": 0.8541,
      "step": 6089
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.268875701829673,
      "learning_rate": 8.545352997407577e-06,
      "loss": 0.7273,
      "step": 6090
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.480713517069899,
      "learning_rate": 8.54484358934174e-06,
      "loss": 0.7951,
      "step": 6091
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.197659632124393,
      "learning_rate": 8.544334107285053e-06,
      "loss": 0.8315,
      "step": 6092
    },
    {
      "epoch": 0.54,
      "grad_norm": 12.783276290298039,
      "learning_rate": 8.543824551248154e-06,
      "loss": 0.8243,
      "step": 6093
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.96486596010229,
      "learning_rate": 8.543314921241682e-06,
      "loss": 0.8482,
      "step": 6094
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.459199548043038,
      "learning_rate": 8.542805217276266e-06,
      "loss": 0.8279,
      "step": 6095
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.084335575283407,
      "learning_rate": 8.542295439362552e-06,
      "loss": 0.8414,
      "step": 6096
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.475444790242986,
      "learning_rate": 8.541785587511177e-06,
      "loss": 0.7853,
      "step": 6097
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.476033841835467,
      "learning_rate": 8.541275661732786e-06,
      "loss": 0.7805,
      "step": 6098
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.288016445468627,
      "learning_rate": 8.54076566203802e-06,
      "loss": 0.8039,
      "step": 6099
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.58086167340294,
      "learning_rate": 8.540255588437524e-06,
      "loss": 0.785,
      "step": 6100
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.484317924137649,
      "learning_rate": 8.539745440941947e-06,
      "loss": 0.8405,
      "step": 6101
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.558743665017648,
      "learning_rate": 8.539235219561936e-06,
      "loss": 0.7911,
      "step": 6102
    },
    {
      "epoch": 0.54,
      "grad_norm": 10.812253016634005,
      "learning_rate": 8.53872492430814e-06,
      "loss": 0.8763,
      "step": 6103
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.892192534241795,
      "learning_rate": 8.53821455519121e-06,
      "loss": 0.7753,
      "step": 6104
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.155576907203239,
      "learning_rate": 8.5377041122218e-06,
      "loss": 0.8438,
      "step": 6105
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.830494636115569,
      "learning_rate": 8.537193595410566e-06,
      "loss": 0.8391,
      "step": 6106
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.187512729162747,
      "learning_rate": 8.536683004768161e-06,
      "loss": 0.8666,
      "step": 6107
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.538826599601231,
      "learning_rate": 8.536172340305242e-06,
      "loss": 0.8379,
      "step": 6108
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.1413526361094615,
      "learning_rate": 8.535661602032472e-06,
      "loss": 0.851,
      "step": 6109
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.246461776283396,
      "learning_rate": 8.535150789960507e-06,
      "loss": 0.8318,
      "step": 6110
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.027629678423712,
      "learning_rate": 8.534639904100011e-06,
      "loss": 0.8214,
      "step": 6111
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.3097926295388405,
      "learning_rate": 8.53412894446165e-06,
      "loss": 0.809,
      "step": 6112
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.688158955421188,
      "learning_rate": 8.533617911056084e-06,
      "loss": 0.808,
      "step": 6113
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.39351617237998,
      "learning_rate": 8.533106803893984e-06,
      "loss": 0.8779,
      "step": 6114
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.3857744163071395,
      "learning_rate": 8.532595622986017e-06,
      "loss": 0.8039,
      "step": 6115
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.132158925511807,
      "learning_rate": 8.532084368342854e-06,
      "loss": 0.7913,
      "step": 6116
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.652570348626114,
      "learning_rate": 8.531573039975164e-06,
      "loss": 0.7603,
      "step": 6117
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.506853503591832,
      "learning_rate": 8.531061637893618e-06,
      "loss": 0.7983,
      "step": 6118
    },
    {
      "epoch": 0.55,
      "grad_norm": 8.030941631291078,
      "learning_rate": 8.530550162108897e-06,
      "loss": 0.8486,
      "step": 6119
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.350807794703884,
      "learning_rate": 8.53003861263167e-06,
      "loss": 0.8908,
      "step": 6120
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.492747925562131,
      "learning_rate": 8.529526989472621e-06,
      "loss": 0.8515,
      "step": 6121
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.870426288603464,
      "learning_rate": 8.529015292642423e-06,
      "loss": 0.7915,
      "step": 6122
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.2031722424627755,
      "learning_rate": 8.528503522151759e-06,
      "loss": 0.9284,
      "step": 6123
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.060135655263216,
      "learning_rate": 8.527991678011313e-06,
      "loss": 0.8799,
      "step": 6124
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.75945972116728,
      "learning_rate": 8.527479760231766e-06,
      "loss": 0.8801,
      "step": 6125
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.52827871296182,
      "learning_rate": 8.526967768823805e-06,
      "loss": 0.7716,
      "step": 6126
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.544286626629943,
      "learning_rate": 8.526455703798113e-06,
      "loss": 0.8145,
      "step": 6127
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.4289019178874,
      "learning_rate": 8.525943565165384e-06,
      "loss": 0.7611,
      "step": 6128
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.9165742674876185,
      "learning_rate": 8.525431352936304e-06,
      "loss": 0.7961,
      "step": 6129
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.806291070497382,
      "learning_rate": 8.524919067121565e-06,
      "loss": 0.7568,
      "step": 6130
    },
    {
      "epoch": 0.55,
      "grad_norm": 9.473074750498004,
      "learning_rate": 8.52440670773186e-06,
      "loss": 0.8254,
      "step": 6131
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.80676068055187,
      "learning_rate": 8.523894274777881e-06,
      "loss": 0.7927,
      "step": 6132
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.1035006938847065,
      "learning_rate": 8.523381768270328e-06,
      "loss": 0.803,
      "step": 6133
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.8932974768295665,
      "learning_rate": 8.522869188219897e-06,
      "loss": 0.8137,
      "step": 6134
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.57753006792163,
      "learning_rate": 8.522356534637289e-06,
      "loss": 0.7931,
      "step": 6135
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.6830923577533525,
      "learning_rate": 8.521843807533198e-06,
      "loss": 0.8466,
      "step": 6136
    },
    {
      "epoch": 0.55,
      "grad_norm": 22.577365496736437,
      "learning_rate": 8.521331006918332e-06,
      "loss": 0.8939,
      "step": 6137
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.240673115275456,
      "learning_rate": 8.520818132803393e-06,
      "loss": 0.9448,
      "step": 6138
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.680152692879237,
      "learning_rate": 8.520305185199086e-06,
      "loss": 0.8448,
      "step": 6139
    },
    {
      "epoch": 0.55,
      "grad_norm": 8.265266659354326,
      "learning_rate": 8.519792164116116e-06,
      "loss": 0.8008,
      "step": 6140
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.171404773664601,
      "learning_rate": 8.519279069565194e-06,
      "loss": 0.7996,
      "step": 6141
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.936926518889804,
      "learning_rate": 8.518765901557029e-06,
      "loss": 0.876,
      "step": 6142
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.5959364192323,
      "learning_rate": 8.518252660102332e-06,
      "loss": 0.7479,
      "step": 6143
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.583298869792633,
      "learning_rate": 8.517739345211815e-06,
      "loss": 0.8352,
      "step": 6144
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.072393972075555,
      "learning_rate": 8.517225956896192e-06,
      "loss": 0.7841,
      "step": 6145
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.71228958728902,
      "learning_rate": 8.516712495166181e-06,
      "loss": 0.7372,
      "step": 6146
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.194409858498816,
      "learning_rate": 8.516198960032498e-06,
      "loss": 0.7673,
      "step": 6147
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.785605798911552,
      "learning_rate": 8.515685351505862e-06,
      "loss": 0.8144,
      "step": 6148
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.988166068397192,
      "learning_rate": 8.515171669596994e-06,
      "loss": 0.8504,
      "step": 6149
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.64038373616684,
      "learning_rate": 8.514657914316614e-06,
      "loss": 0.8506,
      "step": 6150
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.616194665394371,
      "learning_rate": 8.514144085675447e-06,
      "loss": 0.8454,
      "step": 6151
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.6384352366671235,
      "learning_rate": 8.513630183684219e-06,
      "loss": 0.8236,
      "step": 6152
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.714038505224621,
      "learning_rate": 8.513116208353656e-06,
      "loss": 0.849,
      "step": 6153
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.58433448007138,
      "learning_rate": 8.512602159694484e-06,
      "loss": 0.7623,
      "step": 6154
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.92934196169855,
      "learning_rate": 8.512088037717435e-06,
      "loss": 0.8309,
      "step": 6155
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.3977838168335115,
      "learning_rate": 8.511573842433241e-06,
      "loss": 0.7869,
      "step": 6156
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.672238009435888,
      "learning_rate": 8.511059573852631e-06,
      "loss": 0.7694,
      "step": 6157
    },
    {
      "epoch": 0.55,
      "grad_norm": 8.10240728160039,
      "learning_rate": 8.510545231986345e-06,
      "loss": 0.858,
      "step": 6158
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.403313937104579,
      "learning_rate": 8.510030816845112e-06,
      "loss": 0.7655,
      "step": 6159
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.135652116802077,
      "learning_rate": 8.509516328439672e-06,
      "loss": 0.8266,
      "step": 6160
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.356264189961006,
      "learning_rate": 8.509001766780767e-06,
      "loss": 0.8474,
      "step": 6161
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.535713193115595,
      "learning_rate": 8.508487131879133e-06,
      "loss": 0.8218,
      "step": 6162
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.265778929767966,
      "learning_rate": 8.507972423745514e-06,
      "loss": 0.9337,
      "step": 6163
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.651389247458359,
      "learning_rate": 8.507457642390655e-06,
      "loss": 0.7855,
      "step": 6164
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.0412275337831165,
      "learning_rate": 8.506942787825297e-06,
      "loss": 0.7924,
      "step": 6165
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.869582867265301,
      "learning_rate": 8.506427860060188e-06,
      "loss": 0.8657,
      "step": 6166
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.572228809830339,
      "learning_rate": 8.505912859106076e-06,
      "loss": 0.7875,
      "step": 6167
    },
    {
      "epoch": 0.55,
      "grad_norm": 8.238495264105655,
      "learning_rate": 8.505397784973713e-06,
      "loss": 0.876,
      "step": 6168
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.705198294482988,
      "learning_rate": 8.504882637673846e-06,
      "loss": 0.8114,
      "step": 6169
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.651033572783869,
      "learning_rate": 8.50436741721723e-06,
      "loss": 0.8113,
      "step": 6170
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.3885708148308105,
      "learning_rate": 8.50385212361462e-06,
      "loss": 0.9038,
      "step": 6171
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.781686433451358,
      "learning_rate": 8.503336756876768e-06,
      "loss": 0.7905,
      "step": 6172
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.182124498683177,
      "learning_rate": 8.502821317014435e-06,
      "loss": 0.853,
      "step": 6173
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.165236050079116,
      "learning_rate": 8.502305804038378e-06,
      "loss": 0.8581,
      "step": 6174
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.383905639846727,
      "learning_rate": 8.501790217959356e-06,
      "loss": 0.7667,
      "step": 6175
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.788932636278462,
      "learning_rate": 8.501274558788132e-06,
      "loss": 0.8085,
      "step": 6176
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.232506092646649,
      "learning_rate": 8.50075882653547e-06,
      "loss": 0.8351,
      "step": 6177
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.15309952238115,
      "learning_rate": 8.500243021212134e-06,
      "loss": 0.8334,
      "step": 6178
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.25696044281939,
      "learning_rate": 8.499727142828891e-06,
      "loss": 0.827,
      "step": 6179
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.024752377862959,
      "learning_rate": 8.499211191396507e-06,
      "loss": 0.8184,
      "step": 6180
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.502766460356569,
      "learning_rate": 8.498695166925752e-06,
      "loss": 0.8036,
      "step": 6181
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.686933824678985,
      "learning_rate": 8.4981790694274e-06,
      "loss": 0.8471,
      "step": 6182
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.867360482712466,
      "learning_rate": 8.497662898912217e-06,
      "loss": 0.8267,
      "step": 6183
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.896911452303937,
      "learning_rate": 8.497146655390983e-06,
      "loss": 0.8149,
      "step": 6184
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.549454694567512,
      "learning_rate": 8.496630338874472e-06,
      "loss": 0.7623,
      "step": 6185
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.588479665324505,
      "learning_rate": 8.496113949373459e-06,
      "loss": 0.827,
      "step": 6186
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.168992502176361,
      "learning_rate": 8.495597486898725e-06,
      "loss": 0.7979,
      "step": 6187
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.69096513482704,
      "learning_rate": 8.495080951461047e-06,
      "loss": 0.8009,
      "step": 6188
    },
    {
      "epoch": 0.55,
      "grad_norm": 8.279172339248566,
      "learning_rate": 8.494564343071208e-06,
      "loss": 0.7653,
      "step": 6189
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.032240702611898,
      "learning_rate": 8.494047661739993e-06,
      "loss": 0.7463,
      "step": 6190
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.45045263504536,
      "learning_rate": 8.493530907478185e-06,
      "loss": 0.838,
      "step": 6191
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.68817240129019,
      "learning_rate": 8.493014080296568e-06,
      "loss": 0.7658,
      "step": 6192
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.019394605947285,
      "learning_rate": 8.492497180205933e-06,
      "loss": 0.8451,
      "step": 6193
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.488384110892746,
      "learning_rate": 8.491980207217067e-06,
      "loss": 0.8314,
      "step": 6194
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.296688182012467,
      "learning_rate": 8.491463161340763e-06,
      "loss": 0.8682,
      "step": 6195
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.229336561781027,
      "learning_rate": 8.490946042587811e-06,
      "loss": 0.7701,
      "step": 6196
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.73288484022705,
      "learning_rate": 8.490428850969003e-06,
      "loss": 0.8132,
      "step": 6197
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.626764850375793,
      "learning_rate": 8.48991158649514e-06,
      "loss": 0.8013,
      "step": 6198
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.44650280944281,
      "learning_rate": 8.489394249177014e-06,
      "loss": 0.8196,
      "step": 6199
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.544397670626195,
      "learning_rate": 8.488876839025425e-06,
      "loss": 0.8285,
      "step": 6200
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.88041715441373,
      "learning_rate": 8.488359356051173e-06,
      "loss": 0.8467,
      "step": 6201
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.802429633707484,
      "learning_rate": 8.487841800265057e-06,
      "loss": 0.8289,
      "step": 6202
    },
    {
      "epoch": 0.55,
      "grad_norm": 8.732475229772797,
      "learning_rate": 8.487324171677883e-06,
      "loss": 0.7883,
      "step": 6203
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.85431711785811,
      "learning_rate": 8.486806470300454e-06,
      "loss": 0.8985,
      "step": 6204
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.4707884856013695,
      "learning_rate": 8.486288696143575e-06,
      "loss": 0.7989,
      "step": 6205
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.9559208345258545,
      "learning_rate": 8.485770849218056e-06,
      "loss": 0.8539,
      "step": 6206
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.412133779917283,
      "learning_rate": 8.485252929534704e-06,
      "loss": 0.835,
      "step": 6207
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.230043251090192,
      "learning_rate": 8.484734937104328e-06,
      "loss": 0.7344,
      "step": 6208
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.238770667421405,
      "learning_rate": 8.484216871937743e-06,
      "loss": 0.8014,
      "step": 6209
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.909913880190588,
      "learning_rate": 8.48369873404576e-06,
      "loss": 0.8837,
      "step": 6210
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.031842016810103,
      "learning_rate": 8.483180523439196e-06,
      "loss": 0.8274,
      "step": 6211
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.991914581562651,
      "learning_rate": 8.482662240128865e-06,
      "loss": 0.7621,
      "step": 6212
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.547377085438254,
      "learning_rate": 8.48214388412559e-06,
      "loss": 0.7363,
      "step": 6213
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.588381205660118,
      "learning_rate": 8.481625455440185e-06,
      "loss": 0.7416,
      "step": 6214
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.548409260396815,
      "learning_rate": 8.481106954083474e-06,
      "loss": 0.7313,
      "step": 6215
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.92638862992217,
      "learning_rate": 8.480588380066277e-06,
      "loss": 0.8193,
      "step": 6216
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.944868465147303,
      "learning_rate": 8.48006973339942e-06,
      "loss": 0.7554,
      "step": 6217
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.115003838308667,
      "learning_rate": 8.47955101409373e-06,
      "loss": 0.7747,
      "step": 6218
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.907041181367516,
      "learning_rate": 8.479032222160032e-06,
      "loss": 0.826,
      "step": 6219
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.784267422542636,
      "learning_rate": 8.478513357609158e-06,
      "loss": 0.7732,
      "step": 6220
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.527020873792892,
      "learning_rate": 8.477994420451932e-06,
      "loss": 0.8031,
      "step": 6221
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.1419833088213975,
      "learning_rate": 8.477475410699191e-06,
      "loss": 0.7674,
      "step": 6222
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.957203611719857,
      "learning_rate": 8.476956328361765e-06,
      "loss": 0.8481,
      "step": 6223
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.1916148789951775,
      "learning_rate": 8.476437173450491e-06,
      "loss": 0.8378,
      "step": 6224
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.711236751936519,
      "learning_rate": 8.475917945976204e-06,
      "loss": 0.8564,
      "step": 6225
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.818195869576693,
      "learning_rate": 8.475398645949741e-06,
      "loss": 0.8569,
      "step": 6226
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.12118276425607,
      "learning_rate": 8.474879273381944e-06,
      "loss": 0.8582,
      "step": 6227
    },
    {
      "epoch": 0.56,
      "grad_norm": 10.722477354096204,
      "learning_rate": 8.474359828283652e-06,
      "loss": 0.8276,
      "step": 6228
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.743799754666631,
      "learning_rate": 8.473840310665706e-06,
      "loss": 0.7777,
      "step": 6229
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.298126877660536,
      "learning_rate": 8.473320720538951e-06,
      "loss": 0.8561,
      "step": 6230
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.582817716537235,
      "learning_rate": 8.472801057914235e-06,
      "loss": 0.8208,
      "step": 6231
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.381779660405868,
      "learning_rate": 8.472281322802401e-06,
      "loss": 0.8141,
      "step": 6232
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.781511200481694,
      "learning_rate": 8.471761515214296e-06,
      "loss": 0.785,
      "step": 6233
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.846120028090245,
      "learning_rate": 8.471241635160774e-06,
      "loss": 0.8331,
      "step": 6234
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.031494598828061,
      "learning_rate": 8.470721682652684e-06,
      "loss": 0.84,
      "step": 6235
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.626416423506994,
      "learning_rate": 8.470201657700884e-06,
      "loss": 0.7974,
      "step": 6236
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.018376778980985,
      "learning_rate": 8.469681560316219e-06,
      "loss": 0.9541,
      "step": 6237
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.6935782475211445,
      "learning_rate": 8.469161390509553e-06,
      "loss": 0.8348,
      "step": 6238
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.533818457564585,
      "learning_rate": 8.468641148291738e-06,
      "loss": 0.8006,
      "step": 6239
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.899207562638327,
      "learning_rate": 8.468120833673636e-06,
      "loss": 0.7626,
      "step": 6240
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.804851569995742,
      "learning_rate": 8.467600446666106e-06,
      "loss": 0.8134,
      "step": 6241
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.136177462255176,
      "learning_rate": 8.46707998728001e-06,
      "loss": 0.773,
      "step": 6242
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.955011746701454,
      "learning_rate": 8.466559455526215e-06,
      "loss": 0.8498,
      "step": 6243
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.976064590078513,
      "learning_rate": 8.466038851415581e-06,
      "loss": 0.7965,
      "step": 6244
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.862344108008988,
      "learning_rate": 8.465518174958978e-06,
      "loss": 0.7798,
      "step": 6245
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.254069486652891,
      "learning_rate": 8.464997426167271e-06,
      "loss": 0.8783,
      "step": 6246
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.006443546340096,
      "learning_rate": 8.46447660505133e-06,
      "loss": 0.7832,
      "step": 6247
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.554205593482087,
      "learning_rate": 8.46395571162203e-06,
      "loss": 0.8481,
      "step": 6248
    },
    {
      "epoch": 0.56,
      "grad_norm": 8.839629187630445,
      "learning_rate": 8.463434745890238e-06,
      "loss": 0.8388,
      "step": 6249
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.333135076944842,
      "learning_rate": 8.46291370786683e-06,
      "loss": 0.8657,
      "step": 6250
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.441589658680589,
      "learning_rate": 8.462392597562685e-06,
      "loss": 0.8073,
      "step": 6251
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.937790280186652,
      "learning_rate": 8.461871414988673e-06,
      "loss": 0.8211,
      "step": 6252
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.79251490007488,
      "learning_rate": 8.461350160155678e-06,
      "loss": 0.8871,
      "step": 6253
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.394440620719659,
      "learning_rate": 8.460828833074582e-06,
      "loss": 0.7998,
      "step": 6254
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.663932195422399,
      "learning_rate": 8.46030743375626e-06,
      "loss": 0.7737,
      "step": 6255
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.343246134577354,
      "learning_rate": 8.459785962211599e-06,
      "loss": 0.8114,
      "step": 6256
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.80863811071495,
      "learning_rate": 8.459264418451483e-06,
      "loss": 0.8513,
      "step": 6257
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.908366344607159,
      "learning_rate": 8.458742802486798e-06,
      "loss": 0.7853,
      "step": 6258
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.237891202919982,
      "learning_rate": 8.45822111432843e-06,
      "loss": 0.7867,
      "step": 6259
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.058890497441204,
      "learning_rate": 8.45769935398727e-06,
      "loss": 0.7682,
      "step": 6260
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.386244922059968,
      "learning_rate": 8.457177521474209e-06,
      "loss": 0.8003,
      "step": 6261
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.425782499658812,
      "learning_rate": 8.45665561680014e-06,
      "loss": 0.7757,
      "step": 6262
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.548978986857963,
      "learning_rate": 8.456133639975952e-06,
      "loss": 0.8332,
      "step": 6263
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.9658661806508855,
      "learning_rate": 8.455611591012544e-06,
      "loss": 0.8027,
      "step": 6264
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.580460635182092,
      "learning_rate": 8.455089469920812e-06,
      "loss": 0.8049,
      "step": 6265
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.191637957649086,
      "learning_rate": 8.454567276711654e-06,
      "loss": 0.779,
      "step": 6266
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.565770793508458,
      "learning_rate": 8.45404501139597e-06,
      "loss": 0.8187,
      "step": 6267
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.094109439481992,
      "learning_rate": 8.45352267398466e-06,
      "loss": 0.8303,
      "step": 6268
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.4061588126629525,
      "learning_rate": 8.453000264488627e-06,
      "loss": 0.8934,
      "step": 6269
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.247308137596359,
      "learning_rate": 8.452477782918775e-06,
      "loss": 0.7943,
      "step": 6270
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.704847379843028,
      "learning_rate": 8.451955229286012e-06,
      "loss": 0.824,
      "step": 6271
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.774737485666886,
      "learning_rate": 8.451432603601243e-06,
      "loss": 0.8583,
      "step": 6272
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.7858093536546225,
      "learning_rate": 8.450909905875374e-06,
      "loss": 0.8631,
      "step": 6273
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.550783444853573,
      "learning_rate": 8.45038713611932e-06,
      "loss": 0.7459,
      "step": 6274
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.142852263201285,
      "learning_rate": 8.44986429434399e-06,
      "loss": 0.7938,
      "step": 6275
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.5094940287490415,
      "learning_rate": 8.449341380560298e-06,
      "loss": 0.8172,
      "step": 6276
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.64538582441074,
      "learning_rate": 8.44881839477916e-06,
      "loss": 0.7928,
      "step": 6277
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.077133999730289,
      "learning_rate": 8.448295337011489e-06,
      "loss": 0.7691,
      "step": 6278
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.068427885933199,
      "learning_rate": 8.447772207268207e-06,
      "loss": 0.7955,
      "step": 6279
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.315754049812272,
      "learning_rate": 8.44724900556023e-06,
      "loss": 0.8045,
      "step": 6280
    },
    {
      "epoch": 0.56,
      "grad_norm": 8.690280502518323,
      "learning_rate": 8.446725731898478e-06,
      "loss": 0.8633,
      "step": 6281
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.576959038792565,
      "learning_rate": 8.446202386293874e-06,
      "loss": 0.7823,
      "step": 6282
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.064200440107103,
      "learning_rate": 8.445678968757343e-06,
      "loss": 0.8775,
      "step": 6283
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.89167701655723,
      "learning_rate": 8.44515547929981e-06,
      "loss": 0.8276,
      "step": 6284
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.007439869280262,
      "learning_rate": 8.4446319179322e-06,
      "loss": 0.9052,
      "step": 6285
    },
    {
      "epoch": 0.56,
      "grad_norm": 9.12907322371847,
      "learning_rate": 8.444108284665444e-06,
      "loss": 0.789,
      "step": 6286
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.151874107192116,
      "learning_rate": 8.44358457951047e-06,
      "loss": 0.849,
      "step": 6287
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.62119654635346,
      "learning_rate": 8.443060802478207e-06,
      "loss": 0.8,
      "step": 6288
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.009118726127017,
      "learning_rate": 8.442536953579591e-06,
      "loss": 0.7456,
      "step": 6289
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.9040719935820745,
      "learning_rate": 8.442013032825556e-06,
      "loss": 0.767,
      "step": 6290
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.779871130558414,
      "learning_rate": 8.441489040227037e-06,
      "loss": 0.8674,
      "step": 6291
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.231902348059256,
      "learning_rate": 8.44096497579497e-06,
      "loss": 0.7996,
      "step": 6292
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.425774042426578,
      "learning_rate": 8.440440839540296e-06,
      "loss": 0.8075,
      "step": 6293
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.138511337768828,
      "learning_rate": 8.439916631473952e-06,
      "loss": 0.8146,
      "step": 6294
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.224241795700008,
      "learning_rate": 8.439392351606884e-06,
      "loss": 0.8012,
      "step": 6295
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.463608596283664,
      "learning_rate": 8.43886799995003e-06,
      "loss": 0.8532,
      "step": 6296
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.328126953955852,
      "learning_rate": 8.438343576514339e-06,
      "loss": 0.7582,
      "step": 6297
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.414492044123563,
      "learning_rate": 8.437819081310756e-06,
      "loss": 0.8276,
      "step": 6298
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.292027280206943,
      "learning_rate": 8.437294514350227e-06,
      "loss": 0.8552,
      "step": 6299
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.581227539847117,
      "learning_rate": 8.436769875643703e-06,
      "loss": 0.8481,
      "step": 6300
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.683993921880735,
      "learning_rate": 8.436245165202135e-06,
      "loss": 0.7774,
      "step": 6301
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.8277235157999487,
      "learning_rate": 8.435720383036474e-06,
      "loss": 0.7803,
      "step": 6302
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.097736823698492,
      "learning_rate": 8.435195529157676e-06,
      "loss": 0.7996,
      "step": 6303
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.865699572511002,
      "learning_rate": 8.434670603576691e-06,
      "loss": 0.8003,
      "step": 6304
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.537567528908108,
      "learning_rate": 8.43414560630448e-06,
      "loss": 0.8047,
      "step": 6305
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.658066927145337,
      "learning_rate": 8.433620537352001e-06,
      "loss": 0.8066,
      "step": 6306
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.595720075705235,
      "learning_rate": 8.433095396730212e-06,
      "loss": 0.7674,
      "step": 6307
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.492135945545079,
      "learning_rate": 8.432570184450077e-06,
      "loss": 0.8381,
      "step": 6308
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.667860466282736,
      "learning_rate": 8.432044900522553e-06,
      "loss": 0.8035,
      "step": 6309
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.215963728713693,
      "learning_rate": 8.43151954495861e-06,
      "loss": 0.8453,
      "step": 6310
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.7456038803293055,
      "learning_rate": 8.430994117769211e-06,
      "loss": 0.7885,
      "step": 6311
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.235561861721924,
      "learning_rate": 8.430468618965324e-06,
      "loss": 0.738,
      "step": 6312
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.398555423094524,
      "learning_rate": 8.429943048557917e-06,
      "loss": 0.8688,
      "step": 6313
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.7313324190948025,
      "learning_rate": 8.42941740655796e-06,
      "loss": 0.8526,
      "step": 6314
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.636788492540761,
      "learning_rate": 8.428891692976425e-06,
      "loss": 0.8635,
      "step": 6315
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.338508054248021,
      "learning_rate": 8.428365907824285e-06,
      "loss": 0.8344,
      "step": 6316
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.800794171299255,
      "learning_rate": 8.427840051112514e-06,
      "loss": 0.8942,
      "step": 6317
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.289539922395088,
      "learning_rate": 8.42731412285209e-06,
      "loss": 0.8931,
      "step": 6318
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.097839716063886,
      "learning_rate": 8.42678812305399e-06,
      "loss": 0.7746,
      "step": 6319
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.487372301726876,
      "learning_rate": 8.42626205172919e-06,
      "loss": 0.795,
      "step": 6320
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.797470602121593,
      "learning_rate": 8.425735908888676e-06,
      "loss": 0.7905,
      "step": 6321
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.83939571421671,
      "learning_rate": 8.425209694543425e-06,
      "loss": 0.878,
      "step": 6322
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.950947694500436,
      "learning_rate": 8.424683408704423e-06,
      "loss": 0.8285,
      "step": 6323
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.058197370776621,
      "learning_rate": 8.424157051382656e-06,
      "loss": 0.8258,
      "step": 6324
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.95255844847449,
      "learning_rate": 8.423630622589107e-06,
      "loss": 0.8177,
      "step": 6325
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.452879134427941,
      "learning_rate": 8.423104122334768e-06,
      "loss": 0.7669,
      "step": 6326
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.273184348009166,
      "learning_rate": 8.422577550630627e-06,
      "loss": 0.8241,
      "step": 6327
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.155244319566487,
      "learning_rate": 8.422050907487673e-06,
      "loss": 0.7708,
      "step": 6328
    },
    {
      "epoch": 0.56,
      "grad_norm": 8.218260491866683,
      "learning_rate": 8.421524192916903e-06,
      "loss": 0.8075,
      "step": 6329
    },
    {
      "epoch": 0.56,
      "grad_norm": 10.515615296779487,
      "learning_rate": 8.420997406929308e-06,
      "loss": 0.8705,
      "step": 6330
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.926702165901228,
      "learning_rate": 8.420470549535884e-06,
      "loss": 0.8351,
      "step": 6331
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.90293589194997,
      "learning_rate": 8.419943620747628e-06,
      "loss": 0.839,
      "step": 6332
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.6543381595380895,
      "learning_rate": 8.419416620575536e-06,
      "loss": 0.7462,
      "step": 6333
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.336092772487322,
      "learning_rate": 8.418889549030614e-06,
      "loss": 0.8458,
      "step": 6334
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.690894946019437,
      "learning_rate": 8.418362406123856e-06,
      "loss": 0.7713,
      "step": 6335
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.824133202567947,
      "learning_rate": 8.417835191866273e-06,
      "loss": 0.8936,
      "step": 6336
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.407555817901577,
      "learning_rate": 8.417307906268864e-06,
      "loss": 0.8289,
      "step": 6337
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.969687132248885,
      "learning_rate": 8.416780549342636e-06,
      "loss": 0.7947,
      "step": 6338
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.833216048142414,
      "learning_rate": 8.416253121098595e-06,
      "loss": 0.7635,
      "step": 6339
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.050130620307425,
      "learning_rate": 8.415725621547752e-06,
      "loss": 0.7957,
      "step": 6340
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.052182220601051,
      "learning_rate": 8.415198050701119e-06,
      "loss": 0.8724,
      "step": 6341
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.477900014215796,
      "learning_rate": 8.414670408569705e-06,
      "loss": 0.9096,
      "step": 6342
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.42919112174351,
      "learning_rate": 8.414142695164523e-06,
      "loss": 0.8052,
      "step": 6343
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.545578540382481,
      "learning_rate": 8.41361491049659e-06,
      "loss": 0.752,
      "step": 6344
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.875468964415634,
      "learning_rate": 8.413087054576923e-06,
      "loss": 0.8169,
      "step": 6345
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.87391438065856,
      "learning_rate": 8.412559127416536e-06,
      "loss": 0.7759,
      "step": 6346
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.721858255793767,
      "learning_rate": 8.412031129026452e-06,
      "loss": 0.8199,
      "step": 6347
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.697367756939675,
      "learning_rate": 8.41150305941769e-06,
      "loss": 0.8919,
      "step": 6348
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.163665965888667,
      "learning_rate": 8.410974918601272e-06,
      "loss": 0.8399,
      "step": 6349
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.447015633361741,
      "learning_rate": 8.410446706588223e-06,
      "loss": 0.8505,
      "step": 6350
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.504920155126795,
      "learning_rate": 8.409918423389568e-06,
      "loss": 0.8037,
      "step": 6351
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.429953750743542,
      "learning_rate": 8.409390069016331e-06,
      "loss": 0.7912,
      "step": 6352
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.0501594240388314,
      "learning_rate": 8.408861643479546e-06,
      "loss": 0.7864,
      "step": 6353
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.40529439940605,
      "learning_rate": 8.408333146790238e-06,
      "loss": 0.7739,
      "step": 6354
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.5541722409253405,
      "learning_rate": 8.40780457895944e-06,
      "loss": 0.8146,
      "step": 6355
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.9557074984490255,
      "learning_rate": 8.407275939998184e-06,
      "loss": 0.8563,
      "step": 6356
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.214713537587707,
      "learning_rate": 8.406747229917507e-06,
      "loss": 0.7698,
      "step": 6357
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.94358432998389,
      "learning_rate": 8.40621844872844e-06,
      "loss": 0.8071,
      "step": 6358
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.02690489579196,
      "learning_rate": 8.405689596442022e-06,
      "loss": 0.9108,
      "step": 6359
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.8096379239525575,
      "learning_rate": 8.405160673069293e-06,
      "loss": 0.7132,
      "step": 6360
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.149839066726462,
      "learning_rate": 8.404631678621292e-06,
      "loss": 0.7552,
      "step": 6361
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.418012708795107,
      "learning_rate": 8.40410261310906e-06,
      "loss": 0.8363,
      "step": 6362
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.127212407014792,
      "learning_rate": 8.403573476543641e-06,
      "loss": 0.7863,
      "step": 6363
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.283871410444886,
      "learning_rate": 8.40304426893608e-06,
      "loss": 0.8599,
      "step": 6364
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.98981824958954,
      "learning_rate": 8.402514990297421e-06,
      "loss": 0.7673,
      "step": 6365
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.09888553824146,
      "learning_rate": 8.401985640638715e-06,
      "loss": 0.8366,
      "step": 6366
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.939549962831484,
      "learning_rate": 8.401456219971008e-06,
      "loss": 0.8476,
      "step": 6367
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.228795402897187,
      "learning_rate": 8.400926728305353e-06,
      "loss": 0.859,
      "step": 6368
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.684209483462334,
      "learning_rate": 8.400397165652797e-06,
      "loss": 0.7902,
      "step": 6369
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.27385695311404,
      "learning_rate": 8.3998675320244e-06,
      "loss": 0.7436,
      "step": 6370
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.224226627762636,
      "learning_rate": 8.399337827431211e-06,
      "loss": 0.8078,
      "step": 6371
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.55815220215311,
      "learning_rate": 8.398808051884292e-06,
      "loss": 0.8522,
      "step": 6372
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.2698600440447185,
      "learning_rate": 8.398278205394697e-06,
      "loss": 0.7517,
      "step": 6373
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.987968636550121,
      "learning_rate": 8.397748287973486e-06,
      "loss": 0.7384,
      "step": 6374
    },
    {
      "epoch": 0.57,
      "grad_norm": 8.141548708758679,
      "learning_rate": 8.397218299631723e-06,
      "loss": 0.8345,
      "step": 6375
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.38404744235643,
      "learning_rate": 8.396688240380465e-06,
      "loss": 0.7993,
      "step": 6376
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.705991452506625,
      "learning_rate": 8.396158110230778e-06,
      "loss": 0.8506,
      "step": 6377
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.9915806526215265,
      "learning_rate": 8.395627909193731e-06,
      "loss": 0.8211,
      "step": 6378
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.958566420534438,
      "learning_rate": 8.395097637280387e-06,
      "loss": 0.8042,
      "step": 6379
    },
    {
      "epoch": 0.57,
      "grad_norm": 8.927662893097061,
      "learning_rate": 8.394567294501815e-06,
      "loss": 0.827,
      "step": 6380
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.701348690666862,
      "learning_rate": 8.394036880869082e-06,
      "loss": 0.8516,
      "step": 6381
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.996377973099109,
      "learning_rate": 8.393506396393264e-06,
      "loss": 0.8108,
      "step": 6382
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.665861125055204,
      "learning_rate": 8.392975841085432e-06,
      "loss": 0.8452,
      "step": 6383
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.254886442770166,
      "learning_rate": 8.392445214956658e-06,
      "loss": 0.8034,
      "step": 6384
    },
    {
      "epoch": 0.57,
      "grad_norm": 9.228158014432296,
      "learning_rate": 8.39191451801802e-06,
      "loss": 0.8538,
      "step": 6385
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.446224330772276,
      "learning_rate": 8.391383750280594e-06,
      "loss": 0.8555,
      "step": 6386
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.141599537751966,
      "learning_rate": 8.390852911755463e-06,
      "loss": 0.8294,
      "step": 6387
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.611089176272104,
      "learning_rate": 8.3903220024537e-06,
      "loss": 0.782,
      "step": 6388
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.156796784584249,
      "learning_rate": 8.38979102238639e-06,
      "loss": 0.7887,
      "step": 6389
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.340125238868906,
      "learning_rate": 8.389259971564618e-06,
      "loss": 0.7689,
      "step": 6390
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.1012900336155536,
      "learning_rate": 8.388728849999465e-06,
      "loss": 0.8338,
      "step": 6391
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.7312958571210535,
      "learning_rate": 8.388197657702017e-06,
      "loss": 0.8189,
      "step": 6392
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.568835269016552,
      "learning_rate": 8.387666394683365e-06,
      "loss": 0.7722,
      "step": 6393
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.715292791689378,
      "learning_rate": 8.387135060954594e-06,
      "loss": 0.7958,
      "step": 6394
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.903541099100973,
      "learning_rate": 8.3866036565268e-06,
      "loss": 0.7837,
      "step": 6395
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.819768684308915,
      "learning_rate": 8.386072181411068e-06,
      "loss": 0.7627,
      "step": 6396
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.840295527539505,
      "learning_rate": 8.385540635618495e-06,
      "loss": 0.7593,
      "step": 6397
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.2308612624173545,
      "learning_rate": 8.385009019160176e-06,
      "loss": 0.8239,
      "step": 6398
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.007674355162268,
      "learning_rate": 8.384477332047206e-06,
      "loss": 0.7881,
      "step": 6399
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.37681632762346,
      "learning_rate": 8.383945574290684e-06,
      "loss": 0.8603,
      "step": 6400
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.669914659728779,
      "learning_rate": 8.38341374590171e-06,
      "loss": 0.8286,
      "step": 6401
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.217165536361894,
      "learning_rate": 8.382881846891381e-06,
      "loss": 0.8728,
      "step": 6402
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.5367591140674435,
      "learning_rate": 8.382349877270804e-06,
      "loss": 0.7575,
      "step": 6403
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.958038363001682,
      "learning_rate": 8.38181783705108e-06,
      "loss": 0.7964,
      "step": 6404
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.042780481711249,
      "learning_rate": 8.381285726243316e-06,
      "loss": 0.7839,
      "step": 6405
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.241322894572705,
      "learning_rate": 8.380753544858615e-06,
      "loss": 0.7531,
      "step": 6406
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.441367426592209,
      "learning_rate": 8.380221292908088e-06,
      "loss": 0.7648,
      "step": 6407
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.640196404216833,
      "learning_rate": 8.379688970402845e-06,
      "loss": 0.779,
      "step": 6408
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.909939547150019,
      "learning_rate": 8.379156577353995e-06,
      "loss": 0.7921,
      "step": 6409
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.875912809404505,
      "learning_rate": 8.378624113772653e-06,
      "loss": 0.8413,
      "step": 6410
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.9044348475426585,
      "learning_rate": 8.37809157966993e-06,
      "loss": 0.8073,
      "step": 6411
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.827483789523319,
      "learning_rate": 8.377558975056945e-06,
      "loss": 0.7521,
      "step": 6412
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.9629672980840516,
      "learning_rate": 8.377026299944813e-06,
      "loss": 0.8023,
      "step": 6413
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.7907634224069495,
      "learning_rate": 8.376493554344652e-06,
      "loss": 0.8427,
      "step": 6414
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.375826074232661,
      "learning_rate": 8.37596073826758e-06,
      "loss": 0.796,
      "step": 6415
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.70135947496201,
      "learning_rate": 8.375427851724725e-06,
      "loss": 0.7897,
      "step": 6416
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.824913729040646,
      "learning_rate": 8.374894894727203e-06,
      "loss": 0.7885,
      "step": 6417
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.640193494959229,
      "learning_rate": 8.374361867286143e-06,
      "loss": 0.7609,
      "step": 6418
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.732170228556597,
      "learning_rate": 8.373828769412667e-06,
      "loss": 0.8066,
      "step": 6419
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.341596206767716,
      "learning_rate": 8.373295601117903e-06,
      "loss": 0.7864,
      "step": 6420
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.935007139865047,
      "learning_rate": 8.372762362412982e-06,
      "loss": 0.8213,
      "step": 6421
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.9616765403765495,
      "learning_rate": 8.372229053309035e-06,
      "loss": 0.8351,
      "step": 6422
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.972802019605322,
      "learning_rate": 8.37169567381719e-06,
      "loss": 0.8431,
      "step": 6423
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.445837448479996,
      "learning_rate": 8.37116222394858e-06,
      "loss": 0.7908,
      "step": 6424
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.314495008176629,
      "learning_rate": 8.370628703714343e-06,
      "loss": 0.7756,
      "step": 6425
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.617900010467859,
      "learning_rate": 8.370095113125613e-06,
      "loss": 0.768,
      "step": 6426
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.265545606876863,
      "learning_rate": 8.369561452193528e-06,
      "loss": 0.8879,
      "step": 6427
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.291272722547609,
      "learning_rate": 8.369027720929228e-06,
      "loss": 0.7555,
      "step": 6428
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.276792640234346,
      "learning_rate": 8.368493919343849e-06,
      "loss": 0.815,
      "step": 6429
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.101473634051828,
      "learning_rate": 8.36796004744854e-06,
      "loss": 0.8249,
      "step": 6430
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.135722169608094,
      "learning_rate": 8.367426105254438e-06,
      "loss": 0.7666,
      "step": 6431
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.58591992493518,
      "learning_rate": 8.366892092772694e-06,
      "loss": 0.7901,
      "step": 6432
    },
    {
      "epoch": 0.57,
      "grad_norm": 9.911933853041136,
      "learning_rate": 8.366358010014449e-06,
      "loss": 0.7422,
      "step": 6433
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.384447259282951,
      "learning_rate": 8.365823856990851e-06,
      "loss": 0.8115,
      "step": 6434
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.667093295584337,
      "learning_rate": 8.365289633713054e-06,
      "loss": 0.8053,
      "step": 6435
    },
    {
      "epoch": 0.57,
      "grad_norm": 9.936444051628119,
      "learning_rate": 8.364755340192203e-06,
      "loss": 0.8635,
      "step": 6436
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.997035653612605,
      "learning_rate": 8.364220976439454e-06,
      "loss": 0.8575,
      "step": 6437
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.694355146901909,
      "learning_rate": 8.36368654246596e-06,
      "loss": 0.7686,
      "step": 6438
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.170693255894848,
      "learning_rate": 8.363152038282876e-06,
      "loss": 0.8023,
      "step": 6439
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.491048148406624,
      "learning_rate": 8.362617463901358e-06,
      "loss": 0.8591,
      "step": 6440
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.549260190265516,
      "learning_rate": 8.362082819332564e-06,
      "loss": 0.8572,
      "step": 6441
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.957296621966334,
      "learning_rate": 8.361548104587655e-06,
      "loss": 0.7625,
      "step": 6442
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.736309725661769,
      "learning_rate": 8.36101331967779e-06,
      "loss": 0.813,
      "step": 6443
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.4105505574306285,
      "learning_rate": 8.360478464614133e-06,
      "loss": 0.8315,
      "step": 6444
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.704648734830471,
      "learning_rate": 8.359943539407847e-06,
      "loss": 0.8601,
      "step": 6445
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.937704858071197,
      "learning_rate": 8.3594085440701e-06,
      "loss": 0.7721,
      "step": 6446
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.6311327444161865,
      "learning_rate": 8.358873478612054e-06,
      "loss": 0.7438,
      "step": 6447
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.592319380886831,
      "learning_rate": 8.35833834304488e-06,
      "loss": 0.7907,
      "step": 6448
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.225017609578877,
      "learning_rate": 8.357803137379748e-06,
      "loss": 0.8145,
      "step": 6449
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.991006583485945,
      "learning_rate": 8.35726786162783e-06,
      "loss": 0.882,
      "step": 6450
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.7911869172432064,
      "learning_rate": 8.356732515800296e-06,
      "loss": 0.8249,
      "step": 6451
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.2532772536449395,
      "learning_rate": 8.356197099908324e-06,
      "loss": 0.7157,
      "step": 6452
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.129141585902698,
      "learning_rate": 8.355661613963086e-06,
      "loss": 0.7839,
      "step": 6453
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.150798687943961,
      "learning_rate": 8.35512605797576e-06,
      "loss": 0.812,
      "step": 6454
    },
    {
      "epoch": 0.58,
      "grad_norm": 8.143849079140178,
      "learning_rate": 8.354590431957525e-06,
      "loss": 0.8857,
      "step": 6455
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.959695180633051,
      "learning_rate": 8.354054735919563e-06,
      "loss": 0.8227,
      "step": 6456
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.394192951306144,
      "learning_rate": 8.353518969873053e-06,
      "loss": 0.7313,
      "step": 6457
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.683953597729611,
      "learning_rate": 8.352983133829178e-06,
      "loss": 0.8663,
      "step": 6458
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.824641455153864,
      "learning_rate": 8.352447227799125e-06,
      "loss": 0.8154,
      "step": 6459
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.5643920437432675,
      "learning_rate": 8.351911251794076e-06,
      "loss": 0.756,
      "step": 6460
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.248037237801102,
      "learning_rate": 8.351375205825221e-06,
      "loss": 0.839,
      "step": 6461
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.171551040286376,
      "learning_rate": 8.350839089903746e-06,
      "loss": 0.8323,
      "step": 6462
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.5829477893941215,
      "learning_rate": 8.350302904040844e-06,
      "loss": 0.842,
      "step": 6463
    },
    {
      "epoch": 0.58,
      "grad_norm": 8.535719894210468,
      "learning_rate": 8.349766648247707e-06,
      "loss": 0.794,
      "step": 6464
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.718341444884894,
      "learning_rate": 8.349230322535527e-06,
      "loss": 0.8178,
      "step": 6465
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.599520501679836,
      "learning_rate": 8.348693926915497e-06,
      "loss": 0.8189,
      "step": 6466
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.68648673746061,
      "learning_rate": 8.348157461398818e-06,
      "loss": 0.833,
      "step": 6467
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.044219315964691,
      "learning_rate": 8.347620925996682e-06,
      "loss": 0.7835,
      "step": 6468
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.140841195846728,
      "learning_rate": 8.34708432072029e-06,
      "loss": 0.8235,
      "step": 6469
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.790359148330508,
      "learning_rate": 8.346547645580844e-06,
      "loss": 0.7918,
      "step": 6470
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.638324976636239,
      "learning_rate": 8.346010900589545e-06,
      "loss": 0.8259,
      "step": 6471
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.057890831912723,
      "learning_rate": 8.345474085757594e-06,
      "loss": 0.9297,
      "step": 6472
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.707654564421968,
      "learning_rate": 8.3449372010962e-06,
      "loss": 0.832,
      "step": 6473
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.9385205445662805,
      "learning_rate": 8.344400246616566e-06,
      "loss": 0.797,
      "step": 6474
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.21771592661929,
      "learning_rate": 8.343863222329903e-06,
      "loss": 0.7793,
      "step": 6475
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.41163697706573,
      "learning_rate": 8.343326128247415e-06,
      "loss": 0.7621,
      "step": 6476
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.378320776830714,
      "learning_rate": 8.342788964380317e-06,
      "loss": 0.8205,
      "step": 6477
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.346530145607095,
      "learning_rate": 8.34225173073982e-06,
      "loss": 0.8154,
      "step": 6478
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.54969941555116,
      "learning_rate": 8.34171442733714e-06,
      "loss": 0.8578,
      "step": 6479
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.7280633080433385,
      "learning_rate": 8.341177054183488e-06,
      "loss": 0.8438,
      "step": 6480
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.734055764513203,
      "learning_rate": 8.340639611290081e-06,
      "loss": 0.8043,
      "step": 6481
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.038671822979245,
      "learning_rate": 8.340102098668138e-06,
      "loss": 0.7918,
      "step": 6482
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.531316967489235,
      "learning_rate": 8.33956451632888e-06,
      "loss": 0.7713,
      "step": 6483
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.517084735852784,
      "learning_rate": 8.339026864283524e-06,
      "loss": 0.8289,
      "step": 6484
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.732851261457929,
      "learning_rate": 8.338489142543295e-06,
      "loss": 0.7858,
      "step": 6485
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.912376041210346,
      "learning_rate": 8.337951351119417e-06,
      "loss": 0.766,
      "step": 6486
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.116482406632327,
      "learning_rate": 8.337413490023115e-06,
      "loss": 0.8272,
      "step": 6487
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.40173341219443,
      "learning_rate": 8.336875559265616e-06,
      "loss": 0.7807,
      "step": 6488
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.802428829543586,
      "learning_rate": 8.336337558858145e-06,
      "loss": 0.8545,
      "step": 6489
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.608992645290709,
      "learning_rate": 8.335799488811936e-06,
      "loss": 0.7748,
      "step": 6490
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.2054065211028036,
      "learning_rate": 8.335261349138217e-06,
      "loss": 0.767,
      "step": 6491
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.9980232782685405,
      "learning_rate": 8.33472313984822e-06,
      "loss": 0.739,
      "step": 6492
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.975814378235043,
      "learning_rate": 8.334184860953183e-06,
      "loss": 0.7998,
      "step": 6493
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.983901792952384,
      "learning_rate": 8.333646512464337e-06,
      "loss": 0.7189,
      "step": 6494
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.114748489263203,
      "learning_rate": 8.333108094392921e-06,
      "loss": 0.8162,
      "step": 6495
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.5827630354317535,
      "learning_rate": 8.332569606750174e-06,
      "loss": 0.7994,
      "step": 6496
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.353084675836693,
      "learning_rate": 8.332031049547334e-06,
      "loss": 0.8561,
      "step": 6497
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.734688588517768,
      "learning_rate": 8.331492422795641e-06,
      "loss": 0.8808,
      "step": 6498
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.31009392546542,
      "learning_rate": 8.33095372650634e-06,
      "loss": 0.8726,
      "step": 6499
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.028264353462315,
      "learning_rate": 8.330414960690678e-06,
      "loss": 0.8265,
      "step": 6500
    },
    {
      "epoch": 0.58,
      "grad_norm": 8.61583212432705,
      "learning_rate": 8.329876125359894e-06,
      "loss": 0.8301,
      "step": 6501
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.677444933462211,
      "learning_rate": 8.329337220525239e-06,
      "loss": 0.8233,
      "step": 6502
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.134629603052685,
      "learning_rate": 8.328798246197963e-06,
      "loss": 0.8409,
      "step": 6503
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.529457989698098,
      "learning_rate": 8.328259202389311e-06,
      "loss": 0.8596,
      "step": 6504
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.633697046507348,
      "learning_rate": 8.327720089110536e-06,
      "loss": 0.7416,
      "step": 6505
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.830233972929224,
      "learning_rate": 8.327180906372893e-06,
      "loss": 0.7761,
      "step": 6506
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.583793005061743,
      "learning_rate": 8.326641654187634e-06,
      "loss": 0.8874,
      "step": 6507
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.768832534031234,
      "learning_rate": 8.326102332566018e-06,
      "loss": 0.7787,
      "step": 6508
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.291370712953162,
      "learning_rate": 8.325562941519298e-06,
      "loss": 0.8131,
      "step": 6509
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.625343195972422,
      "learning_rate": 8.325023481058736e-06,
      "loss": 0.8022,
      "step": 6510
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.572211766666693,
      "learning_rate": 8.324483951195587e-06,
      "loss": 0.8231,
      "step": 6511
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.9988273279010595,
      "learning_rate": 8.32394435194112e-06,
      "loss": 0.7395,
      "step": 6512
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.217974979390241,
      "learning_rate": 8.323404683306593e-06,
      "loss": 0.8671,
      "step": 6513
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.915139144135979,
      "learning_rate": 8.32286494530327e-06,
      "loss": 0.8467,
      "step": 6514
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.001017258201932,
      "learning_rate": 8.322325137942419e-06,
      "loss": 0.7665,
      "step": 6515
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.342533541231736,
      "learning_rate": 8.321785261235306e-06,
      "loss": 0.8267,
      "step": 6516
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.3482228951260335,
      "learning_rate": 8.321245315193202e-06,
      "loss": 0.7672,
      "step": 6517
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.744080882937791,
      "learning_rate": 8.320705299827375e-06,
      "loss": 0.7885,
      "step": 6518
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.07888679015369,
      "learning_rate": 8.320165215149094e-06,
      "loss": 0.7703,
      "step": 6519
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.111108903218273,
      "learning_rate": 8.319625061169639e-06,
      "loss": 0.8772,
      "step": 6520
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.708294355193285,
      "learning_rate": 8.319084837900278e-06,
      "loss": 0.7912,
      "step": 6521
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.4843478805659975,
      "learning_rate": 8.318544545352292e-06,
      "loss": 0.889,
      "step": 6522
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.872404165735018,
      "learning_rate": 8.318004183536954e-06,
      "loss": 0.8186,
      "step": 6523
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.873862626441227,
      "learning_rate": 8.317463752465547e-06,
      "loss": 0.7619,
      "step": 6524
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.882270191757441,
      "learning_rate": 8.316923252149346e-06,
      "loss": 0.8179,
      "step": 6525
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.089408342294823,
      "learning_rate": 8.31638268259964e-06,
      "loss": 0.8365,
      "step": 6526
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.304414575466239,
      "learning_rate": 8.315842043827706e-06,
      "loss": 0.7742,
      "step": 6527
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.828734931507414,
      "learning_rate": 8.315301335844832e-06,
      "loss": 0.864,
      "step": 6528
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.059427539851561,
      "learning_rate": 8.314760558662303e-06,
      "loss": 0.7831,
      "step": 6529
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.470687775111974,
      "learning_rate": 8.314219712291404e-06,
      "loss": 0.8481,
      "step": 6530
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.887516896248597,
      "learning_rate": 8.31367879674343e-06,
      "loss": 0.8643,
      "step": 6531
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.948859012274711,
      "learning_rate": 8.313137812029666e-06,
      "loss": 0.7929,
      "step": 6532
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.464864060439881,
      "learning_rate": 8.312596758161407e-06,
      "loss": 0.9184,
      "step": 6533
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.815801619198798,
      "learning_rate": 8.312055635149944e-06,
      "loss": 0.7589,
      "step": 6534
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.192147414731345,
      "learning_rate": 8.311514443006573e-06,
      "loss": 0.8125,
      "step": 6535
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.373342729756691,
      "learning_rate": 8.31097318174259e-06,
      "loss": 0.7977,
      "step": 6536
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.715739803551229,
      "learning_rate": 8.310431851369294e-06,
      "loss": 0.8349,
      "step": 6537
    },
    {
      "epoch": 0.58,
      "grad_norm": 9.481738427196262,
      "learning_rate": 8.309890451897982e-06,
      "loss": 0.8154,
      "step": 6538
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.6560271841873835,
      "learning_rate": 8.309348983339956e-06,
      "loss": 0.7811,
      "step": 6539
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.732504390195648,
      "learning_rate": 8.308807445706516e-06,
      "loss": 0.7952,
      "step": 6540
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.231198196644329,
      "learning_rate": 8.308265839008966e-06,
      "loss": 0.7602,
      "step": 6541
    },
    {
      "epoch": 0.58,
      "grad_norm": 12.410109134962266,
      "learning_rate": 8.307724163258615e-06,
      "loss": 0.834,
      "step": 6542
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.794573680802504,
      "learning_rate": 8.307182418466763e-06,
      "loss": 0.8595,
      "step": 6543
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.485580975051904,
      "learning_rate": 8.30664060464472e-06,
      "loss": 0.7603,
      "step": 6544
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.159299325673465,
      "learning_rate": 8.306098721803798e-06,
      "loss": 0.7333,
      "step": 6545
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.134051897162804,
      "learning_rate": 8.305556769955306e-06,
      "loss": 0.7571,
      "step": 6546
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.4387829989685486,
      "learning_rate": 8.305014749110554e-06,
      "loss": 0.8225,
      "step": 6547
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.976061637504289,
      "learning_rate": 8.304472659280857e-06,
      "loss": 0.7759,
      "step": 6548
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.723480524687842,
      "learning_rate": 8.303930500477529e-06,
      "loss": 0.7435,
      "step": 6549
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.58041545972474,
      "learning_rate": 8.303388272711888e-06,
      "loss": 0.7996,
      "step": 6550
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.20484338679722,
      "learning_rate": 8.302845975995253e-06,
      "loss": 0.8176,
      "step": 6551
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.120971864626568,
      "learning_rate": 8.302303610338938e-06,
      "loss": 0.7057,
      "step": 6552
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.999208906832063,
      "learning_rate": 8.30176117575427e-06,
      "loss": 0.7692,
      "step": 6553
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.665590544651028,
      "learning_rate": 8.301218672252566e-06,
      "loss": 0.803,
      "step": 6554
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.178398959873913,
      "learning_rate": 8.300676099845155e-06,
      "loss": 0.8508,
      "step": 6555
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.723911213655337,
      "learning_rate": 8.300133458543355e-06,
      "loss": 0.7705,
      "step": 6556
    },
    {
      "epoch": 0.58,
      "grad_norm": 11.610498832410464,
      "learning_rate": 8.299590748358498e-06,
      "loss": 0.8026,
      "step": 6557
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.553347931007273,
      "learning_rate": 8.299047969301909e-06,
      "loss": 0.8452,
      "step": 6558
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.582438837790492,
      "learning_rate": 8.298505121384921e-06,
      "loss": 0.7952,
      "step": 6559
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.80838132082343,
      "learning_rate": 8.29796220461886e-06,
      "loss": 0.8029,
      "step": 6560
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.739244175538246,
      "learning_rate": 8.297419219015061e-06,
      "loss": 0.8365,
      "step": 6561
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.942084906265972,
      "learning_rate": 8.296876164584857e-06,
      "loss": 0.8446,
      "step": 6562
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.807822739495269,
      "learning_rate": 8.296333041339583e-06,
      "loss": 0.8448,
      "step": 6563
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.947927735300665,
      "learning_rate": 8.295789849290576e-06,
      "loss": 0.805,
      "step": 6564
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.074972465693616,
      "learning_rate": 8.295246588449173e-06,
      "loss": 0.7948,
      "step": 6565
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.848212603002914,
      "learning_rate": 8.294703258826713e-06,
      "loss": 0.8491,
      "step": 6566
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.899751485370875,
      "learning_rate": 8.29415986043454e-06,
      "loss": 0.8213,
      "step": 6567
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.454746230223329,
      "learning_rate": 8.293616393283993e-06,
      "loss": 0.7672,
      "step": 6568
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.802700134043347,
      "learning_rate": 8.293072857386418e-06,
      "loss": 0.8535,
      "step": 6569
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.926781242722968,
      "learning_rate": 8.292529252753156e-06,
      "loss": 0.8611,
      "step": 6570
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.4874655517034086,
      "learning_rate": 8.291985579395557e-06,
      "loss": 0.8021,
      "step": 6571
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.207468659424794,
      "learning_rate": 8.29144183732497e-06,
      "loss": 0.8298,
      "step": 6572
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.667501823944266,
      "learning_rate": 8.290898026552741e-06,
      "loss": 0.8057,
      "step": 6573
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.953874694255466,
      "learning_rate": 8.290354147090224e-06,
      "loss": 0.7828,
      "step": 6574
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.408994035527277,
      "learning_rate": 8.28981019894877e-06,
      "loss": 0.7925,
      "step": 6575
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.85017060127885,
      "learning_rate": 8.289266182139733e-06,
      "loss": 0.7461,
      "step": 6576
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.279692542980102,
      "learning_rate": 8.288722096674467e-06,
      "loss": 0.7939,
      "step": 6577
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.595810416365484,
      "learning_rate": 8.288177942564328e-06,
      "loss": 0.8551,
      "step": 6578
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.868336945382938,
      "learning_rate": 8.287633719820676e-06,
      "loss": 0.83,
      "step": 6579
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.242467994452417,
      "learning_rate": 8.287089428454871e-06,
      "loss": 0.71,
      "step": 6580
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.779013079652278,
      "learning_rate": 8.286545068478273e-06,
      "loss": 0.857,
      "step": 6581
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.398748400131399,
      "learning_rate": 8.286000639902243e-06,
      "loss": 0.8035,
      "step": 6582
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.7198623724677775,
      "learning_rate": 8.285456142738147e-06,
      "loss": 0.9221,
      "step": 6583
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.759520170769188,
      "learning_rate": 8.284911576997347e-06,
      "loss": 0.884,
      "step": 6584
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.740732078653209,
      "learning_rate": 8.284366942691214e-06,
      "loss": 0.7553,
      "step": 6585
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.713463540220353,
      "learning_rate": 8.283822239831115e-06,
      "loss": 0.835,
      "step": 6586
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.855677004617308,
      "learning_rate": 8.283277468428415e-06,
      "loss": 0.7376,
      "step": 6587
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.858814708665648,
      "learning_rate": 8.282732628494488e-06,
      "loss": 0.7397,
      "step": 6588
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.719057212376901,
      "learning_rate": 8.282187720040709e-06,
      "loss": 0.8219,
      "step": 6589
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.6395074729455414,
      "learning_rate": 8.281642743078446e-06,
      "loss": 0.7813,
      "step": 6590
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.541469521269135,
      "learning_rate": 8.281097697619081e-06,
      "loss": 0.8406,
      "step": 6591
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.884737864752866,
      "learning_rate": 8.280552583673984e-06,
      "loss": 0.8345,
      "step": 6592
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.499590329327538,
      "learning_rate": 8.280007401254538e-06,
      "loss": 0.7671,
      "step": 6593
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.190829480422459,
      "learning_rate": 8.27946215037212e-06,
      "loss": 0.7951,
      "step": 6594
    },
    {
      "epoch": 0.59,
      "grad_norm": 8.274918934765552,
      "learning_rate": 8.27891683103811e-06,
      "loss": 0.7627,
      "step": 6595
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.379027127731314,
      "learning_rate": 8.278371443263896e-06,
      "loss": 0.766,
      "step": 6596
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.635292838572133,
      "learning_rate": 8.277825987060853e-06,
      "loss": 0.9053,
      "step": 6597
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.844070144557614,
      "learning_rate": 8.277280462440373e-06,
      "loss": 0.8411,
      "step": 6598
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.604134346489943,
      "learning_rate": 8.276734869413841e-06,
      "loss": 0.7926,
      "step": 6599
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.785936076991452,
      "learning_rate": 8.276189207992644e-06,
      "loss": 0.8065,
      "step": 6600
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.642862173334842,
      "learning_rate": 8.275643478188171e-06,
      "loss": 0.8644,
      "step": 6601
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.757278464706172,
      "learning_rate": 8.275097680011814e-06,
      "loss": 0.7179,
      "step": 6602
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.2841357704698275,
      "learning_rate": 8.274551813474967e-06,
      "loss": 0.7387,
      "step": 6603
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.11286194695604,
      "learning_rate": 8.274005878589022e-06,
      "loss": 0.8544,
      "step": 6604
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.944702120962299,
      "learning_rate": 8.273459875365373e-06,
      "loss": 0.7966,
      "step": 6605
    },
    {
      "epoch": 0.59,
      "grad_norm": 8.208674474299395,
      "learning_rate": 8.272913803815416e-06,
      "loss": 0.7806,
      "step": 6606
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.782127872674019,
      "learning_rate": 8.272367663950554e-06,
      "loss": 0.7816,
      "step": 6607
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.806825303416563,
      "learning_rate": 8.271821455782183e-06,
      "loss": 0.7986,
      "step": 6608
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.511392308463543,
      "learning_rate": 8.271275179321703e-06,
      "loss": 0.736,
      "step": 6609
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.747936205121432,
      "learning_rate": 8.27072883458052e-06,
      "loss": 0.8258,
      "step": 6610
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.906652403069464,
      "learning_rate": 8.270182421570033e-06,
      "loss": 0.8188,
      "step": 6611
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.375205317797918,
      "learning_rate": 8.26963594030165e-06,
      "loss": 0.8278,
      "step": 6612
    },
    {
      "epoch": 0.59,
      "grad_norm": 8.282678593955008,
      "learning_rate": 8.269089390786777e-06,
      "loss": 0.8085,
      "step": 6613
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.390239736488926,
      "learning_rate": 8.268542773036824e-06,
      "loss": 0.8489,
      "step": 6614
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.3383272549989,
      "learning_rate": 8.267996087063195e-06,
      "loss": 0.8234,
      "step": 6615
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.721851340043212,
      "learning_rate": 8.267449332877306e-06,
      "loss": 0.8013,
      "step": 6616
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.621121792658276,
      "learning_rate": 8.266902510490568e-06,
      "loss": 0.8127,
      "step": 6617
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.682026688576515,
      "learning_rate": 8.266355619914396e-06,
      "loss": 0.8168,
      "step": 6618
    },
    {
      "epoch": 0.59,
      "grad_norm": 9.431706679601259,
      "learning_rate": 8.265808661160202e-06,
      "loss": 0.7328,
      "step": 6619
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.144964951155161,
      "learning_rate": 8.265261634239406e-06,
      "loss": 0.7982,
      "step": 6620
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.766080727778055,
      "learning_rate": 8.264714539163422e-06,
      "loss": 0.7852,
      "step": 6621
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.109365197726513,
      "learning_rate": 8.264167375943672e-06,
      "loss": 0.7644,
      "step": 6622
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.12097901732077,
      "learning_rate": 8.263620144591578e-06,
      "loss": 0.8737,
      "step": 6623
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.777288872528583,
      "learning_rate": 8.263072845118559e-06,
      "loss": 0.8129,
      "step": 6624
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.603961641106768,
      "learning_rate": 8.262525477536041e-06,
      "loss": 0.8181,
      "step": 6625
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.005165253134522,
      "learning_rate": 8.261978041855448e-06,
      "loss": 0.8076,
      "step": 6626
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.721641058798318,
      "learning_rate": 8.261430538088207e-06,
      "loss": 0.8277,
      "step": 6627
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.237456798111674,
      "learning_rate": 8.260882966245747e-06,
      "loss": 0.7833,
      "step": 6628
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.609641387274582,
      "learning_rate": 8.260335326339494e-06,
      "loss": 0.793,
      "step": 6629
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.963122227898637,
      "learning_rate": 8.259787618380883e-06,
      "loss": 0.7623,
      "step": 6630
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.263384809869215,
      "learning_rate": 8.259239842381344e-06,
      "loss": 0.8125,
      "step": 6631
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.880514913643217,
      "learning_rate": 8.258691998352312e-06,
      "loss": 0.7854,
      "step": 6632
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.651145854310553,
      "learning_rate": 8.25814408630522e-06,
      "loss": 0.7953,
      "step": 6633
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.022278224337511,
      "learning_rate": 8.257596106251504e-06,
      "loss": 0.8067,
      "step": 6634
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.653026920230497,
      "learning_rate": 8.257048058202604e-06,
      "loss": 0.8286,
      "step": 6635
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.765057926171833,
      "learning_rate": 8.25649994216996e-06,
      "loss": 0.7636,
      "step": 6636
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.067601513269242,
      "learning_rate": 8.255951758165011e-06,
      "loss": 0.804,
      "step": 6637
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.841170074062014,
      "learning_rate": 8.255403506199197e-06,
      "loss": 0.896,
      "step": 6638
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.597488678848482,
      "learning_rate": 8.254855186283966e-06,
      "loss": 0.8866,
      "step": 6639
    },
    {
      "epoch": 0.59,
      "grad_norm": 8.042683158537608,
      "learning_rate": 8.254306798430762e-06,
      "loss": 0.7454,
      "step": 6640
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.960232751667454,
      "learning_rate": 8.25375834265103e-06,
      "loss": 0.7775,
      "step": 6641
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.99675161414685,
      "learning_rate": 8.253209818956216e-06,
      "loss": 0.7526,
      "step": 6642
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.165206654541226,
      "learning_rate": 8.252661227357775e-06,
      "loss": 0.7703,
      "step": 6643
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.257575390620387,
      "learning_rate": 8.25211256786715e-06,
      "loss": 0.7814,
      "step": 6644
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.5837036576443495,
      "learning_rate": 8.2515638404958e-06,
      "loss": 0.7544,
      "step": 6645
    },
    {
      "epoch": 0.59,
      "grad_norm": 9.352524573543771,
      "learning_rate": 8.251015045255173e-06,
      "loss": 0.8217,
      "step": 6646
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.3024615945389035,
      "learning_rate": 8.25046618215673e-06,
      "loss": 0.8029,
      "step": 6647
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.7904224927651216,
      "learning_rate": 8.249917251211924e-06,
      "loss": 0.7755,
      "step": 6648
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.8479459083797725,
      "learning_rate": 8.249368252432208e-06,
      "loss": 0.8233,
      "step": 6649
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.914573051603051,
      "learning_rate": 8.248819185829049e-06,
      "loss": 0.7875,
      "step": 6650
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.206075963016387,
      "learning_rate": 8.248270051413905e-06,
      "loss": 0.7594,
      "step": 6651
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.489629207134546,
      "learning_rate": 8.247720849198235e-06,
      "loss": 0.7963,
      "step": 6652
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.3653055692873926,
      "learning_rate": 8.247171579193505e-06,
      "loss": 0.8408,
      "step": 6653
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.459434549544171,
      "learning_rate": 8.24662224141118e-06,
      "loss": 0.7536,
      "step": 6654
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.951661783722108,
      "learning_rate": 8.246072835862726e-06,
      "loss": 0.7502,
      "step": 6655
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.73667383494703,
      "learning_rate": 8.245523362559611e-06,
      "loss": 0.7662,
      "step": 6656
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.066635551806379,
      "learning_rate": 8.244973821513302e-06,
      "loss": 0.8196,
      "step": 6657
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.3349799020133135,
      "learning_rate": 8.244424212735272e-06,
      "loss": 0.8205,
      "step": 6658
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.136730473794988,
      "learning_rate": 8.24387453623699e-06,
      "loss": 0.8338,
      "step": 6659
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.026718960353829,
      "learning_rate": 8.243324792029931e-06,
      "loss": 0.8125,
      "step": 6660
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.98620905558427,
      "learning_rate": 8.242774980125573e-06,
      "loss": 0.8538,
      "step": 6661
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.807019931117183,
      "learning_rate": 8.242225100535386e-06,
      "loss": 0.8561,
      "step": 6662
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.631264662860794,
      "learning_rate": 8.241675153270852e-06,
      "loss": 0.774,
      "step": 6663
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.805289238645786,
      "learning_rate": 8.241125138343447e-06,
      "loss": 0.7928,
      "step": 6664
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.050954164609158,
      "learning_rate": 8.240575055764653e-06,
      "loss": 0.8406,
      "step": 6665
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.2786998129264875,
      "learning_rate": 8.24002490554595e-06,
      "loss": 0.8402,
      "step": 6666
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.003252502321802,
      "learning_rate": 8.239474687698825e-06,
      "loss": 0.7612,
      "step": 6667
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.67285773600483,
      "learning_rate": 8.23892440223476e-06,
      "loss": 0.8109,
      "step": 6668
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.834436215495525,
      "learning_rate": 8.23837404916524e-06,
      "loss": 0.7623,
      "step": 6669
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.312077964674626,
      "learning_rate": 8.237823628501756e-06,
      "loss": 0.7683,
      "step": 6670
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.041893802694508,
      "learning_rate": 8.237273140255792e-06,
      "loss": 0.8054,
      "step": 6671
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.309996305541788,
      "learning_rate": 8.236722584438841e-06,
      "loss": 0.7876,
      "step": 6672
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.907666052710539,
      "learning_rate": 8.236171961062394e-06,
      "loss": 0.7323,
      "step": 6673
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.5653490628792825,
      "learning_rate": 8.235621270137945e-06,
      "loss": 0.8625,
      "step": 6674
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.005983595096594,
      "learning_rate": 8.235070511676989e-06,
      "loss": 0.7886,
      "step": 6675
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.04460968680593,
      "learning_rate": 8.23451968569102e-06,
      "loss": 0.7985,
      "step": 6676
    },
    {
      "epoch": 0.6,
      "grad_norm": 9.962846117098081,
      "learning_rate": 8.233968792191535e-06,
      "loss": 0.8491,
      "step": 6677
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.7943553165130695,
      "learning_rate": 8.233417831190036e-06,
      "loss": 0.8181,
      "step": 6678
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.687193703900281,
      "learning_rate": 8.232866802698019e-06,
      "loss": 0.8123,
      "step": 6679
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.871630135025604,
      "learning_rate": 8.232315706726988e-06,
      "loss": 0.7926,
      "step": 6680
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.588484284722919,
      "learning_rate": 8.231764543288445e-06,
      "loss": 0.8622,
      "step": 6681
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.3656298795301165,
      "learning_rate": 8.231213312393895e-06,
      "loss": 0.7683,
      "step": 6682
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.2690455673925625,
      "learning_rate": 8.230662014054841e-06,
      "loss": 0.7745,
      "step": 6683
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.109352543813534,
      "learning_rate": 8.230110648282797e-06,
      "loss": 0.7916,
      "step": 6684
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.03225231892724,
      "learning_rate": 8.229559215089264e-06,
      "loss": 0.8575,
      "step": 6685
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.516175683311602,
      "learning_rate": 8.229007714485755e-06,
      "loss": 0.8144,
      "step": 6686
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.817765330982432,
      "learning_rate": 8.228456146483781e-06,
      "loss": 0.829,
      "step": 6687
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.103612679052471,
      "learning_rate": 8.227904511094858e-06,
      "loss": 0.7881,
      "step": 6688
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.809276732097919,
      "learning_rate": 8.227352808330495e-06,
      "loss": 0.8235,
      "step": 6689
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.3951427517731,
      "learning_rate": 8.22680103820221e-06,
      "loss": 0.774,
      "step": 6690
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.068680907143205,
      "learning_rate": 8.226249200721521e-06,
      "loss": 0.7712,
      "step": 6691
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.945854506324684,
      "learning_rate": 8.225697295899944e-06,
      "loss": 0.8623,
      "step": 6692
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.068690926975485,
      "learning_rate": 8.225145323749e-06,
      "loss": 0.8218,
      "step": 6693
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.559512495772458,
      "learning_rate": 8.224593284280212e-06,
      "loss": 0.7543,
      "step": 6694
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.843239391864627,
      "learning_rate": 8.224041177505097e-06,
      "loss": 0.8194,
      "step": 6695
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.371643149671071,
      "learning_rate": 8.223489003435188e-06,
      "loss": 0.8096,
      "step": 6696
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.565507795930152,
      "learning_rate": 8.222936762082003e-06,
      "loss": 0.8431,
      "step": 6697
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.050620507692319,
      "learning_rate": 8.222384453457071e-06,
      "loss": 0.7427,
      "step": 6698
    },
    {
      "epoch": 0.6,
      "grad_norm": 8.695908403104138,
      "learning_rate": 8.22183207757192e-06,
      "loss": 0.7872,
      "step": 6699
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.7861713277509152,
      "learning_rate": 8.221279634438082e-06,
      "loss": 0.7969,
      "step": 6700
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.910134269406934,
      "learning_rate": 8.220727124067085e-06,
      "loss": 0.8585,
      "step": 6701
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.651640446394922,
      "learning_rate": 8.220174546470462e-06,
      "loss": 0.7407,
      "step": 6702
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.748679608248464,
      "learning_rate": 8.219621901659746e-06,
      "loss": 0.7871,
      "step": 6703
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.9950266831591845,
      "learning_rate": 8.219069189646476e-06,
      "loss": 0.8275,
      "step": 6704
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.492172347895253,
      "learning_rate": 8.218516410442188e-06,
      "loss": 0.7952,
      "step": 6705
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.634356187379448,
      "learning_rate": 8.217963564058415e-06,
      "loss": 0.8303,
      "step": 6706
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.120977597631844,
      "learning_rate": 8.217410650506702e-06,
      "loss": 0.812,
      "step": 6707
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.693817833452381,
      "learning_rate": 8.216857669798586e-06,
      "loss": 0.8169,
      "step": 6708
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.748880775794269,
      "learning_rate": 8.216304621945612e-06,
      "loss": 0.8332,
      "step": 6709
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.251488163643184,
      "learning_rate": 8.215751506959322e-06,
      "loss": 0.8978,
      "step": 6710
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.850339033425616,
      "learning_rate": 8.21519832485126e-06,
      "loss": 0.8384,
      "step": 6711
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.213198379777557,
      "learning_rate": 8.214645075632979e-06,
      "loss": 0.7318,
      "step": 6712
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.686815530653178,
      "learning_rate": 8.214091759316018e-06,
      "loss": 0.8532,
      "step": 6713
    },
    {
      "epoch": 0.6,
      "grad_norm": 8.030411411361563,
      "learning_rate": 8.213538375911932e-06,
      "loss": 0.8281,
      "step": 6714
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.429598453757664,
      "learning_rate": 8.212984925432269e-06,
      "loss": 0.7948,
      "step": 6715
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.908624621308087,
      "learning_rate": 8.212431407888583e-06,
      "loss": 0.8071,
      "step": 6716
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.787754853773987,
      "learning_rate": 8.211877823292425e-06,
      "loss": 0.7795,
      "step": 6717
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.8359609715332157,
      "learning_rate": 8.211324171655353e-06,
      "loss": 0.7596,
      "step": 6718
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.6647921998729585,
      "learning_rate": 8.210770452988921e-06,
      "loss": 0.8103,
      "step": 6719
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.0627295307032085,
      "learning_rate": 8.210216667304687e-06,
      "loss": 0.8575,
      "step": 6720
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.4554958528325725,
      "learning_rate": 8.20966281461421e-06,
      "loss": 0.833,
      "step": 6721
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.80899943466147,
      "learning_rate": 8.20910889492905e-06,
      "loss": 0.8324,
      "step": 6722
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.014812576735678,
      "learning_rate": 8.20855490826077e-06,
      "loss": 0.8156,
      "step": 6723
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.317628828456329,
      "learning_rate": 8.208000854620934e-06,
      "loss": 0.8244,
      "step": 6724
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.876650035677475,
      "learning_rate": 8.207446734021105e-06,
      "loss": 0.7035,
      "step": 6725
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.364311990472134,
      "learning_rate": 8.206892546472851e-06,
      "loss": 0.7942,
      "step": 6726
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.692485332649723,
      "learning_rate": 8.206338291987736e-06,
      "loss": 0.8462,
      "step": 6727
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.9780671260189995,
      "learning_rate": 8.205783970577333e-06,
      "loss": 0.868,
      "step": 6728
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.814032493004836,
      "learning_rate": 8.20522958225321e-06,
      "loss": 0.884,
      "step": 6729
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.162143467255809,
      "learning_rate": 8.204675127026935e-06,
      "loss": 0.7841,
      "step": 6730
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.948846005157953,
      "learning_rate": 8.204120604910088e-06,
      "loss": 0.7491,
      "step": 6731
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.936337767331207,
      "learning_rate": 8.20356601591424e-06,
      "loss": 0.8757,
      "step": 6732
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.661185428597538,
      "learning_rate": 8.203011360050967e-06,
      "loss": 0.8267,
      "step": 6733
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.9811674016598415,
      "learning_rate": 8.202456637331846e-06,
      "loss": 0.8248,
      "step": 6734
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.988377044071307,
      "learning_rate": 8.201901847768456e-06,
      "loss": 0.8355,
      "step": 6735
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.1582446685505685,
      "learning_rate": 8.201346991372379e-06,
      "loss": 0.8723,
      "step": 6736
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.540308234829635,
      "learning_rate": 8.200792068155192e-06,
      "loss": 0.8265,
      "step": 6737
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.5769686675761765,
      "learning_rate": 8.20023707812848e-06,
      "loss": 0.7978,
      "step": 6738
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.218094294102233,
      "learning_rate": 8.199682021303829e-06,
      "loss": 0.7873,
      "step": 6739
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.420048223544867,
      "learning_rate": 8.199126897692823e-06,
      "loss": 0.7766,
      "step": 6740
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.785156796671875,
      "learning_rate": 8.198571707307048e-06,
      "loss": 0.8748,
      "step": 6741
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.455397181145248,
      "learning_rate": 8.198016450158093e-06,
      "loss": 0.782,
      "step": 6742
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.134533703428839,
      "learning_rate": 8.19746112625755e-06,
      "loss": 0.8322,
      "step": 6743
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.079899482854342,
      "learning_rate": 8.196905735617005e-06,
      "loss": 0.7369,
      "step": 6744
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.162233867231676,
      "learning_rate": 8.196350278248056e-06,
      "loss": 0.8376,
      "step": 6745
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.772519797278695,
      "learning_rate": 8.195794754162294e-06,
      "loss": 0.8033,
      "step": 6746
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.726606906995816,
      "learning_rate": 8.195239163371316e-06,
      "loss": 0.7983,
      "step": 6747
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.691049559504579,
      "learning_rate": 8.194683505886718e-06,
      "loss": 0.8081,
      "step": 6748
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.723086282878302,
      "learning_rate": 8.194127781720097e-06,
      "loss": 0.8912,
      "step": 6749
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.756419890225413,
      "learning_rate": 8.193571990883056e-06,
      "loss": 0.8155,
      "step": 6750
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.007087562229614,
      "learning_rate": 8.19301613338719e-06,
      "loss": 0.757,
      "step": 6751
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.714328572336348,
      "learning_rate": 8.192460209244108e-06,
      "loss": 0.8089,
      "step": 6752
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.549118555905508,
      "learning_rate": 8.19190421846541e-06,
      "loss": 0.8166,
      "step": 6753
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.496844533475255,
      "learning_rate": 8.1913481610627e-06,
      "loss": 0.7505,
      "step": 6754
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.722586243374113,
      "learning_rate": 8.190792037047587e-06,
      "loss": 0.7691,
      "step": 6755
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.805282775228774,
      "learning_rate": 8.190235846431678e-06,
      "loss": 0.8194,
      "step": 6756
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.547130797908869,
      "learning_rate": 8.189679589226582e-06,
      "loss": 0.7464,
      "step": 6757
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.972212872332395,
      "learning_rate": 8.18912326544391e-06,
      "loss": 0.8163,
      "step": 6758
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.2739354416947055,
      "learning_rate": 8.188566875095274e-06,
      "loss": 0.8682,
      "step": 6759
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.997055205978728,
      "learning_rate": 8.188010418192286e-06,
      "loss": 0.8246,
      "step": 6760
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.8199179865655974,
      "learning_rate": 8.187453894746564e-06,
      "loss": 0.7799,
      "step": 6761
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.922491042635821,
      "learning_rate": 8.186897304769722e-06,
      "loss": 0.8579,
      "step": 6762
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.550495885303634,
      "learning_rate": 8.186340648273377e-06,
      "loss": 0.8568,
      "step": 6763
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.150879273873694,
      "learning_rate": 8.18578392526915e-06,
      "loss": 0.8603,
      "step": 6764
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.929845725570779,
      "learning_rate": 8.18522713576866e-06,
      "loss": 0.775,
      "step": 6765
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.025039800120546,
      "learning_rate": 8.184670279783528e-06,
      "loss": 0.8393,
      "step": 6766
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.0803708695159875,
      "learning_rate": 8.18411335732538e-06,
      "loss": 0.7927,
      "step": 6767
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.728429169558053,
      "learning_rate": 8.183556368405836e-06,
      "loss": 0.7339,
      "step": 6768
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.0377435319472195,
      "learning_rate": 8.182999313036528e-06,
      "loss": 0.8049,
      "step": 6769
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.44357289244365,
      "learning_rate": 8.182442191229078e-06,
      "loss": 0.7969,
      "step": 6770
    },
    {
      "epoch": 0.6,
      "grad_norm": 8.514304423351478,
      "learning_rate": 8.181885002995118e-06,
      "loss": 0.8213,
      "step": 6771
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.891871592294824,
      "learning_rate": 8.181327748346275e-06,
      "loss": 0.7534,
      "step": 6772
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.741781741770471,
      "learning_rate": 8.180770427294182e-06,
      "loss": 0.7519,
      "step": 6773
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.824199954517697,
      "learning_rate": 8.180213039850475e-06,
      "loss": 0.861,
      "step": 6774
    },
    {
      "epoch": 0.6,
      "grad_norm": 8.077784606921792,
      "learning_rate": 8.179655586026783e-06,
      "loss": 0.7766,
      "step": 6775
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.857431538899155,
      "learning_rate": 8.179098065834745e-06,
      "loss": 0.8575,
      "step": 6776
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.907148971750793,
      "learning_rate": 8.178540479285996e-06,
      "loss": 0.7528,
      "step": 6777
    },
    {
      "epoch": 0.6,
      "grad_norm": 10.173039719142778,
      "learning_rate": 8.177982826392177e-06,
      "loss": 0.7967,
      "step": 6778
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.143584074938996,
      "learning_rate": 8.177425107164925e-06,
      "loss": 0.8233,
      "step": 6779
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.122294221530041,
      "learning_rate": 8.176867321615882e-06,
      "loss": 0.8168,
      "step": 6780
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.092034472665681,
      "learning_rate": 8.176309469756691e-06,
      "loss": 0.7618,
      "step": 6781
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.702818327295792,
      "learning_rate": 8.175751551598995e-06,
      "loss": 0.7966,
      "step": 6782
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.110430875160217,
      "learning_rate": 8.175193567154442e-06,
      "loss": 0.8258,
      "step": 6783
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.509539918378062,
      "learning_rate": 8.174635516434677e-06,
      "loss": 0.789,
      "step": 6784
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.562237676943552,
      "learning_rate": 8.174077399451348e-06,
      "loss": 0.8354,
      "step": 6785
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.094386956191116,
      "learning_rate": 8.173519216216103e-06,
      "loss": 0.7404,
      "step": 6786
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.5645275635483555,
      "learning_rate": 8.172960966740594e-06,
      "loss": 0.8072,
      "step": 6787
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.650625379002032,
      "learning_rate": 8.172402651036474e-06,
      "loss": 0.7476,
      "step": 6788
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.971573122752251,
      "learning_rate": 8.171844269115397e-06,
      "loss": 0.8244,
      "step": 6789
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.977950625599968,
      "learning_rate": 8.171285820989015e-06,
      "loss": 0.7774,
      "step": 6790
    },
    {
      "epoch": 0.61,
      "grad_norm": 9.290647430441261,
      "learning_rate": 8.170727306668988e-06,
      "loss": 0.808,
      "step": 6791
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.563718367702113,
      "learning_rate": 8.170168726166974e-06,
      "loss": 0.7826,
      "step": 6792
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.600539263370696,
      "learning_rate": 8.169610079494628e-06,
      "loss": 0.8571,
      "step": 6793
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.083525114299053,
      "learning_rate": 8.169051366663612e-06,
      "loss": 0.7563,
      "step": 6794
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.806583175030543,
      "learning_rate": 8.16849258768559e-06,
      "loss": 0.8494,
      "step": 6795
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.210277699486852,
      "learning_rate": 8.167933742572225e-06,
      "loss": 0.7641,
      "step": 6796
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.462452776356719,
      "learning_rate": 8.16737483133518e-06,
      "loss": 0.789,
      "step": 6797
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.2495499126198775,
      "learning_rate": 8.166815853986123e-06,
      "loss": 0.784,
      "step": 6798
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.473298786178799,
      "learning_rate": 8.16625681053672e-06,
      "loss": 0.768,
      "step": 6799
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.556264823653086,
      "learning_rate": 8.165697700998638e-06,
      "loss": 0.8124,
      "step": 6800
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.7636546967405575,
      "learning_rate": 8.16513852538355e-06,
      "loss": 0.7941,
      "step": 6801
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.364103953875404,
      "learning_rate": 8.164579283703127e-06,
      "loss": 0.7869,
      "step": 6802
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.577240186213648,
      "learning_rate": 8.164019975969043e-06,
      "loss": 0.8963,
      "step": 6803
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.090283014326774,
      "learning_rate": 8.16346060219297e-06,
      "loss": 0.7774,
      "step": 6804
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.241281539375794,
      "learning_rate": 8.162901162386586e-06,
      "loss": 0.8818,
      "step": 6805
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.144286080216064,
      "learning_rate": 8.162341656561565e-06,
      "loss": 0.8013,
      "step": 6806
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.877036976707335,
      "learning_rate": 8.161782084729588e-06,
      "loss": 0.8055,
      "step": 6807
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.668694898820878,
      "learning_rate": 8.161222446902334e-06,
      "loss": 0.8517,
      "step": 6808
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.32381708234665,
      "learning_rate": 8.160662743091485e-06,
      "loss": 0.8482,
      "step": 6809
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.426583113177476,
      "learning_rate": 8.160102973308723e-06,
      "loss": 0.8474,
      "step": 6810
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.504120440284222,
      "learning_rate": 8.159543137565731e-06,
      "loss": 0.8336,
      "step": 6811
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.303393997002978,
      "learning_rate": 8.158983235874196e-06,
      "loss": 0.7589,
      "step": 6812
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.827552303619555,
      "learning_rate": 8.158423268245805e-06,
      "loss": 0.8221,
      "step": 6813
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.509162292212347,
      "learning_rate": 8.157863234692244e-06,
      "loss": 0.8244,
      "step": 6814
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.1233057919885185,
      "learning_rate": 8.157303135225204e-06,
      "loss": 0.798,
      "step": 6815
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.495505914142259,
      "learning_rate": 8.156742969856375e-06,
      "loss": 0.7629,
      "step": 6816
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.231185287709736,
      "learning_rate": 8.156182738597449e-06,
      "loss": 0.9044,
      "step": 6817
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.220836908956301,
      "learning_rate": 8.155622441460122e-06,
      "loss": 0.7852,
      "step": 6818
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.041208639860685,
      "learning_rate": 8.155062078456085e-06,
      "loss": 0.7411,
      "step": 6819
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.912235470977246,
      "learning_rate": 8.15450164959704e-06,
      "loss": 0.8394,
      "step": 6820
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.420277143623738,
      "learning_rate": 8.153941154894678e-06,
      "loss": 0.7558,
      "step": 6821
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.861708346173799,
      "learning_rate": 8.153380594360702e-06,
      "loss": 0.741,
      "step": 6822
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.84953062076325,
      "learning_rate": 8.152819968006812e-06,
      "loss": 0.7214,
      "step": 6823
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.7404295635205695,
      "learning_rate": 8.15225927584471e-06,
      "loss": 0.8035,
      "step": 6824
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.575158836595933,
      "learning_rate": 8.1516985178861e-06,
      "loss": 0.7956,
      "step": 6825
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.075126304521108,
      "learning_rate": 8.151137694142684e-06,
      "loss": 0.7828,
      "step": 6826
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.2820124381509865,
      "learning_rate": 8.15057680462617e-06,
      "loss": 0.7978,
      "step": 6827
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.983107723039327,
      "learning_rate": 8.150015849348265e-06,
      "loss": 0.8165,
      "step": 6828
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.273383760305386,
      "learning_rate": 8.149454828320678e-06,
      "loss": 0.7939,
      "step": 6829
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.251481852111514,
      "learning_rate": 8.148893741555118e-06,
      "loss": 0.7814,
      "step": 6830
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.533898758373996,
      "learning_rate": 8.148332589063297e-06,
      "loss": 0.7927,
      "step": 6831
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.315357139495274,
      "learning_rate": 8.147771370856929e-06,
      "loss": 0.8732,
      "step": 6832
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.888643359440391,
      "learning_rate": 8.147210086947725e-06,
      "loss": 0.763,
      "step": 6833
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.143635920893837,
      "learning_rate": 8.146648737347404e-06,
      "loss": 0.7956,
      "step": 6834
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.103803573234262,
      "learning_rate": 8.146087322067682e-06,
      "loss": 0.8029,
      "step": 6835
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.985284850030705,
      "learning_rate": 8.145525841120278e-06,
      "loss": 0.7638,
      "step": 6836
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.5194660585594955,
      "learning_rate": 8.144964294516909e-06,
      "loss": 0.7877,
      "step": 6837
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.520263103842538,
      "learning_rate": 8.144402682269296e-06,
      "loss": 0.8234,
      "step": 6838
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.9904163951965215,
      "learning_rate": 8.143841004389168e-06,
      "loss": 0.7854,
      "step": 6839
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.369234158703639,
      "learning_rate": 8.14327926088824e-06,
      "loss": 0.7867,
      "step": 6840
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.690422618381915,
      "learning_rate": 8.142717451778242e-06,
      "loss": 0.8314,
      "step": 6841
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.6109704626641514,
      "learning_rate": 8.1421555770709e-06,
      "loss": 0.816,
      "step": 6842
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.839103517246004,
      "learning_rate": 8.14159363677794e-06,
      "loss": 0.7963,
      "step": 6843
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.719017452422228,
      "learning_rate": 8.141031630911095e-06,
      "loss": 0.7266,
      "step": 6844
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.59964376667567,
      "learning_rate": 8.140469559482092e-06,
      "loss": 0.7151,
      "step": 6845
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.648427732564185,
      "learning_rate": 8.139907422502666e-06,
      "loss": 0.8053,
      "step": 6846
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.824229723552528,
      "learning_rate": 8.139345219984546e-06,
      "loss": 0.808,
      "step": 6847
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.963415150439329,
      "learning_rate": 8.138782951939473e-06,
      "loss": 0.8125,
      "step": 6848
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.611454503742912,
      "learning_rate": 8.138220618379179e-06,
      "loss": 0.8663,
      "step": 6849
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.904752700901874,
      "learning_rate": 8.1376582193154e-06,
      "loss": 0.8021,
      "step": 6850
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.725328702630351,
      "learning_rate": 8.137095754759878e-06,
      "loss": 0.8336,
      "step": 6851
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.6699802046968335,
      "learning_rate": 8.136533224724354e-06,
      "loss": 0.8291,
      "step": 6852
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.314859010449418,
      "learning_rate": 8.135970629220567e-06,
      "loss": 0.8737,
      "step": 6853
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.622577967365904,
      "learning_rate": 8.135407968260258e-06,
      "loss": 0.853,
      "step": 6854
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.950566508529718,
      "learning_rate": 8.134845241855178e-06,
      "loss": 0.9329,
      "step": 6855
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.779800606728138,
      "learning_rate": 8.134282450017067e-06,
      "loss": 0.7486,
      "step": 6856
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.164634793368161,
      "learning_rate": 8.133719592757673e-06,
      "loss": 0.8316,
      "step": 6857
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.662442881375417,
      "learning_rate": 8.133156670088747e-06,
      "loss": 0.8398,
      "step": 6858
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.357576422627782,
      "learning_rate": 8.132593682022037e-06,
      "loss": 0.7757,
      "step": 6859
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.582341819156483,
      "learning_rate": 8.132030628569292e-06,
      "loss": 0.7918,
      "step": 6860
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.50008046140628,
      "learning_rate": 8.131467509742267e-06,
      "loss": 0.7525,
      "step": 6861
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.272915891838216,
      "learning_rate": 8.130904325552717e-06,
      "loss": 0.7832,
      "step": 6862
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.646948385609108,
      "learning_rate": 8.130341076012396e-06,
      "loss": 0.8402,
      "step": 6863
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.2661541941171635,
      "learning_rate": 8.12977776113306e-06,
      "loss": 0.8862,
      "step": 6864
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.003583840613984,
      "learning_rate": 8.129214380926466e-06,
      "loss": 0.8464,
      "step": 6865
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.381261050874977,
      "learning_rate": 8.128650935404376e-06,
      "loss": 0.8136,
      "step": 6866
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.54366513678714,
      "learning_rate": 8.128087424578548e-06,
      "loss": 0.8031,
      "step": 6867
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.590172837204918,
      "learning_rate": 8.127523848460746e-06,
      "loss": 0.7808,
      "step": 6868
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.662201052328477,
      "learning_rate": 8.126960207062734e-06,
      "loss": 0.824,
      "step": 6869
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.5167848768471295,
      "learning_rate": 8.126396500396275e-06,
      "loss": 0.7918,
      "step": 6870
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.81065370020721,
      "learning_rate": 8.125832728473136e-06,
      "loss": 0.8308,
      "step": 6871
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.127690825568373,
      "learning_rate": 8.125268891305083e-06,
      "loss": 0.7917,
      "step": 6872
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.312659700105328,
      "learning_rate": 8.124704988903886e-06,
      "loss": 0.7825,
      "step": 6873
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.27607470677544,
      "learning_rate": 8.124141021281318e-06,
      "loss": 0.7204,
      "step": 6874
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.275915745151287,
      "learning_rate": 8.123576988449144e-06,
      "loss": 0.7385,
      "step": 6875
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.913608185297566,
      "learning_rate": 8.123012890419145e-06,
      "loss": 0.8451,
      "step": 6876
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.371533331276638,
      "learning_rate": 8.122448727203088e-06,
      "loss": 0.8647,
      "step": 6877
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.801931362257748,
      "learning_rate": 8.121884498812755e-06,
      "loss": 0.833,
      "step": 6878
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.780468494552725,
      "learning_rate": 8.121320205259918e-06,
      "loss": 0.802,
      "step": 6879
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.546743786599118,
      "learning_rate": 8.120755846556356e-06,
      "loss": 0.8544,
      "step": 6880
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.154814399333846,
      "learning_rate": 8.120191422713852e-06,
      "loss": 0.7827,
      "step": 6881
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.4609596957640045,
      "learning_rate": 8.119626933744185e-06,
      "loss": 0.7515,
      "step": 6882
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.115323064992698,
      "learning_rate": 8.119062379659135e-06,
      "loss": 0.8391,
      "step": 6883
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.339099166158037,
      "learning_rate": 8.118497760470492e-06,
      "loss": 0.8104,
      "step": 6884
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.640247223517294,
      "learning_rate": 8.117933076190037e-06,
      "loss": 0.7984,
      "step": 6885
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.2658884939225175,
      "learning_rate": 8.117368326829555e-06,
      "loss": 0.7895,
      "step": 6886
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.997831738657247,
      "learning_rate": 8.116803512400836e-06,
      "loss": 0.8183,
      "step": 6887
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.690688217996239,
      "learning_rate": 8.116238632915672e-06,
      "loss": 0.7754,
      "step": 6888
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.315484525702357,
      "learning_rate": 8.11567368838585e-06,
      "loss": 0.7402,
      "step": 6889
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.346423521077991,
      "learning_rate": 8.115108678823163e-06,
      "loss": 0.7768,
      "step": 6890
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.013819994425285,
      "learning_rate": 8.114543604239404e-06,
      "loss": 0.7731,
      "step": 6891
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.631999179450564,
      "learning_rate": 8.113978464646368e-06,
      "loss": 0.7658,
      "step": 6892
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.443977734159103,
      "learning_rate": 8.113413260055852e-06,
      "loss": 0.8557,
      "step": 6893
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.393525082723005,
      "learning_rate": 8.11284799047965e-06,
      "loss": 0.7758,
      "step": 6894
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.81174470702288,
      "learning_rate": 8.112282655929566e-06,
      "loss": 0.7732,
      "step": 6895
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.599158599692469,
      "learning_rate": 8.111717256417396e-06,
      "loss": 0.7886,
      "step": 6896
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.996027548143773,
      "learning_rate": 8.111151791954944e-06,
      "loss": 0.8364,
      "step": 6897
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.633620807994186,
      "learning_rate": 8.11058626255401e-06,
      "loss": 0.8174,
      "step": 6898
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.92331886638172,
      "learning_rate": 8.110020668226401e-06,
      "loss": 0.8033,
      "step": 6899
    },
    {
      "epoch": 0.62,
      "grad_norm": 10.210251747134011,
      "learning_rate": 8.109455008983919e-06,
      "loss": 0.8682,
      "step": 6900
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.153554570934183,
      "learning_rate": 8.108889284838377e-06,
      "loss": 0.8647,
      "step": 6901
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.095209242551938,
      "learning_rate": 8.108323495801577e-06,
      "loss": 0.7837,
      "step": 6902
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.081346769070026,
      "learning_rate": 8.107757641885333e-06,
      "loss": 0.8002,
      "step": 6903
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.8573624924735475,
      "learning_rate": 8.107191723101452e-06,
      "loss": 0.8411,
      "step": 6904
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.0430957202438815,
      "learning_rate": 8.106625739461748e-06,
      "loss": 0.7544,
      "step": 6905
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.17503847955542,
      "learning_rate": 8.106059690978038e-06,
      "loss": 0.7827,
      "step": 6906
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.664366730544066,
      "learning_rate": 8.105493577662131e-06,
      "loss": 0.824,
      "step": 6907
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.557421645602761,
      "learning_rate": 8.104927399525847e-06,
      "loss": 0.8958,
      "step": 6908
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.954758798922541,
      "learning_rate": 8.104361156581006e-06,
      "loss": 0.7815,
      "step": 6909
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.7029777278349165,
      "learning_rate": 8.103794848839422e-06,
      "loss": 0.8143,
      "step": 6910
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.124038698323085,
      "learning_rate": 8.103228476312919e-06,
      "loss": 0.667,
      "step": 6911
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.279534586050763,
      "learning_rate": 8.102662039013317e-06,
      "loss": 0.8412,
      "step": 6912
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.205922362220809,
      "learning_rate": 8.102095536952438e-06,
      "loss": 0.8195,
      "step": 6913
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.988199418043433,
      "learning_rate": 8.10152897014211e-06,
      "loss": 0.7739,
      "step": 6914
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.073357368037343,
      "learning_rate": 8.100962338594157e-06,
      "loss": 0.8021,
      "step": 6915
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.990205614124662,
      "learning_rate": 8.100395642320406e-06,
      "loss": 0.7871,
      "step": 6916
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.394887112714275,
      "learning_rate": 8.099828881332686e-06,
      "loss": 0.7476,
      "step": 6917
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.768547687627983,
      "learning_rate": 8.099262055642824e-06,
      "loss": 0.7648,
      "step": 6918
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.9642161602317705,
      "learning_rate": 8.098695165262656e-06,
      "loss": 0.8223,
      "step": 6919
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.796981725855802,
      "learning_rate": 8.098128210204012e-06,
      "loss": 0.7987,
      "step": 6920
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.702603305336524,
      "learning_rate": 8.097561190478728e-06,
      "loss": 0.8386,
      "step": 6921
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.533722035063568,
      "learning_rate": 8.096994106098636e-06,
      "loss": 0.8168,
      "step": 6922
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.135957973632429,
      "learning_rate": 8.096426957075574e-06,
      "loss": 0.7984,
      "step": 6923
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.866694709323272,
      "learning_rate": 8.095859743421381e-06,
      "loss": 0.81,
      "step": 6924
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.32213808354182,
      "learning_rate": 8.095292465147897e-06,
      "loss": 0.789,
      "step": 6925
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.7277875063639785,
      "learning_rate": 8.09472512226696e-06,
      "loss": 0.8294,
      "step": 6926
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.545443037200798,
      "learning_rate": 8.094157714790413e-06,
      "loss": 0.8521,
      "step": 6927
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.61571860077136,
      "learning_rate": 8.0935902427301e-06,
      "loss": 0.8807,
      "step": 6928
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.322962944322857,
      "learning_rate": 8.093022706097866e-06,
      "loss": 0.7581,
      "step": 6929
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.644755943220788,
      "learning_rate": 8.092455104905557e-06,
      "loss": 0.7437,
      "step": 6930
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.5746084370535085,
      "learning_rate": 8.09188743916502e-06,
      "loss": 0.8115,
      "step": 6931
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.296235367060424,
      "learning_rate": 8.091319708888103e-06,
      "loss": 0.7614,
      "step": 6932
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.144345425633739,
      "learning_rate": 8.090751914086657e-06,
      "loss": 0.833,
      "step": 6933
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.128175621627171,
      "learning_rate": 8.090184054772534e-06,
      "loss": 0.7565,
      "step": 6934
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.813601358074685,
      "learning_rate": 8.089616130957585e-06,
      "loss": 0.7906,
      "step": 6935
    },
    {
      "epoch": 0.62,
      "grad_norm": 8.180421075103032,
      "learning_rate": 8.089048142653667e-06,
      "loss": 0.8108,
      "step": 6936
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.891601226164926,
      "learning_rate": 8.088480089872633e-06,
      "loss": 0.8914,
      "step": 6937
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.571865654652032,
      "learning_rate": 8.087911972626342e-06,
      "loss": 0.8394,
      "step": 6938
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.303675364258027,
      "learning_rate": 8.08734379092665e-06,
      "loss": 0.8698,
      "step": 6939
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.756576248684503,
      "learning_rate": 8.086775544785417e-06,
      "loss": 0.7857,
      "step": 6940
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.298888694930089,
      "learning_rate": 8.086207234214505e-06,
      "loss": 0.7454,
      "step": 6941
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.373510938220605,
      "learning_rate": 8.085638859225775e-06,
      "loss": 0.7796,
      "step": 6942
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.316286544153729,
      "learning_rate": 8.085070419831092e-06,
      "loss": 0.7773,
      "step": 6943
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.015787438423901,
      "learning_rate": 8.08450191604232e-06,
      "loss": 0.8078,
      "step": 6944
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.881251841329082,
      "learning_rate": 8.083933347871325e-06,
      "loss": 0.7755,
      "step": 6945
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.003916918990588,
      "learning_rate": 8.083364715329976e-06,
      "loss": 0.7666,
      "step": 6946
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.362655848918575,
      "learning_rate": 8.082796018430142e-06,
      "loss": 0.8167,
      "step": 6947
    },
    {
      "epoch": 0.62,
      "grad_norm": 8.310372583636804,
      "learning_rate": 8.08222725718369e-06,
      "loss": 0.8206,
      "step": 6948
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.778996741694583,
      "learning_rate": 8.081658431602495e-06,
      "loss": 0.8258,
      "step": 6949
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.56039996098141,
      "learning_rate": 8.081089541698428e-06,
      "loss": 0.7602,
      "step": 6950
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.865085658706781,
      "learning_rate": 8.080520587483366e-06,
      "loss": 0.8212,
      "step": 6951
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.027833013730606,
      "learning_rate": 8.079951568969183e-06,
      "loss": 0.8003,
      "step": 6952
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.428786319954096,
      "learning_rate": 8.079382486167753e-06,
      "loss": 0.8202,
      "step": 6953
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.339472826038203,
      "learning_rate": 8.07881333909096e-06,
      "loss": 0.8908,
      "step": 6954
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.590374295838503,
      "learning_rate": 8.07824412775068e-06,
      "loss": 0.7839,
      "step": 6955
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.094366051187591,
      "learning_rate": 8.077674852158795e-06,
      "loss": 0.8216,
      "step": 6956
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.2496740817483225,
      "learning_rate": 8.07710551232719e-06,
      "loss": 0.758,
      "step": 6957
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.399303174182608,
      "learning_rate": 8.076536108267743e-06,
      "loss": 0.8259,
      "step": 6958
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.37455476070981,
      "learning_rate": 8.075966639992342e-06,
      "loss": 0.8402,
      "step": 6959
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.32986596248156,
      "learning_rate": 8.075397107512876e-06,
      "loss": 0.764,
      "step": 6960
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.453865737659811,
      "learning_rate": 8.07482751084123e-06,
      "loss": 0.8024,
      "step": 6961
    },
    {
      "epoch": 0.62,
      "grad_norm": 8.52957073829763,
      "learning_rate": 8.074257849989293e-06,
      "loss": 0.809,
      "step": 6962
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.527078502512016,
      "learning_rate": 8.073688124968955e-06,
      "loss": 0.8165,
      "step": 6963
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.030371922159815,
      "learning_rate": 8.07311833579211e-06,
      "loss": 0.7702,
      "step": 6964
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.979139956783152,
      "learning_rate": 8.07254848247065e-06,
      "loss": 0.9268,
      "step": 6965
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.931818972339798,
      "learning_rate": 8.071978565016468e-06,
      "loss": 0.853,
      "step": 6966
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.31965954621122,
      "learning_rate": 8.071408583441462e-06,
      "loss": 0.801,
      "step": 6967
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.995403120231001,
      "learning_rate": 8.070838537757526e-06,
      "loss": 0.7903,
      "step": 6968
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.59954487126442,
      "learning_rate": 8.070268427976562e-06,
      "loss": 0.7634,
      "step": 6969
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.682777497293176,
      "learning_rate": 8.069698254110468e-06,
      "loss": 0.8763,
      "step": 6970
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.993497868050582,
      "learning_rate": 8.069128016171147e-06,
      "loss": 0.7872,
      "step": 6971
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.094022367626832,
      "learning_rate": 8.068557714170498e-06,
      "loss": 0.7709,
      "step": 6972
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.010277345778064,
      "learning_rate": 8.067987348120427e-06,
      "loss": 0.785,
      "step": 6973
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.259601463129585,
      "learning_rate": 8.06741691803284e-06,
      "loss": 0.7683,
      "step": 6974
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.276894797465551,
      "learning_rate": 8.066846423919642e-06,
      "loss": 0.7652,
      "step": 6975
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.168862560500962,
      "learning_rate": 8.066275865792741e-06,
      "loss": 0.8275,
      "step": 6976
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.192228496971852,
      "learning_rate": 8.065705243664045e-06,
      "loss": 0.8005,
      "step": 6977
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.7627853852827515,
      "learning_rate": 8.065134557545468e-06,
      "loss": 0.8165,
      "step": 6978
    },
    {
      "epoch": 0.62,
      "grad_norm": 9.448146352857938,
      "learning_rate": 8.06456380744892e-06,
      "loss": 0.8638,
      "step": 6979
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.109266598211198,
      "learning_rate": 8.063992993386311e-06,
      "loss": 0.8211,
      "step": 6980
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.556512022133831,
      "learning_rate": 8.06342211536956e-06,
      "loss": 0.7668,
      "step": 6981
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.163084457007714,
      "learning_rate": 8.062851173410582e-06,
      "loss": 0.8003,
      "step": 6982
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.362397295399718,
      "learning_rate": 8.06228016752129e-06,
      "loss": 0.8017,
      "step": 6983
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.417428878720762,
      "learning_rate": 8.06170909771361e-06,
      "loss": 0.7566,
      "step": 6984
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.115304861950151,
      "learning_rate": 8.061137963999455e-06,
      "loss": 0.7407,
      "step": 6985
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.542282605303808,
      "learning_rate": 8.06056676639075e-06,
      "loss": 0.9014,
      "step": 6986
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.806480335211902,
      "learning_rate": 8.059995504899415e-06,
      "loss": 0.7989,
      "step": 6987
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.119646499339333,
      "learning_rate": 8.059424179537376e-06,
      "loss": 0.9001,
      "step": 6988
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.022981879479564,
      "learning_rate": 8.058852790316557e-06,
      "loss": 0.7774,
      "step": 6989
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.326752012408423,
      "learning_rate": 8.058281337248884e-06,
      "loss": 0.7773,
      "step": 6990
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.27137035487564,
      "learning_rate": 8.057709820346287e-06,
      "loss": 0.7625,
      "step": 6991
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.079449025463701,
      "learning_rate": 8.057138239620693e-06,
      "loss": 0.8139,
      "step": 6992
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.062453255358927,
      "learning_rate": 8.056566595084034e-06,
      "loss": 0.7979,
      "step": 6993
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.7117081009231025,
      "learning_rate": 8.05599488674824e-06,
      "loss": 0.7816,
      "step": 6994
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.9722134823029234,
      "learning_rate": 8.055423114625246e-06,
      "loss": 0.8481,
      "step": 6995
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.221278278133744,
      "learning_rate": 8.054851278726985e-06,
      "loss": 0.7564,
      "step": 6996
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.254355286044848,
      "learning_rate": 8.054279379065395e-06,
      "loss": 0.8121,
      "step": 6997
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.393736860176314,
      "learning_rate": 8.05370741565241e-06,
      "loss": 0.8305,
      "step": 6998
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.7463245610590095,
      "learning_rate": 8.05313538849997e-06,
      "loss": 0.749,
      "step": 6999
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.7156110096443715,
      "learning_rate": 8.052563297620017e-06,
      "loss": 0.8256,
      "step": 7000
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.847610923145393,
      "learning_rate": 8.051991143024488e-06,
      "loss": 0.7734,
      "step": 7001
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.874496945779397,
      "learning_rate": 8.05141892472533e-06,
      "loss": 0.8338,
      "step": 7002
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.126585432122027,
      "learning_rate": 8.050846642734484e-06,
      "loss": 0.8316,
      "step": 7003
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.788278386563417,
      "learning_rate": 8.050274297063894e-06,
      "loss": 0.755,
      "step": 7004
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.354138131514921,
      "learning_rate": 8.049701887725509e-06,
      "loss": 0.7373,
      "step": 7005
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.830904952011638,
      "learning_rate": 8.049129414731276e-06,
      "loss": 0.811,
      "step": 7006
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.716719742526026,
      "learning_rate": 8.048556878093145e-06,
      "loss": 0.774,
      "step": 7007
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.835406280812972,
      "learning_rate": 8.047984277823064e-06,
      "loss": 0.7893,
      "step": 7008
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.423963106554698,
      "learning_rate": 8.047411613932986e-06,
      "loss": 0.9137,
      "step": 7009
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.9766995976438695,
      "learning_rate": 8.046838886434865e-06,
      "loss": 0.7635,
      "step": 7010
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.346345911486274,
      "learning_rate": 8.046266095340653e-06,
      "loss": 0.8763,
      "step": 7011
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.718229827045991,
      "learning_rate": 8.045693240662309e-06,
      "loss": 0.7753,
      "step": 7012
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.069104316441401,
      "learning_rate": 8.045120322411789e-06,
      "loss": 0.85,
      "step": 7013
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.36903341643116,
      "learning_rate": 8.04454734060105e-06,
      "loss": 0.8455,
      "step": 7014
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.04164733465387,
      "learning_rate": 8.043974295242052e-06,
      "loss": 0.808,
      "step": 7015
    },
    {
      "epoch": 0.63,
      "grad_norm": 9.52087450889189,
      "learning_rate": 8.043401186346758e-06,
      "loss": 0.7793,
      "step": 7016
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.578440221600089,
      "learning_rate": 8.04282801392713e-06,
      "loss": 0.7606,
      "step": 7017
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.431670872794137,
      "learning_rate": 8.042254777995129e-06,
      "loss": 0.7496,
      "step": 7018
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.767166350642211,
      "learning_rate": 8.041681478562722e-06,
      "loss": 0.7245,
      "step": 7019
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.755335391898238,
      "learning_rate": 8.041108115641876e-06,
      "loss": 0.6964,
      "step": 7020
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.5996637492447165,
      "learning_rate": 8.040534689244558e-06,
      "loss": 0.8686,
      "step": 7021
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.291805282630803,
      "learning_rate": 8.039961199382737e-06,
      "loss": 0.8295,
      "step": 7022
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.988289567559861,
      "learning_rate": 8.039387646068384e-06,
      "loss": 0.83,
      "step": 7023
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.7152843342582855,
      "learning_rate": 8.038814029313469e-06,
      "loss": 0.7755,
      "step": 7024
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.8196808429672995,
      "learning_rate": 8.038240349129967e-06,
      "loss": 0.8208,
      "step": 7025
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.791741879972288,
      "learning_rate": 8.03766660552985e-06,
      "loss": 0.836,
      "step": 7026
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.244375499668143,
      "learning_rate": 8.037092798525096e-06,
      "loss": 0.8224,
      "step": 7027
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.592336215535681,
      "learning_rate": 8.036518928127681e-06,
      "loss": 0.7325,
      "step": 7028
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.114217574329579,
      "learning_rate": 8.035944994349582e-06,
      "loss": 0.7263,
      "step": 7029
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.9744858274176575,
      "learning_rate": 8.035370997202782e-06,
      "loss": 0.8185,
      "step": 7030
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.795482905861157,
      "learning_rate": 8.03479693669926e-06,
      "loss": 0.7557,
      "step": 7031
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.602858180056339,
      "learning_rate": 8.034222812850996e-06,
      "loss": 0.8134,
      "step": 7032
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.391495932907588,
      "learning_rate": 8.033648625669975e-06,
      "loss": 0.8293,
      "step": 7033
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.411199003633842,
      "learning_rate": 8.033074375168184e-06,
      "loss": 0.7533,
      "step": 7034
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.6619719788390155,
      "learning_rate": 8.032500061357606e-06,
      "loss": 0.8253,
      "step": 7035
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.839745859656985,
      "learning_rate": 8.031925684250234e-06,
      "loss": 0.8253,
      "step": 7036
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.718863463448694,
      "learning_rate": 8.03135124385805e-06,
      "loss": 0.8816,
      "step": 7037
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.933706627198956,
      "learning_rate": 8.03077674019305e-06,
      "loss": 0.8363,
      "step": 7038
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.792455413883634,
      "learning_rate": 8.03020217326722e-06,
      "loss": 0.8069,
      "step": 7039
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.537959413519084,
      "learning_rate": 8.029627543092556e-06,
      "loss": 0.769,
      "step": 7040
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.155058850312842,
      "learning_rate": 8.029052849681055e-06,
      "loss": 0.7921,
      "step": 7041
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.550400233954453,
      "learning_rate": 8.028478093044705e-06,
      "loss": 0.7928,
      "step": 7042
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.371479964405629,
      "learning_rate": 8.02790327319551e-06,
      "loss": 0.863,
      "step": 7043
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.743055698624136,
      "learning_rate": 8.027328390145463e-06,
      "loss": 0.7525,
      "step": 7044
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.9074166791069604,
      "learning_rate": 8.026753443906568e-06,
      "loss": 0.8038,
      "step": 7045
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.515769909609539,
      "learning_rate": 8.026178434490822e-06,
      "loss": 0.8469,
      "step": 7046
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.977402205530207,
      "learning_rate": 8.025603361910227e-06,
      "loss": 0.8832,
      "step": 7047
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.86357245747819,
      "learning_rate": 8.02502822617679e-06,
      "loss": 0.7804,
      "step": 7048
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.394674841555729,
      "learning_rate": 8.024453027302512e-06,
      "loss": 0.8342,
      "step": 7049
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.509559317939442,
      "learning_rate": 8.0238777652994e-06,
      "loss": 0.866,
      "step": 7050
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.099844168997305,
      "learning_rate": 8.023302440179465e-06,
      "loss": 0.8004,
      "step": 7051
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.972225297610281,
      "learning_rate": 8.02272705195471e-06,
      "loss": 0.8459,
      "step": 7052
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.539912404481116,
      "learning_rate": 8.022151600637146e-06,
      "loss": 0.7894,
      "step": 7053
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.973858503146333,
      "learning_rate": 8.02157608623879e-06,
      "loss": 0.8396,
      "step": 7054
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.46430670236725,
      "learning_rate": 8.021000508771648e-06,
      "loss": 0.8133,
      "step": 7055
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.467361944901082,
      "learning_rate": 8.020424868247735e-06,
      "loss": 0.8718,
      "step": 7056
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.255701349071418,
      "learning_rate": 8.01984916467907e-06,
      "loss": 0.724,
      "step": 7057
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.765384480155345,
      "learning_rate": 8.019273398077664e-06,
      "loss": 0.852,
      "step": 7058
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.966234130880603,
      "learning_rate": 8.01869756845554e-06,
      "loss": 0.8092,
      "step": 7059
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.976389750085673,
      "learning_rate": 8.018121675824715e-06,
      "loss": 0.7081,
      "step": 7060
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.83501919620551,
      "learning_rate": 8.01754572019721e-06,
      "loss": 0.7809,
      "step": 7061
    },
    {
      "epoch": 0.63,
      "grad_norm": 9.036397957382544,
      "learning_rate": 8.016969701585045e-06,
      "loss": 0.7693,
      "step": 7062
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.288783439739562,
      "learning_rate": 8.016393620000246e-06,
      "loss": 0.8437,
      "step": 7063
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.043949685566679,
      "learning_rate": 8.015817475454835e-06,
      "loss": 0.8359,
      "step": 7064
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.843052176563591,
      "learning_rate": 8.015241267960841e-06,
      "loss": 0.7996,
      "step": 7065
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.77622263400318,
      "learning_rate": 8.014664997530286e-06,
      "loss": 0.7947,
      "step": 7066
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.939327262265271,
      "learning_rate": 8.0140886641752e-06,
      "loss": 0.8101,
      "step": 7067
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.558395161053029,
      "learning_rate": 8.013512267907617e-06,
      "loss": 0.7287,
      "step": 7068
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.170562339925411,
      "learning_rate": 8.012935808739563e-06,
      "loss": 0.8459,
      "step": 7069
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.953126610185626,
      "learning_rate": 8.012359286683074e-06,
      "loss": 0.7659,
      "step": 7070
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.539972862389098,
      "learning_rate": 8.01178270175018e-06,
      "loss": 0.7979,
      "step": 7071
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.36841507519751,
      "learning_rate": 8.01120605395292e-06,
      "loss": 0.8496,
      "step": 7072
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.084842810517694,
      "learning_rate": 8.010629343303326e-06,
      "loss": 0.8623,
      "step": 7073
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.87893568290355,
      "learning_rate": 8.010052569813438e-06,
      "loss": 0.7758,
      "step": 7074
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.383253508386346,
      "learning_rate": 8.009475733495296e-06,
      "loss": 0.7515,
      "step": 7075
    },
    {
      "epoch": 0.63,
      "grad_norm": 9.667473670864817,
      "learning_rate": 8.008898834360936e-06,
      "loss": 0.8792,
      "step": 7076
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.293729966908839,
      "learning_rate": 8.008321872422404e-06,
      "loss": 0.8575,
      "step": 7077
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.219686596223907,
      "learning_rate": 8.00774484769174e-06,
      "loss": 0.6861,
      "step": 7078
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.769179728538904,
      "learning_rate": 8.007167760180992e-06,
      "loss": 0.8385,
      "step": 7079
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.122907223576569,
      "learning_rate": 8.0065906099022e-06,
      "loss": 0.7308,
      "step": 7080
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.593321514469411,
      "learning_rate": 8.006013396867414e-06,
      "loss": 0.8087,
      "step": 7081
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.434738309630223,
      "learning_rate": 8.00543612108868e-06,
      "loss": 0.8116,
      "step": 7082
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.444906310592417,
      "learning_rate": 8.00485878257805e-06,
      "loss": 0.7667,
      "step": 7083
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.62291175052663,
      "learning_rate": 8.004281381347574e-06,
      "loss": 0.8221,
      "step": 7084
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.1095544185532855,
      "learning_rate": 8.003703917409302e-06,
      "loss": 0.7835,
      "step": 7085
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.319947699797402,
      "learning_rate": 8.003126390775288e-06,
      "loss": 0.7676,
      "step": 7086
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.186718700001167,
      "learning_rate": 8.00254880145759e-06,
      "loss": 0.8396,
      "step": 7087
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.4041124070561235,
      "learning_rate": 8.001971149468262e-06,
      "loss": 0.7785,
      "step": 7088
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.041241862971773,
      "learning_rate": 8.001393434819358e-06,
      "loss": 0.8328,
      "step": 7089
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.752196509495697,
      "learning_rate": 8.00081565752294e-06,
      "loss": 0.8131,
      "step": 7090
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.472559279782797,
      "learning_rate": 8.000237817591066e-06,
      "loss": 0.8317,
      "step": 7091
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.599518534580797,
      "learning_rate": 7.999659915035798e-06,
      "loss": 0.7877,
      "step": 7092
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.7328134871082295,
      "learning_rate": 7.999081949869199e-06,
      "loss": 0.8108,
      "step": 7093
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.216479736063277,
      "learning_rate": 7.998503922103332e-06,
      "loss": 0.8569,
      "step": 7094
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.392998955474488,
      "learning_rate": 7.997925831750262e-06,
      "loss": 0.8236,
      "step": 7095
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.364324899972853,
      "learning_rate": 7.997347678822056e-06,
      "loss": 0.8317,
      "step": 7096
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.226467488240026,
      "learning_rate": 7.996769463330781e-06,
      "loss": 0.8252,
      "step": 7097
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.827149300415938,
      "learning_rate": 7.996191185288507e-06,
      "loss": 0.791,
      "step": 7098
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.388159486164026,
      "learning_rate": 7.995612844707301e-06,
      "loss": 0.7265,
      "step": 7099
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.9674713816651535,
      "learning_rate": 7.99503444159924e-06,
      "loss": 0.801,
      "step": 7100
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.137809615954206,
      "learning_rate": 7.994455975976395e-06,
      "loss": 0.7572,
      "step": 7101
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.357335441831199,
      "learning_rate": 7.993877447850836e-06,
      "loss": 0.7985,
      "step": 7102
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.9980945499030165,
      "learning_rate": 7.993298857234644e-06,
      "loss": 0.843,
      "step": 7103
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.872885667832998,
      "learning_rate": 7.992720204139892e-06,
      "loss": 0.7373,
      "step": 7104
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.867299428915381,
      "learning_rate": 7.99214148857866e-06,
      "loss": 0.8037,
      "step": 7105
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.015789022639338,
      "learning_rate": 7.991562710563028e-06,
      "loss": 0.731,
      "step": 7106
    },
    {
      "epoch": 0.63,
      "grad_norm": 9.137653169603446,
      "learning_rate": 7.990983870105073e-06,
      "loss": 0.7742,
      "step": 7107
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.992130198680315,
      "learning_rate": 7.990404967216885e-06,
      "loss": 0.7359,
      "step": 7108
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.6488929766309,
      "learning_rate": 7.98982600191054e-06,
      "loss": 0.7936,
      "step": 7109
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.2024639625078954,
      "learning_rate": 7.989246974198121e-06,
      "loss": 0.8464,
      "step": 7110
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.729778313052926,
      "learning_rate": 7.988667884091723e-06,
      "loss": 0.9203,
      "step": 7111
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.959284467179346,
      "learning_rate": 7.988088731603425e-06,
      "loss": 0.785,
      "step": 7112
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.692437993750479,
      "learning_rate": 7.98750951674532e-06,
      "loss": 0.8785,
      "step": 7113
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.396835201093804,
      "learning_rate": 7.986930239529496e-06,
      "loss": 0.7663,
      "step": 7114
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.830505230831576,
      "learning_rate": 7.986350899968043e-06,
      "loss": 0.7886,
      "step": 7115
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.575023790757853,
      "learning_rate": 7.985771498073057e-06,
      "loss": 0.7786,
      "step": 7116
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.552919942851462,
      "learning_rate": 7.985192033856628e-06,
      "loss": 0.8478,
      "step": 7117
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.538397680427876,
      "learning_rate": 7.984612507330854e-06,
      "loss": 0.8623,
      "step": 7118
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.305844739752404,
      "learning_rate": 7.98403291850783e-06,
      "loss": 0.7577,
      "step": 7119
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.534933049691424,
      "learning_rate": 7.983453267399654e-06,
      "loss": 0.8346,
      "step": 7120
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.447465584127629,
      "learning_rate": 7.982873554018424e-06,
      "loss": 0.7828,
      "step": 7121
    },
    {
      "epoch": 0.64,
      "grad_norm": 8.55093278414744,
      "learning_rate": 7.982293778376242e-06,
      "loss": 0.8793,
      "step": 7122
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.8021097373060195,
      "learning_rate": 7.981713940485207e-06,
      "loss": 0.7564,
      "step": 7123
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.967270342504546,
      "learning_rate": 7.981134040357424e-06,
      "loss": 0.7775,
      "step": 7124
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.93978524199754,
      "learning_rate": 7.980554078004996e-06,
      "loss": 0.8057,
      "step": 7125
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.761814276188255,
      "learning_rate": 7.979974053440029e-06,
      "loss": 0.6866,
      "step": 7126
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.342386558952088,
      "learning_rate": 7.979393966674631e-06,
      "loss": 0.7961,
      "step": 7127
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.177919145147326,
      "learning_rate": 7.978813817720907e-06,
      "loss": 0.831,
      "step": 7128
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.844519158442585,
      "learning_rate": 7.97823360659097e-06,
      "loss": 0.7735,
      "step": 7129
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.620381075913567,
      "learning_rate": 7.977653333296927e-06,
      "loss": 0.794,
      "step": 7130
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.494716805813692,
      "learning_rate": 7.977072997850892e-06,
      "loss": 0.806,
      "step": 7131
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.77216118495672,
      "learning_rate": 7.976492600264979e-06,
      "loss": 0.7877,
      "step": 7132
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.7127219834052845,
      "learning_rate": 7.9759121405513e-06,
      "loss": 0.7327,
      "step": 7133
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.47182389795068,
      "learning_rate": 7.975331618721973e-06,
      "loss": 0.7809,
      "step": 7134
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.382731441418061,
      "learning_rate": 7.974751034789113e-06,
      "loss": 0.7955,
      "step": 7135
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.76941928410649,
      "learning_rate": 7.974170388764842e-06,
      "loss": 0.8206,
      "step": 7136
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.785592700142208,
      "learning_rate": 7.973589680661276e-06,
      "loss": 0.7956,
      "step": 7137
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.978132435679496,
      "learning_rate": 7.973008910490536e-06,
      "loss": 0.8359,
      "step": 7138
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.099315767011291,
      "learning_rate": 7.972428078264748e-06,
      "loss": 0.8504,
      "step": 7139
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.438212352377424,
      "learning_rate": 7.971847183996033e-06,
      "loss": 0.7854,
      "step": 7140
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.9110520860490405,
      "learning_rate": 7.971266227696517e-06,
      "loss": 0.8315,
      "step": 7141
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.950218407590673,
      "learning_rate": 7.970685209378324e-06,
      "loss": 0.7437,
      "step": 7142
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.940572093958994,
      "learning_rate": 7.970104129053582e-06,
      "loss": 0.7831,
      "step": 7143
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.0754034272261785,
      "learning_rate": 7.969522986734424e-06,
      "loss": 0.8969,
      "step": 7144
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.768471675956912,
      "learning_rate": 7.968941782432975e-06,
      "loss": 0.8067,
      "step": 7145
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.853950591500136,
      "learning_rate": 7.968360516161367e-06,
      "loss": 0.8059,
      "step": 7146
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.586888013477863,
      "learning_rate": 7.967779187931735e-06,
      "loss": 0.8075,
      "step": 7147
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.690843801807002,
      "learning_rate": 7.967197797756212e-06,
      "loss": 0.7416,
      "step": 7148
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.251863217520145,
      "learning_rate": 7.966616345646932e-06,
      "loss": 0.7835,
      "step": 7149
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.399052024106859,
      "learning_rate": 7.966034831616034e-06,
      "loss": 0.7996,
      "step": 7150
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.975569850836193,
      "learning_rate": 7.965453255675653e-06,
      "loss": 0.7327,
      "step": 7151
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.236391260689289,
      "learning_rate": 7.96487161783793e-06,
      "loss": 0.8013,
      "step": 7152
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.450463150716186,
      "learning_rate": 7.964289918115006e-06,
      "loss": 0.7505,
      "step": 7153
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.39168306416975,
      "learning_rate": 7.96370815651902e-06,
      "loss": 0.8323,
      "step": 7154
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.906180339071581,
      "learning_rate": 7.963126333062116e-06,
      "loss": 0.7732,
      "step": 7155
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.680217760636456,
      "learning_rate": 7.962544447756441e-06,
      "loss": 0.8168,
      "step": 7156
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.222176341716212,
      "learning_rate": 7.961962500614138e-06,
      "loss": 0.8515,
      "step": 7157
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.865733563190256,
      "learning_rate": 7.961380491647355e-06,
      "loss": 0.8076,
      "step": 7158
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.6567732195155,
      "learning_rate": 7.960798420868238e-06,
      "loss": 0.7329,
      "step": 7159
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.901595974019546,
      "learning_rate": 7.96021628828894e-06,
      "loss": 0.8335,
      "step": 7160
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.381971946494121,
      "learning_rate": 7.959634093921609e-06,
      "loss": 0.8489,
      "step": 7161
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.105244537796855,
      "learning_rate": 7.959051837778396e-06,
      "loss": 0.8827,
      "step": 7162
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.780793919285622,
      "learning_rate": 7.958469519871457e-06,
      "loss": 0.7392,
      "step": 7163
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.990536252121846,
      "learning_rate": 7.957887140212948e-06,
      "loss": 0.7902,
      "step": 7164
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.787813708942032,
      "learning_rate": 7.95730469881502e-06,
      "loss": 0.7737,
      "step": 7165
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.186932483286272,
      "learning_rate": 7.956722195689835e-06,
      "loss": 0.8428,
      "step": 7166
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.910863737449839,
      "learning_rate": 7.956139630849546e-06,
      "loss": 0.7763,
      "step": 7167
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.702711309318253,
      "learning_rate": 7.955557004306318e-06,
      "loss": 0.7925,
      "step": 7168
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.324992145622875,
      "learning_rate": 7.954974316072311e-06,
      "loss": 0.8021,
      "step": 7169
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.4009170535411855,
      "learning_rate": 7.954391566159685e-06,
      "loss": 0.8006,
      "step": 7170
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.056026641183625,
      "learning_rate": 7.953808754580604e-06,
      "loss": 0.8234,
      "step": 7171
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.412695805105385,
      "learning_rate": 7.953225881347235e-06,
      "loss": 0.7703,
      "step": 7172
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.5636363758463565,
      "learning_rate": 7.952642946471744e-06,
      "loss": 0.8035,
      "step": 7173
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.543112604106601,
      "learning_rate": 7.952059949966298e-06,
      "loss": 0.7971,
      "step": 7174
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.853332910493852,
      "learning_rate": 7.951476891843065e-06,
      "loss": 0.7686,
      "step": 7175
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.377433609287862,
      "learning_rate": 7.950893772114214e-06,
      "loss": 0.8062,
      "step": 7176
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.631972114388808,
      "learning_rate": 7.95031059079192e-06,
      "loss": 0.8118,
      "step": 7177
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.106129396870192,
      "learning_rate": 7.949727347888353e-06,
      "loss": 0.7807,
      "step": 7178
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.311860196863349,
      "learning_rate": 7.949144043415687e-06,
      "loss": 0.826,
      "step": 7179
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.692610663551311,
      "learning_rate": 7.948560677386098e-06,
      "loss": 0.7331,
      "step": 7180
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.346531093810068,
      "learning_rate": 7.94797724981176e-06,
      "loss": 0.8613,
      "step": 7181
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.386050339431628,
      "learning_rate": 7.947393760704856e-06,
      "loss": 0.8655,
      "step": 7182
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.611064510648886,
      "learning_rate": 7.94681021007756e-06,
      "loss": 0.8095,
      "step": 7183
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.417359526456947,
      "learning_rate": 7.946226597942055e-06,
      "loss": 0.8369,
      "step": 7184
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.542607137790846,
      "learning_rate": 7.945642924310521e-06,
      "loss": 0.7616,
      "step": 7185
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.50589412908159,
      "learning_rate": 7.945059189195143e-06,
      "loss": 0.7993,
      "step": 7186
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.1084154940957065,
      "learning_rate": 7.944475392608102e-06,
      "loss": 0.7756,
      "step": 7187
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.908818041742095,
      "learning_rate": 7.943891534561589e-06,
      "loss": 0.8473,
      "step": 7188
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.335015459320931,
      "learning_rate": 7.943307615067784e-06,
      "loss": 0.8245,
      "step": 7189
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.6170813584678285,
      "learning_rate": 7.94272363413888e-06,
      "loss": 0.7809,
      "step": 7190
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.974828555791969,
      "learning_rate": 7.942139591787063e-06,
      "loss": 0.7836,
      "step": 7191
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.132435752450243,
      "learning_rate": 7.941555488024526e-06,
      "loss": 0.7296,
      "step": 7192
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.111289958115326,
      "learning_rate": 7.940971322863462e-06,
      "loss": 0.7903,
      "step": 7193
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.7686044624207975,
      "learning_rate": 7.94038709631606e-06,
      "loss": 0.7968,
      "step": 7194
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.99693170637764,
      "learning_rate": 7.939802808394516e-06,
      "loss": 0.7899,
      "step": 7195
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.297872839370755,
      "learning_rate": 7.939218459111029e-06,
      "loss": 0.7568,
      "step": 7196
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.95202702075069,
      "learning_rate": 7.938634048477791e-06,
      "loss": 0.7843,
      "step": 7197
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.656631366917541,
      "learning_rate": 7.938049576507004e-06,
      "loss": 0.752,
      "step": 7198
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.9900320414658026,
      "learning_rate": 7.937465043210866e-06,
      "loss": 0.8317,
      "step": 7199
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.769767306004276,
      "learning_rate": 7.936880448601578e-06,
      "loss": 0.7528,
      "step": 7200
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.868983632651134,
      "learning_rate": 7.936295792691344e-06,
      "loss": 0.7688,
      "step": 7201
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.421767324804065,
      "learning_rate": 7.935711075492365e-06,
      "loss": 0.7483,
      "step": 7202
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.2945783246476825,
      "learning_rate": 7.935126297016844e-06,
      "loss": 0.7656,
      "step": 7203
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.1082692023643235,
      "learning_rate": 7.934541457276992e-06,
      "loss": 0.7032,
      "step": 7204
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.34025015321995,
      "learning_rate": 7.933956556285012e-06,
      "loss": 0.8338,
      "step": 7205
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.310280839226349,
      "learning_rate": 7.933371594053115e-06,
      "loss": 0.7911,
      "step": 7206
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.501390053569232,
      "learning_rate": 7.932786570593509e-06,
      "loss": 0.7938,
      "step": 7207
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.805010391872277,
      "learning_rate": 7.932201485918407e-06,
      "loss": 0.787,
      "step": 7208
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.614903038580794,
      "learning_rate": 7.931616340040022e-06,
      "loss": 0.7444,
      "step": 7209
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.487685373681917,
      "learning_rate": 7.931031132970563e-06,
      "loss": 0.8572,
      "step": 7210
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.792600653136466,
      "learning_rate": 7.930445864722249e-06,
      "loss": 0.7888,
      "step": 7211
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.002078894549036,
      "learning_rate": 7.929860535307297e-06,
      "loss": 0.8518,
      "step": 7212
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.813766776569422,
      "learning_rate": 7.92927514473792e-06,
      "loss": 0.7909,
      "step": 7213
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.1413793123651335,
      "learning_rate": 7.92868969302634e-06,
      "loss": 0.7461,
      "step": 7214
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.9898223051449255,
      "learning_rate": 7.928104180184778e-06,
      "loss": 0.7993,
      "step": 7215
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.1012715641085835,
      "learning_rate": 7.927518606225452e-06,
      "loss": 0.8368,
      "step": 7216
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.230024986936379,
      "learning_rate": 7.926932971160585e-06,
      "loss": 0.8135,
      "step": 7217
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.97867935896615,
      "learning_rate": 7.926347275002405e-06,
      "loss": 0.7696,
      "step": 7218
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.449945213885118,
      "learning_rate": 7.925761517763132e-06,
      "loss": 0.8082,
      "step": 7219
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.346071806605596,
      "learning_rate": 7.925175699454996e-06,
      "loss": 0.8435,
      "step": 7220
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.242269329615449,
      "learning_rate": 7.924589820090222e-06,
      "loss": 0.7813,
      "step": 7221
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.971169568964333,
      "learning_rate": 7.924003879681042e-06,
      "loss": 0.8019,
      "step": 7222
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.129299271237798,
      "learning_rate": 7.923417878239683e-06,
      "loss": 0.8018,
      "step": 7223
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.128031418502038,
      "learning_rate": 7.92283181577838e-06,
      "loss": 0.7805,
      "step": 7224
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.116397491987267,
      "learning_rate": 7.922245692309362e-06,
      "loss": 0.7987,
      "step": 7225
    },
    {
      "epoch": 0.64,
      "grad_norm": 8.012571814523596,
      "learning_rate": 7.921659507844865e-06,
      "loss": 0.7627,
      "step": 7226
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.819239789309228,
      "learning_rate": 7.921073262397125e-06,
      "loss": 0.7735,
      "step": 7227
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.792327577246019,
      "learning_rate": 7.920486955978376e-06,
      "loss": 0.7777,
      "step": 7228
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.647934133552474,
      "learning_rate": 7.919900588600857e-06,
      "loss": 0.825,
      "step": 7229
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.445092138065056,
      "learning_rate": 7.91931416027681e-06,
      "loss": 0.7275,
      "step": 7230
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.434968001415219,
      "learning_rate": 7.918727671018474e-06,
      "loss": 0.8563,
      "step": 7231
    },
    {
      "epoch": 0.65,
      "grad_norm": 7.549055789434388,
      "learning_rate": 7.918141120838088e-06,
      "loss": 0.8247,
      "step": 7232
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.687548484692197,
      "learning_rate": 7.917554509747895e-06,
      "loss": 0.7086,
      "step": 7233
    },
    {
      "epoch": 0.65,
      "grad_norm": 8.13273184801703,
      "learning_rate": 7.916967837760142e-06,
      "loss": 0.8045,
      "step": 7234
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.080516609874037,
      "learning_rate": 7.916381104887074e-06,
      "loss": 0.7507,
      "step": 7235
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.124350716765973,
      "learning_rate": 7.91579431114094e-06,
      "loss": 0.8155,
      "step": 7236
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.58882956296973,
      "learning_rate": 7.91520745653398e-06,
      "loss": 0.6804,
      "step": 7237
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.434060716063205,
      "learning_rate": 7.914620541078453e-06,
      "loss": 0.8209,
      "step": 7238
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.748709634359352,
      "learning_rate": 7.914033564786603e-06,
      "loss": 0.844,
      "step": 7239
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.224734810466471,
      "learning_rate": 7.913446527670683e-06,
      "loss": 0.789,
      "step": 7240
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.947488832221033,
      "learning_rate": 7.912859429742948e-06,
      "loss": 0.8002,
      "step": 7241
    },
    {
      "epoch": 0.65,
      "grad_norm": 8.392794057546384,
      "learning_rate": 7.912272271015653e-06,
      "loss": 0.8401,
      "step": 7242
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.373573646816047,
      "learning_rate": 7.91168505150105e-06,
      "loss": 0.8033,
      "step": 7243
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.544997526908821,
      "learning_rate": 7.911097771211397e-06,
      "loss": 0.8019,
      "step": 7244
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.881580523577525,
      "learning_rate": 7.910510430158957e-06,
      "loss": 0.7811,
      "step": 7245
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.257777905054153,
      "learning_rate": 7.909923028355983e-06,
      "loss": 0.8702,
      "step": 7246
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.099734851019966,
      "learning_rate": 7.909335565814739e-06,
      "loss": 0.7359,
      "step": 7247
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.481762950154281,
      "learning_rate": 7.908748042547487e-06,
      "loss": 0.8612,
      "step": 7248
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.735974967263165,
      "learning_rate": 7.908160458566489e-06,
      "loss": 0.7944,
      "step": 7249
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.916726684682544,
      "learning_rate": 7.90757281388401e-06,
      "loss": 0.8167,
      "step": 7250
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.9880982841065142,
      "learning_rate": 7.906985108512316e-06,
      "loss": 0.7675,
      "step": 7251
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.94238101407038,
      "learning_rate": 7.906397342463674e-06,
      "loss": 0.7874,
      "step": 7252
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.848934951080192,
      "learning_rate": 7.905809515750353e-06,
      "loss": 0.8023,
      "step": 7253
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.736213250254177,
      "learning_rate": 7.905221628384624e-06,
      "loss": 0.8805,
      "step": 7254
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.910024518987341,
      "learning_rate": 7.904633680378753e-06,
      "loss": 0.8168,
      "step": 7255
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.2350365781298684,
      "learning_rate": 7.904045671745016e-06,
      "loss": 0.6997,
      "step": 7256
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.7653602170746305,
      "learning_rate": 7.903457602495683e-06,
      "loss": 0.7601,
      "step": 7257
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.015273138663125,
      "learning_rate": 7.902869472643033e-06,
      "loss": 0.8212,
      "step": 7258
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.990322082149091,
      "learning_rate": 7.90228128219934e-06,
      "loss": 0.797,
      "step": 7259
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.116367455695627,
      "learning_rate": 7.90169303117688e-06,
      "loss": 0.8275,
      "step": 7260
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.369276659295666,
      "learning_rate": 7.901104719587935e-06,
      "loss": 0.8567,
      "step": 7261
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.2335048746974335,
      "learning_rate": 7.90051634744478e-06,
      "loss": 0.8202,
      "step": 7262
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.156657758970991,
      "learning_rate": 7.8999279147597e-06,
      "loss": 0.8024,
      "step": 7263
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.2729147372795975,
      "learning_rate": 7.899339421544973e-06,
      "loss": 0.8001,
      "step": 7264
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.203240280290182,
      "learning_rate": 7.898750867812888e-06,
      "loss": 0.7611,
      "step": 7265
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.36285217702354,
      "learning_rate": 7.898162253575726e-06,
      "loss": 0.7528,
      "step": 7266
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.719140862340553,
      "learning_rate": 7.897573578845773e-06,
      "loss": 0.8106,
      "step": 7267
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.265776888296912,
      "learning_rate": 7.896984843635317e-06,
      "loss": 0.8529,
      "step": 7268
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.9567604515626345,
      "learning_rate": 7.896396047956648e-06,
      "loss": 0.8039,
      "step": 7269
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.710371368529087,
      "learning_rate": 7.895807191822052e-06,
      "loss": 0.7792,
      "step": 7270
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.038424414351003,
      "learning_rate": 7.895218275243825e-06,
      "loss": 0.7992,
      "step": 7271
    },
    {
      "epoch": 0.65,
      "grad_norm": 8.344977035259465,
      "learning_rate": 7.894629298234256e-06,
      "loss": 0.7372,
      "step": 7272
    },
    {
      "epoch": 0.65,
      "grad_norm": 7.23354358448285,
      "learning_rate": 7.89404026080564e-06,
      "loss": 0.825,
      "step": 7273
    },
    {
      "epoch": 0.65,
      "grad_norm": 8.05871983947589,
      "learning_rate": 7.89345116297027e-06,
      "loss": 0.8189,
      "step": 7274
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.141599957985732,
      "learning_rate": 7.892862004740445e-06,
      "loss": 0.7563,
      "step": 7275
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.305056291007838,
      "learning_rate": 7.89227278612846e-06,
      "loss": 0.7334,
      "step": 7276
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.732819083054324,
      "learning_rate": 7.891683507146614e-06,
      "loss": 0.7513,
      "step": 7277
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.659585730234145,
      "learning_rate": 7.89109416780721e-06,
      "loss": 0.8193,
      "step": 7278
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.812057731981562,
      "learning_rate": 7.890504768122544e-06,
      "loss": 0.712,
      "step": 7279
    },
    {
      "epoch": 0.65,
      "grad_norm": 7.525918061121079,
      "learning_rate": 7.889915308104922e-06,
      "loss": 0.7251,
      "step": 7280
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.44733136122289,
      "learning_rate": 7.889325787766649e-06,
      "loss": 0.8394,
      "step": 7281
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.198110113395652,
      "learning_rate": 7.888736207120025e-06,
      "loss": 0.817,
      "step": 7282
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.822806229816122,
      "learning_rate": 7.88814656617736e-06,
      "loss": 0.8216,
      "step": 7283
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.076727873866262,
      "learning_rate": 7.887556864950959e-06,
      "loss": 0.8642,
      "step": 7284
    },
    {
      "epoch": 0.65,
      "grad_norm": 9.332622396357536,
      "learning_rate": 7.886967103453132e-06,
      "loss": 0.8173,
      "step": 7285
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.335076820512628,
      "learning_rate": 7.88637728169619e-06,
      "loss": 0.7558,
      "step": 7286
    },
    {
      "epoch": 0.65,
      "grad_norm": 8.162635549762367,
      "learning_rate": 7.885787399692443e-06,
      "loss": 0.8483,
      "step": 7287
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.013498006173284,
      "learning_rate": 7.885197457454204e-06,
      "loss": 0.7623,
      "step": 7288
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.362446750102223,
      "learning_rate": 7.884607454993786e-06,
      "loss": 0.8284,
      "step": 7289
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.461025240924296,
      "learning_rate": 7.884017392323507e-06,
      "loss": 0.8018,
      "step": 7290
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.609227283145032,
      "learning_rate": 7.883427269455679e-06,
      "loss": 0.8331,
      "step": 7291
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.91373396400926,
      "learning_rate": 7.88283708640262e-06,
      "loss": 0.8236,
      "step": 7292
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.248486336807874,
      "learning_rate": 7.882246843176653e-06,
      "loss": 0.7922,
      "step": 7293
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.495533169212878,
      "learning_rate": 7.881656539790093e-06,
      "loss": 0.7792,
      "step": 7294
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.70612395545589,
      "learning_rate": 7.881066176255264e-06,
      "loss": 0.7407,
      "step": 7295
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.744159844952703,
      "learning_rate": 7.88047575258449e-06,
      "loss": 0.7817,
      "step": 7296
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.8452625134196206,
      "learning_rate": 7.879885268790088e-06,
      "loss": 0.8309,
      "step": 7297
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.239506705738267,
      "learning_rate": 7.879294724884392e-06,
      "loss": 0.7538,
      "step": 7298
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.23430136958058,
      "learning_rate": 7.878704120879724e-06,
      "loss": 0.8008,
      "step": 7299
    },
    {
      "epoch": 0.65,
      "grad_norm": 7.241654515200779,
      "learning_rate": 7.87811345678841e-06,
      "loss": 0.836,
      "step": 7300
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.9556640012837665,
      "learning_rate": 7.87752273262278e-06,
      "loss": 0.8102,
      "step": 7301
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.578975126501275,
      "learning_rate": 7.876931948395165e-06,
      "loss": 0.8802,
      "step": 7302
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.8071491566931845,
      "learning_rate": 7.8763411041179e-06,
      "loss": 0.7898,
      "step": 7303
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.060252415692179,
      "learning_rate": 7.875750199803309e-06,
      "loss": 0.8203,
      "step": 7304
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.395421557449016,
      "learning_rate": 7.875159235463731e-06,
      "loss": 0.7633,
      "step": 7305
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.3678283660052974,
      "learning_rate": 7.874568211111502e-06,
      "loss": 0.7553,
      "step": 7306
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.470882638072655,
      "learning_rate": 7.873977126758957e-06,
      "loss": 0.8363,
      "step": 7307
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.571162588805288,
      "learning_rate": 7.873385982418432e-06,
      "loss": 0.8218,
      "step": 7308
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.605982042840392,
      "learning_rate": 7.872794778102265e-06,
      "loss": 0.8643,
      "step": 7309
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.148067799623685,
      "learning_rate": 7.872203513822802e-06,
      "loss": 0.6854,
      "step": 7310
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.339483046941087,
      "learning_rate": 7.87161218959238e-06,
      "loss": 0.7489,
      "step": 7311
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.070236153899968,
      "learning_rate": 7.871020805423342e-06,
      "loss": 0.8052,
      "step": 7312
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.2775659088580635,
      "learning_rate": 7.87042936132803e-06,
      "loss": 0.8152,
      "step": 7313
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.076219946801336,
      "learning_rate": 7.869837857318794e-06,
      "loss": 0.8201,
      "step": 7314
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.15062480925235,
      "learning_rate": 7.869246293407978e-06,
      "loss": 0.8046,
      "step": 7315
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.317952062872707,
      "learning_rate": 7.868654669607926e-06,
      "loss": 0.7586,
      "step": 7316
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.249601190619039,
      "learning_rate": 7.868062985930993e-06,
      "loss": 0.8936,
      "step": 7317
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.936380091318307,
      "learning_rate": 7.867471242389524e-06,
      "loss": 0.7353,
      "step": 7318
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.958910690182678,
      "learning_rate": 7.866879438995874e-06,
      "loss": 0.8762,
      "step": 7319
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.296027589945574,
      "learning_rate": 7.866287575762395e-06,
      "loss": 0.7643,
      "step": 7320
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.274585012651619,
      "learning_rate": 7.865695652701437e-06,
      "loss": 0.8474,
      "step": 7321
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.248878039963356,
      "learning_rate": 7.865103669825359e-06,
      "loss": 0.8126,
      "step": 7322
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.8428638624909714,
      "learning_rate": 7.864511627146517e-06,
      "loss": 0.8058,
      "step": 7323
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.000210861059267,
      "learning_rate": 7.863919524677268e-06,
      "loss": 0.7731,
      "step": 7324
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.933925084999985,
      "learning_rate": 7.86332736242997e-06,
      "loss": 0.7899,
      "step": 7325
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.256431592684385,
      "learning_rate": 7.862735140416986e-06,
      "loss": 0.7574,
      "step": 7326
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.627357158107828,
      "learning_rate": 7.862142858650674e-06,
      "loss": 0.7406,
      "step": 7327
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.835428255300179,
      "learning_rate": 7.861550517143397e-06,
      "loss": 0.7707,
      "step": 7328
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.724149272160683,
      "learning_rate": 7.860958115907519e-06,
      "loss": 0.704,
      "step": 7329
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.124178195445194,
      "learning_rate": 7.860365654955407e-06,
      "loss": 0.7959,
      "step": 7330
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.296760982164352,
      "learning_rate": 7.859773134299427e-06,
      "loss": 0.8126,
      "step": 7331
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.607890626205313,
      "learning_rate": 7.859180553951944e-06,
      "loss": 0.775,
      "step": 7332
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.3619845723020045,
      "learning_rate": 7.858587913925329e-06,
      "loss": 0.7477,
      "step": 7333
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.842663520612056,
      "learning_rate": 7.857995214231952e-06,
      "loss": 0.7456,
      "step": 7334
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.9579059556835485,
      "learning_rate": 7.857402454884181e-06,
      "loss": 0.7927,
      "step": 7335
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.492023148069855,
      "learning_rate": 7.856809635894395e-06,
      "loss": 0.7861,
      "step": 7336
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.309008058821417,
      "learning_rate": 7.856216757274963e-06,
      "loss": 0.7895,
      "step": 7337
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.907688791225626,
      "learning_rate": 7.85562381903826e-06,
      "loss": 0.8757,
      "step": 7338
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.457530281619961,
      "learning_rate": 7.855030821196666e-06,
      "loss": 0.7745,
      "step": 7339
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.323453649106245,
      "learning_rate": 7.854437763762555e-06,
      "loss": 0.7268,
      "step": 7340
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.278759093217568,
      "learning_rate": 7.853844646748305e-06,
      "loss": 0.8042,
      "step": 7341
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.781913976451842,
      "learning_rate": 7.853251470166299e-06,
      "loss": 0.8369,
      "step": 7342
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.497147122941115,
      "learning_rate": 7.852658234028919e-06,
      "loss": 0.8042,
      "step": 7343
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.864820526591365,
      "learning_rate": 7.852064938348542e-06,
      "loss": 0.8177,
      "step": 7344
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.978160050925628,
      "learning_rate": 7.851471583137557e-06,
      "loss": 0.7876,
      "step": 7345
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.9300065661324535,
      "learning_rate": 7.850878168408349e-06,
      "loss": 0.7757,
      "step": 7346
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.970402507752749,
      "learning_rate": 7.8502846941733e-06,
      "loss": 0.8426,
      "step": 7347
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.325681388590815,
      "learning_rate": 7.8496911604448e-06,
      "loss": 0.8354,
      "step": 7348
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.947281744804757,
      "learning_rate": 7.84909756723524e-06,
      "loss": 0.8198,
      "step": 7349
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.0900812278997085,
      "learning_rate": 7.848503914557007e-06,
      "loss": 0.7669,
      "step": 7350
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.654326716114816,
      "learning_rate": 7.847910202422491e-06,
      "loss": 0.8442,
      "step": 7351
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.708785376643339,
      "learning_rate": 7.847316430844087e-06,
      "loss": 0.7857,
      "step": 7352
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.1618957974259825,
      "learning_rate": 7.84672259983419e-06,
      "loss": 0.8453,
      "step": 7353
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.501265237871065,
      "learning_rate": 7.846128709405192e-06,
      "loss": 0.7608,
      "step": 7354
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.696238247978455,
      "learning_rate": 7.845534759569488e-06,
      "loss": 0.8208,
      "step": 7355
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.623232923079764,
      "learning_rate": 7.84494075033948e-06,
      "loss": 0.7881,
      "step": 7356
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.237142003767484,
      "learning_rate": 7.844346681727564e-06,
      "loss": 0.7713,
      "step": 7357
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.056016373178894,
      "learning_rate": 7.843752553746139e-06,
      "loss": 0.7781,
      "step": 7358
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.051125020974088,
      "learning_rate": 7.843158366407608e-06,
      "loss": 0.7735,
      "step": 7359
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.220364196747527,
      "learning_rate": 7.842564119724372e-06,
      "loss": 0.7977,
      "step": 7360
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.1960591554155515,
      "learning_rate": 7.841969813708835e-06,
      "loss": 0.6877,
      "step": 7361
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.271859734805716,
      "learning_rate": 7.841375448373403e-06,
      "loss": 0.7946,
      "step": 7362
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.730182374400626,
      "learning_rate": 7.84078102373048e-06,
      "loss": 0.8084,
      "step": 7363
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.633062131008258,
      "learning_rate": 7.840186539792474e-06,
      "loss": 0.7502,
      "step": 7364
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.80858529836988,
      "learning_rate": 7.839591996571793e-06,
      "loss": 0.7994,
      "step": 7365
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.372162763752777,
      "learning_rate": 7.83899739408085e-06,
      "loss": 0.7344,
      "step": 7366
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.259772707999068,
      "learning_rate": 7.838402732332053e-06,
      "loss": 0.6923,
      "step": 7367
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.159662510739582,
      "learning_rate": 7.837808011337815e-06,
      "loss": 0.8355,
      "step": 7368
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.443209326393017,
      "learning_rate": 7.83721323111055e-06,
      "loss": 0.7362,
      "step": 7369
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.036826139213099,
      "learning_rate": 7.836618391662671e-06,
      "loss": 0.7695,
      "step": 7370
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.557108750179034,
      "learning_rate": 7.836023493006595e-06,
      "loss": 0.8334,
      "step": 7371
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.878773726073794,
      "learning_rate": 7.83542853515474e-06,
      "loss": 0.7855,
      "step": 7372
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.462918774804227,
      "learning_rate": 7.834833518119524e-06,
      "loss": 0.8511,
      "step": 7373
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.8977479928547325,
      "learning_rate": 7.834238441913366e-06,
      "loss": 0.8534,
      "step": 7374
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.438329995486269,
      "learning_rate": 7.833643306548688e-06,
      "loss": 0.8383,
      "step": 7375
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.479095024010013,
      "learning_rate": 7.833048112037911e-06,
      "loss": 0.7664,
      "step": 7376
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.604747839881378,
      "learning_rate": 7.832452858393459e-06,
      "loss": 0.8273,
      "step": 7377
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.057427305150213,
      "learning_rate": 7.831857545627756e-06,
      "loss": 0.8936,
      "step": 7378
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.029733113100272,
      "learning_rate": 7.831262173753229e-06,
      "loss": 0.804,
      "step": 7379
    },
    {
      "epoch": 0.66,
      "grad_norm": 10.794482306609783,
      "learning_rate": 7.830666742782305e-06,
      "loss": 0.8274,
      "step": 7380
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.4876841709064275,
      "learning_rate": 7.830071252727412e-06,
      "loss": 0.7576,
      "step": 7381
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.745694631135,
      "learning_rate": 7.829475703600979e-06,
      "loss": 0.8292,
      "step": 7382
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.262324673310644,
      "learning_rate": 7.828880095415435e-06,
      "loss": 0.773,
      "step": 7383
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.418367446227206,
      "learning_rate": 7.828284428183217e-06,
      "loss": 0.8672,
      "step": 7384
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.020411379843967,
      "learning_rate": 7.827688701916754e-06,
      "loss": 0.7699,
      "step": 7385
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.6224461938085994,
      "learning_rate": 7.827092916628482e-06,
      "loss": 0.7855,
      "step": 7386
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.566688437953154,
      "learning_rate": 7.826497072330836e-06,
      "loss": 0.861,
      "step": 7387
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.080779276092116,
      "learning_rate": 7.825901169036255e-06,
      "loss": 0.8022,
      "step": 7388
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.300548287793895,
      "learning_rate": 7.825305206757175e-06,
      "loss": 0.7872,
      "step": 7389
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.450541149050415,
      "learning_rate": 7.824709185506036e-06,
      "loss": 0.7452,
      "step": 7390
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.483459206662198,
      "learning_rate": 7.82411310529528e-06,
      "loss": 0.844,
      "step": 7391
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.623636863336024,
      "learning_rate": 7.823516966137346e-06,
      "loss": 0.8186,
      "step": 7392
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.047842117818669,
      "learning_rate": 7.82292076804468e-06,
      "loss": 0.7967,
      "step": 7393
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.9096519901035505,
      "learning_rate": 7.822324511029724e-06,
      "loss": 0.8453,
      "step": 7394
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.552917856349465,
      "learning_rate": 7.821728195104925e-06,
      "loss": 0.7676,
      "step": 7395
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.581830366408834,
      "learning_rate": 7.821131820282729e-06,
      "loss": 0.764,
      "step": 7396
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.9754073774863885,
      "learning_rate": 7.820535386575584e-06,
      "loss": 0.7997,
      "step": 7397
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.17223771946283,
      "learning_rate": 7.819938893995942e-06,
      "loss": 0.802,
      "step": 7398
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.3008576456931085,
      "learning_rate": 7.81934234255625e-06,
      "loss": 0.7824,
      "step": 7399
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.861907931011364,
      "learning_rate": 7.81874573226896e-06,
      "loss": 0.7564,
      "step": 7400
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.185863865137588,
      "learning_rate": 7.818149063146524e-06,
      "loss": 0.8043,
      "step": 7401
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.367895952970188,
      "learning_rate": 7.817552335201401e-06,
      "loss": 0.8444,
      "step": 7402
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.395208810649217,
      "learning_rate": 7.816955548446044e-06,
      "loss": 0.82,
      "step": 7403
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.423413306877656,
      "learning_rate": 7.816358702892907e-06,
      "loss": 0.7631,
      "step": 7404
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.400677541344206,
      "learning_rate": 7.81576179855445e-06,
      "loss": 0.7906,
      "step": 7405
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.3878550346033345,
      "learning_rate": 7.815164835443132e-06,
      "loss": 0.8297,
      "step": 7406
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.049159133847353,
      "learning_rate": 7.814567813571415e-06,
      "loss": 0.7839,
      "step": 7407
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.525131506003655,
      "learning_rate": 7.813970732951756e-06,
      "loss": 0.8564,
      "step": 7408
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.151103755559571,
      "learning_rate": 7.81337359359662e-06,
      "loss": 0.8385,
      "step": 7409
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.845025771948788,
      "learning_rate": 7.812776395518474e-06,
      "loss": 0.7912,
      "step": 7410
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.696669846916749,
      "learning_rate": 7.81217913872978e-06,
      "loss": 0.8103,
      "step": 7411
    },
    {
      "epoch": 0.66,
      "grad_norm": 9.52335312768078,
      "learning_rate": 7.811581823243003e-06,
      "loss": 0.8263,
      "step": 7412
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.483031821265717,
      "learning_rate": 7.810984449070616e-06,
      "loss": 0.8597,
      "step": 7413
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.00420843402275,
      "learning_rate": 7.810387016225082e-06,
      "loss": 0.8941,
      "step": 7414
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.699197019486055,
      "learning_rate": 7.809789524718875e-06,
      "loss": 0.8253,
      "step": 7415
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.5767196391763285,
      "learning_rate": 7.809191974564464e-06,
      "loss": 0.8272,
      "step": 7416
    },
    {
      "epoch": 0.66,
      "grad_norm": 8.343788810953756,
      "learning_rate": 7.808594365774323e-06,
      "loss": 0.7648,
      "step": 7417
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.169236519126121,
      "learning_rate": 7.807996698360926e-06,
      "loss": 0.8837,
      "step": 7418
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.411817226572251,
      "learning_rate": 7.807398972336747e-06,
      "loss": 0.7743,
      "step": 7419
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.034283214494565,
      "learning_rate": 7.806801187714262e-06,
      "loss": 0.7705,
      "step": 7420
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.098984401105007,
      "learning_rate": 7.80620334450595e-06,
      "loss": 0.7011,
      "step": 7421
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.542192010732576,
      "learning_rate": 7.805605442724286e-06,
      "loss": 0.7623,
      "step": 7422
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.482261402943133,
      "learning_rate": 7.805007482381755e-06,
      "loss": 0.7543,
      "step": 7423
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.410804410738877,
      "learning_rate": 7.804409463490836e-06,
      "loss": 0.9009,
      "step": 7424
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.267569412633374,
      "learning_rate": 7.80381138606401e-06,
      "loss": 0.7779,
      "step": 7425
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.83310431261269,
      "learning_rate": 7.803213250113761e-06,
      "loss": 0.7864,
      "step": 7426
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.060449039044572,
      "learning_rate": 7.802615055652577e-06,
      "loss": 0.8205,
      "step": 7427
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.092746689300448,
      "learning_rate": 7.802016802692937e-06,
      "loss": 0.7702,
      "step": 7428
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.3893760787985485,
      "learning_rate": 7.801418491247336e-06,
      "loss": 0.8443,
      "step": 7429
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.115477280380238,
      "learning_rate": 7.800820121328259e-06,
      "loss": 0.7831,
      "step": 7430
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.994712298719262,
      "learning_rate": 7.800221692948194e-06,
      "loss": 0.7969,
      "step": 7431
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.802075469705073,
      "learning_rate": 7.799623206119634e-06,
      "loss": 0.8362,
      "step": 7432
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.757693675074636,
      "learning_rate": 7.799024660855072e-06,
      "loss": 0.7597,
      "step": 7433
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.942355293544224,
      "learning_rate": 7.798426057166998e-06,
      "loss": 0.7899,
      "step": 7434
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.3246267469651105,
      "learning_rate": 7.797827395067909e-06,
      "loss": 0.7243,
      "step": 7435
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.637842070657123,
      "learning_rate": 7.7972286745703e-06,
      "loss": 0.763,
      "step": 7436
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.507162777412712,
      "learning_rate": 7.796629895686669e-06,
      "loss": 0.7411,
      "step": 7437
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.92817626786634,
      "learning_rate": 7.796031058429513e-06,
      "loss": 0.8309,
      "step": 7438
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.472058410044816,
      "learning_rate": 7.795432162811331e-06,
      "loss": 0.8117,
      "step": 7439
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.170945857249831,
      "learning_rate": 7.794833208844627e-06,
      "loss": 0.8012,
      "step": 7440
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.4650173980939165,
      "learning_rate": 7.794234196541898e-06,
      "loss": 0.856,
      "step": 7441
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.742274310277386,
      "learning_rate": 7.79363512591565e-06,
      "loss": 0.7294,
      "step": 7442
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.641570779242813,
      "learning_rate": 7.79303599697839e-06,
      "loss": 0.868,
      "step": 7443
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.387013757077916,
      "learning_rate": 7.792436809742617e-06,
      "loss": 0.7884,
      "step": 7444
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.591329216991061,
      "learning_rate": 7.791837564220838e-06,
      "loss": 0.7999,
      "step": 7445
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.7735798418799895,
      "learning_rate": 7.791238260425568e-06,
      "loss": 0.7094,
      "step": 7446
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.338181744880929,
      "learning_rate": 7.790638898369312e-06,
      "loss": 0.765,
      "step": 7447
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.812979902052341,
      "learning_rate": 7.790039478064579e-06,
      "loss": 0.7765,
      "step": 7448
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.609783915680946,
      "learning_rate": 7.789439999523883e-06,
      "loss": 0.7937,
      "step": 7449
    },
    {
      "epoch": 0.66,
      "grad_norm": 10.027366712764692,
      "learning_rate": 7.788840462759736e-06,
      "loss": 0.8173,
      "step": 7450
    },
    {
      "epoch": 0.66,
      "grad_norm": 10.224786400190451,
      "learning_rate": 7.788240867784648e-06,
      "loss": 0.8742,
      "step": 7451
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.074150603589031,
      "learning_rate": 7.787641214611142e-06,
      "loss": 0.824,
      "step": 7452
    },
    {
      "epoch": 0.66,
      "grad_norm": 8.056629318735109,
      "learning_rate": 7.787041503251731e-06,
      "loss": 0.7922,
      "step": 7453
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.8789505612724815,
      "learning_rate": 7.78644173371893e-06,
      "loss": 0.7773,
      "step": 7454
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.038437330754451,
      "learning_rate": 7.785841906025261e-06,
      "loss": 0.8071,
      "step": 7455
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.315954357882153,
      "learning_rate": 7.785242020183243e-06,
      "loss": 0.8025,
      "step": 7456
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.409234553980946,
      "learning_rate": 7.784642076205396e-06,
      "loss": 0.753,
      "step": 7457
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.810209668398402,
      "learning_rate": 7.784042074104246e-06,
      "loss": 0.7644,
      "step": 7458
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.880103522416976,
      "learning_rate": 7.783442013892313e-06,
      "loss": 0.7407,
      "step": 7459
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.452319557353951,
      "learning_rate": 7.782841895582125e-06,
      "loss": 0.7745,
      "step": 7460
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.719329245029103,
      "learning_rate": 7.782241719186206e-06,
      "loss": 0.8035,
      "step": 7461
    },
    {
      "epoch": 0.67,
      "grad_norm": 8.243621961409922,
      "learning_rate": 7.781641484717085e-06,
      "loss": 0.7511,
      "step": 7462
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.055682719721051,
      "learning_rate": 7.78104119218729e-06,
      "loss": 0.7768,
      "step": 7463
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.941090867236563,
      "learning_rate": 7.780440841609347e-06,
      "loss": 0.7124,
      "step": 7464
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.720256993964334,
      "learning_rate": 7.779840432995793e-06,
      "loss": 0.8388,
      "step": 7465
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.335830485627826,
      "learning_rate": 7.779239966359158e-06,
      "loss": 0.7909,
      "step": 7466
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.353500822085093,
      "learning_rate": 7.778639441711975e-06,
      "loss": 0.7893,
      "step": 7467
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.026685698017558,
      "learning_rate": 7.778038859066778e-06,
      "loss": 0.7928,
      "step": 7468
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.555662163050832,
      "learning_rate": 7.777438218436104e-06,
      "loss": 0.8814,
      "step": 7469
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.523573302708189,
      "learning_rate": 7.77683751983249e-06,
      "loss": 0.7544,
      "step": 7470
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.371150779549276,
      "learning_rate": 7.776236763268474e-06,
      "loss": 0.7943,
      "step": 7471
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.322112233881904,
      "learning_rate": 7.775635948756594e-06,
      "loss": 0.7966,
      "step": 7472
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.860584656954639,
      "learning_rate": 7.775035076309395e-06,
      "loss": 0.7757,
      "step": 7473
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.46546717247418,
      "learning_rate": 7.774434145939414e-06,
      "loss": 0.8296,
      "step": 7474
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.9721070630061295,
      "learning_rate": 7.773833157659197e-06,
      "loss": 0.776,
      "step": 7475
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.8483853880028205,
      "learning_rate": 7.773232111481285e-06,
      "loss": 0.8011,
      "step": 7476
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.615022459926625,
      "learning_rate": 7.772631007418229e-06,
      "loss": 0.8053,
      "step": 7477
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.224253464396829,
      "learning_rate": 7.77202984548257e-06,
      "loss": 0.7753,
      "step": 7478
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.284333998130052,
      "learning_rate": 7.771428625686861e-06,
      "loss": 0.7504,
      "step": 7479
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.905304618450035,
      "learning_rate": 7.770827348043648e-06,
      "loss": 0.7692,
      "step": 7480
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.87468582521652,
      "learning_rate": 7.77022601256548e-06,
      "loss": 0.7123,
      "step": 7481
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.888654115754701,
      "learning_rate": 7.769624619264914e-06,
      "loss": 0.8174,
      "step": 7482
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.664896813535755,
      "learning_rate": 7.769023168154498e-06,
      "loss": 0.7847,
      "step": 7483
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.471761227961322,
      "learning_rate": 7.768421659246787e-06,
      "loss": 0.7603,
      "step": 7484
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.098265012694944,
      "learning_rate": 7.767820092554338e-06,
      "loss": 0.7946,
      "step": 7485
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.559392064449401,
      "learning_rate": 7.767218468089704e-06,
      "loss": 0.8049,
      "step": 7486
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.059632232186752,
      "learning_rate": 7.766616785865445e-06,
      "loss": 0.7467,
      "step": 7487
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.545976807050508,
      "learning_rate": 7.766015045894121e-06,
      "loss": 0.7692,
      "step": 7488
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.9923207491867045,
      "learning_rate": 7.76541324818829e-06,
      "loss": 0.778,
      "step": 7489
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.339795473067474,
      "learning_rate": 7.764811392760513e-06,
      "loss": 0.8989,
      "step": 7490
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.574408205552939,
      "learning_rate": 7.764209479623352e-06,
      "loss": 0.7727,
      "step": 7491
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.294976042389653,
      "learning_rate": 7.763607508789373e-06,
      "loss": 0.7312,
      "step": 7492
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.021326947507948,
      "learning_rate": 7.763005480271139e-06,
      "loss": 0.768,
      "step": 7493
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.924394715015294,
      "learning_rate": 7.762403394081217e-06,
      "loss": 0.8924,
      "step": 7494
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.939244244236905,
      "learning_rate": 7.761801250232173e-06,
      "loss": 0.789,
      "step": 7495
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.897915601585195,
      "learning_rate": 7.761199048736577e-06,
      "loss": 0.9073,
      "step": 7496
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.388472526516305,
      "learning_rate": 7.760596789606997e-06,
      "loss": 0.9327,
      "step": 7497
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.483017563935484,
      "learning_rate": 7.759994472856004e-06,
      "loss": 0.816,
      "step": 7498
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.7708611944107275,
      "learning_rate": 7.759392098496173e-06,
      "loss": 0.7574,
      "step": 7499
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.199861090391169,
      "learning_rate": 7.758789666540073e-06,
      "loss": 0.788,
      "step": 7500
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.2514675308991485,
      "learning_rate": 7.75818717700028e-06,
      "loss": 0.7373,
      "step": 7501
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.2848284098652405,
      "learning_rate": 7.757584629889371e-06,
      "loss": 0.7811,
      "step": 7502
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.055339403271627,
      "learning_rate": 7.756982025219921e-06,
      "loss": 0.8211,
      "step": 7503
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.30030436019759,
      "learning_rate": 7.756379363004511e-06,
      "loss": 0.7305,
      "step": 7504
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.288245604262926,
      "learning_rate": 7.755776643255716e-06,
      "loss": 0.7964,
      "step": 7505
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.399847616934398,
      "learning_rate": 7.75517386598612e-06,
      "loss": 0.8381,
      "step": 7506
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.072708138317721,
      "learning_rate": 7.754571031208303e-06,
      "loss": 0.8298,
      "step": 7507
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.969950788593255,
      "learning_rate": 7.753968138934847e-06,
      "loss": 0.7781,
      "step": 7508
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.8941821459111083,
      "learning_rate": 7.753365189178338e-06,
      "loss": 0.7307,
      "step": 7509
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.2469069051309,
      "learning_rate": 7.752762181951362e-06,
      "loss": 0.826,
      "step": 7510
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.800080248190455,
      "learning_rate": 7.7521591172665e-06,
      "loss": 0.8099,
      "step": 7511
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.605009777553003,
      "learning_rate": 7.751555995136345e-06,
      "loss": 0.8097,
      "step": 7512
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.740456209253779,
      "learning_rate": 7.750952815573488e-06,
      "loss": 0.7803,
      "step": 7513
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.563251671931665,
      "learning_rate": 7.750349578590512e-06,
      "loss": 0.8143,
      "step": 7514
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.676702229494565,
      "learning_rate": 7.749746284200012e-06,
      "loss": 0.8346,
      "step": 7515
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.681368341302862,
      "learning_rate": 7.74914293241458e-06,
      "loss": 0.7919,
      "step": 7516
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.234965002351286,
      "learning_rate": 7.74853952324681e-06,
      "loss": 0.7802,
      "step": 7517
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.145384294399352,
      "learning_rate": 7.747936056709297e-06,
      "loss": 0.7165,
      "step": 7518
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.408297329191453,
      "learning_rate": 7.747332532814638e-06,
      "loss": 0.8071,
      "step": 7519
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.3917498614157715,
      "learning_rate": 7.746728951575427e-06,
      "loss": 0.7524,
      "step": 7520
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.936195231288735,
      "learning_rate": 7.746125313004265e-06,
      "loss": 0.8054,
      "step": 7521
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.043188812864851,
      "learning_rate": 7.745521617113751e-06,
      "loss": 0.8053,
      "step": 7522
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.346574341615688,
      "learning_rate": 7.744917863916487e-06,
      "loss": 0.7975,
      "step": 7523
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.546262937543969,
      "learning_rate": 7.74431405342507e-06,
      "loss": 0.7468,
      "step": 7524
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.695233951053181,
      "learning_rate": 7.743710185652111e-06,
      "loss": 0.7974,
      "step": 7525
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.4771025845503445,
      "learning_rate": 7.74310626061021e-06,
      "loss": 0.7589,
      "step": 7526
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.121654099150269,
      "learning_rate": 7.742502278311972e-06,
      "loss": 0.7545,
      "step": 7527
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.8252625151652415,
      "learning_rate": 7.741898238770005e-06,
      "loss": 0.8069,
      "step": 7528
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.724082015941043,
      "learning_rate": 7.741294141996919e-06,
      "loss": 0.7036,
      "step": 7529
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.8175877473392905,
      "learning_rate": 7.740689988005318e-06,
      "loss": 0.7586,
      "step": 7530
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.325114379816399,
      "learning_rate": 7.740085776807817e-06,
      "loss": 0.8575,
      "step": 7531
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.934851664176571,
      "learning_rate": 7.739481508417024e-06,
      "loss": 0.8395,
      "step": 7532
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.877481984389563,
      "learning_rate": 7.738877182845557e-06,
      "loss": 0.8353,
      "step": 7533
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.92148218063268,
      "learning_rate": 7.738272800106026e-06,
      "loss": 0.779,
      "step": 7534
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.2079132865382745,
      "learning_rate": 7.737668360211047e-06,
      "loss": 0.8278,
      "step": 7535
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.319059423578033,
      "learning_rate": 7.737063863173236e-06,
      "loss": 0.7888,
      "step": 7536
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.314627972334559,
      "learning_rate": 7.736459309005211e-06,
      "loss": 0.7394,
      "step": 7537
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.9155800687412485,
      "learning_rate": 7.735854697719591e-06,
      "loss": 0.8245,
      "step": 7538
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.212602533842416,
      "learning_rate": 7.735250029328994e-06,
      "loss": 0.7774,
      "step": 7539
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.712850254882772,
      "learning_rate": 7.734645303846045e-06,
      "loss": 0.7865,
      "step": 7540
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.122854507520816,
      "learning_rate": 7.734040521283363e-06,
      "loss": 0.6799,
      "step": 7541
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.198791734124865,
      "learning_rate": 7.733435681653571e-06,
      "loss": 0.7939,
      "step": 7542
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.410299578072398,
      "learning_rate": 7.732830784969299e-06,
      "loss": 0.8284,
      "step": 7543
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.983794122480936,
      "learning_rate": 7.732225831243166e-06,
      "loss": 0.874,
      "step": 7544
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.8959525825583015,
      "learning_rate": 7.731620820487804e-06,
      "loss": 0.8404,
      "step": 7545
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.3005387383297515,
      "learning_rate": 7.731015752715837e-06,
      "loss": 0.8193,
      "step": 7546
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.124731011279562,
      "learning_rate": 7.7304106279399e-06,
      "loss": 0.7781,
      "step": 7547
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.4392462130905495,
      "learning_rate": 7.72980544617262e-06,
      "loss": 0.765,
      "step": 7548
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.752957594871339,
      "learning_rate": 7.729200207426628e-06,
      "loss": 0.7225,
      "step": 7549
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.470868470252671,
      "learning_rate": 7.72859491171456e-06,
      "loss": 0.7905,
      "step": 7550
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.92229975792626,
      "learning_rate": 7.727989559049048e-06,
      "loss": 0.7857,
      "step": 7551
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.300262247435067,
      "learning_rate": 7.727384149442729e-06,
      "loss": 0.8547,
      "step": 7552
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.226955130946679,
      "learning_rate": 7.726778682908239e-06,
      "loss": 0.7752,
      "step": 7553
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.623105866623984,
      "learning_rate": 7.726173159458213e-06,
      "loss": 0.8225,
      "step": 7554
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.235355999002692,
      "learning_rate": 7.725567579105295e-06,
      "loss": 0.8277,
      "step": 7555
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.986799402747926,
      "learning_rate": 7.72496194186212e-06,
      "loss": 0.7906,
      "step": 7556
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.16504428184052,
      "learning_rate": 7.724356247741335e-06,
      "loss": 0.7392,
      "step": 7557
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.3687765881955345,
      "learning_rate": 7.723750496755578e-06,
      "loss": 0.7648,
      "step": 7558
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.9853025667174045,
      "learning_rate": 7.723144688917494e-06,
      "loss": 0.7954,
      "step": 7559
    },
    {
      "epoch": 0.67,
      "grad_norm": 8.10396475214905,
      "learning_rate": 7.72253882423973e-06,
      "loss": 0.8374,
      "step": 7560
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.353069816884688,
      "learning_rate": 7.721932902734929e-06,
      "loss": 0.8016,
      "step": 7561
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.4135523640419345,
      "learning_rate": 7.721326924415739e-06,
      "loss": 0.7364,
      "step": 7562
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.8622763979619656,
      "learning_rate": 7.720720889294811e-06,
      "loss": 0.7258,
      "step": 7563
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.744100049264777,
      "learning_rate": 7.720114797384791e-06,
      "loss": 0.7474,
      "step": 7564
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.697409236372265,
      "learning_rate": 7.719508648698332e-06,
      "loss": 0.7554,
      "step": 7565
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.553193568860331,
      "learning_rate": 7.718902443248086e-06,
      "loss": 0.7675,
      "step": 7566
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.969497420222376,
      "learning_rate": 7.718296181046705e-06,
      "loss": 0.7864,
      "step": 7567
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.961430525761238,
      "learning_rate": 7.717689862106844e-06,
      "loss": 0.8709,
      "step": 7568
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.3796229010381165,
      "learning_rate": 7.71708348644116e-06,
      "loss": 0.7679,
      "step": 7569
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.654873177011836,
      "learning_rate": 7.716477054062308e-06,
      "loss": 0.8204,
      "step": 7570
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.328227211049488,
      "learning_rate": 7.715870564982947e-06,
      "loss": 0.7799,
      "step": 7571
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.327583361305452,
      "learning_rate": 7.715264019215734e-06,
      "loss": 0.8037,
      "step": 7572
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.132519949781146,
      "learning_rate": 7.714657416773332e-06,
      "loss": 0.913,
      "step": 7573
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.397600225276254,
      "learning_rate": 7.714050757668403e-06,
      "loss": 0.8172,
      "step": 7574
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.754757228857548,
      "learning_rate": 7.713444041913604e-06,
      "loss": 0.7836,
      "step": 7575
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.95741302463087,
      "learning_rate": 7.712837269521608e-06,
      "loss": 0.8251,
      "step": 7576
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.723652098968439,
      "learning_rate": 7.712230440505072e-06,
      "loss": 0.7561,
      "step": 7577
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.007896900117036,
      "learning_rate": 7.711623554876666e-06,
      "loss": 0.832,
      "step": 7578
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.478615476850892,
      "learning_rate": 7.711016612649056e-06,
      "loss": 0.7798,
      "step": 7579
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.728711126555095,
      "learning_rate": 7.71040961383491e-06,
      "loss": 0.8031,
      "step": 7580
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.894156152622334,
      "learning_rate": 7.709802558446902e-06,
      "loss": 0.8448,
      "step": 7581
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.537927602880863,
      "learning_rate": 7.709195446497698e-06,
      "loss": 0.7758,
      "step": 7582
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.113675294480688,
      "learning_rate": 7.708588277999974e-06,
      "loss": 0.7387,
      "step": 7583
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.2939288985472555,
      "learning_rate": 7.7079810529664e-06,
      "loss": 0.7805,
      "step": 7584
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.5966073155328235,
      "learning_rate": 7.70737377140965e-06,
      "loss": 0.8126,
      "step": 7585
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.111686144306963,
      "learning_rate": 7.706766433342406e-06,
      "loss": 0.8508,
      "step": 7586
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.223499834489987,
      "learning_rate": 7.706159038777336e-06,
      "loss": 0.7827,
      "step": 7587
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.757637621094529,
      "learning_rate": 7.705551587727125e-06,
      "loss": 0.7415,
      "step": 7588
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.901784190731154,
      "learning_rate": 7.704944080204449e-06,
      "loss": 0.7484,
      "step": 7589
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.571368981992699,
      "learning_rate": 7.704336516221989e-06,
      "loss": 0.8625,
      "step": 7590
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.124041521442684,
      "learning_rate": 7.703728895792428e-06,
      "loss": 0.8718,
      "step": 7591
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.470537489038532,
      "learning_rate": 7.703121218928445e-06,
      "loss": 0.829,
      "step": 7592
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.302213240823461,
      "learning_rate": 7.702513485642727e-06,
      "loss": 0.8686,
      "step": 7593
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.7221579850980735,
      "learning_rate": 7.701905695947959e-06,
      "loss": 0.8127,
      "step": 7594
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.666509447114909,
      "learning_rate": 7.701297849856825e-06,
      "loss": 0.7666,
      "step": 7595
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.848931954086172,
      "learning_rate": 7.700689947382017e-06,
      "loss": 0.8003,
      "step": 7596
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.795657633408427,
      "learning_rate": 7.700081988536219e-06,
      "loss": 0.8089,
      "step": 7597
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.526746166157627,
      "learning_rate": 7.699473973332123e-06,
      "loss": 0.7647,
      "step": 7598
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.495556037668056,
      "learning_rate": 7.69886590178242e-06,
      "loss": 0.8053,
      "step": 7599
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.623456839049134,
      "learning_rate": 7.6982577738998e-06,
      "loss": 0.7484,
      "step": 7600
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.387415474864637,
      "learning_rate": 7.69764958969696e-06,
      "loss": 0.7736,
      "step": 7601
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.472397997316965,
      "learning_rate": 7.697041349186593e-06,
      "loss": 0.7808,
      "step": 7602
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.439914276946039,
      "learning_rate": 7.696433052381392e-06,
      "loss": 0.7855,
      "step": 7603
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.669193703937444,
      "learning_rate": 7.695824699294056e-06,
      "loss": 0.7698,
      "step": 7604
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.934326363998305,
      "learning_rate": 7.695216289937287e-06,
      "loss": 0.8101,
      "step": 7605
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.5523741864640925,
      "learning_rate": 7.694607824323777e-06,
      "loss": 0.7511,
      "step": 7606
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.796093406215694,
      "learning_rate": 7.69399930246623e-06,
      "loss": 0.8755,
      "step": 7607
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.828821947883862,
      "learning_rate": 7.693390724377348e-06,
      "loss": 0.8129,
      "step": 7608
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.164511882673904,
      "learning_rate": 7.692782090069832e-06,
      "loss": 0.7354,
      "step": 7609
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.277087166028941,
      "learning_rate": 7.692173399556388e-06,
      "loss": 0.8081,
      "step": 7610
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.1231503211088745,
      "learning_rate": 7.69156465284972e-06,
      "loss": 0.8269,
      "step": 7611
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.121707824418647,
      "learning_rate": 7.690955849962534e-06,
      "loss": 0.7238,
      "step": 7612
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.42702493899873,
      "learning_rate": 7.690346990907538e-06,
      "loss": 0.7923,
      "step": 7613
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.322468672201752,
      "learning_rate": 7.68973807569744e-06,
      "loss": 0.7404,
      "step": 7614
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.664416811582935,
      "learning_rate": 7.68912910434495e-06,
      "loss": 0.781,
      "step": 7615
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.232354803837314,
      "learning_rate": 7.68852007686278e-06,
      "loss": 0.8114,
      "step": 7616
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.493728268618887,
      "learning_rate": 7.68791099326364e-06,
      "loss": 0.7949,
      "step": 7617
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.537499436920661,
      "learning_rate": 7.687301853560245e-06,
      "loss": 0.8621,
      "step": 7618
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.298623530139193,
      "learning_rate": 7.686692657765309e-06,
      "loss": 0.7395,
      "step": 7619
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.331878587684508,
      "learning_rate": 7.686083405891547e-06,
      "loss": 0.7121,
      "step": 7620
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.314164479993293,
      "learning_rate": 7.685474097951678e-06,
      "loss": 0.8244,
      "step": 7621
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.927201992974882,
      "learning_rate": 7.684864733958416e-06,
      "loss": 0.7753,
      "step": 7622
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.387592043297697,
      "learning_rate": 7.684255313924482e-06,
      "loss": 0.7903,
      "step": 7623
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.0540792298762405,
      "learning_rate": 7.6836458378626e-06,
      "loss": 0.8249,
      "step": 7624
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.276759778654334,
      "learning_rate": 7.683036305785485e-06,
      "loss": 0.8355,
      "step": 7625
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.962711074640727,
      "learning_rate": 7.682426717705864e-06,
      "loss": 0.7472,
      "step": 7626
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.952867844412923,
      "learning_rate": 7.68181707363646e-06,
      "loss": 0.7695,
      "step": 7627
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.008032312956153,
      "learning_rate": 7.681207373589998e-06,
      "loss": 0.8715,
      "step": 7628
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.717460136143103,
      "learning_rate": 7.680597617579204e-06,
      "loss": 0.8474,
      "step": 7629
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.223582580563883,
      "learning_rate": 7.679987805616804e-06,
      "loss": 0.7835,
      "step": 7630
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.970207876097407,
      "learning_rate": 7.679377937715529e-06,
      "loss": 0.7749,
      "step": 7631
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.6929326020658175,
      "learning_rate": 7.678768013888106e-06,
      "loss": 0.7902,
      "step": 7632
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.095250175947519,
      "learning_rate": 7.678158034147266e-06,
      "loss": 0.8595,
      "step": 7633
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.5298173502012276,
      "learning_rate": 7.677547998505746e-06,
      "loss": 0.7931,
      "step": 7634
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.559817785456491,
      "learning_rate": 7.676937906976272e-06,
      "loss": 0.7747,
      "step": 7635
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.501244567035963,
      "learning_rate": 7.676327759571583e-06,
      "loss": 0.8002,
      "step": 7636
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.7674458629614955,
      "learning_rate": 7.675717556304412e-06,
      "loss": 0.7882,
      "step": 7637
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.689030379873994,
      "learning_rate": 7.675107297187498e-06,
      "loss": 0.8238,
      "step": 7638
    },
    {
      "epoch": 0.68,
      "grad_norm": 11.13753738789413,
      "learning_rate": 7.674496982233576e-06,
      "loss": 0.8597,
      "step": 7639
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.102618248338764,
      "learning_rate": 7.673886611455388e-06,
      "loss": 0.7381,
      "step": 7640
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.108141908473352,
      "learning_rate": 7.673276184865672e-06,
      "loss": 0.7809,
      "step": 7641
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.755688625291136,
      "learning_rate": 7.67266570247717e-06,
      "loss": 0.8006,
      "step": 7642
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.188839246472388,
      "learning_rate": 7.672055164302624e-06,
      "loss": 0.7609,
      "step": 7643
    },
    {
      "epoch": 0.68,
      "grad_norm": 8.677550120517731,
      "learning_rate": 7.67144457035478e-06,
      "loss": 0.7007,
      "step": 7644
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.533281272204738,
      "learning_rate": 7.670833920646378e-06,
      "loss": 0.8053,
      "step": 7645
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.68738023831253,
      "learning_rate": 7.67022321519017e-06,
      "loss": 0.7849,
      "step": 7646
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.280047136552515,
      "learning_rate": 7.669612453998899e-06,
      "loss": 0.8099,
      "step": 7647
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.006534606727385,
      "learning_rate": 7.669001637085313e-06,
      "loss": 0.8352,
      "step": 7648
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.347157511480133,
      "learning_rate": 7.668390764462163e-06,
      "loss": 0.7394,
      "step": 7649
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.683916945860286,
      "learning_rate": 7.667779836142201e-06,
      "loss": 0.7896,
      "step": 7650
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.351700824810665,
      "learning_rate": 7.667168852138178e-06,
      "loss": 0.8193,
      "step": 7651
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.98005612380919,
      "learning_rate": 7.666557812462844e-06,
      "loss": 0.8473,
      "step": 7652
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.027058815739654,
      "learning_rate": 7.665946717128955e-06,
      "loss": 0.7079,
      "step": 7653
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.810606430972879,
      "learning_rate": 7.66533556614927e-06,
      "loss": 0.8989,
      "step": 7654
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.806155762886301,
      "learning_rate": 7.664724359536538e-06,
      "loss": 0.7961,
      "step": 7655
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.405791466740857,
      "learning_rate": 7.664113097303525e-06,
      "loss": 0.7249,
      "step": 7656
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.169939066397133,
      "learning_rate": 7.663501779462982e-06,
      "loss": 0.8094,
      "step": 7657
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.399083939796335,
      "learning_rate": 7.662890406027673e-06,
      "loss": 0.8183,
      "step": 7658
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.382317761372601,
      "learning_rate": 7.662278977010359e-06,
      "loss": 0.8083,
      "step": 7659
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.885688943320695,
      "learning_rate": 7.661667492423801e-06,
      "loss": 0.8369,
      "step": 7660
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.2572005973740294,
      "learning_rate": 7.661055952280762e-06,
      "loss": 0.7978,
      "step": 7661
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.567625991294728,
      "learning_rate": 7.66044435659401e-06,
      "loss": 0.8259,
      "step": 7662
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.931050309975084,
      "learning_rate": 7.659832705376307e-06,
      "loss": 0.7678,
      "step": 7663
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.445643445279682,
      "learning_rate": 7.659220998640423e-06,
      "loss": 0.6865,
      "step": 7664
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.911501034946819,
      "learning_rate": 7.658609236399121e-06,
      "loss": 0.8047,
      "step": 7665
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.8075100553659365,
      "learning_rate": 7.657997418665176e-06,
      "loss": 0.8405,
      "step": 7666
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.747992427245949,
      "learning_rate": 7.657385545451355e-06,
      "loss": 0.7816,
      "step": 7667
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.909697654096059,
      "learning_rate": 7.656773616770429e-06,
      "loss": 0.7786,
      "step": 7668
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.4223571957077725,
      "learning_rate": 7.656161632635174e-06,
      "loss": 0.7798,
      "step": 7669
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.903839068644271,
      "learning_rate": 7.65554959305836e-06,
      "loss": 0.7862,
      "step": 7670
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.134746282712212,
      "learning_rate": 7.654937498052766e-06,
      "loss": 0.7988,
      "step": 7671
    },
    {
      "epoch": 0.68,
      "grad_norm": 8.424630766692243,
      "learning_rate": 7.654325347631165e-06,
      "loss": 0.8057,
      "step": 7672
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.262187716173571,
      "learning_rate": 7.653713141806334e-06,
      "loss": 0.8335,
      "step": 7673
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.283608874631508,
      "learning_rate": 7.653100880591054e-06,
      "loss": 0.7552,
      "step": 7674
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.384746040310324,
      "learning_rate": 7.652488563998103e-06,
      "loss": 0.8317,
      "step": 7675
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.871801048686925,
      "learning_rate": 7.651876192040262e-06,
      "loss": 0.7437,
      "step": 7676
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.164777094118929,
      "learning_rate": 7.651263764730313e-06,
      "loss": 0.7768,
      "step": 7677
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.840105261437366,
      "learning_rate": 7.65065128208104e-06,
      "loss": 0.7928,
      "step": 7678
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.29190650242915,
      "learning_rate": 7.650038744105226e-06,
      "loss": 0.8085,
      "step": 7679
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.289565494745867,
      "learning_rate": 7.649426150815656e-06,
      "loss": 0.7966,
      "step": 7680
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.826725641641609,
      "learning_rate": 7.648813502225117e-06,
      "loss": 0.7999,
      "step": 7681
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.729576577581364,
      "learning_rate": 7.648200798346397e-06,
      "loss": 0.822,
      "step": 7682
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.978937697706007,
      "learning_rate": 7.647588039192286e-06,
      "loss": 0.7612,
      "step": 7683
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.410845851931574,
      "learning_rate": 7.646975224775573e-06,
      "loss": 0.8525,
      "step": 7684
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.081322451979987,
      "learning_rate": 7.646362355109047e-06,
      "loss": 0.8499,
      "step": 7685
    },
    {
      "epoch": 0.69,
      "grad_norm": 8.224663423942024,
      "learning_rate": 7.645749430205504e-06,
      "loss": 0.7714,
      "step": 7686
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.78383570324417,
      "learning_rate": 7.645136450077735e-06,
      "loss": 0.7617,
      "step": 7687
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.694176786141585,
      "learning_rate": 7.644523414738535e-06,
      "loss": 0.7775,
      "step": 7688
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.6513403183789634,
      "learning_rate": 7.6439103242007e-06,
      "loss": 0.7672,
      "step": 7689
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.804358370771572,
      "learning_rate": 7.643297178477027e-06,
      "loss": 0.8118,
      "step": 7690
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.53979341748493,
      "learning_rate": 7.642683977580315e-06,
      "loss": 0.8309,
      "step": 7691
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.062637954352693,
      "learning_rate": 7.642070721523363e-06,
      "loss": 0.7343,
      "step": 7692
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.13969727212979,
      "learning_rate": 7.641457410318969e-06,
      "loss": 0.7687,
      "step": 7693
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.541801383950333,
      "learning_rate": 7.640844043979938e-06,
      "loss": 0.7322,
      "step": 7694
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.488669727563981,
      "learning_rate": 7.640230622519069e-06,
      "loss": 0.7476,
      "step": 7695
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.18995436725868,
      "learning_rate": 7.63961714594917e-06,
      "loss": 0.7792,
      "step": 7696
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.93590377495648,
      "learning_rate": 7.639003614283041e-06,
      "loss": 0.7273,
      "step": 7697
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.155477598379602,
      "learning_rate": 7.638390027533493e-06,
      "loss": 0.7782,
      "step": 7698
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.577477857257526,
      "learning_rate": 7.63777638571333e-06,
      "loss": 0.811,
      "step": 7699
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.589771586501075,
      "learning_rate": 7.637162688835363e-06,
      "loss": 0.7271,
      "step": 7700
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.179842247270028,
      "learning_rate": 7.636548936912398e-06,
      "loss": 0.7791,
      "step": 7701
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.172001204925341,
      "learning_rate": 7.635935129957248e-06,
      "loss": 0.7463,
      "step": 7702
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.1689630207134645,
      "learning_rate": 7.635321267982727e-06,
      "loss": 0.7943,
      "step": 7703
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.789810060525493,
      "learning_rate": 7.634707351001645e-06,
      "loss": 0.8055,
      "step": 7704
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.711283655182005,
      "learning_rate": 7.634093379026816e-06,
      "loss": 0.7667,
      "step": 7705
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.424836307989746,
      "learning_rate": 7.633479352071055e-06,
      "loss": 0.8493,
      "step": 7706
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.819645503478086,
      "learning_rate": 7.632865270147184e-06,
      "loss": 0.754,
      "step": 7707
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.229314484424215,
      "learning_rate": 7.632251133268014e-06,
      "loss": 0.7243,
      "step": 7708
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.700344300412818,
      "learning_rate": 7.631636941446365e-06,
      "loss": 0.8331,
      "step": 7709
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.221842818951836,
      "learning_rate": 7.63102269469506e-06,
      "loss": 0.7916,
      "step": 7710
    },
    {
      "epoch": 0.69,
      "grad_norm": 8.931643693120213,
      "learning_rate": 7.630408393026917e-06,
      "loss": 0.8178,
      "step": 7711
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.21715536672453,
      "learning_rate": 7.629794036454758e-06,
      "loss": 0.7423,
      "step": 7712
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.183891450762643,
      "learning_rate": 7.629179624991409e-06,
      "loss": 0.756,
      "step": 7713
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.247314115658371,
      "learning_rate": 7.628565158649695e-06,
      "loss": 0.8435,
      "step": 7714
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.637357420647921,
      "learning_rate": 7.627950637442438e-06,
      "loss": 0.7799,
      "step": 7715
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.755633476780596,
      "learning_rate": 7.627336061382467e-06,
      "loss": 0.7856,
      "step": 7716
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.680912597047716,
      "learning_rate": 7.626721430482609e-06,
      "loss": 0.7868,
      "step": 7717
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.3313107631818974,
      "learning_rate": 7.626106744755693e-06,
      "loss": 0.7486,
      "step": 7718
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.728963927086245,
      "learning_rate": 7.6254920042145495e-06,
      "loss": 0.8309,
      "step": 7719
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.792120401205434,
      "learning_rate": 7.624877208872013e-06,
      "loss": 0.8733,
      "step": 7720
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.096236477487446,
      "learning_rate": 7.624262358740911e-06,
      "loss": 0.7895,
      "step": 7721
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.945059947329456,
      "learning_rate": 7.623647453834078e-06,
      "loss": 0.7874,
      "step": 7722
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.737551051257854,
      "learning_rate": 7.623032494164353e-06,
      "loss": 0.8228,
      "step": 7723
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.3206134388419875,
      "learning_rate": 7.622417479744566e-06,
      "loss": 0.8513,
      "step": 7724
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.131212065934153,
      "learning_rate": 7.62180241058756e-06,
      "loss": 0.7844,
      "step": 7725
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.961527462552656,
      "learning_rate": 7.621187286706169e-06,
      "loss": 0.8085,
      "step": 7726
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.913261636565339,
      "learning_rate": 7.6205721081132335e-06,
      "loss": 0.8332,
      "step": 7727
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.714761347633321,
      "learning_rate": 7.619956874821595e-06,
      "loss": 0.7663,
      "step": 7728
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.9023365376783214,
      "learning_rate": 7.619341586844094e-06,
      "loss": 0.8579,
      "step": 7729
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.657651671712657,
      "learning_rate": 7.618726244193573e-06,
      "loss": 0.7932,
      "step": 7730
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.7280969010376,
      "learning_rate": 7.618110846882876e-06,
      "loss": 0.8253,
      "step": 7731
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.828053340995662,
      "learning_rate": 7.617495394924849e-06,
      "loss": 0.8108,
      "step": 7732
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.959620632630165,
      "learning_rate": 7.61687988833234e-06,
      "loss": 0.7201,
      "step": 7733
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.7440344499426965,
      "learning_rate": 7.6162643271181926e-06,
      "loss": 0.8018,
      "step": 7734
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.620902132049789,
      "learning_rate": 7.615648711295256e-06,
      "loss": 0.7865,
      "step": 7735
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.833026730175112,
      "learning_rate": 7.615033040876383e-06,
      "loss": 0.699,
      "step": 7736
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.625427238744532,
      "learning_rate": 7.614417315874421e-06,
      "loss": 0.7538,
      "step": 7737
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.792546750856068,
      "learning_rate": 7.613801536302221e-06,
      "loss": 0.7692,
      "step": 7738
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.883082508959225,
      "learning_rate": 7.613185702172641e-06,
      "loss": 0.795,
      "step": 7739
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.434478610232027,
      "learning_rate": 7.612569813498531e-06,
      "loss": 0.7463,
      "step": 7740
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.2185370021833615,
      "learning_rate": 7.611953870292747e-06,
      "loss": 0.8249,
      "step": 7741
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.966976924185312,
      "learning_rate": 7.611337872568148e-06,
      "loss": 0.7383,
      "step": 7742
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.842655661245851,
      "learning_rate": 7.610721820337587e-06,
      "loss": 0.8197,
      "step": 7743
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.2676029234463515,
      "learning_rate": 7.610105713613927e-06,
      "loss": 0.7648,
      "step": 7744
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.207206808581557,
      "learning_rate": 7.6094895524100274e-06,
      "loss": 0.8527,
      "step": 7745
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.3358658383946755,
      "learning_rate": 7.608873336738746e-06,
      "loss": 0.7896,
      "step": 7746
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.209204834858133,
      "learning_rate": 7.608257066612947e-06,
      "loss": 0.7746,
      "step": 7747
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.592089915783581,
      "learning_rate": 7.6076407420454946e-06,
      "loss": 0.8349,
      "step": 7748
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.100964353577968,
      "learning_rate": 7.607024363049251e-06,
      "loss": 0.8254,
      "step": 7749
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.5881070712212955,
      "learning_rate": 7.606407929637085e-06,
      "loss": 0.7915,
      "step": 7750
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.2208729311664035,
      "learning_rate": 7.6057914418218615e-06,
      "loss": 0.8326,
      "step": 7751
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.627568394650325,
      "learning_rate": 7.605174899616448e-06,
      "loss": 0.9071,
      "step": 7752
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.168531721188039,
      "learning_rate": 7.604558303033712e-06,
      "loss": 0.8359,
      "step": 7753
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.275327588170875,
      "learning_rate": 7.603941652086526e-06,
      "loss": 0.8292,
      "step": 7754
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.087752838058639,
      "learning_rate": 7.603324946787762e-06,
      "loss": 0.7906,
      "step": 7755
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.854889335384123,
      "learning_rate": 7.602708187150289e-06,
      "loss": 0.8505,
      "step": 7756
    },
    {
      "epoch": 0.69,
      "grad_norm": 18.084879787185653,
      "learning_rate": 7.602091373186984e-06,
      "loss": 0.7773,
      "step": 7757
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.328897127579517,
      "learning_rate": 7.6014745049107195e-06,
      "loss": 0.7242,
      "step": 7758
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.799291711416466,
      "learning_rate": 7.600857582334371e-06,
      "loss": 0.8106,
      "step": 7759
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.941821490468286,
      "learning_rate": 7.600240605470817e-06,
      "loss": 0.7951,
      "step": 7760
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.385817494026444,
      "learning_rate": 7.599623574332935e-06,
      "loss": 0.7476,
      "step": 7761
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.759030226030275,
      "learning_rate": 7.599006488933604e-06,
      "loss": 0.8143,
      "step": 7762
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.113537488768801,
      "learning_rate": 7.598389349285704e-06,
      "loss": 0.7944,
      "step": 7763
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.672178896154457,
      "learning_rate": 7.597772155402116e-06,
      "loss": 0.803,
      "step": 7764
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.26967571706157,
      "learning_rate": 7.597154907295725e-06,
      "loss": 0.7334,
      "step": 7765
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.11640031887388,
      "learning_rate": 7.596537604979413e-06,
      "loss": 0.8036,
      "step": 7766
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.6740098195767565,
      "learning_rate": 7.595920248466062e-06,
      "loss": 0.8553,
      "step": 7767
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.7840054777453025,
      "learning_rate": 7.595302837768564e-06,
      "loss": 0.8113,
      "step": 7768
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.818079305901796,
      "learning_rate": 7.594685372899801e-06,
      "loss": 0.8293,
      "step": 7769
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.686436701133617,
      "learning_rate": 7.594067853872664e-06,
      "loss": 0.7271,
      "step": 7770
    },
    {
      "epoch": 0.69,
      "grad_norm": 9.848466977834743,
      "learning_rate": 7.593450280700041e-06,
      "loss": 0.8555,
      "step": 7771
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.8253302609235265,
      "learning_rate": 7.5928326533948225e-06,
      "loss": 0.8364,
      "step": 7772
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.211660282281385,
      "learning_rate": 7.5922149719699e-06,
      "loss": 0.7742,
      "step": 7773
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.355644647754552,
      "learning_rate": 7.5915972364381685e-06,
      "loss": 0.7853,
      "step": 7774
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.678127314965765,
      "learning_rate": 7.59097944681252e-06,
      "loss": 0.7663,
      "step": 7775
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.8325556515400865,
      "learning_rate": 7.590361603105849e-06,
      "loss": 0.8324,
      "step": 7776
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.503552599136412,
      "learning_rate": 7.589743705331053e-06,
      "loss": 0.819,
      "step": 7777
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.336523492574578,
      "learning_rate": 7.589125753501028e-06,
      "loss": 0.7745,
      "step": 7778
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.68392287809145,
      "learning_rate": 7.588507747628673e-06,
      "loss": 0.7829,
      "step": 7779
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.061711967122832,
      "learning_rate": 7.587889687726888e-06,
      "loss": 0.7629,
      "step": 7780
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.026018695630051,
      "learning_rate": 7.587271573808572e-06,
      "loss": 0.748,
      "step": 7781
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.029616924310662,
      "learning_rate": 7.586653405886629e-06,
      "loss": 0.7501,
      "step": 7782
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.826568665358502,
      "learning_rate": 7.586035183973961e-06,
      "loss": 0.7551,
      "step": 7783
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.540810077037549,
      "learning_rate": 7.585416908083472e-06,
      "loss": 0.809,
      "step": 7784
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.176376722276423,
      "learning_rate": 7.584798578228066e-06,
      "loss": 0.7427,
      "step": 7785
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.373850758402862,
      "learning_rate": 7.58418019442065e-06,
      "loss": 0.7973,
      "step": 7786
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.032555205817951,
      "learning_rate": 7.5835617566741335e-06,
      "loss": 0.7707,
      "step": 7787
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.030417923008486,
      "learning_rate": 7.582943265001421e-06,
      "loss": 0.7903,
      "step": 7788
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.846774119583063,
      "learning_rate": 7.582324719415426e-06,
      "loss": 0.7739,
      "step": 7789
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.880265095166864,
      "learning_rate": 7.5817061199290575e-06,
      "loss": 0.7871,
      "step": 7790
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.7944368654209635,
      "learning_rate": 7.581087466555227e-06,
      "loss": 0.8025,
      "step": 7791
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.232343469797591,
      "learning_rate": 7.580468759306848e-06,
      "loss": 0.7778,
      "step": 7792
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.2334531641192585,
      "learning_rate": 7.579849998196836e-06,
      "loss": 0.8013,
      "step": 7793
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.6092815089931305,
      "learning_rate": 7.579231183238105e-06,
      "loss": 0.8444,
      "step": 7794
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.2275733104394915,
      "learning_rate": 7.578612314443569e-06,
      "loss": 0.8052,
      "step": 7795
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.009702547382585,
      "learning_rate": 7.577993391826151e-06,
      "loss": 0.8452,
      "step": 7796
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.579621334836527,
      "learning_rate": 7.577374415398765e-06,
      "loss": 0.77,
      "step": 7797
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.748482527571452,
      "learning_rate": 7.576755385174332e-06,
      "loss": 0.8617,
      "step": 7798
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.743221940491091,
      "learning_rate": 7.576136301165775e-06,
      "loss": 0.794,
      "step": 7799
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.683305017055746,
      "learning_rate": 7.575517163386014e-06,
      "loss": 0.8156,
      "step": 7800
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.118925260296863,
      "learning_rate": 7.574897971847972e-06,
      "loss": 0.7384,
      "step": 7801
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.810778123809922,
      "learning_rate": 7.5742787265645735e-06,
      "loss": 0.8114,
      "step": 7802
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.231998514406285,
      "learning_rate": 7.5736594275487455e-06,
      "loss": 0.8192,
      "step": 7803
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.253144963502085,
      "learning_rate": 7.573040074813411e-06,
      "loss": 0.7795,
      "step": 7804
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.2943829258620285,
      "learning_rate": 7.572420668371501e-06,
      "loss": 0.7774,
      "step": 7805
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.991360882166256,
      "learning_rate": 7.571801208235943e-06,
      "loss": 0.8585,
      "step": 7806
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.944324112962722,
      "learning_rate": 7.571181694419667e-06,
      "loss": 0.7451,
      "step": 7807
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.223429317209248,
      "learning_rate": 7.570562126935603e-06,
      "loss": 0.7214,
      "step": 7808
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.310989535856731,
      "learning_rate": 7.569942505796686e-06,
      "loss": 0.8144,
      "step": 7809
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.655584430258291,
      "learning_rate": 7.569322831015846e-06,
      "loss": 0.7899,
      "step": 7810
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.548310198115499,
      "learning_rate": 7.568703102606019e-06,
      "loss": 0.7645,
      "step": 7811
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.025484086164654,
      "learning_rate": 7.56808332058014e-06,
      "loss": 0.8405,
      "step": 7812
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.553719643712117,
      "learning_rate": 7.567463484951145e-06,
      "loss": 0.806,
      "step": 7813
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.9407775653348756,
      "learning_rate": 7.566843595731972e-06,
      "loss": 0.7731,
      "step": 7814
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.111956448056105,
      "learning_rate": 7.566223652935562e-06,
      "loss": 0.7457,
      "step": 7815
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.065972336617041,
      "learning_rate": 7.565603656574852e-06,
      "loss": 0.8364,
      "step": 7816
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.85008685665357,
      "learning_rate": 7.564983606662784e-06,
      "loss": 0.8258,
      "step": 7817
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.96120824788405,
      "learning_rate": 7.564363503212302e-06,
      "loss": 0.7288,
      "step": 7818
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.900400499346372,
      "learning_rate": 7.563743346236346e-06,
      "loss": 0.7361,
      "step": 7819
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.480278003015133,
      "learning_rate": 7.5631231357478626e-06,
      "loss": 0.8,
      "step": 7820
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.690685501591544,
      "learning_rate": 7.562502871759798e-06,
      "loss": 0.775,
      "step": 7821
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.629694590265112,
      "learning_rate": 7.561882554285098e-06,
      "loss": 0.7434,
      "step": 7822
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.49386658263311,
      "learning_rate": 7.561262183336708e-06,
      "loss": 0.7466,
      "step": 7823
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.132165412094173,
      "learning_rate": 7.560641758927581e-06,
      "loss": 0.8012,
      "step": 7824
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.848971329412663,
      "learning_rate": 7.560021281070664e-06,
      "loss": 0.8425,
      "step": 7825
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.65638998851288,
      "learning_rate": 7.559400749778908e-06,
      "loss": 0.7534,
      "step": 7826
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.673181954886724,
      "learning_rate": 7.558780165065267e-06,
      "loss": 0.8019,
      "step": 7827
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.164732762403855,
      "learning_rate": 7.558159526942694e-06,
      "loss": 0.7756,
      "step": 7828
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.305652724124263,
      "learning_rate": 7.557538835424145e-06,
      "loss": 0.7935,
      "step": 7829
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.788069478258926,
      "learning_rate": 7.556918090522572e-06,
      "loss": 0.8343,
      "step": 7830
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.232431297646364,
      "learning_rate": 7.556297292250933e-06,
      "loss": 0.7977,
      "step": 7831
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.921999653919874,
      "learning_rate": 7.555676440622186e-06,
      "loss": 0.8036,
      "step": 7832
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.6719370516591825,
      "learning_rate": 7.555055535649291e-06,
      "loss": 0.7283,
      "step": 7833
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.411359595028725,
      "learning_rate": 7.554434577345206e-06,
      "loss": 0.8391,
      "step": 7834
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.117088039428969,
      "learning_rate": 7.553813565722895e-06,
      "loss": 0.8813,
      "step": 7835
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.904147082684149,
      "learning_rate": 7.553192500795317e-06,
      "loss": 0.8002,
      "step": 7836
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.272842172200147,
      "learning_rate": 7.552571382575439e-06,
      "loss": 0.787,
      "step": 7837
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.628148595888817,
      "learning_rate": 7.551950211076223e-06,
      "loss": 0.7911,
      "step": 7838
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.283511854800911,
      "learning_rate": 7.551328986310633e-06,
      "loss": 0.8145,
      "step": 7839
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.734395681903562,
      "learning_rate": 7.550707708291638e-06,
      "loss": 0.7842,
      "step": 7840
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.066661527741526,
      "learning_rate": 7.550086377032208e-06,
      "loss": 0.87,
      "step": 7841
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.003928525813774,
      "learning_rate": 7.5494649925453075e-06,
      "loss": 0.8112,
      "step": 7842
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.396889539308282,
      "learning_rate": 7.548843554843909e-06,
      "loss": 0.7555,
      "step": 7843
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.6887202503333,
      "learning_rate": 7.548222063940984e-06,
      "loss": 0.7666,
      "step": 7844
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.454635316584767,
      "learning_rate": 7.547600519849503e-06,
      "loss": 0.8203,
      "step": 7845
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.393008114544564,
      "learning_rate": 7.54697892258244e-06,
      "loss": 0.8499,
      "step": 7846
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.177286907921559,
      "learning_rate": 7.546357272152771e-06,
      "loss": 0.8685,
      "step": 7847
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.54538618675291,
      "learning_rate": 7.545735568573469e-06,
      "loss": 0.7642,
      "step": 7848
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.9196611634493435,
      "learning_rate": 7.545113811857512e-06,
      "loss": 0.7952,
      "step": 7849
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.488990762588969,
      "learning_rate": 7.544492002017878e-06,
      "loss": 0.7784,
      "step": 7850
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.95866428236915,
      "learning_rate": 7.543870139067548e-06,
      "loss": 0.7896,
      "step": 7851
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.87871819005342,
      "learning_rate": 7.543248223019497e-06,
      "loss": 0.8516,
      "step": 7852
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.517430052820431,
      "learning_rate": 7.54262625388671e-06,
      "loss": 0.7722,
      "step": 7853
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.586733022871997,
      "learning_rate": 7.542004231682169e-06,
      "loss": 0.8378,
      "step": 7854
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.618817242462028,
      "learning_rate": 7.541382156418855e-06,
      "loss": 0.8183,
      "step": 7855
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.528276905007494,
      "learning_rate": 7.5407600281097544e-06,
      "loss": 0.8203,
      "step": 7856
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.528647657234929,
      "learning_rate": 7.540137846767854e-06,
      "loss": 0.8097,
      "step": 7857
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.96543787915373,
      "learning_rate": 7.539515612406138e-06,
      "loss": 0.7758,
      "step": 7858
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.105050539097277,
      "learning_rate": 7.538893325037594e-06,
      "loss": 0.7603,
      "step": 7859
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.553920191776604,
      "learning_rate": 7.538270984675213e-06,
      "loss": 0.7852,
      "step": 7860
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.36374936013212,
      "learning_rate": 7.537648591331983e-06,
      "loss": 0.7072,
      "step": 7861
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.6439729130044025,
      "learning_rate": 7.537026145020896e-06,
      "loss": 0.8075,
      "step": 7862
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.480528788658741,
      "learning_rate": 7.536403645754946e-06,
      "loss": 0.796,
      "step": 7863
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.435124620731268,
      "learning_rate": 7.535781093547123e-06,
      "loss": 0.7872,
      "step": 7864
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.475068686046202,
      "learning_rate": 7.5351584884104235e-06,
      "loss": 0.7615,
      "step": 7865
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.438740466815408,
      "learning_rate": 7.534535830357843e-06,
      "loss": 0.7907,
      "step": 7866
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.122997814134779,
      "learning_rate": 7.533913119402376e-06,
      "loss": 0.8247,
      "step": 7867
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.119934743495468,
      "learning_rate": 7.533290355557023e-06,
      "loss": 0.8722,
      "step": 7868
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.567769982017343,
      "learning_rate": 7.532667538834781e-06,
      "loss": 0.8343,
      "step": 7869
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.563206758886511,
      "learning_rate": 7.53204466924865e-06,
      "loss": 0.822,
      "step": 7870
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.70998211484728,
      "learning_rate": 7.531421746811632e-06,
      "loss": 0.8287,
      "step": 7871
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.102254035007287,
      "learning_rate": 7.530798771536731e-06,
      "loss": 0.8824,
      "step": 7872
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.1040125668783665,
      "learning_rate": 7.530175743436946e-06,
      "loss": 0.7198,
      "step": 7873
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.9469505659596695,
      "learning_rate": 7.5295526625252836e-06,
      "loss": 0.7612,
      "step": 7874
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.124614989267985,
      "learning_rate": 7.528929528814748e-06,
      "loss": 0.8149,
      "step": 7875
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.092041180365186,
      "learning_rate": 7.528306342318348e-06,
      "loss": 0.8282,
      "step": 7876
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.435438301347454,
      "learning_rate": 7.527683103049089e-06,
      "loss": 0.7912,
      "step": 7877
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.451859472193517,
      "learning_rate": 7.527059811019981e-06,
      "loss": 0.7708,
      "step": 7878
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.509187525046455,
      "learning_rate": 7.526436466244035e-06,
      "loss": 0.8397,
      "step": 7879
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.95157469949699,
      "learning_rate": 7.52581306873426e-06,
      "loss": 0.7483,
      "step": 7880
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.16849512622031,
      "learning_rate": 7.525189618503668e-06,
      "loss": 0.7549,
      "step": 7881
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.993666015908522,
      "learning_rate": 7.5245661155652724e-06,
      "loss": 0.78,
      "step": 7882
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.346253725950219,
      "learning_rate": 7.5239425599320894e-06,
      "loss": 0.829,
      "step": 7883
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.797758638284803,
      "learning_rate": 7.5233189516171315e-06,
      "loss": 0.7772,
      "step": 7884
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.765972617280916,
      "learning_rate": 7.522695290633417e-06,
      "loss": 0.8346,
      "step": 7885
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.024752305518248,
      "learning_rate": 7.522071576993962e-06,
      "loss": 0.7341,
      "step": 7886
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.554926243065915,
      "learning_rate": 7.521447810711788e-06,
      "loss": 0.7788,
      "step": 7887
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.874739691479034,
      "learning_rate": 7.52082399179991e-06,
      "loss": 0.7933,
      "step": 7888
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.02207266181155,
      "learning_rate": 7.520200120271355e-06,
      "loss": 0.8124,
      "step": 7889
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.656656284374934,
      "learning_rate": 7.519576196139139e-06,
      "loss": 0.7464,
      "step": 7890
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.477325351809031,
      "learning_rate": 7.518952219416289e-06,
      "loss": 0.7343,
      "step": 7891
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.231243750365789,
      "learning_rate": 7.518328190115826e-06,
      "loss": 0.805,
      "step": 7892
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.332267489773912,
      "learning_rate": 7.5177041082507795e-06,
      "loss": 0.7758,
      "step": 7893
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.075942177353138,
      "learning_rate": 7.517079973834174e-06,
      "loss": 0.7812,
      "step": 7894
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.941485705644206,
      "learning_rate": 7.516455786879034e-06,
      "loss": 0.8448,
      "step": 7895
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.3544069235507665,
      "learning_rate": 7.515831547398391e-06,
      "loss": 0.7579,
      "step": 7896
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.445919834064377,
      "learning_rate": 7.515207255405275e-06,
      "loss": 0.7285,
      "step": 7897
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.8643161684477825,
      "learning_rate": 7.5145829109127145e-06,
      "loss": 0.7455,
      "step": 7898
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.027416982330641,
      "learning_rate": 7.513958513933742e-06,
      "loss": 0.8154,
      "step": 7899
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.335692405155143,
      "learning_rate": 7.513334064481392e-06,
      "loss": 0.7861,
      "step": 7900
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.436777758612885,
      "learning_rate": 7.512709562568699e-06,
      "loss": 0.8011,
      "step": 7901
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.217070995792567,
      "learning_rate": 7.512085008208695e-06,
      "loss": 0.7622,
      "step": 7902
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.628476670564611,
      "learning_rate": 7.5114604014144175e-06,
      "loss": 0.8211,
      "step": 7903
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.36981244611042,
      "learning_rate": 7.510835742198905e-06,
      "loss": 0.7753,
      "step": 7904
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.7674256043102075,
      "learning_rate": 7.510211030575194e-06,
      "loss": 0.861,
      "step": 7905
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.99404613953194,
      "learning_rate": 7.509586266556325e-06,
      "loss": 0.8203,
      "step": 7906
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.035907106404633,
      "learning_rate": 7.508961450155341e-06,
      "loss": 0.7645,
      "step": 7907
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.219645012423327,
      "learning_rate": 7.508336581385278e-06,
      "loss": 0.7595,
      "step": 7908
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.614879847842888,
      "learning_rate": 7.507711660259185e-06,
      "loss": 0.78,
      "step": 7909
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.923404295236724,
      "learning_rate": 7.507086686790101e-06,
      "loss": 0.7263,
      "step": 7910
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.731790421706726,
      "learning_rate": 7.506461660991074e-06,
      "loss": 0.8505,
      "step": 7911
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.478198151652288,
      "learning_rate": 7.505836582875148e-06,
      "loss": 0.8013,
      "step": 7912
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.871592158150468,
      "learning_rate": 7.50521145245537e-06,
      "loss": 0.7092,
      "step": 7913
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.0648864151901885,
      "learning_rate": 7.504586269744791e-06,
      "loss": 0.7975,
      "step": 7914
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.953803857014598,
      "learning_rate": 7.5039610347564576e-06,
      "loss": 0.7763,
      "step": 7915
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.186578028640509,
      "learning_rate": 7.503335747503422e-06,
      "loss": 0.8263,
      "step": 7916
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.251882100897378,
      "learning_rate": 7.502710407998734e-06,
      "loss": 0.7691,
      "step": 7917
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.388343339629617,
      "learning_rate": 7.502085016255448e-06,
      "loss": 0.7964,
      "step": 7918
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.09830814613878,
      "learning_rate": 7.501459572286616e-06,
      "loss": 0.85,
      "step": 7919
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.6296564597298735,
      "learning_rate": 7.5008340761052944e-06,
      "loss": 0.7587,
      "step": 7920
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.040971949524659,
      "learning_rate": 7.5002085277245375e-06,
      "loss": 0.8399,
      "step": 7921
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.929842743114392,
      "learning_rate": 7.499582927157403e-06,
      "loss": 0.7502,
      "step": 7922
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.145787935809559,
      "learning_rate": 7.498957274416949e-06,
      "loss": 0.7673,
      "step": 7923
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.200613696788193,
      "learning_rate": 7.498331569516235e-06,
      "loss": 0.8217,
      "step": 7924
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.689162556241866,
      "learning_rate": 7.49770581246832e-06,
      "loss": 0.8217,
      "step": 7925
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.2903860817715636,
      "learning_rate": 7.497080003286266e-06,
      "loss": 0.7791,
      "step": 7926
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.195338927654314,
      "learning_rate": 7.4964541419831364e-06,
      "loss": 0.8093,
      "step": 7927
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.605212134074036,
      "learning_rate": 7.495828228571992e-06,
      "loss": 0.7745,
      "step": 7928
    },
    {
      "epoch": 0.71,
      "grad_norm": 8.270806770763496,
      "learning_rate": 7.495202263065899e-06,
      "loss": 0.8336,
      "step": 7929
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.918842157779109,
      "learning_rate": 7.494576245477927e-06,
      "loss": 0.8571,
      "step": 7930
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.4266868681253575,
      "learning_rate": 7.493950175821135e-06,
      "loss": 0.7889,
      "step": 7931
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.126392829036127,
      "learning_rate": 7.493324054108595e-06,
      "loss": 0.829,
      "step": 7932
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.879861413354143,
      "learning_rate": 7.492697880353378e-06,
      "loss": 0.8686,
      "step": 7933
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.412309973767169,
      "learning_rate": 7.492071654568549e-06,
      "loss": 0.8097,
      "step": 7934
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.587588824285416,
      "learning_rate": 7.491445376767183e-06,
      "loss": 0.8115,
      "step": 7935
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.476274925425593,
      "learning_rate": 7.4908190469623524e-06,
      "loss": 0.7675,
      "step": 7936
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.316033836344861,
      "learning_rate": 7.490192665167128e-06,
      "loss": 0.774,
      "step": 7937
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.116460361254473,
      "learning_rate": 7.489566231394584e-06,
      "loss": 0.7694,
      "step": 7938
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.910833239388987,
      "learning_rate": 7.488939745657797e-06,
      "loss": 0.7995,
      "step": 7939
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.509383481856893,
      "learning_rate": 7.488313207969844e-06,
      "loss": 0.7622,
      "step": 7940
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.4769049527856914,
      "learning_rate": 7.487686618343802e-06,
      "loss": 0.765,
      "step": 7941
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.886538184029175,
      "learning_rate": 7.487059976792751e-06,
      "loss": 0.7678,
      "step": 7942
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.033860959358567,
      "learning_rate": 7.486433283329768e-06,
      "loss": 0.7665,
      "step": 7943
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.310937782136536,
      "learning_rate": 7.485806537967936e-06,
      "loss": 0.751,
      "step": 7944
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.268179878899627,
      "learning_rate": 7.485179740720336e-06,
      "loss": 0.7493,
      "step": 7945
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.663235897141652,
      "learning_rate": 7.484552891600052e-06,
      "loss": 0.7498,
      "step": 7946
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.1224345744818605,
      "learning_rate": 7.483925990620166e-06,
      "loss": 0.7968,
      "step": 7947
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.088512957347719,
      "learning_rate": 7.483299037793766e-06,
      "loss": 0.7898,
      "step": 7948
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.548352327984504,
      "learning_rate": 7.482672033133937e-06,
      "loss": 0.7375,
      "step": 7949
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.896448641265528,
      "learning_rate": 7.4820449766537665e-06,
      "loss": 0.9026,
      "step": 7950
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.519729857381605,
      "learning_rate": 7.481417868366342e-06,
      "loss": 0.7458,
      "step": 7951
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.735707612904966,
      "learning_rate": 7.480790708284755e-06,
      "loss": 0.7829,
      "step": 7952
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.072438189754723,
      "learning_rate": 7.480163496422094e-06,
      "loss": 0.7868,
      "step": 7953
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.975894387896344,
      "learning_rate": 7.47953623279145e-06,
      "loss": 0.7786,
      "step": 7954
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.259798562731409,
      "learning_rate": 7.478908917405919e-06,
      "loss": 0.8591,
      "step": 7955
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.34747026775682,
      "learning_rate": 7.478281550278592e-06,
      "loss": 0.8661,
      "step": 7956
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.683883994461897,
      "learning_rate": 7.4776541314225646e-06,
      "loss": 0.7179,
      "step": 7957
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.370563087366387,
      "learning_rate": 7.4770266608509355e-06,
      "loss": 0.7912,
      "step": 7958
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.958175318516089,
      "learning_rate": 7.476399138576798e-06,
      "loss": 0.7434,
      "step": 7959
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.121702412802429,
      "learning_rate": 7.475771564613252e-06,
      "loss": 0.7562,
      "step": 7960
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.232632216492475,
      "learning_rate": 7.475143938973395e-06,
      "loss": 0.7316,
      "step": 7961
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.677583787445041,
      "learning_rate": 7.474516261670331e-06,
      "loss": 0.8221,
      "step": 7962
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.2085394295605925,
      "learning_rate": 7.4738885327171574e-06,
      "loss": 0.7669,
      "step": 7963
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.9359241498103135,
      "learning_rate": 7.473260752126979e-06,
      "loss": 0.8438,
      "step": 7964
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.183368338108208,
      "learning_rate": 7.472632919912899e-06,
      "loss": 0.8234,
      "step": 7965
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.45842404253519,
      "learning_rate": 7.472005036088022e-06,
      "loss": 0.8004,
      "step": 7966
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.9117025194193635,
      "learning_rate": 7.4713771006654525e-06,
      "loss": 0.7703,
      "step": 7967
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.344896461286907,
      "learning_rate": 7.4707491136583e-06,
      "loss": 0.737,
      "step": 7968
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.940715703060105,
      "learning_rate": 7.470121075079668e-06,
      "loss": 0.7417,
      "step": 7969
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.753066274510542,
      "learning_rate": 7.469492984942669e-06,
      "loss": 0.7349,
      "step": 7970
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.4388063057399165,
      "learning_rate": 7.468864843260413e-06,
      "loss": 0.8499,
      "step": 7971
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.469337012215663,
      "learning_rate": 7.468236650046009e-06,
      "loss": 0.802,
      "step": 7972
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.605774912796959,
      "learning_rate": 7.467608405312571e-06,
      "loss": 0.7791,
      "step": 7973
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.57428060986952,
      "learning_rate": 7.466980109073212e-06,
      "loss": 0.8383,
      "step": 7974
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.494400537280058,
      "learning_rate": 7.466351761341045e-06,
      "loss": 0.7812,
      "step": 7975
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.5001810282970025,
      "learning_rate": 7.4657233621291854e-06,
      "loss": 0.8066,
      "step": 7976
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.5320189647704145,
      "learning_rate": 7.465094911450752e-06,
      "loss": 0.8033,
      "step": 7977
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.844897158238273,
      "learning_rate": 7.46446640931886e-06,
      "loss": 0.7423,
      "step": 7978
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.058934531548048,
      "learning_rate": 7.463837855746629e-06,
      "loss": 0.8144,
      "step": 7979
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.718618247789218,
      "learning_rate": 7.463209250747179e-06,
      "loss": 0.7641,
      "step": 7980
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.273851375582472,
      "learning_rate": 7.46258059433363e-06,
      "loss": 0.7871,
      "step": 7981
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.334867709882106,
      "learning_rate": 7.461951886519103e-06,
      "loss": 0.8523,
      "step": 7982
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.617835352315898,
      "learning_rate": 7.461323127316723e-06,
      "loss": 0.8118,
      "step": 7983
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.701130267559413,
      "learning_rate": 7.460694316739613e-06,
      "loss": 0.8166,
      "step": 7984
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.16149481946943,
      "learning_rate": 7.460065454800898e-06,
      "loss": 0.7683,
      "step": 7985
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.514484970689203,
      "learning_rate": 7.459436541513703e-06,
      "loss": 0.808,
      "step": 7986
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.438039719745585,
      "learning_rate": 7.4588075768911574e-06,
      "loss": 0.7754,
      "step": 7987
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.428282140089798,
      "learning_rate": 7.4581785609463875e-06,
      "loss": 0.7777,
      "step": 7988
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.118732774029786,
      "learning_rate": 7.457549493692524e-06,
      "loss": 0.8185,
      "step": 7989
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.816571939645837,
      "learning_rate": 7.456920375142697e-06,
      "loss": 0.8015,
      "step": 7990
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.238664078366661,
      "learning_rate": 7.456291205310036e-06,
      "loss": 0.8317,
      "step": 7991
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.792155842965622,
      "learning_rate": 7.455661984207677e-06,
      "loss": 0.7715,
      "step": 7992
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.103944548272625,
      "learning_rate": 7.455032711848752e-06,
      "loss": 0.7797,
      "step": 7993
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.425875869950076,
      "learning_rate": 7.4544033882463936e-06,
      "loss": 0.8122,
      "step": 7994
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.224675412885934,
      "learning_rate": 7.453774013413743e-06,
      "loss": 0.8473,
      "step": 7995
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.163577630186194,
      "learning_rate": 7.453144587363931e-06,
      "loss": 0.7615,
      "step": 7996
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.967016705994163,
      "learning_rate": 7.452515110110099e-06,
      "loss": 0.7238,
      "step": 7997
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.545385444486682,
      "learning_rate": 7.451885581665383e-06,
      "loss": 0.8065,
      "step": 7998
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.0476525158702605,
      "learning_rate": 7.451256002042927e-06,
      "loss": 0.7832,
      "step": 7999
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.551306847669112,
      "learning_rate": 7.45062637125587e-06,
      "loss": 0.8126,
      "step": 8000
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.024414540058065,
      "learning_rate": 7.449996689317354e-06,
      "loss": 0.7829,
      "step": 8001
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.904464708256639,
      "learning_rate": 7.449366956240523e-06,
      "loss": 0.823,
      "step": 8002
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.907929397456919,
      "learning_rate": 7.448737172038521e-06,
      "loss": 0.7938,
      "step": 8003
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.5429783286907055,
      "learning_rate": 7.448107336724491e-06,
      "loss": 0.7785,
      "step": 8004
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.005685999013478,
      "learning_rate": 7.447477450311583e-06,
      "loss": 0.7851,
      "step": 8005
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.185831030034902,
      "learning_rate": 7.446847512812943e-06,
      "loss": 0.8478,
      "step": 8006
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.372254665939628,
      "learning_rate": 7.44621752424172e-06,
      "loss": 0.8494,
      "step": 8007
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.477188119150958,
      "learning_rate": 7.445587484611063e-06,
      "loss": 0.7814,
      "step": 8008
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.64414601924335,
      "learning_rate": 7.444957393934124e-06,
      "loss": 0.8799,
      "step": 8009
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.991734478414799,
      "learning_rate": 7.444327252224053e-06,
      "loss": 0.8146,
      "step": 8010
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.094979933536866,
      "learning_rate": 7.443697059494004e-06,
      "loss": 0.7555,
      "step": 8011
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.661851772396895,
      "learning_rate": 7.443066815757129e-06,
      "loss": 0.7087,
      "step": 8012
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.191510730363255,
      "learning_rate": 7.442436521026586e-06,
      "loss": 0.7501,
      "step": 8013
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.822495016134685,
      "learning_rate": 7.441806175315528e-06,
      "loss": 0.7888,
      "step": 8014
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.9978825236928675,
      "learning_rate": 7.441175778637115e-06,
      "loss": 0.7264,
      "step": 8015
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.757944816888849,
      "learning_rate": 7.440545331004503e-06,
      "loss": 0.7752,
      "step": 8016
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.991364706193056,
      "learning_rate": 7.439914832430852e-06,
      "loss": 0.802,
      "step": 8017
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.032419924250394,
      "learning_rate": 7.439284282929322e-06,
      "loss": 0.7597,
      "step": 8018
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.924327138063203,
      "learning_rate": 7.438653682513077e-06,
      "loss": 0.8228,
      "step": 8019
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.830989757755557,
      "learning_rate": 7.438023031195274e-06,
      "loss": 0.7566,
      "step": 8020
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.097135696112259,
      "learning_rate": 7.437392328989079e-06,
      "loss": 0.8224,
      "step": 8021
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.399269588750674,
      "learning_rate": 7.436761575907658e-06,
      "loss": 0.7175,
      "step": 8022
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.368379706559058,
      "learning_rate": 7.4361307719641765e-06,
      "loss": 0.7709,
      "step": 8023
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.2676705058070095,
      "learning_rate": 7.435499917171799e-06,
      "loss": 0.8164,
      "step": 8024
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.695651484389013,
      "learning_rate": 7.434869011543695e-06,
      "loss": 0.7614,
      "step": 8025
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.442745625473054,
      "learning_rate": 7.434238055093033e-06,
      "loss": 0.7897,
      "step": 8026
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.800147121056146,
      "learning_rate": 7.43360704783298e-06,
      "loss": 0.7836,
      "step": 8027
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.103996772789028,
      "learning_rate": 7.432975989776713e-06,
      "loss": 0.9125,
      "step": 8028
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.709939589985719,
      "learning_rate": 7.432344880937398e-06,
      "loss": 0.8184,
      "step": 8029
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.161755132978119,
      "learning_rate": 7.4317137213282105e-06,
      "loss": 0.7484,
      "step": 8030
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.816584053101581,
      "learning_rate": 7.431082510962325e-06,
      "loss": 0.7499,
      "step": 8031
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.521452395539106,
      "learning_rate": 7.430451249852917e-06,
      "loss": 0.8247,
      "step": 8032
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.070655321601379,
      "learning_rate": 7.42981993801316e-06,
      "loss": 0.7511,
      "step": 8033
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.221753564987714,
      "learning_rate": 7.429188575456233e-06,
      "loss": 0.8182,
      "step": 8034
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.247898204509015,
      "learning_rate": 7.428557162195315e-06,
      "loss": 0.7805,
      "step": 8035
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.742003127800784,
      "learning_rate": 7.427925698243586e-06,
      "loss": 0.8711,
      "step": 8036
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.924523221611583,
      "learning_rate": 7.427294183614222e-06,
      "loss": 0.775,
      "step": 8037
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.886684759122131,
      "learning_rate": 7.426662618320411e-06,
      "loss": 0.7602,
      "step": 8038
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.811448495610583,
      "learning_rate": 7.42603100237533e-06,
      "loss": 0.7956,
      "step": 8039
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.331838056102882,
      "learning_rate": 7.425399335792165e-06,
      "loss": 0.8584,
      "step": 8040
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.55502569266274,
      "learning_rate": 7.424767618584101e-06,
      "loss": 0.7394,
      "step": 8041
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.264047961932086,
      "learning_rate": 7.424135850764323e-06,
      "loss": 0.7571,
      "step": 8042
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.271421482802664,
      "learning_rate": 7.423504032346018e-06,
      "loss": 0.8584,
      "step": 8043
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.875133446570406,
      "learning_rate": 7.422872163342373e-06,
      "loss": 0.7992,
      "step": 8044
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.214581793257712,
      "learning_rate": 7.422240243766578e-06,
      "loss": 0.7901,
      "step": 8045
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.4851103066679485,
      "learning_rate": 7.421608273631821e-06,
      "loss": 0.805,
      "step": 8046
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.729076233876768,
      "learning_rate": 7.420976252951297e-06,
      "loss": 0.774,
      "step": 8047
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.366001637431151,
      "learning_rate": 7.420344181738195e-06,
      "loss": 0.7874,
      "step": 8048
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.390072893948882,
      "learning_rate": 7.419712060005709e-06,
      "loss": 0.7848,
      "step": 8049
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.284758521567236,
      "learning_rate": 7.419079887767031e-06,
      "loss": 0.7245,
      "step": 8050
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.365093470384115,
      "learning_rate": 7.41844766503536e-06,
      "loss": 0.7591,
      "step": 8051
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.430665002476206,
      "learning_rate": 7.417815391823889e-06,
      "loss": 0.6807,
      "step": 8052
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.243914104753792,
      "learning_rate": 7.417183068145817e-06,
      "loss": 0.8249,
      "step": 8053
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.7619871268356055,
      "learning_rate": 7.4165506940143415e-06,
      "loss": 0.7392,
      "step": 8054
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.069915985809673,
      "learning_rate": 7.415918269442663e-06,
      "loss": 0.7983,
      "step": 8055
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.207341731924211,
      "learning_rate": 7.4152857944439815e-06,
      "loss": 0.7607,
      "step": 8056
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.757325688357405,
      "learning_rate": 7.414653269031499e-06,
      "loss": 0.7802,
      "step": 8057
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.460560833637711,
      "learning_rate": 7.414020693218415e-06,
      "loss": 0.8178,
      "step": 8058
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.033523413576593,
      "learning_rate": 7.4133880670179385e-06,
      "loss": 0.8409,
      "step": 8059
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.023159347672714,
      "learning_rate": 7.412755390443271e-06,
      "loss": 0.871,
      "step": 8060
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.562324662321809,
      "learning_rate": 7.412122663507617e-06,
      "loss": 0.787,
      "step": 8061
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.991271620341486,
      "learning_rate": 7.411489886224186e-06,
      "loss": 0.8641,
      "step": 8062
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.450840664430811,
      "learning_rate": 7.4108570586061846e-06,
      "loss": 0.8252,
      "step": 8063
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.452824198281665,
      "learning_rate": 7.410224180666821e-06,
      "loss": 0.7534,
      "step": 8064
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.082834713896673,
      "learning_rate": 7.409591252419306e-06,
      "loss": 0.7968,
      "step": 8065
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.530098289703796,
      "learning_rate": 7.408958273876851e-06,
      "loss": 0.7647,
      "step": 8066
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.033100009270034,
      "learning_rate": 7.408325245052669e-06,
      "loss": 0.7412,
      "step": 8067
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.557487312823466,
      "learning_rate": 7.40769216595997e-06,
      "loss": 0.8212,
      "step": 8068
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.900700131930921,
      "learning_rate": 7.40705903661197e-06,
      "loss": 0.759,
      "step": 8069
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.525619028396252,
      "learning_rate": 7.406425857021884e-06,
      "loss": 0.7373,
      "step": 8070
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.215475095677604,
      "learning_rate": 7.405792627202928e-06,
      "loss": 0.8002,
      "step": 8071
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.872485857116215,
      "learning_rate": 7.405159347168319e-06,
      "loss": 0.8114,
      "step": 8072
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.689126977154325,
      "learning_rate": 7.4045260169312775e-06,
      "loss": 0.8426,
      "step": 8073
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.645605479448751,
      "learning_rate": 7.403892636505021e-06,
      "loss": 0.7489,
      "step": 8074
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.429986074623284,
      "learning_rate": 7.403259205902769e-06,
      "loss": 0.8179,
      "step": 8075
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.5515264168482625,
      "learning_rate": 7.402625725137745e-06,
      "loss": 0.7886,
      "step": 8076
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.767561530613094,
      "learning_rate": 7.40199219422317e-06,
      "loss": 0.7815,
      "step": 8077
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.896408479255329,
      "learning_rate": 7.401358613172267e-06,
      "loss": 0.7952,
      "step": 8078
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.4332076164625,
      "learning_rate": 7.400724981998264e-06,
      "loss": 0.82,
      "step": 8079
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.736632020424963,
      "learning_rate": 7.400091300714384e-06,
      "loss": 0.7324,
      "step": 8080
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.650131794427687,
      "learning_rate": 7.3994575693338536e-06,
      "loss": 0.8035,
      "step": 8081
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.24862959608484,
      "learning_rate": 7.398823787869902e-06,
      "loss": 0.8062,
      "step": 8082
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.949614691939611,
      "learning_rate": 7.398189956335757e-06,
      "loss": 0.8102,
      "step": 8083
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.010469125602329,
      "learning_rate": 7.397556074744648e-06,
      "loss": 0.7887,
      "step": 8084
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.928153021689938,
      "learning_rate": 7.396922143109806e-06,
      "loss": 0.7623,
      "step": 8085
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.225091879996679,
      "learning_rate": 7.396288161444465e-06,
      "loss": 0.7358,
      "step": 8086
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.831578777871091,
      "learning_rate": 7.3956541297618554e-06,
      "loss": 0.7937,
      "step": 8087
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.573984753831123,
      "learning_rate": 7.3950200480752125e-06,
      "loss": 0.85,
      "step": 8088
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.3831684880060715,
      "learning_rate": 7.394385916397772e-06,
      "loss": 0.8154,
      "step": 8089
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.097601028992834,
      "learning_rate": 7.393751734742768e-06,
      "loss": 0.863,
      "step": 8090
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.797569781526311,
      "learning_rate": 7.393117503123438e-06,
      "loss": 0.8042,
      "step": 8091
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.5974693941746105,
      "learning_rate": 7.392483221553023e-06,
      "loss": 0.7763,
      "step": 8092
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.876360920060236,
      "learning_rate": 7.391848890044758e-06,
      "loss": 0.7489,
      "step": 8093
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.474114196917781,
      "learning_rate": 7.391214508611886e-06,
      "loss": 0.7918,
      "step": 8094
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.36036327977871,
      "learning_rate": 7.390580077267649e-06,
      "loss": 0.7892,
      "step": 8095
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.373289081161579,
      "learning_rate": 7.389945596025289e-06,
      "loss": 0.7818,
      "step": 8096
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.522423534688371,
      "learning_rate": 7.3893110648980455e-06,
      "loss": 0.8074,
      "step": 8097
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.9853846354372955,
      "learning_rate": 7.388676483899167e-06,
      "loss": 0.8494,
      "step": 8098
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.905254239154744,
      "learning_rate": 7.3880418530419005e-06,
      "loss": 0.7931,
      "step": 8099
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.079028929520829,
      "learning_rate": 7.3874071723394865e-06,
      "loss": 0.7707,
      "step": 8100
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.878756442282105,
      "learning_rate": 7.386772441805179e-06,
      "loss": 0.8079,
      "step": 8101
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.054177897286903,
      "learning_rate": 7.386137661452222e-06,
      "loss": 0.8909,
      "step": 8102
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.8070255214040443,
      "learning_rate": 7.385502831293868e-06,
      "loss": 0.8159,
      "step": 8103
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.85498944391718,
      "learning_rate": 7.3848679513433665e-06,
      "loss": 0.7975,
      "step": 8104
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.086668663920648,
      "learning_rate": 7.384233021613969e-06,
      "loss": 0.7573,
      "step": 8105
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.170362369744039,
      "learning_rate": 7.383598042118928e-06,
      "loss": 0.794,
      "step": 8106
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.96305497172723,
      "learning_rate": 7.382963012871498e-06,
      "loss": 0.7547,
      "step": 8107
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.0219327973450145,
      "learning_rate": 7.382327933884934e-06,
      "loss": 0.7656,
      "step": 8108
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.508180235186709,
      "learning_rate": 7.3816928051724915e-06,
      "loss": 0.8169,
      "step": 8109
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.009409383704669,
      "learning_rate": 7.3810576267474276e-06,
      "loss": 0.7714,
      "step": 8110
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.895419659055805,
      "learning_rate": 7.380422398623e-06,
      "loss": 0.7436,
      "step": 8111
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.017394689451529,
      "learning_rate": 7.379787120812467e-06,
      "loss": 0.7944,
      "step": 8112
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.139692936981693,
      "learning_rate": 7.379151793329091e-06,
      "loss": 0.8596,
      "step": 8113
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.492424084958046,
      "learning_rate": 7.378516416186131e-06,
      "loss": 0.8167,
      "step": 8114
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.163571427289567,
      "learning_rate": 7.37788098939685e-06,
      "loss": 0.7745,
      "step": 8115
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.56024607509027,
      "learning_rate": 7.377245512974509e-06,
      "loss": 0.7268,
      "step": 8116
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.460747330606476,
      "learning_rate": 7.376609986932377e-06,
      "loss": 0.7669,
      "step": 8117
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.568018710852844,
      "learning_rate": 7.375974411283715e-06,
      "loss": 0.7892,
      "step": 8118
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.05694562092402,
      "learning_rate": 7.37533878604179e-06,
      "loss": 0.7418,
      "step": 8119
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.636941744156954,
      "learning_rate": 7.374703111219868e-06,
      "loss": 0.8294,
      "step": 8120
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.605860803618198,
      "learning_rate": 7.3740673868312226e-06,
      "loss": 0.7617,
      "step": 8121
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.257943659706699,
      "learning_rate": 7.373431612889117e-06,
      "loss": 0.7898,
      "step": 8122
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.114357954971109,
      "learning_rate": 7.3727957894068255e-06,
      "loss": 0.8347,
      "step": 8123
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.666580088384638,
      "learning_rate": 7.372159916397619e-06,
      "loss": 0.7464,
      "step": 8124
    },
    {
      "epoch": 0.72,
      "grad_norm": 8.450746272545596,
      "learning_rate": 7.371523993874767e-06,
      "loss": 0.8168,
      "step": 8125
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.383294505620616,
      "learning_rate": 7.3708880218515464e-06,
      "loss": 0.7939,
      "step": 8126
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.138773240577823,
      "learning_rate": 7.370252000341231e-06,
      "loss": 0.7847,
      "step": 8127
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.676041973607977,
      "learning_rate": 7.369615929357095e-06,
      "loss": 0.7214,
      "step": 8128
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.399500133804675,
      "learning_rate": 7.368979808912415e-06,
      "loss": 0.7378,
      "step": 8129
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.188150923110593,
      "learning_rate": 7.368343639020472e-06,
      "loss": 0.7962,
      "step": 8130
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.751843690438909,
      "learning_rate": 7.36770741969454e-06,
      "loss": 0.8084,
      "step": 8131
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.116544426286438,
      "learning_rate": 7.367071150947902e-06,
      "loss": 0.8465,
      "step": 8132
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.752078045604564,
      "learning_rate": 7.366434832793836e-06,
      "loss": 0.8067,
      "step": 8133
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.560948144841868,
      "learning_rate": 7.365798465245627e-06,
      "loss": 0.7136,
      "step": 8134
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.9771654726093075,
      "learning_rate": 7.3651620483165555e-06,
      "loss": 0.7784,
      "step": 8135
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.411138094075645,
      "learning_rate": 7.364525582019905e-06,
      "loss": 0.7858,
      "step": 8136
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.360833997413353,
      "learning_rate": 7.363889066368962e-06,
      "loss": 0.7581,
      "step": 8137
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.6924788277530025,
      "learning_rate": 7.363252501377012e-06,
      "loss": 0.8068,
      "step": 8138
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.429921911217902,
      "learning_rate": 7.362615887057341e-06,
      "loss": 0.7824,
      "step": 8139
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.25015939684266,
      "learning_rate": 7.361979223423237e-06,
      "loss": 0.8334,
      "step": 8140
    },
    {
      "epoch": 0.73,
      "grad_norm": 9.39274350546816,
      "learning_rate": 7.36134251048799e-06,
      "loss": 0.8137,
      "step": 8141
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.706601558361519,
      "learning_rate": 7.360705748264888e-06,
      "loss": 0.7938,
      "step": 8142
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.8880271821776367,
      "learning_rate": 7.3600689367672255e-06,
      "loss": 0.7763,
      "step": 8143
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.124075217787345,
      "learning_rate": 7.359432076008291e-06,
      "loss": 0.8068,
      "step": 8144
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.925306416161488,
      "learning_rate": 7.3587951660013775e-06,
      "loss": 0.8039,
      "step": 8145
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.026750889156244,
      "learning_rate": 7.358158206759784e-06,
      "loss": 0.7589,
      "step": 8146
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.050649136075178,
      "learning_rate": 7.357521198296801e-06,
      "loss": 0.8721,
      "step": 8147
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.921946267799854,
      "learning_rate": 7.356884140625724e-06,
      "loss": 0.6958,
      "step": 8148
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.754425094493915,
      "learning_rate": 7.3562470337598544e-06,
      "loss": 0.7722,
      "step": 8149
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.1375326585252035,
      "learning_rate": 7.355609877712487e-06,
      "loss": 0.7722,
      "step": 8150
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.270026953886901,
      "learning_rate": 7.354972672496922e-06,
      "loss": 0.8242,
      "step": 8151
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.956239758151039,
      "learning_rate": 7.354335418126461e-06,
      "loss": 0.8421,
      "step": 8152
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.524715051282911,
      "learning_rate": 7.353698114614401e-06,
      "loss": 0.7988,
      "step": 8153
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.246882918529215,
      "learning_rate": 7.353060761974051e-06,
      "loss": 0.7344,
      "step": 8154
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.153823344463887,
      "learning_rate": 7.352423360218709e-06,
      "loss": 0.7302,
      "step": 8155
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.82420779525798,
      "learning_rate": 7.3517859093616815e-06,
      "loss": 0.7752,
      "step": 8156
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.429377116574914,
      "learning_rate": 7.351148409416272e-06,
      "loss": 0.7353,
      "step": 8157
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.480967619913319,
      "learning_rate": 7.35051086039579e-06,
      "loss": 0.7279,
      "step": 8158
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.23828478101436,
      "learning_rate": 7.349873262313541e-06,
      "loss": 0.8069,
      "step": 8159
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.181121934820009,
      "learning_rate": 7.349235615182833e-06,
      "loss": 0.8229,
      "step": 8160
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.479877518489671,
      "learning_rate": 7.348597919016978e-06,
      "loss": 0.7423,
      "step": 8161
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.937150441700704,
      "learning_rate": 7.347960173829283e-06,
      "loss": 0.8332,
      "step": 8162
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.61873402128602,
      "learning_rate": 7.347322379633063e-06,
      "loss": 0.8333,
      "step": 8163
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.009773550000838,
      "learning_rate": 7.346684536441627e-06,
      "loss": 0.7234,
      "step": 8164
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.751044588983,
      "learning_rate": 7.3460466442682925e-06,
      "loss": 0.7445,
      "step": 8165
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.04958946728549,
      "learning_rate": 7.34540870312637e-06,
      "loss": 0.8324,
      "step": 8166
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.220895890956877,
      "learning_rate": 7.344770713029178e-06,
      "loss": 0.7754,
      "step": 8167
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.098962774753667,
      "learning_rate": 7.344132673990032e-06,
      "loss": 0.8167,
      "step": 8168
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.188955759673362,
      "learning_rate": 7.343494586022252e-06,
      "loss": 0.7944,
      "step": 8169
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.281147425714562,
      "learning_rate": 7.342856449139152e-06,
      "loss": 0.7211,
      "step": 8170
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.262832867239455,
      "learning_rate": 7.342218263354055e-06,
      "loss": 0.7757,
      "step": 8171
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.34352971900085,
      "learning_rate": 7.341580028680282e-06,
      "loss": 0.7379,
      "step": 8172
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.878759422939054,
      "learning_rate": 7.340941745131154e-06,
      "loss": 0.8073,
      "step": 8173
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.726836511678035,
      "learning_rate": 7.3403034127199915e-06,
      "loss": 0.8296,
      "step": 8174
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.618128115119916,
      "learning_rate": 7.339665031460124e-06,
      "loss": 0.7737,
      "step": 8175
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.386523619329982,
      "learning_rate": 7.33902660136487e-06,
      "loss": 0.7319,
      "step": 8176
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.9633759418296055,
      "learning_rate": 7.338388122447559e-06,
      "loss": 0.793,
      "step": 8177
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.086040039350024,
      "learning_rate": 7.337749594721517e-06,
      "loss": 0.7752,
      "step": 8178
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.313619430119193,
      "learning_rate": 7.337111018200071e-06,
      "loss": 0.8243,
      "step": 8179
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.514620412827863,
      "learning_rate": 7.336472392896551e-06,
      "loss": 0.8052,
      "step": 8180
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.75484674856618,
      "learning_rate": 7.335833718824288e-06,
      "loss": 0.7219,
      "step": 8181
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.022613627224324,
      "learning_rate": 7.33519499599661e-06,
      "loss": 0.7937,
      "step": 8182
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.336702845776959,
      "learning_rate": 7.334556224426851e-06,
      "loss": 0.7473,
      "step": 8183
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.9264764653449395,
      "learning_rate": 7.3339174041283435e-06,
      "loss": 0.7896,
      "step": 8184
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.517621043911835,
      "learning_rate": 7.333278535114421e-06,
      "loss": 0.7836,
      "step": 8185
    },
    {
      "epoch": 0.73,
      "grad_norm": 8.255672475787565,
      "learning_rate": 7.332639617398418e-06,
      "loss": 0.7545,
      "step": 8186
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.313940753626284,
      "learning_rate": 7.332000650993674e-06,
      "loss": 0.8306,
      "step": 8187
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.043628525359937,
      "learning_rate": 7.331361635913521e-06,
      "loss": 0.8038,
      "step": 8188
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.432803833918143,
      "learning_rate": 7.330722572171299e-06,
      "loss": 0.7357,
      "step": 8189
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.452159314785163,
      "learning_rate": 7.330083459780349e-06,
      "loss": 0.7858,
      "step": 8190
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.455305135730963,
      "learning_rate": 7.32944429875401e-06,
      "loss": 0.7786,
      "step": 8191
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.827093264614046,
      "learning_rate": 7.328805089105622e-06,
      "loss": 0.8125,
      "step": 8192
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.617177401438749,
      "learning_rate": 7.3281658308485275e-06,
      "loss": 0.7723,
      "step": 8193
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.895723488884495,
      "learning_rate": 7.32752652399607e-06,
      "loss": 0.7858,
      "step": 8194
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.253474884294706,
      "learning_rate": 7.326887168561594e-06,
      "loss": 0.7833,
      "step": 8195
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.553497574157832,
      "learning_rate": 7.326247764558443e-06,
      "loss": 0.7596,
      "step": 8196
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.409867943097077,
      "learning_rate": 7.325608311999966e-06,
      "loss": 0.7846,
      "step": 8197
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.389838011765209,
      "learning_rate": 7.324968810899508e-06,
      "loss": 0.8353,
      "step": 8198
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.398505278605998,
      "learning_rate": 7.324329261270417e-06,
      "loss": 0.771,
      "step": 8199
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.011640528397752,
      "learning_rate": 7.323689663126044e-06,
      "loss": 0.8059,
      "step": 8200
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.026970729297101,
      "learning_rate": 7.323050016479736e-06,
      "loss": 0.8159,
      "step": 8201
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.670324547392072,
      "learning_rate": 7.322410321344846e-06,
      "loss": 0.7766,
      "step": 8202
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.806274148249362,
      "learning_rate": 7.321770577734729e-06,
      "loss": 0.7961,
      "step": 8203
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.629675460350589,
      "learning_rate": 7.321130785662736e-06,
      "loss": 0.7861,
      "step": 8204
    },
    {
      "epoch": 0.73,
      "grad_norm": 9.62343965694151,
      "learning_rate": 7.320490945142219e-06,
      "loss": 0.7361,
      "step": 8205
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.226919766783721,
      "learning_rate": 7.319851056186536e-06,
      "loss": 0.8629,
      "step": 8206
    },
    {
      "epoch": 0.73,
      "grad_norm": 8.258630049767156,
      "learning_rate": 7.319211118809043e-06,
      "loss": 0.8151,
      "step": 8207
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.4765162352314025,
      "learning_rate": 7.318571133023096e-06,
      "loss": 0.6623,
      "step": 8208
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.144531713469088,
      "learning_rate": 7.317931098842054e-06,
      "loss": 0.7927,
      "step": 8209
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.184474694383772,
      "learning_rate": 7.317291016279278e-06,
      "loss": 0.758,
      "step": 8210
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.396792159301318,
      "learning_rate": 7.3166508853481265e-06,
      "loss": 0.7809,
      "step": 8211
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.407328161941519,
      "learning_rate": 7.31601070606196e-06,
      "loss": 0.8261,
      "step": 8212
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.699366166844901,
      "learning_rate": 7.315370478434144e-06,
      "loss": 0.7752,
      "step": 8213
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.998094230871877,
      "learning_rate": 7.314730202478038e-06,
      "loss": 0.791,
      "step": 8214
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.479004902472032,
      "learning_rate": 7.314089878207008e-06,
      "loss": 0.825,
      "step": 8215
    },
    {
      "epoch": 0.73,
      "grad_norm": 8.001385622532375,
      "learning_rate": 7.313449505634421e-06,
      "loss": 0.861,
      "step": 8216
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.514787635929313,
      "learning_rate": 7.312809084773641e-06,
      "loss": 0.7687,
      "step": 8217
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.286280660477105,
      "learning_rate": 7.3121686156380354e-06,
      "loss": 0.844,
      "step": 8218
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.837593908439315,
      "learning_rate": 7.311528098240975e-06,
      "loss": 0.7651,
      "step": 8219
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.86543775712285,
      "learning_rate": 7.310887532595828e-06,
      "loss": 0.8092,
      "step": 8220
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.334627167643357,
      "learning_rate": 7.310246918715964e-06,
      "loss": 0.7645,
      "step": 8221
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.094707964612209,
      "learning_rate": 7.3096062566147554e-06,
      "loss": 0.7294,
      "step": 8222
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.9148127637503025,
      "learning_rate": 7.308965546305573e-06,
      "loss": 0.7821,
      "step": 8223
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.430021899578567,
      "learning_rate": 7.3083247878017914e-06,
      "loss": 0.772,
      "step": 8224
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.482422037390778,
      "learning_rate": 7.307683981116787e-06,
      "loss": 0.7808,
      "step": 8225
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.087836271555243,
      "learning_rate": 7.307043126263932e-06,
      "loss": 0.7416,
      "step": 8226
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.070878468122808,
      "learning_rate": 7.306402223256604e-06,
      "loss": 0.7476,
      "step": 8227
    },
    {
      "epoch": 0.73,
      "grad_norm": 8.749045045204904,
      "learning_rate": 7.3057612721081785e-06,
      "loss": 0.7187,
      "step": 8228
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.3930924104658295,
      "learning_rate": 7.305120272832039e-06,
      "loss": 0.744,
      "step": 8229
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.111530002366528,
      "learning_rate": 7.304479225441561e-06,
      "loss": 0.7952,
      "step": 8230
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.684668024561761,
      "learning_rate": 7.3038381299501245e-06,
      "loss": 0.7456,
      "step": 8231
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.689876949463993,
      "learning_rate": 7.303196986371114e-06,
      "loss": 0.7704,
      "step": 8232
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.270557586515891,
      "learning_rate": 7.30255579471791e-06,
      "loss": 0.7943,
      "step": 8233
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.670306618048739,
      "learning_rate": 7.301914555003896e-06,
      "loss": 0.7939,
      "step": 8234
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.825631473379009,
      "learning_rate": 7.301273267242458e-06,
      "loss": 0.6778,
      "step": 8235
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.848727848154191,
      "learning_rate": 7.300631931446978e-06,
      "loss": 0.8059,
      "step": 8236
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.056150144197105,
      "learning_rate": 7.2999905476308465e-06,
      "loss": 0.7837,
      "step": 8237
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.335078503327824,
      "learning_rate": 7.299349115807448e-06,
      "loss": 0.7806,
      "step": 8238
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.214139373350839,
      "learning_rate": 7.298707635990173e-06,
      "loss": 0.7914,
      "step": 8239
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.444822233434615,
      "learning_rate": 7.298066108192409e-06,
      "loss": 0.8222,
      "step": 8240
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.80980767552241,
      "learning_rate": 7.2974245324275495e-06,
      "loss": 0.7735,
      "step": 8241
    },
    {
      "epoch": 0.74,
      "grad_norm": 8.92577650398868,
      "learning_rate": 7.296782908708984e-06,
      "loss": 0.8211,
      "step": 8242
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.547194658509991,
      "learning_rate": 7.296141237050104e-06,
      "loss": 0.8115,
      "step": 8243
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.05520107636626,
      "learning_rate": 7.295499517464304e-06,
      "loss": 0.7453,
      "step": 8244
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.884272134058154,
      "learning_rate": 7.29485774996498e-06,
      "loss": 0.7609,
      "step": 8245
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.568628495034229,
      "learning_rate": 7.294215934565525e-06,
      "loss": 0.8129,
      "step": 8246
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.867750842044337,
      "learning_rate": 7.293574071279337e-06,
      "loss": 0.8114,
      "step": 8247
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.492842197924874,
      "learning_rate": 7.292932160119814e-06,
      "loss": 0.7539,
      "step": 8248
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.9680380616880475,
      "learning_rate": 7.292290201100351e-06,
      "loss": 0.8569,
      "step": 8249
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.4373297380795735,
      "learning_rate": 7.291648194234353e-06,
      "loss": 0.734,
      "step": 8250
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.943839619023628,
      "learning_rate": 7.291006139535216e-06,
      "loss": 0.7609,
      "step": 8251
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.78657726247374,
      "learning_rate": 7.290364037016343e-06,
      "loss": 0.7456,
      "step": 8252
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.730714651251813,
      "learning_rate": 7.289721886691136e-06,
      "loss": 0.7415,
      "step": 8253
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.920655195668091,
      "learning_rate": 7.289079688573002e-06,
      "loss": 0.8746,
      "step": 8254
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.100700638119718,
      "learning_rate": 7.28843744267534e-06,
      "loss": 0.8321,
      "step": 8255
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.600113981041793,
      "learning_rate": 7.287795149011559e-06,
      "loss": 0.7663,
      "step": 8256
    },
    {
      "epoch": 0.74,
      "grad_norm": 9.765073908816124,
      "learning_rate": 7.2871528075950626e-06,
      "loss": 0.8414,
      "step": 8257
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.542197844300765,
      "learning_rate": 7.286510418439262e-06,
      "loss": 0.8065,
      "step": 8258
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.551351258624759,
      "learning_rate": 7.285867981557563e-06,
      "loss": 0.8763,
      "step": 8259
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.5642115416911535,
      "learning_rate": 7.285225496963377e-06,
      "loss": 0.8296,
      "step": 8260
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.536206175448339,
      "learning_rate": 7.284582964670112e-06,
      "loss": 0.8203,
      "step": 8261
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.003951357070649,
      "learning_rate": 7.283940384691181e-06,
      "loss": 0.7495,
      "step": 8262
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.678447676034929,
      "learning_rate": 7.283297757039996e-06,
      "loss": 0.7951,
      "step": 8263
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.06897896765556,
      "learning_rate": 7.282655081729971e-06,
      "loss": 0.7912,
      "step": 8264
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.104381861032628,
      "learning_rate": 7.282012358774519e-06,
      "loss": 0.8079,
      "step": 8265
    },
    {
      "epoch": 0.74,
      "grad_norm": 10.492019584871516,
      "learning_rate": 7.281369588187058e-06,
      "loss": 0.8701,
      "step": 8266
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.781702773911845,
      "learning_rate": 7.2807267699810035e-06,
      "loss": 0.8559,
      "step": 8267
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.35518996228275,
      "learning_rate": 7.28008390416977e-06,
      "loss": 0.7923,
      "step": 8268
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.939295055465337,
      "learning_rate": 7.279440990766781e-06,
      "loss": 0.7855,
      "step": 8269
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.555794425119006,
      "learning_rate": 7.278798029785452e-06,
      "loss": 0.8316,
      "step": 8270
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.815948646560741,
      "learning_rate": 7.278155021239205e-06,
      "loss": 0.7295,
      "step": 8271
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.144362607227897,
      "learning_rate": 7.277511965141459e-06,
      "loss": 0.7853,
      "step": 8272
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.213283979141265,
      "learning_rate": 7.2768688615056416e-06,
      "loss": 0.8109,
      "step": 8273
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.6177343103774815,
      "learning_rate": 7.27622571034517e-06,
      "loss": 0.7801,
      "step": 8274
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.729287431402088,
      "learning_rate": 7.275582511673473e-06,
      "loss": 0.8309,
      "step": 8275
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.313679522605185,
      "learning_rate": 7.274939265503975e-06,
      "loss": 0.7962,
      "step": 8276
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.065568880518448,
      "learning_rate": 7.274295971850102e-06,
      "loss": 0.8099,
      "step": 8277
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.36026109988822,
      "learning_rate": 7.273652630725279e-06,
      "loss": 0.8384,
      "step": 8278
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.450842053470196,
      "learning_rate": 7.2730092421429384e-06,
      "loss": 0.7207,
      "step": 8279
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.889531149092267,
      "learning_rate": 7.272365806116508e-06,
      "loss": 0.8569,
      "step": 8280
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.386139379552158,
      "learning_rate": 7.271722322659417e-06,
      "loss": 0.7769,
      "step": 8281
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.155952634226903,
      "learning_rate": 7.271078791785097e-06,
      "loss": 0.7587,
      "step": 8282
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.964535463160814,
      "learning_rate": 7.270435213506981e-06,
      "loss": 0.7375,
      "step": 8283
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.9868275580688515,
      "learning_rate": 7.269791587838504e-06,
      "loss": 0.8224,
      "step": 8284
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.182215152717567,
      "learning_rate": 7.269147914793095e-06,
      "loss": 0.7067,
      "step": 8285
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.207456478193949,
      "learning_rate": 7.268504194384196e-06,
      "loss": 0.7964,
      "step": 8286
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.7156438986789615,
      "learning_rate": 7.267860426625237e-06,
      "loss": 0.7899,
      "step": 8287
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.997498136531247,
      "learning_rate": 7.26721661152966e-06,
      "loss": 0.797,
      "step": 8288
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.81284517993646,
      "learning_rate": 7.266572749110901e-06,
      "loss": 0.7381,
      "step": 8289
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.4735471210760895,
      "learning_rate": 7.265928839382399e-06,
      "loss": 0.7415,
      "step": 8290
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.297761158358614,
      "learning_rate": 7.265284882357596e-06,
      "loss": 0.7919,
      "step": 8291
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.714953384695105,
      "learning_rate": 7.264640878049931e-06,
      "loss": 0.8531,
      "step": 8292
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.333871927312308,
      "learning_rate": 7.263996826472848e-06,
      "loss": 0.7717,
      "step": 8293
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.301446077504863,
      "learning_rate": 7.263352727639788e-06,
      "loss": 0.8516,
      "step": 8294
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.0921639161196595,
      "learning_rate": 7.262708581564196e-06,
      "loss": 0.8347,
      "step": 8295
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.955115848699971,
      "learning_rate": 7.26206438825952e-06,
      "loss": 0.7869,
      "step": 8296
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.241591799274524,
      "learning_rate": 7.261420147739202e-06,
      "loss": 0.7625,
      "step": 8297
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.854121834151947,
      "learning_rate": 7.260775860016691e-06,
      "loss": 0.8807,
      "step": 8298
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.91167508145503,
      "learning_rate": 7.260131525105436e-06,
      "loss": 0.7106,
      "step": 8299
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.867536414253444,
      "learning_rate": 7.259487143018884e-06,
      "loss": 0.743,
      "step": 8300
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.477922929120844,
      "learning_rate": 7.258842713770487e-06,
      "loss": 0.8052,
      "step": 8301
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.990262324693517,
      "learning_rate": 7.258198237373694e-06,
      "loss": 0.7262,
      "step": 8302
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.859882223332972,
      "learning_rate": 7.257553713841958e-06,
      "loss": 0.7423,
      "step": 8303
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.088863299287504,
      "learning_rate": 7.256909143188733e-06,
      "loss": 0.7933,
      "step": 8304
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.118562571875256,
      "learning_rate": 7.256264525427472e-06,
      "loss": 0.7897,
      "step": 8305
    },
    {
      "epoch": 0.74,
      "grad_norm": 10.586078210268099,
      "learning_rate": 7.25561986057163e-06,
      "loss": 0.7604,
      "step": 8306
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.47781222536973,
      "learning_rate": 7.254975148634663e-06,
      "loss": 0.7635,
      "step": 8307
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.0372527152208315,
      "learning_rate": 7.254330389630028e-06,
      "loss": 0.7566,
      "step": 8308
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.434800256519501,
      "learning_rate": 7.2536855835711835e-06,
      "loss": 0.7928,
      "step": 8309
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.547799212745193,
      "learning_rate": 7.253040730471588e-06,
      "loss": 0.7672,
      "step": 8310
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.1258534537314295,
      "learning_rate": 7.2523958303447015e-06,
      "loss": 0.7641,
      "step": 8311
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.879963092008109,
      "learning_rate": 7.2517508832039854e-06,
      "loss": 0.7799,
      "step": 8312
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.082152480610103,
      "learning_rate": 7.251105889062899e-06,
      "loss": 0.7915,
      "step": 8313
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.740330182518155,
      "learning_rate": 7.250460847934907e-06,
      "loss": 0.7964,
      "step": 8314
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.206982947354539,
      "learning_rate": 7.249815759833475e-06,
      "loss": 0.853,
      "step": 8315
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.386502128638249,
      "learning_rate": 7.249170624772064e-06,
      "loss": 0.7444,
      "step": 8316
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.47890943013288,
      "learning_rate": 7.248525442764143e-06,
      "loss": 0.8339,
      "step": 8317
    },
    {
      "epoch": 0.74,
      "grad_norm": 8.792987110329758,
      "learning_rate": 7.247880213823178e-06,
      "loss": 0.7854,
      "step": 8318
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.07755200826069,
      "learning_rate": 7.247234937962635e-06,
      "loss": 0.7944,
      "step": 8319
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.257484652885722,
      "learning_rate": 7.246589615195984e-06,
      "loss": 0.7665,
      "step": 8320
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.818876149323978,
      "learning_rate": 7.245944245536698e-06,
      "loss": 0.8659,
      "step": 8321
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.996314975722696,
      "learning_rate": 7.2452988289982405e-06,
      "loss": 0.8176,
      "step": 8322
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.7302282132724445,
      "learning_rate": 7.244653365594089e-06,
      "loss": 0.7654,
      "step": 8323
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.409073160446657,
      "learning_rate": 7.2440078553377135e-06,
      "loss": 0.8151,
      "step": 8324
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.896642578481406,
      "learning_rate": 7.243362298242588e-06,
      "loss": 0.7133,
      "step": 8325
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.080283589731239,
      "learning_rate": 7.2427166943221875e-06,
      "loss": 0.8341,
      "step": 8326
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.038394674870549,
      "learning_rate": 7.242071043589989e-06,
      "loss": 0.817,
      "step": 8327
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.664615757285098,
      "learning_rate": 7.241425346059466e-06,
      "loss": 0.7472,
      "step": 8328
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.445515093722944,
      "learning_rate": 7.2407796017440975e-06,
      "loss": 0.7538,
      "step": 8329
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.722247303594539,
      "learning_rate": 7.240133810657363e-06,
      "loss": 0.8149,
      "step": 8330
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.100761950454002,
      "learning_rate": 7.23948797281274e-06,
      "loss": 0.7438,
      "step": 8331
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.444312098701101,
      "learning_rate": 7.23884208822371e-06,
      "loss": 0.7998,
      "step": 8332
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.55869478480386,
      "learning_rate": 7.238196156903754e-06,
      "loss": 0.8334,
      "step": 8333
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.892199754993843,
      "learning_rate": 7.2375501788663574e-06,
      "loss": 0.9232,
      "step": 8334
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.977625103021928,
      "learning_rate": 7.236904154125e-06,
      "loss": 0.7941,
      "step": 8335
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.3258147420106114,
      "learning_rate": 7.236258082693165e-06,
      "loss": 0.7737,
      "step": 8336
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.1943668532581055,
      "learning_rate": 7.235611964584341e-06,
      "loss": 0.743,
      "step": 8337
    },
    {
      "epoch": 0.74,
      "grad_norm": 8.751534400072075,
      "learning_rate": 7.2349657998120135e-06,
      "loss": 0.7453,
      "step": 8338
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.7831565416906106,
      "learning_rate": 7.234319588389669e-06,
      "loss": 0.7246,
      "step": 8339
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.662634411961692,
      "learning_rate": 7.233673330330797e-06,
      "loss": 0.7993,
      "step": 8340
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.1490217949167105,
      "learning_rate": 7.233027025648886e-06,
      "loss": 0.753,
      "step": 8341
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.062131627271469,
      "learning_rate": 7.232380674357425e-06,
      "loss": 0.8385,
      "step": 8342
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.266242047399319,
      "learning_rate": 7.231734276469905e-06,
      "loss": 0.8341,
      "step": 8343
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.623009683654718,
      "learning_rate": 7.231087831999822e-06,
      "loss": 0.7532,
      "step": 8344
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.336365418647049,
      "learning_rate": 7.230441340960666e-06,
      "loss": 0.7439,
      "step": 8345
    },
    {
      "epoch": 0.74,
      "grad_norm": 9.050153365356373,
      "learning_rate": 7.229794803365931e-06,
      "loss": 0.8587,
      "step": 8346
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.221961658720766,
      "learning_rate": 7.229148219229112e-06,
      "loss": 0.7488,
      "step": 8347
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.185613535631729,
      "learning_rate": 7.228501588563707e-06,
      "loss": 0.7979,
      "step": 8348
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.970222451934138,
      "learning_rate": 7.227854911383211e-06,
      "loss": 0.7659,
      "step": 8349
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.387784850195836,
      "learning_rate": 7.227208187701122e-06,
      "loss": 0.7848,
      "step": 8350
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.438330868539399,
      "learning_rate": 7.22656141753094e-06,
      "loss": 0.8169,
      "step": 8351
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.223597107351562,
      "learning_rate": 7.225914600886164e-06,
      "loss": 0.79,
      "step": 8352
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.336398098488384,
      "learning_rate": 7.225267737780297e-06,
      "loss": 0.8299,
      "step": 8353
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.666711439008565,
      "learning_rate": 7.224620828226838e-06,
      "loss": 0.7139,
      "step": 8354
    },
    {
      "epoch": 0.75,
      "grad_norm": 8.378117430766114,
      "learning_rate": 7.223973872239292e-06,
      "loss": 0.7609,
      "step": 8355
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.664315752685664,
      "learning_rate": 7.223326869831162e-06,
      "loss": 0.7853,
      "step": 8356
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.189116361289905,
      "learning_rate": 7.222679821015952e-06,
      "loss": 0.7652,
      "step": 8357
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.933663059972507,
      "learning_rate": 7.222032725807168e-06,
      "loss": 0.7612,
      "step": 8358
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.224587589646313,
      "learning_rate": 7.221385584218318e-06,
      "loss": 0.7162,
      "step": 8359
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.3009386411949455,
      "learning_rate": 7.220738396262907e-06,
      "loss": 0.8257,
      "step": 8360
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.627393102759216,
      "learning_rate": 7.220091161954445e-06,
      "loss": 0.7583,
      "step": 8361
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.447362463435873,
      "learning_rate": 7.219443881306445e-06,
      "loss": 0.7487,
      "step": 8362
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.1954855729349685,
      "learning_rate": 7.218796554332412e-06,
      "loss": 0.7881,
      "step": 8363
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.735313710259194,
      "learning_rate": 7.21814918104586e-06,
      "loss": 0.7549,
      "step": 8364
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.36333590645361,
      "learning_rate": 7.217501761460303e-06,
      "loss": 0.755,
      "step": 8365
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.030231107760213,
      "learning_rate": 7.216854295589251e-06,
      "loss": 0.7928,
      "step": 8366
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.5258319501116,
      "learning_rate": 7.216206783446222e-06,
      "loss": 0.7428,
      "step": 8367
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.629756917956121,
      "learning_rate": 7.215559225044729e-06,
      "loss": 0.7716,
      "step": 8368
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.813427559292424,
      "learning_rate": 7.2149116203982905e-06,
      "loss": 0.7704,
      "step": 8369
    },
    {
      "epoch": 0.75,
      "grad_norm": 8.131204386503361,
      "learning_rate": 7.214263969520422e-06,
      "loss": 0.8797,
      "step": 8370
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.7882775545446306,
      "learning_rate": 7.213616272424641e-06,
      "loss": 0.7792,
      "step": 8371
    },
    {
      "epoch": 0.75,
      "grad_norm": 9.33117526773447,
      "learning_rate": 7.212968529124469e-06,
      "loss": 0.849,
      "step": 8372
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.802091795408552,
      "learning_rate": 7.212320739633425e-06,
      "loss": 0.7722,
      "step": 8373
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.941900574028413,
      "learning_rate": 7.21167290396503e-06,
      "loss": 0.8287,
      "step": 8374
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.386806106809522,
      "learning_rate": 7.211025022132807e-06,
      "loss": 0.8305,
      "step": 8375
    },
    {
      "epoch": 0.75,
      "grad_norm": 8.35052619983981,
      "learning_rate": 7.2103770941502795e-06,
      "loss": 0.7508,
      "step": 8376
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.606942277935321,
      "learning_rate": 7.2097291200309704e-06,
      "loss": 0.8042,
      "step": 8377
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.121855885853477,
      "learning_rate": 7.209081099788404e-06,
      "loss": 0.7285,
      "step": 8378
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.702280579456126,
      "learning_rate": 7.2084330334361095e-06,
      "loss": 0.7317,
      "step": 8379
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.559239209528451,
      "learning_rate": 7.20778492098761e-06,
      "loss": 0.8188,
      "step": 8380
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.046253534632505,
      "learning_rate": 7.207136762456436e-06,
      "loss": 0.7898,
      "step": 8381
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.36679177196306,
      "learning_rate": 7.206488557856116e-06,
      "loss": 0.7433,
      "step": 8382
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.5797347217738285,
      "learning_rate": 7.205840307200178e-06,
      "loss": 0.7433,
      "step": 8383
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.733312199559989,
      "learning_rate": 7.205192010502156e-06,
      "loss": 0.7846,
      "step": 8384
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.403596344087073,
      "learning_rate": 7.204543667775581e-06,
      "loss": 0.8124,
      "step": 8385
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.266665246256145,
      "learning_rate": 7.203895279033982e-06,
      "loss": 0.8335,
      "step": 8386
    },
    {
      "epoch": 0.75,
      "grad_norm": 8.660108354890642,
      "learning_rate": 7.2032468442908965e-06,
      "loss": 0.7213,
      "step": 8387
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.906848130125493,
      "learning_rate": 7.202598363559859e-06,
      "loss": 0.8056,
      "step": 8388
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.399929098048576,
      "learning_rate": 7.2019498368544025e-06,
      "loss": 0.7797,
      "step": 8389
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.499543402175248,
      "learning_rate": 7.201301264188066e-06,
      "loss": 0.7857,
      "step": 8390
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.60553211878301,
      "learning_rate": 7.200652645574389e-06,
      "loss": 0.759,
      "step": 8391
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.198775320010959,
      "learning_rate": 7.200003981026904e-06,
      "loss": 0.8393,
      "step": 8392
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.7636545866125175,
      "learning_rate": 7.199355270559155e-06,
      "loss": 0.8372,
      "step": 8393
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.72186868799078,
      "learning_rate": 7.198706514184683e-06,
      "loss": 0.8962,
      "step": 8394
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.340620939695928,
      "learning_rate": 7.198057711917025e-06,
      "loss": 0.7516,
      "step": 8395
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.393906384017675,
      "learning_rate": 7.197408863769726e-06,
      "loss": 0.6771,
      "step": 8396
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.149218752732445,
      "learning_rate": 7.19675996975633e-06,
      "loss": 0.7409,
      "step": 8397
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.876031133364807,
      "learning_rate": 7.19611102989038e-06,
      "loss": 0.7105,
      "step": 8398
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.399813958680154,
      "learning_rate": 7.195462044185424e-06,
      "loss": 0.8024,
      "step": 8399
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.87287201639954,
      "learning_rate": 7.194813012655003e-06,
      "loss": 0.7435,
      "step": 8400
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.643938396183751,
      "learning_rate": 7.194163935312668e-06,
      "loss": 0.8285,
      "step": 8401
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.195344351184833,
      "learning_rate": 7.193514812171965e-06,
      "loss": 0.7266,
      "step": 8402
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.942468657388759,
      "learning_rate": 7.192865643246445e-06,
      "loss": 0.7201,
      "step": 8403
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.173437328858069,
      "learning_rate": 7.192216428549658e-06,
      "loss": 0.8078,
      "step": 8404
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.206392573938052,
      "learning_rate": 7.191567168095152e-06,
      "loss": 0.8537,
      "step": 8405
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.1855961535176816,
      "learning_rate": 7.190917861896482e-06,
      "loss": 0.7877,
      "step": 8406
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.489091860694121,
      "learning_rate": 7.190268509967199e-06,
      "loss": 0.7551,
      "step": 8407
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.753403007819794,
      "learning_rate": 7.189619112320858e-06,
      "loss": 0.8196,
      "step": 8408
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.235309327023768,
      "learning_rate": 7.188969668971013e-06,
      "loss": 0.8115,
      "step": 8409
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.647414374190883,
      "learning_rate": 7.1883201799312194e-06,
      "loss": 0.7742,
      "step": 8410
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.741160106877535,
      "learning_rate": 7.1876706452150345e-06,
      "loss": 0.7848,
      "step": 8411
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.090642518223347,
      "learning_rate": 7.187021064836016e-06,
      "loss": 0.7159,
      "step": 8412
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.715633455717547,
      "learning_rate": 7.186371438807723e-06,
      "loss": 0.7866,
      "step": 8413
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.082965754532854,
      "learning_rate": 7.185721767143714e-06,
      "loss": 0.8743,
      "step": 8414
    },
    {
      "epoch": 0.75,
      "grad_norm": 8.312972622029012,
      "learning_rate": 7.185072049857549e-06,
      "loss": 0.7529,
      "step": 8415
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.315061212449712,
      "learning_rate": 7.184422286962791e-06,
      "loss": 0.7726,
      "step": 8416
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.58604553366292,
      "learning_rate": 7.183772478473001e-06,
      "loss": 0.7226,
      "step": 8417
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.5427728660284545,
      "learning_rate": 7.1831226244017424e-06,
      "loss": 0.7252,
      "step": 8418
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.162894030921875,
      "learning_rate": 7.18247272476258e-06,
      "loss": 0.8715,
      "step": 8419
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.370564914665877,
      "learning_rate": 7.181822779569081e-06,
      "loss": 0.7635,
      "step": 8420
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.203296995969408,
      "learning_rate": 7.181172788834809e-06,
      "loss": 0.7392,
      "step": 8421
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.97900302669416,
      "learning_rate": 7.18052275257333e-06,
      "loss": 0.785,
      "step": 8422
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.675128764213535,
      "learning_rate": 7.179872670798216e-06,
      "loss": 0.8616,
      "step": 8423
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.7744205415312395,
      "learning_rate": 7.179222543523033e-06,
      "loss": 0.7754,
      "step": 8424
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.6572016044584235,
      "learning_rate": 7.178572370761349e-06,
      "loss": 0.8073,
      "step": 8425
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.099172005829504,
      "learning_rate": 7.177922152526742e-06,
      "loss": 0.7836,
      "step": 8426
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.9088207888392965,
      "learning_rate": 7.177271888832778e-06,
      "loss": 0.7586,
      "step": 8427
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.870768131746314,
      "learning_rate": 7.17662157969303e-06,
      "loss": 0.723,
      "step": 8428
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.99519362145291,
      "learning_rate": 7.1759712251210746e-06,
      "loss": 0.7421,
      "step": 8429
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.67553997574567,
      "learning_rate": 7.175320825130485e-06,
      "loss": 0.7291,
      "step": 8430
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.150899260068749,
      "learning_rate": 7.174670379734837e-06,
      "loss": 0.7692,
      "step": 8431
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.070696800612766,
      "learning_rate": 7.174019888947708e-06,
      "loss": 0.8351,
      "step": 8432
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.5400558586834565,
      "learning_rate": 7.173369352782674e-06,
      "loss": 0.8047,
      "step": 8433
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.820875532404524,
      "learning_rate": 7.172718771253314e-06,
      "loss": 0.7878,
      "step": 8434
    },
    {
      "epoch": 0.75,
      "grad_norm": 8.009984336825047,
      "learning_rate": 7.1720681443732095e-06,
      "loss": 0.7392,
      "step": 8435
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.3255603849172095,
      "learning_rate": 7.171417472155939e-06,
      "loss": 0.8376,
      "step": 8436
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.51828593771627,
      "learning_rate": 7.170766754615083e-06,
      "loss": 0.8052,
      "step": 8437
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.658179066851794,
      "learning_rate": 7.170115991764224e-06,
      "loss": 0.7575,
      "step": 8438
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.397547032504536,
      "learning_rate": 7.1694651836169485e-06,
      "loss": 0.8623,
      "step": 8439
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.4366466022221935,
      "learning_rate": 7.168814330186836e-06,
      "loss": 0.7142,
      "step": 8440
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.857981067454428,
      "learning_rate": 7.1681634314874745e-06,
      "loss": 0.7473,
      "step": 8441
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.166990378189652,
      "learning_rate": 7.167512487532452e-06,
      "loss": 0.7494,
      "step": 8442
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.056175245528765,
      "learning_rate": 7.166861498335351e-06,
      "loss": 0.847,
      "step": 8443
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.112171346922174,
      "learning_rate": 7.166210463909762e-06,
      "loss": 0.7872,
      "step": 8444
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.459483718106011,
      "learning_rate": 7.165559384269274e-06,
      "loss": 0.7604,
      "step": 8445
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.418166920909698,
      "learning_rate": 7.164908259427474e-06,
      "loss": 0.7273,
      "step": 8446
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.605621286106605,
      "learning_rate": 7.164257089397957e-06,
      "loss": 0.7759,
      "step": 8447
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.14773209596183,
      "learning_rate": 7.163605874194313e-06,
      "loss": 0.7403,
      "step": 8448
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.726234560755668,
      "learning_rate": 7.162954613830136e-06,
      "loss": 0.7008,
      "step": 8449
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.386883210181328,
      "learning_rate": 7.162303308319015e-06,
      "loss": 0.8311,
      "step": 8450
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.752366303785898,
      "learning_rate": 7.161651957674549e-06,
      "loss": 0.7321,
      "step": 8451
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.975674626194347,
      "learning_rate": 7.161000561910334e-06,
      "loss": 0.7969,
      "step": 8452
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.274405938659868,
      "learning_rate": 7.160349121039963e-06,
      "loss": 0.7913,
      "step": 8453
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.2526794059224615,
      "learning_rate": 7.1596976350770345e-06,
      "loss": 0.767,
      "step": 8454
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.315204047428044,
      "learning_rate": 7.15904610403515e-06,
      "loss": 0.8677,
      "step": 8455
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.771898498310723,
      "learning_rate": 7.1583945279279035e-06,
      "loss": 0.7169,
      "step": 8456
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.118943353037846,
      "learning_rate": 7.157742906768901e-06,
      "loss": 0.7094,
      "step": 8457
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.512675401770207,
      "learning_rate": 7.157091240571739e-06,
      "loss": 0.7862,
      "step": 8458
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.773519712059057,
      "learning_rate": 7.1564395293500215e-06,
      "loss": 0.7443,
      "step": 8459
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.874952173545283,
      "learning_rate": 7.155787773117351e-06,
      "loss": 0.7725,
      "step": 8460
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.6770917743246585,
      "learning_rate": 7.155135971887333e-06,
      "loss": 0.7689,
      "step": 8461
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.378333364800352,
      "learning_rate": 7.15448412567357e-06,
      "loss": 0.8108,
      "step": 8462
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.863528728826642,
      "learning_rate": 7.153832234489668e-06,
      "loss": 0.7155,
      "step": 8463
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.461846907910876,
      "learning_rate": 7.1531802983492375e-06,
      "loss": 0.7728,
      "step": 8464
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.716510714810709,
      "learning_rate": 7.1525283172658835e-06,
      "loss": 0.7485,
      "step": 8465
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.720187914618092,
      "learning_rate": 7.1518762912532135e-06,
      "loss": 0.7997,
      "step": 8466
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.034561180222616,
      "learning_rate": 7.151224220324838e-06,
      "loss": 0.7599,
      "step": 8467
    },
    {
      "epoch": 0.76,
      "grad_norm": 9.558538981505068,
      "learning_rate": 7.15057210449437e-06,
      "loss": 0.7671,
      "step": 8468
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.175672235907901,
      "learning_rate": 7.149919943775417e-06,
      "loss": 0.7859,
      "step": 8469
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.522737869656957,
      "learning_rate": 7.149267738181595e-06,
      "loss": 0.7887,
      "step": 8470
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.777875392533825,
      "learning_rate": 7.148615487726516e-06,
      "loss": 0.7595,
      "step": 8471
    },
    {
      "epoch": 0.76,
      "grad_norm": 8.958412887625077,
      "learning_rate": 7.147963192423792e-06,
      "loss": 0.6681,
      "step": 8472
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.093942370918403,
      "learning_rate": 7.147310852287042e-06,
      "loss": 0.7954,
      "step": 8473
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.676659061586891,
      "learning_rate": 7.14665846732988e-06,
      "loss": 0.8388,
      "step": 8474
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.05849182330456,
      "learning_rate": 7.1460060375659235e-06,
      "loss": 0.7733,
      "step": 8475
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.707363705403732,
      "learning_rate": 7.14535356300879e-06,
      "loss": 0.7536,
      "step": 8476
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.725243142186797,
      "learning_rate": 7.1447010436720995e-06,
      "loss": 0.8603,
      "step": 8477
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.674095185953578,
      "learning_rate": 7.144048479569473e-06,
      "loss": 0.8038,
      "step": 8478
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.813837239402594,
      "learning_rate": 7.143395870714529e-06,
      "loss": 0.8272,
      "step": 8479
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.627530831496922,
      "learning_rate": 7.14274321712089e-06,
      "loss": 0.7733,
      "step": 8480
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.180235797761122,
      "learning_rate": 7.142090518802179e-06,
      "loss": 0.8048,
      "step": 8481
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.810392961081586,
      "learning_rate": 7.1414377757720174e-06,
      "loss": 0.7787,
      "step": 8482
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.0093732827784345,
      "learning_rate": 7.140784988044035e-06,
      "loss": 0.75,
      "step": 8483
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.96579821492246,
      "learning_rate": 7.140132155631852e-06,
      "loss": 0.8204,
      "step": 8484
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.790222016878875,
      "learning_rate": 7.1394792785491e-06,
      "loss": 0.7943,
      "step": 8485
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.2891076406851525,
      "learning_rate": 7.138826356809401e-06,
      "loss": 0.7356,
      "step": 8486
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.216554410122482,
      "learning_rate": 7.138173390426386e-06,
      "loss": 0.7485,
      "step": 8487
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.5008756713437466,
      "learning_rate": 7.1375203794136835e-06,
      "loss": 0.7346,
      "step": 8488
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.851773969638721,
      "learning_rate": 7.1368673237849255e-06,
      "loss": 0.7382,
      "step": 8489
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.634542090323374,
      "learning_rate": 7.136214223553741e-06,
      "loss": 0.837,
      "step": 8490
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.457605433860655,
      "learning_rate": 7.135561078733762e-06,
      "loss": 0.7631,
      "step": 8491
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.249307173382591,
      "learning_rate": 7.134907889338624e-06,
      "loss": 0.7842,
      "step": 8492
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.543722500432254,
      "learning_rate": 7.13425465538196e-06,
      "loss": 0.7841,
      "step": 8493
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.617535438363643,
      "learning_rate": 7.133601376877401e-06,
      "loss": 0.8343,
      "step": 8494
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.455328476326754,
      "learning_rate": 7.132948053838588e-06,
      "loss": 0.7958,
      "step": 8495
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.104735781674468,
      "learning_rate": 7.132294686279156e-06,
      "loss": 0.8018,
      "step": 8496
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.643810418479859,
      "learning_rate": 7.131641274212739e-06,
      "loss": 0.8163,
      "step": 8497
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.479496718470921,
      "learning_rate": 7.13098781765298e-06,
      "loss": 0.8104,
      "step": 8498
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.268275739039591,
      "learning_rate": 7.130334316613518e-06,
      "loss": 0.8628,
      "step": 8499
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.670354202759074,
      "learning_rate": 7.129680771107993e-06,
      "loss": 0.7803,
      "step": 8500
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.169012136892693,
      "learning_rate": 7.1290271811500435e-06,
      "loss": 0.79,
      "step": 8501
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.1034808987322196,
      "learning_rate": 7.128373546753315e-06,
      "loss": 0.7146,
      "step": 8502
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.351634929741886,
      "learning_rate": 7.1277198679314505e-06,
      "loss": 0.8274,
      "step": 8503
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.936849846257414,
      "learning_rate": 7.127066144698094e-06,
      "loss": 0.7809,
      "step": 8504
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.152009938821192,
      "learning_rate": 7.126412377066887e-06,
      "loss": 0.7596,
      "step": 8505
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.634665984975528,
      "learning_rate": 7.125758565051482e-06,
      "loss": 0.7339,
      "step": 8506
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.502896720373686,
      "learning_rate": 7.12510470866552e-06,
      "loss": 0.7912,
      "step": 8507
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.887819195733592,
      "learning_rate": 7.124450807922652e-06,
      "loss": 0.7714,
      "step": 8508
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.235649482623136,
      "learning_rate": 7.1237968628365265e-06,
      "loss": 0.7268,
      "step": 8509
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.44630660574832,
      "learning_rate": 7.123142873420791e-06,
      "loss": 0.7434,
      "step": 8510
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.817105570382496,
      "learning_rate": 7.1224888396890976e-06,
      "loss": 0.8501,
      "step": 8511
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.9859593258733215,
      "learning_rate": 7.121834761655099e-06,
      "loss": 0.7083,
      "step": 8512
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.111927861188702,
      "learning_rate": 7.1211806393324455e-06,
      "loss": 0.7878,
      "step": 8513
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.183424458084783,
      "learning_rate": 7.120526472734792e-06,
      "loss": 0.7597,
      "step": 8514
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.963072336820858,
      "learning_rate": 7.119872261875793e-06,
      "loss": 0.949,
      "step": 8515
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.929950443843143,
      "learning_rate": 7.1192180067691015e-06,
      "loss": 0.8407,
      "step": 8516
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.971827437964212,
      "learning_rate": 7.118563707428376e-06,
      "loss": 0.8882,
      "step": 8517
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.187536348717181,
      "learning_rate": 7.117909363867272e-06,
      "loss": 0.7884,
      "step": 8518
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.323254326933913,
      "learning_rate": 7.117254976099449e-06,
      "loss": 0.801,
      "step": 8519
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.096611662124351,
      "learning_rate": 7.1166005441385645e-06,
      "loss": 0.8163,
      "step": 8520
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.658168271688718,
      "learning_rate": 7.115946067998279e-06,
      "loss": 0.8485,
      "step": 8521
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.890814005756847,
      "learning_rate": 7.115291547692255e-06,
      "loss": 0.7807,
      "step": 8522
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.694899351141687,
      "learning_rate": 7.114636983234149e-06,
      "loss": 0.7722,
      "step": 8523
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.935215439162954,
      "learning_rate": 7.113982374637629e-06,
      "loss": 0.8293,
      "step": 8524
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.038824396197224,
      "learning_rate": 7.113327721916357e-06,
      "loss": 0.7608,
      "step": 8525
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.8085117551591,
      "learning_rate": 7.112673025083995e-06,
      "loss": 0.7068,
      "step": 8526
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.728400284551263,
      "learning_rate": 7.1120182841542105e-06,
      "loss": 0.8353,
      "step": 8527
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.059079466575563,
      "learning_rate": 7.111363499140671e-06,
      "loss": 0.7988,
      "step": 8528
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.312661212356789,
      "learning_rate": 7.1107086700570424e-06,
      "loss": 0.721,
      "step": 8529
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.384001896849415,
      "learning_rate": 7.1100537969169915e-06,
      "loss": 0.7767,
      "step": 8530
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.842111866777924,
      "learning_rate": 7.1093988797341894e-06,
      "loss": 0.7492,
      "step": 8531
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.7817112128253045,
      "learning_rate": 7.108743918522304e-06,
      "loss": 0.7867,
      "step": 8532
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.9655484699768255,
      "learning_rate": 7.1080889132950085e-06,
      "loss": 0.8597,
      "step": 8533
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.828116595509748,
      "learning_rate": 7.107433864065974e-06,
      "loss": 0.7983,
      "step": 8534
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.8309106248896,
      "learning_rate": 7.10677877084887e-06,
      "loss": 0.8142,
      "step": 8535
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.5137516192390805,
      "learning_rate": 7.106123633657376e-06,
      "loss": 0.8563,
      "step": 8536
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.45346241124193,
      "learning_rate": 7.105468452505162e-06,
      "loss": 0.748,
      "step": 8537
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.100384463048517,
      "learning_rate": 7.104813227405906e-06,
      "loss": 0.8028,
      "step": 8538
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.661358651635437,
      "learning_rate": 7.104157958373281e-06,
      "loss": 0.7707,
      "step": 8539
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.246023624020909,
      "learning_rate": 7.103502645420969e-06,
      "loss": 0.8398,
      "step": 8540
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.715079535467981,
      "learning_rate": 7.1028472885626455e-06,
      "loss": 0.7598,
      "step": 8541
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.717089517300284,
      "learning_rate": 7.102191887811989e-06,
      "loss": 0.8387,
      "step": 8542
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.69655335215486,
      "learning_rate": 7.101536443182682e-06,
      "loss": 0.7868,
      "step": 8543
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.333463136737821,
      "learning_rate": 7.1008809546884025e-06,
      "loss": 0.726,
      "step": 8544
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.198239033693332,
      "learning_rate": 7.100225422342834e-06,
      "loss": 0.7821,
      "step": 8545
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.836914752628593,
      "learning_rate": 7.0995698461596605e-06,
      "loss": 0.7062,
      "step": 8546
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.392028595616593,
      "learning_rate": 7.098914226152565e-06,
      "loss": 0.8411,
      "step": 8547
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.350330200689718,
      "learning_rate": 7.09825856233523e-06,
      "loss": 0.8146,
      "step": 8548
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.9815562477129203,
      "learning_rate": 7.097602854721342e-06,
      "loss": 0.7682,
      "step": 8549
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.946033016792804,
      "learning_rate": 7.09694710332459e-06,
      "loss": 0.7847,
      "step": 8550
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.128925130269883,
      "learning_rate": 7.096291308158658e-06,
      "loss": 0.7918,
      "step": 8551
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.986311349672925,
      "learning_rate": 7.095635469237236e-06,
      "loss": 0.7069,
      "step": 8552
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.776740352378233,
      "learning_rate": 7.094979586574015e-06,
      "loss": 0.8008,
      "step": 8553
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.729044131115055,
      "learning_rate": 7.09432366018268e-06,
      "loss": 0.7736,
      "step": 8554
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.550740652826757,
      "learning_rate": 7.0936676900769265e-06,
      "loss": 0.7456,
      "step": 8555
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.152281059750707,
      "learning_rate": 7.093011676270445e-06,
      "loss": 0.7775,
      "step": 8556
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.515939408253878,
      "learning_rate": 7.092355618776928e-06,
      "loss": 0.8608,
      "step": 8557
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.413621189887851,
      "learning_rate": 7.091699517610071e-06,
      "loss": 0.905,
      "step": 8558
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.566346870551573,
      "learning_rate": 7.091043372783566e-06,
      "loss": 0.8565,
      "step": 8559
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.904929806734705,
      "learning_rate": 7.090387184311109e-06,
      "loss": 0.8537,
      "step": 8560
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.3133411270966615,
      "learning_rate": 7.0897309522064e-06,
      "loss": 0.7002,
      "step": 8561
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.692325836663377,
      "learning_rate": 7.089074676483131e-06,
      "loss": 0.85,
      "step": 8562
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.762263907932586,
      "learning_rate": 7.088418357155005e-06,
      "loss": 0.8481,
      "step": 8563
    },
    {
      "epoch": 0.76,
      "grad_norm": 8.68480988313835,
      "learning_rate": 7.0877619942357175e-06,
      "loss": 0.7459,
      "step": 8564
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.486321572169034,
      "learning_rate": 7.087105587738974e-06,
      "loss": 0.7993,
      "step": 8565
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.879630862506185,
      "learning_rate": 7.0864491376784695e-06,
      "loss": 0.7834,
      "step": 8566
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.275618272592122,
      "learning_rate": 7.085792644067908e-06,
      "loss": 0.8631,
      "step": 8567
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.220604526418978,
      "learning_rate": 7.085136106920994e-06,
      "loss": 0.7502,
      "step": 8568
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.636024979973599,
      "learning_rate": 7.08447952625143e-06,
      "loss": 0.7028,
      "step": 8569
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.000667657306148,
      "learning_rate": 7.083822902072921e-06,
      "loss": 0.7956,
      "step": 8570
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.584325135583474,
      "learning_rate": 7.083166234399171e-06,
      "loss": 0.8134,
      "step": 8571
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.396436541197673,
      "learning_rate": 7.08250952324389e-06,
      "loss": 0.8333,
      "step": 8572
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.8237575569700475,
      "learning_rate": 7.0818527686207825e-06,
      "loss": 0.7463,
      "step": 8573
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.523882520458294,
      "learning_rate": 7.081195970543558e-06,
      "loss": 0.8178,
      "step": 8574
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.329425755392627,
      "learning_rate": 7.080539129025923e-06,
      "loss": 0.7855,
      "step": 8575
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.498254644790025,
      "learning_rate": 7.079882244081593e-06,
      "loss": 0.7779,
      "step": 8576
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.113938844245298,
      "learning_rate": 7.079225315724274e-06,
      "loss": 0.8186,
      "step": 8577
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.600071114258012,
      "learning_rate": 7.07856834396768e-06,
      "loss": 0.748,
      "step": 8578
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.751627150762267,
      "learning_rate": 7.077911328825526e-06,
      "loss": 0.8818,
      "step": 8579
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.7875619149795225,
      "learning_rate": 7.077254270311522e-06,
      "loss": 0.7476,
      "step": 8580
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.06552994467639,
      "learning_rate": 7.076597168439384e-06,
      "loss": 0.8232,
      "step": 8581
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.691219882983257,
      "learning_rate": 7.075940023222829e-06,
      "loss": 0.8193,
      "step": 8582
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.1557247622652875,
      "learning_rate": 7.075282834675571e-06,
      "loss": 0.7811,
      "step": 8583
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.65841640148906,
      "learning_rate": 7.074625602811328e-06,
      "loss": 0.7634,
      "step": 8584
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.562064231868681,
      "learning_rate": 7.073968327643821e-06,
      "loss": 0.7267,
      "step": 8585
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.949169552730887,
      "learning_rate": 7.073311009186767e-06,
      "loss": 0.7577,
      "step": 8586
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.093271328328505,
      "learning_rate": 7.0726536474538845e-06,
      "loss": 0.7986,
      "step": 8587
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.38847598055717,
      "learning_rate": 7.071996242458896e-06,
      "loss": 0.855,
      "step": 8588
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.145626620864722,
      "learning_rate": 7.0713387942155244e-06,
      "loss": 0.8214,
      "step": 8589
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.523382980242347,
      "learning_rate": 7.070681302737491e-06,
      "loss": 0.7947,
      "step": 8590
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.464788026919231,
      "learning_rate": 7.070023768038521e-06,
      "loss": 0.8611,
      "step": 8591
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.182775229591709,
      "learning_rate": 7.069366190132337e-06,
      "loss": 0.8602,
      "step": 8592
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.5656475760619095,
      "learning_rate": 7.0687085690326665e-06,
      "loss": 0.8155,
      "step": 8593
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.092929266916508,
      "learning_rate": 7.068050904753235e-06,
      "loss": 0.7111,
      "step": 8594
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.611102673194427,
      "learning_rate": 7.06739319730777e-06,
      "loss": 0.7579,
      "step": 8595
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.247220213622029,
      "learning_rate": 7.0667354467100005e-06,
      "loss": 0.7649,
      "step": 8596
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.5368421150959835,
      "learning_rate": 7.0660776529736515e-06,
      "loss": 0.7424,
      "step": 8597
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.1500636926629815,
      "learning_rate": 7.065419816112459e-06,
      "loss": 0.8097,
      "step": 8598
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.496015454738108,
      "learning_rate": 7.06476193614015e-06,
      "loss": 0.7345,
      "step": 8599
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.089033973743254,
      "learning_rate": 7.064104013070456e-06,
      "loss": 0.8238,
      "step": 8600
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.916422414102349,
      "learning_rate": 7.063446046917115e-06,
      "loss": 0.7188,
      "step": 8601
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.891597314170852,
      "learning_rate": 7.062788037693853e-06,
      "loss": 0.8338,
      "step": 8602
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.462826394889445,
      "learning_rate": 7.062129985414409e-06,
      "loss": 0.7428,
      "step": 8603
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.952845164652138,
      "learning_rate": 7.061471890092518e-06,
      "loss": 0.7741,
      "step": 8604
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.121399445356057,
      "learning_rate": 7.060813751741915e-06,
      "loss": 0.819,
      "step": 8605
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.048621406064104,
      "learning_rate": 7.060155570376338e-06,
      "loss": 0.7891,
      "step": 8606
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.0979410676740935,
      "learning_rate": 7.059497346009526e-06,
      "loss": 0.7276,
      "step": 8607
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.357736330048923,
      "learning_rate": 7.058839078655218e-06,
      "loss": 0.8298,
      "step": 8608
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.482643473817038,
      "learning_rate": 7.058180768327152e-06,
      "loss": 0.7523,
      "step": 8609
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.729775901916862,
      "learning_rate": 7.057522415039069e-06,
      "loss": 0.7947,
      "step": 8610
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.802095487790016,
      "learning_rate": 7.0568640188047125e-06,
      "loss": 0.8191,
      "step": 8611
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.182879769284677,
      "learning_rate": 7.056205579637823e-06,
      "loss": 0.7168,
      "step": 8612
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.138959148716554,
      "learning_rate": 7.055547097552146e-06,
      "loss": 0.7918,
      "step": 8613
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.072039282970667,
      "learning_rate": 7.054888572561425e-06,
      "loss": 0.8097,
      "step": 8614
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.532254596574271,
      "learning_rate": 7.054230004679405e-06,
      "loss": 0.7503,
      "step": 8615
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.136702503002688,
      "learning_rate": 7.053571393919832e-06,
      "loss": 0.7364,
      "step": 8616
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.528163500152788,
      "learning_rate": 7.052912740296454e-06,
      "loss": 0.8539,
      "step": 8617
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.263385065017905,
      "learning_rate": 7.052254043823017e-06,
      "loss": 0.7728,
      "step": 8618
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.729961703149936,
      "learning_rate": 7.051595304513273e-06,
      "loss": 0.7315,
      "step": 8619
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.691112824850399,
      "learning_rate": 7.050936522380969e-06,
      "loss": 0.721,
      "step": 8620
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.620644213768923,
      "learning_rate": 7.050277697439856e-06,
      "loss": 0.7921,
      "step": 8621
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.280583418432948,
      "learning_rate": 7.049618829703687e-06,
      "loss": 0.7353,
      "step": 8622
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.996966779740856,
      "learning_rate": 7.048959919186212e-06,
      "loss": 0.7876,
      "step": 8623
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.5323939902869395,
      "learning_rate": 7.048300965901188e-06,
      "loss": 0.7576,
      "step": 8624
    },
    {
      "epoch": 0.77,
      "grad_norm": 8.880897610274847,
      "learning_rate": 7.047641969862365e-06,
      "loss": 0.7819,
      "step": 8625
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.270349612871482,
      "learning_rate": 7.046982931083501e-06,
      "loss": 0.8385,
      "step": 8626
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.902789953742002,
      "learning_rate": 7.046323849578353e-06,
      "loss": 0.744,
      "step": 8627
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.0564871859159375,
      "learning_rate": 7.045664725360673e-06,
      "loss": 0.787,
      "step": 8628
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.53756858752295,
      "learning_rate": 7.045005558444224e-06,
      "loss": 0.82,
      "step": 8629
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.130417939927797,
      "learning_rate": 7.044346348842762e-06,
      "loss": 0.7521,
      "step": 8630
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.3123738127346485,
      "learning_rate": 7.043687096570046e-06,
      "loss": 0.8334,
      "step": 8631
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.7379462398617695,
      "learning_rate": 7.043027801639838e-06,
      "loss": 0.791,
      "step": 8632
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.45203931612486,
      "learning_rate": 7.0423684640659e-06,
      "loss": 0.761,
      "step": 8633
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.18012118780697,
      "learning_rate": 7.041709083861991e-06,
      "loss": 0.7799,
      "step": 8634
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.773637842530373,
      "learning_rate": 7.041049661041877e-06,
      "loss": 0.772,
      "step": 8635
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.0902460332241235,
      "learning_rate": 7.040390195619322e-06,
      "loss": 0.8057,
      "step": 8636
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.105430368696957,
      "learning_rate": 7.03973068760809e-06,
      "loss": 0.8016,
      "step": 8637
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.520327770118005,
      "learning_rate": 7.039071137021945e-06,
      "loss": 0.8248,
      "step": 8638
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.225895786650584,
      "learning_rate": 7.038411543874657e-06,
      "loss": 0.7956,
      "step": 8639
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.3790384956305415,
      "learning_rate": 7.0377519081799905e-06,
      "loss": 0.7586,
      "step": 8640
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.964020171373996,
      "learning_rate": 7.0370922299517165e-06,
      "loss": 0.7678,
      "step": 8641
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.9934738503145875,
      "learning_rate": 7.036432509203605e-06,
      "loss": 0.755,
      "step": 8642
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.822118563807853,
      "learning_rate": 7.035772745949422e-06,
      "loss": 0.7679,
      "step": 8643
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.916371019277149,
      "learning_rate": 7.035112940202943e-06,
      "loss": 0.8005,
      "step": 8644
    },
    {
      "epoch": 0.77,
      "grad_norm": 8.992323098987782,
      "learning_rate": 7.034453091977937e-06,
      "loss": 0.7527,
      "step": 8645
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.754907902936067,
      "learning_rate": 7.033793201288179e-06,
      "loss": 0.8082,
      "step": 8646
    },
    {
      "epoch": 0.77,
      "grad_norm": 8.430636231759152,
      "learning_rate": 7.033133268147441e-06,
      "loss": 0.695,
      "step": 8647
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.933098131757555,
      "learning_rate": 7.0324732925694985e-06,
      "loss": 0.7427,
      "step": 8648
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.106873071408378,
      "learning_rate": 7.031813274568128e-06,
      "loss": 0.8342,
      "step": 8649
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.19025691478056,
      "learning_rate": 7.031153214157105e-06,
      "loss": 0.7102,
      "step": 8650
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.673780188951984,
      "learning_rate": 7.030493111350207e-06,
      "loss": 0.786,
      "step": 8651
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.853396973020694,
      "learning_rate": 7.029832966161211e-06,
      "loss": 0.797,
      "step": 8652
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.125287075266279,
      "learning_rate": 7.029172778603897e-06,
      "loss": 0.7447,
      "step": 8653
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.124244550306576,
      "learning_rate": 7.0285125486920445e-06,
      "loss": 0.7118,
      "step": 8654
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.816983436509502,
      "learning_rate": 7.027852276439437e-06,
      "loss": 0.8066,
      "step": 8655
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.226018845279421,
      "learning_rate": 7.027191961859851e-06,
      "loss": 0.714,
      "step": 8656
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.908459435049178,
      "learning_rate": 7.026531604967075e-06,
      "loss": 0.8056,
      "step": 8657
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.7192084888365455,
      "learning_rate": 7.02587120577489e-06,
      "loss": 0.7691,
      "step": 8658
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.939117880984281,
      "learning_rate": 7.025210764297079e-06,
      "loss": 0.7032,
      "step": 8659
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.939471382052788,
      "learning_rate": 7.024550280547429e-06,
      "loss": 0.7931,
      "step": 8660
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.878422788902638,
      "learning_rate": 7.023889754539725e-06,
      "loss": 0.7165,
      "step": 8661
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.0033554366942194,
      "learning_rate": 7.023229186287755e-06,
      "loss": 0.7479,
      "step": 8662
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.5569034881191754,
      "learning_rate": 7.022568575805307e-06,
      "loss": 0.7389,
      "step": 8663
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.9509160532028895,
      "learning_rate": 7.021907923106167e-06,
      "loss": 0.815,
      "step": 8664
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.25896838350508,
      "learning_rate": 7.021247228204129e-06,
      "loss": 0.7763,
      "step": 8665
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.276753145928765,
      "learning_rate": 7.020586491112982e-06,
      "loss": 0.8098,
      "step": 8666
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.636041684877542,
      "learning_rate": 7.0199257118465155e-06,
      "loss": 0.7827,
      "step": 8667
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.49553482801673,
      "learning_rate": 7.019264890418524e-06,
      "loss": 0.8846,
      "step": 8668
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.822765626685625,
      "learning_rate": 7.018604026842801e-06,
      "loss": 0.7964,
      "step": 8669
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.175701294105965,
      "learning_rate": 7.0179431211331386e-06,
      "loss": 0.7491,
      "step": 8670
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.2235715110235015,
      "learning_rate": 7.017282173303333e-06,
      "loss": 0.7993,
      "step": 8671
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.33419695549831,
      "learning_rate": 7.01662118336718e-06,
      "loss": 0.7743,
      "step": 8672
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.052462389922944,
      "learning_rate": 7.0159601513384754e-06,
      "loss": 0.6812,
      "step": 8673
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.058178882174749,
      "learning_rate": 7.015299077231018e-06,
      "loss": 0.712,
      "step": 8674
    },
    {
      "epoch": 0.77,
      "grad_norm": 8.555893050241734,
      "learning_rate": 7.014637961058606e-06,
      "loss": 0.753,
      "step": 8675
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.9600673390640155,
      "learning_rate": 7.0139768028350386e-06,
      "loss": 0.8019,
      "step": 8676
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.5125027073245025,
      "learning_rate": 7.0133156025741154e-06,
      "loss": 0.8116,
      "step": 8677
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.348857441653793,
      "learning_rate": 7.012654360289638e-06,
      "loss": 0.7326,
      "step": 8678
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.13290288490775,
      "learning_rate": 7.011993075995409e-06,
      "loss": 0.792,
      "step": 8679
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.337607642857867,
      "learning_rate": 7.01133174970523e-06,
      "loss": 0.8457,
      "step": 8680
    },
    {
      "epoch": 0.77,
      "grad_norm": 8.385371325897804,
      "learning_rate": 7.010670381432907e-06,
      "loss": 0.8372,
      "step": 8681
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.107359164932941,
      "learning_rate": 7.010008971192241e-06,
      "loss": 0.7403,
      "step": 8682
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.847199859293259,
      "learning_rate": 7.0093475189970405e-06,
      "loss": 0.7612,
      "step": 8683
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.893853895487424,
      "learning_rate": 7.008686024861112e-06,
      "loss": 0.7599,
      "step": 8684
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.1372292083109805,
      "learning_rate": 7.00802448879826e-06,
      "loss": 0.8457,
      "step": 8685
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.69470677546294,
      "learning_rate": 7.007362910822295e-06,
      "loss": 0.7352,
      "step": 8686
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.722329329951982,
      "learning_rate": 7.0067012909470265e-06,
      "loss": 0.725,
      "step": 8687
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.371226386585198,
      "learning_rate": 7.006039629186262e-06,
      "loss": 0.8201,
      "step": 8688
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.101062986796609,
      "learning_rate": 7.005377925553813e-06,
      "loss": 0.85,
      "step": 8689
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.578162078671996,
      "learning_rate": 7.004716180063493e-06,
      "loss": 0.7668,
      "step": 8690
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.477909131391436,
      "learning_rate": 7.004054392729112e-06,
      "loss": 0.7142,
      "step": 8691
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.6778229110072544,
      "learning_rate": 7.003392563564483e-06,
      "loss": 0.6994,
      "step": 8692
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.096129419029837,
      "learning_rate": 7.0027306925834234e-06,
      "loss": 0.8052,
      "step": 8693
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.507391940768327,
      "learning_rate": 7.002068779799746e-06,
      "loss": 0.6833,
      "step": 8694
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.121084023074429,
      "learning_rate": 7.0014068252272675e-06,
      "loss": 0.7241,
      "step": 8695
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.677515719216135,
      "learning_rate": 7.000744828879804e-06,
      "loss": 0.8228,
      "step": 8696
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.334503914127166,
      "learning_rate": 7.000082790771174e-06,
      "loss": 0.8295,
      "step": 8697
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.7588246250853015,
      "learning_rate": 6.999420710915196e-06,
      "loss": 0.8003,
      "step": 8698
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.705119526264606,
      "learning_rate": 6.998758589325688e-06,
      "loss": 0.7771,
      "step": 8699
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.765272004935141,
      "learning_rate": 6.998096426016474e-06,
      "loss": 0.6978,
      "step": 8700
    },
    {
      "epoch": 0.78,
      "grad_norm": 9.402101225700655,
      "learning_rate": 6.997434221001371e-06,
      "loss": 0.8194,
      "step": 8701
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.764403352485659,
      "learning_rate": 6.996771974294204e-06,
      "loss": 0.8163,
      "step": 8702
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.9566025479288776,
      "learning_rate": 6.996109685908795e-06,
      "loss": 0.7431,
      "step": 8703
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.029321315819416,
      "learning_rate": 6.995447355858967e-06,
      "loss": 0.7899,
      "step": 8704
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.296764723478985,
      "learning_rate": 6.994784984158544e-06,
      "loss": 0.6696,
      "step": 8705
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.551627925667924,
      "learning_rate": 6.994122570821357e-06,
      "loss": 0.7338,
      "step": 8706
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.776684488490045,
      "learning_rate": 6.9934601158612244e-06,
      "loss": 0.7699,
      "step": 8707
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.978215355176727,
      "learning_rate": 6.992797619291979e-06,
      "loss": 0.8188,
      "step": 8708
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.294305712239087,
      "learning_rate": 6.992135081127448e-06,
      "loss": 0.7643,
      "step": 8709
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.639776438866828,
      "learning_rate": 6.99147250138146e-06,
      "loss": 0.8239,
      "step": 8710
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.533655469431066,
      "learning_rate": 6.990809880067843e-06,
      "loss": 0.7438,
      "step": 8711
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.64018741551741,
      "learning_rate": 6.9901472172004316e-06,
      "loss": 0.7896,
      "step": 8712
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.763957521268249,
      "learning_rate": 6.989484512793054e-06,
      "loss": 0.802,
      "step": 8713
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.3284355846393785,
      "learning_rate": 6.988821766859545e-06,
      "loss": 0.7593,
      "step": 8714
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.049808314967526,
      "learning_rate": 6.988158979413737e-06,
      "loss": 0.8177,
      "step": 8715
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.531596241688552,
      "learning_rate": 6.987496150469465e-06,
      "loss": 0.826,
      "step": 8716
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.413820365584003,
      "learning_rate": 6.986833280040561e-06,
      "loss": 0.8329,
      "step": 8717
    },
    {
      "epoch": 0.78,
      "grad_norm": 9.754624102743538,
      "learning_rate": 6.986170368140865e-06,
      "loss": 0.8406,
      "step": 8718
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.209833937037458,
      "learning_rate": 6.985507414784213e-06,
      "loss": 0.7923,
      "step": 8719
    },
    {
      "epoch": 0.78,
      "grad_norm": 8.062856941827409,
      "learning_rate": 6.98484441998444e-06,
      "loss": 0.7498,
      "step": 8720
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.441369272968425,
      "learning_rate": 6.984181383755386e-06,
      "loss": 0.7754,
      "step": 8721
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.382854422908348,
      "learning_rate": 6.983518306110894e-06,
      "loss": 0.7601,
      "step": 8722
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.266080920277388,
      "learning_rate": 6.9828551870647996e-06,
      "loss": 0.7247,
      "step": 8723
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.455564339574714,
      "learning_rate": 6.982192026630945e-06,
      "loss": 0.7349,
      "step": 8724
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.532274823209501,
      "learning_rate": 6.981528824823174e-06,
      "loss": 0.8564,
      "step": 8725
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.76292770846291,
      "learning_rate": 6.980865581655327e-06,
      "loss": 0.7521,
      "step": 8726
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.468150988718446,
      "learning_rate": 6.9802022971412485e-06,
      "loss": 0.7983,
      "step": 8727
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.948115928090318,
      "learning_rate": 6.979538971294785e-06,
      "loss": 0.7209,
      "step": 8728
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.785185389673297,
      "learning_rate": 6.97887560412978e-06,
      "loss": 0.7822,
      "step": 8729
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.467324154195559,
      "learning_rate": 6.97821219566008e-06,
      "loss": 0.7361,
      "step": 8730
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.566096646457569,
      "learning_rate": 6.9775487458995335e-06,
      "loss": 0.772,
      "step": 8731
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.196555339629652,
      "learning_rate": 6.976885254861988e-06,
      "loss": 0.7656,
      "step": 8732
    },
    {
      "epoch": 0.78,
      "grad_norm": 10.022248240386226,
      "learning_rate": 6.9762217225612915e-06,
      "loss": 0.8759,
      "step": 8733
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.551940366881301,
      "learning_rate": 6.975558149011293e-06,
      "loss": 0.8734,
      "step": 8734
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.730763051793409,
      "learning_rate": 6.974894534225847e-06,
      "loss": 0.7713,
      "step": 8735
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.068940664713013,
      "learning_rate": 6.974230878218801e-06,
      "loss": 0.9028,
      "step": 8736
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.660510451386671,
      "learning_rate": 6.973567181004008e-06,
      "loss": 0.7339,
      "step": 8737
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.118537617527327,
      "learning_rate": 6.9729034425953254e-06,
      "loss": 0.7708,
      "step": 8738
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.233936864827721,
      "learning_rate": 6.972239663006602e-06,
      "loss": 0.7718,
      "step": 8739
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.8112506944001,
      "learning_rate": 6.971575842251695e-06,
      "loss": 0.7404,
      "step": 8740
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.507437468598479,
      "learning_rate": 6.970911980344461e-06,
      "loss": 0.7399,
      "step": 8741
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.193477075567995,
      "learning_rate": 6.970248077298753e-06,
      "loss": 0.7499,
      "step": 8742
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.883516293777517,
      "learning_rate": 6.969584133128432e-06,
      "loss": 0.805,
      "step": 8743
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.209645475991228,
      "learning_rate": 6.968920147847356e-06,
      "loss": 0.8618,
      "step": 8744
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.421834644241645,
      "learning_rate": 6.968256121469385e-06,
      "loss": 0.7323,
      "step": 8745
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.828737317836777,
      "learning_rate": 6.9675920540083765e-06,
      "loss": 0.7422,
      "step": 8746
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.675284800111492,
      "learning_rate": 6.966927945478193e-06,
      "loss": 0.8761,
      "step": 8747
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.435909920368553,
      "learning_rate": 6.966263795892697e-06,
      "loss": 0.7874,
      "step": 8748
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.567047397223353,
      "learning_rate": 6.965599605265749e-06,
      "loss": 0.7885,
      "step": 8749
    },
    {
      "epoch": 0.78,
      "grad_norm": 8.367958307212975,
      "learning_rate": 6.964935373611213e-06,
      "loss": 0.8004,
      "step": 8750
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.908323854964598,
      "learning_rate": 6.964271100942956e-06,
      "loss": 0.8192,
      "step": 8751
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.03702210421218,
      "learning_rate": 6.9636067872748426e-06,
      "loss": 0.6516,
      "step": 8752
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.162900459124763,
      "learning_rate": 6.962942432620736e-06,
      "loss": 0.7893,
      "step": 8753
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.560150329844022,
      "learning_rate": 6.962278036994505e-06,
      "loss": 0.8093,
      "step": 8754
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.650219526719628,
      "learning_rate": 6.961613600410018e-06,
      "loss": 0.7464,
      "step": 8755
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.308691334088479,
      "learning_rate": 6.960949122881141e-06,
      "loss": 0.7305,
      "step": 8756
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.943569185117365,
      "learning_rate": 6.960284604421748e-06,
      "loss": 0.7926,
      "step": 8757
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.478436457526467,
      "learning_rate": 6.959620045045705e-06,
      "loss": 0.807,
      "step": 8758
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.063341609290813,
      "learning_rate": 6.958955444766886e-06,
      "loss": 0.8299,
      "step": 8759
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.325802888834849,
      "learning_rate": 6.958290803599163e-06,
      "loss": 0.847,
      "step": 8760
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.64414140597867,
      "learning_rate": 6.957626121556407e-06,
      "loss": 0.7967,
      "step": 8761
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.166698153645633,
      "learning_rate": 6.956961398652491e-06,
      "loss": 0.7698,
      "step": 8762
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.855507639401478,
      "learning_rate": 6.956296634901294e-06,
      "loss": 0.8143,
      "step": 8763
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.919010968465467,
      "learning_rate": 6.955631830316688e-06,
      "loss": 0.8048,
      "step": 8764
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.842304388563717,
      "learning_rate": 6.95496698491255e-06,
      "loss": 0.7787,
      "step": 8765
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.941451591399437,
      "learning_rate": 6.954302098702759e-06,
      "loss": 0.815,
      "step": 8766
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.44591927973432,
      "learning_rate": 6.953637171701191e-06,
      "loss": 0.7754,
      "step": 8767
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.633333560082455,
      "learning_rate": 6.952972203921724e-06,
      "loss": 0.7836,
      "step": 8768
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.136685079305134,
      "learning_rate": 6.9523071953782395e-06,
      "loss": 0.7482,
      "step": 8769
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.033587900965573,
      "learning_rate": 6.9516421460846184e-06,
      "loss": 0.8036,
      "step": 8770
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.0574686474447885,
      "learning_rate": 6.950977056054742e-06,
      "loss": 0.8191,
      "step": 8771
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.732988301155716,
      "learning_rate": 6.950311925302491e-06,
      "loss": 0.7989,
      "step": 8772
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.256539639709001,
      "learning_rate": 6.949646753841751e-06,
      "loss": 0.7719,
      "step": 8773
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.4758730067637655,
      "learning_rate": 6.948981541686404e-06,
      "loss": 0.9168,
      "step": 8774
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.5139770382602205,
      "learning_rate": 6.948316288850335e-06,
      "loss": 0.7832,
      "step": 8775
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.543695192470939,
      "learning_rate": 6.9476509953474315e-06,
      "loss": 0.7497,
      "step": 8776
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.2995260797640595,
      "learning_rate": 6.946985661191578e-06,
      "loss": 0.7931,
      "step": 8777
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.209498930988276,
      "learning_rate": 6.946320286396661e-06,
      "loss": 0.7922,
      "step": 8778
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.542852672214769,
      "learning_rate": 6.945654870976573e-06,
      "loss": 0.839,
      "step": 8779
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.40461669508864,
      "learning_rate": 6.944989414945199e-06,
      "loss": 0.795,
      "step": 8780
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.130349986596892,
      "learning_rate": 6.9443239183164315e-06,
      "loss": 0.7071,
      "step": 8781
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.456066348319619,
      "learning_rate": 6.9436583811041594e-06,
      "loss": 0.8628,
      "step": 8782
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.467129934819941,
      "learning_rate": 6.942992803322276e-06,
      "loss": 0.7533,
      "step": 8783
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.95311823118279,
      "learning_rate": 6.9423271849846715e-06,
      "loss": 0.8677,
      "step": 8784
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.504575244194517,
      "learning_rate": 6.941661526105241e-06,
      "loss": 0.7732,
      "step": 8785
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.732323672663427,
      "learning_rate": 6.94099582669788e-06,
      "loss": 0.7114,
      "step": 8786
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.66520220537095,
      "learning_rate": 6.940330086776479e-06,
      "loss": 0.7567,
      "step": 8787
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.834789963194458,
      "learning_rate": 6.9396643063549396e-06,
      "loss": 0.8162,
      "step": 8788
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.330490237684976,
      "learning_rate": 6.938998485447155e-06,
      "loss": 0.807,
      "step": 8789
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.928042639929164,
      "learning_rate": 6.938332624067021e-06,
      "loss": 0.7897,
      "step": 8790
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.411202479440829,
      "learning_rate": 6.9376667222284396e-06,
      "loss": 0.856,
      "step": 8791
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.612793208902624,
      "learning_rate": 6.93700077994531e-06,
      "loss": 0.8439,
      "step": 8792
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.7009731309011595,
      "learning_rate": 6.9363347972315296e-06,
      "loss": 0.7928,
      "step": 8793
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.281401842302976,
      "learning_rate": 6.935668774101001e-06,
      "loss": 0.7002,
      "step": 8794
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.278266443587289,
      "learning_rate": 6.935002710567627e-06,
      "loss": 0.8154,
      "step": 8795
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.71029046591003,
      "learning_rate": 6.9343366066453085e-06,
      "loss": 0.7565,
      "step": 8796
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.472494049169806,
      "learning_rate": 6.933670462347949e-06,
      "loss": 0.784,
      "step": 8797
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.857073054840742,
      "learning_rate": 6.933004277689453e-06,
      "loss": 0.7878,
      "step": 8798
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.404655289151616,
      "learning_rate": 6.932338052683727e-06,
      "loss": 0.8191,
      "step": 8799
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.986714499533353,
      "learning_rate": 6.931671787344674e-06,
      "loss": 0.7696,
      "step": 8800
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.3868838052409,
      "learning_rate": 6.931005481686205e-06,
      "loss": 0.8247,
      "step": 8801
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.741346870898593,
      "learning_rate": 6.930339135722224e-06,
      "loss": 0.7446,
      "step": 8802
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.124305963310402,
      "learning_rate": 6.9296727494666415e-06,
      "loss": 0.8559,
      "step": 8803
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.898810014295062,
      "learning_rate": 6.929006322933365e-06,
      "loss": 0.7697,
      "step": 8804
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.077895530198855,
      "learning_rate": 6.928339856136308e-06,
      "loss": 0.7527,
      "step": 8805
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.709622136248156,
      "learning_rate": 6.927673349089378e-06,
      "loss": 0.746,
      "step": 8806
    },
    {
      "epoch": 0.79,
      "grad_norm": 8.868017333718155,
      "learning_rate": 6.927006801806488e-06,
      "loss": 0.7979,
      "step": 8807
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.906422964647263,
      "learning_rate": 6.926340214301553e-06,
      "loss": 0.8176,
      "step": 8808
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.7829421598384885,
      "learning_rate": 6.925673586588483e-06,
      "loss": 0.8388,
      "step": 8809
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.777912231394991,
      "learning_rate": 6.925006918681195e-06,
      "loss": 0.824,
      "step": 8810
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.730755976975412,
      "learning_rate": 6.924340210593603e-06,
      "loss": 0.7625,
      "step": 8811
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.091987943009204,
      "learning_rate": 6.923673462339623e-06,
      "loss": 0.7113,
      "step": 8812
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.467338480497293,
      "learning_rate": 6.923006673933172e-06,
      "loss": 0.8136,
      "step": 8813
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.608862602545761,
      "learning_rate": 6.922339845388168e-06,
      "loss": 0.726,
      "step": 8814
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.22275630090508,
      "learning_rate": 6.921672976718531e-06,
      "loss": 0.8577,
      "step": 8815
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.633835086760427,
      "learning_rate": 6.9210060679381775e-06,
      "loss": 0.7464,
      "step": 8816
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.8231285378359425,
      "learning_rate": 6.92033911906103e-06,
      "loss": 0.7356,
      "step": 8817
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.1750410190413545,
      "learning_rate": 6.91967213010101e-06,
      "loss": 0.7983,
      "step": 8818
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.665783978184943,
      "learning_rate": 6.919005101072036e-06,
      "loss": 0.8561,
      "step": 8819
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.898461084643215,
      "learning_rate": 6.918338031988033e-06,
      "loss": 0.7535,
      "step": 8820
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.668587970354986,
      "learning_rate": 6.917670922862927e-06,
      "loss": 0.7117,
      "step": 8821
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.1654646425059365,
      "learning_rate": 6.917003773710638e-06,
      "loss": 0.8317,
      "step": 8822
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.649197230220908,
      "learning_rate": 6.9163365845450935e-06,
      "loss": 0.7817,
      "step": 8823
    },
    {
      "epoch": 0.79,
      "grad_norm": 11.817527077694836,
      "learning_rate": 6.915669355380221e-06,
      "loss": 0.7654,
      "step": 8824
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.358470673489723,
      "learning_rate": 6.915002086229945e-06,
      "loss": 0.7206,
      "step": 8825
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.664089814451144,
      "learning_rate": 6.914334777108195e-06,
      "loss": 0.7532,
      "step": 8826
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.540520861662685,
      "learning_rate": 6.913667428028899e-06,
      "loss": 0.8378,
      "step": 8827
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.2923948765217315,
      "learning_rate": 6.913000039005984e-06,
      "loss": 0.7165,
      "step": 8828
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.026892445475728,
      "learning_rate": 6.912332610053384e-06,
      "loss": 0.8451,
      "step": 8829
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.3736731870898184,
      "learning_rate": 6.911665141185029e-06,
      "loss": 0.7592,
      "step": 8830
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.614078356065484,
      "learning_rate": 6.910997632414851e-06,
      "loss": 0.8143,
      "step": 8831
    },
    {
      "epoch": 0.79,
      "grad_norm": 9.541582947547187,
      "learning_rate": 6.910330083756782e-06,
      "loss": 0.809,
      "step": 8832
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.316803259142336,
      "learning_rate": 6.909662495224755e-06,
      "loss": 0.762,
      "step": 8833
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.141846870131795,
      "learning_rate": 6.908994866832708e-06,
      "loss": 0.8004,
      "step": 8834
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.988982666523254,
      "learning_rate": 6.908327198594571e-06,
      "loss": 0.7429,
      "step": 8835
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.873681096149504,
      "learning_rate": 6.907659490524285e-06,
      "loss": 0.7773,
      "step": 8836
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.126116146002044,
      "learning_rate": 6.906991742635784e-06,
      "loss": 0.7854,
      "step": 8837
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.701297209385712,
      "learning_rate": 6.906323954943005e-06,
      "loss": 0.7458,
      "step": 8838
    },
    {
      "epoch": 0.79,
      "grad_norm": 10.645633836258698,
      "learning_rate": 6.905656127459891e-06,
      "loss": 0.7721,
      "step": 8839
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.670065991810353,
      "learning_rate": 6.9049882602003785e-06,
      "loss": 0.8292,
      "step": 8840
    },
    {
      "epoch": 0.79,
      "grad_norm": 10.413841629902224,
      "learning_rate": 6.904320353178405e-06,
      "loss": 0.7232,
      "step": 8841
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.733560025497825,
      "learning_rate": 6.903652406407917e-06,
      "loss": 0.7515,
      "step": 8842
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.7902116307298925,
      "learning_rate": 6.902984419902854e-06,
      "loss": 0.8874,
      "step": 8843
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.134001651217793,
      "learning_rate": 6.9023163936771576e-06,
      "loss": 0.7634,
      "step": 8844
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.787607320533095,
      "learning_rate": 6.901648327744772e-06,
      "loss": 0.8858,
      "step": 8845
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.240502814256087,
      "learning_rate": 6.900980222119644e-06,
      "loss": 0.7909,
      "step": 8846
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.727571605730573,
      "learning_rate": 6.9003120768157164e-06,
      "loss": 0.8182,
      "step": 8847
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.015008356436154,
      "learning_rate": 6.899643891846935e-06,
      "loss": 0.8081,
      "step": 8848
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.7859972740346075,
      "learning_rate": 6.898975667227248e-06,
      "loss": 0.823,
      "step": 8849
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.112589762501796,
      "learning_rate": 6.898307402970602e-06,
      "loss": 0.8174,
      "step": 8850
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.51806833146756,
      "learning_rate": 6.897639099090946e-06,
      "loss": 0.7351,
      "step": 8851
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.159084598276891,
      "learning_rate": 6.89697075560223e-06,
      "loss": 0.7442,
      "step": 8852
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.430648302734665,
      "learning_rate": 6.896302372518405e-06,
      "loss": 0.8117,
      "step": 8853
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.948527504877078,
      "learning_rate": 6.8956339498534195e-06,
      "loss": 0.8313,
      "step": 8854
    },
    {
      "epoch": 0.79,
      "grad_norm": 8.397941337666921,
      "learning_rate": 6.8949654876212266e-06,
      "loss": 0.809,
      "step": 8855
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.595203649862717,
      "learning_rate": 6.89429698583578e-06,
      "loss": 0.7442,
      "step": 8856
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.242395434669381,
      "learning_rate": 6.893628444511032e-06,
      "loss": 0.7872,
      "step": 8857
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.823324318802721,
      "learning_rate": 6.892959863660936e-06,
      "loss": 0.8197,
      "step": 8858
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.9555766658064595,
      "learning_rate": 6.89229124329945e-06,
      "loss": 0.8091,
      "step": 8859
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.209856740723686,
      "learning_rate": 6.891622583440528e-06,
      "loss": 0.7723,
      "step": 8860
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.731593551671059,
      "learning_rate": 6.890953884098126e-06,
      "loss": 0.8181,
      "step": 8861
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.090682803796809,
      "learning_rate": 6.890285145286204e-06,
      "loss": 0.7107,
      "step": 8862
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.3658584898223625,
      "learning_rate": 6.889616367018718e-06,
      "loss": 0.8744,
      "step": 8863
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.682068313771945,
      "learning_rate": 6.888947549309629e-06,
      "loss": 0.8495,
      "step": 8864
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.688367314210548,
      "learning_rate": 6.888278692172898e-06,
      "loss": 0.7528,
      "step": 8865
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.611292152284029,
      "learning_rate": 6.887609795622483e-06,
      "loss": 0.7842,
      "step": 8866
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.366402696231333,
      "learning_rate": 6.886940859672348e-06,
      "loss": 0.7769,
      "step": 8867
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.899610003297146,
      "learning_rate": 6.886271884336455e-06,
      "loss": 0.6996,
      "step": 8868
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.473031758982985,
      "learning_rate": 6.885602869628767e-06,
      "loss": 0.7763,
      "step": 8869
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.562399444836806,
      "learning_rate": 6.884933815563248e-06,
      "loss": 0.7129,
      "step": 8870
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.584762898108932,
      "learning_rate": 6.8842647221538636e-06,
      "loss": 0.6779,
      "step": 8871
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.381184035995556,
      "learning_rate": 6.88359558941458e-06,
      "loss": 0.7709,
      "step": 8872
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.193934288718516,
      "learning_rate": 6.882926417359363e-06,
      "loss": 0.7255,
      "step": 8873
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.302543589833199,
      "learning_rate": 6.882257206002182e-06,
      "loss": 0.7794,
      "step": 8874
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.414893970703186,
      "learning_rate": 6.881587955357004e-06,
      "loss": 0.7877,
      "step": 8875
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.961303311252845,
      "learning_rate": 6.8809186654377965e-06,
      "loss": 0.7433,
      "step": 8876
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.395797558105373,
      "learning_rate": 6.880249336258531e-06,
      "loss": 0.8405,
      "step": 8877
    },
    {
      "epoch": 0.79,
      "grad_norm": 10.028552049256819,
      "learning_rate": 6.879579967833179e-06,
      "loss": 0.8459,
      "step": 8878
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.35603644048776,
      "learning_rate": 6.878910560175711e-06,
      "loss": 0.7909,
      "step": 8879
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.043525622385374,
      "learning_rate": 6.8782411133000996e-06,
      "loss": 0.8457,
      "step": 8880
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.7872326607346745,
      "learning_rate": 6.877571627220319e-06,
      "loss": 0.7811,
      "step": 8881
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.339116775467267,
      "learning_rate": 6.876902101950342e-06,
      "loss": 0.7154,
      "step": 8882
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.974058998628582,
      "learning_rate": 6.876232537504144e-06,
      "loss": 0.7677,
      "step": 8883
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.307690896891668,
      "learning_rate": 6.875562933895701e-06,
      "loss": 0.8688,
      "step": 8884
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.386930169088093,
      "learning_rate": 6.874893291138989e-06,
      "loss": 0.6666,
      "step": 8885
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.902590431555418,
      "learning_rate": 6.874223609247986e-06,
      "loss": 0.8165,
      "step": 8886
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.688379942097481,
      "learning_rate": 6.873553888236668e-06,
      "loss": 0.8051,
      "step": 8887
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.73798120555801,
      "learning_rate": 6.8728841281190165e-06,
      "loss": 0.7984,
      "step": 8888
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.048348432750598,
      "learning_rate": 6.8722143289090115e-06,
      "loss": 0.7772,
      "step": 8889
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.791599070703299,
      "learning_rate": 6.871544490620631e-06,
      "loss": 0.7653,
      "step": 8890
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.405147110748098,
      "learning_rate": 6.870874613267858e-06,
      "loss": 0.773,
      "step": 8891
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.53296302045722,
      "learning_rate": 6.870204696864675e-06,
      "loss": 0.7126,
      "step": 8892
    },
    {
      "epoch": 0.79,
      "grad_norm": 13.013739820235273,
      "learning_rate": 6.869534741425064e-06,
      "loss": 0.7558,
      "step": 8893
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.670897671238176,
      "learning_rate": 6.868864746963011e-06,
      "loss": 0.7869,
      "step": 8894
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.644936375254622,
      "learning_rate": 6.8681947134924975e-06,
      "loss": 0.8094,
      "step": 8895
    },
    {
      "epoch": 0.79,
      "grad_norm": 20.66760019811634,
      "learning_rate": 6.867524641027512e-06,
      "loss": 0.7754,
      "step": 8896
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.138747392442604,
      "learning_rate": 6.866854529582039e-06,
      "loss": 0.7803,
      "step": 8897
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.3292578971826945,
      "learning_rate": 6.866184379170065e-06,
      "loss": 0.7842,
      "step": 8898
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.077089406738545,
      "learning_rate": 6.86551418980558e-06,
      "loss": 0.7857,
      "step": 8899
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.6169371722239285,
      "learning_rate": 6.864843961502572e-06,
      "loss": 0.8175,
      "step": 8900
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.419929075311483,
      "learning_rate": 6.864173694275029e-06,
      "loss": 0.7561,
      "step": 8901
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.395812639113104,
      "learning_rate": 6.863503388136943e-06,
      "loss": 0.7612,
      "step": 8902
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.790073210301413,
      "learning_rate": 6.862833043102305e-06,
      "loss": 0.8127,
      "step": 8903
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.843780880290494,
      "learning_rate": 6.8621626591851075e-06,
      "loss": 0.8348,
      "step": 8904
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.746742677137354,
      "learning_rate": 6.861492236399341e-06,
      "loss": 0.7421,
      "step": 8905
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.109303414387534,
      "learning_rate": 6.860821774759001e-06,
      "loss": 0.7962,
      "step": 8906
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.779669334769916,
      "learning_rate": 6.860151274278083e-06,
      "loss": 0.7701,
      "step": 8907
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.995411299322764,
      "learning_rate": 6.859480734970579e-06,
      "loss": 0.7368,
      "step": 8908
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.823442556594654,
      "learning_rate": 6.858810156850488e-06,
      "loss": 0.7846,
      "step": 8909
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.920303439107068,
      "learning_rate": 6.8581395399318065e-06,
      "loss": 0.8259,
      "step": 8910
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.134068283460694,
      "learning_rate": 6.8574688842285305e-06,
      "loss": 0.7834,
      "step": 8911
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.3324489682602225,
      "learning_rate": 6.856798189754658e-06,
      "loss": 0.6993,
      "step": 8912
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.0907137176613,
      "learning_rate": 6.856127456524192e-06,
      "loss": 0.7753,
      "step": 8913
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.151887232807324,
      "learning_rate": 6.855456684551129e-06,
      "loss": 0.8716,
      "step": 8914
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.524889361807719,
      "learning_rate": 6.85478587384947e-06,
      "loss": 0.8064,
      "step": 8915
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.791308791110216,
      "learning_rate": 6.854115024433219e-06,
      "loss": 0.7246,
      "step": 8916
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.251966476631001,
      "learning_rate": 6.853444136316377e-06,
      "loss": 0.8134,
      "step": 8917
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.9375206300566616,
      "learning_rate": 6.852773209512948e-06,
      "loss": 0.7485,
      "step": 8918
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.51650228553902,
      "learning_rate": 6.852102244036936e-06,
      "loss": 0.7385,
      "step": 8919
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.187012126624098,
      "learning_rate": 6.8514312399023456e-06,
      "loss": 0.7923,
      "step": 8920
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.081527449869237,
      "learning_rate": 6.8507601971231815e-06,
      "loss": 0.7753,
      "step": 8921
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.906807068896219,
      "learning_rate": 6.850089115713453e-06,
      "loss": 0.7828,
      "step": 8922
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.507539535443064,
      "learning_rate": 6.849417995687164e-06,
      "loss": 0.7662,
      "step": 8923
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.939277765104757,
      "learning_rate": 6.8487468370583256e-06,
      "loss": 0.8193,
      "step": 8924
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.349033175401351,
      "learning_rate": 6.848075639840945e-06,
      "loss": 0.7462,
      "step": 8925
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.400337614633899,
      "learning_rate": 6.847404404049034e-06,
      "loss": 0.7594,
      "step": 8926
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.934260008253583,
      "learning_rate": 6.846733129696601e-06,
      "loss": 0.807,
      "step": 8927
    },
    {
      "epoch": 0.8,
      "grad_norm": 10.872961336248737,
      "learning_rate": 6.846061816797658e-06,
      "loss": 0.7582,
      "step": 8928
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.163037942597244,
      "learning_rate": 6.845390465366217e-06,
      "loss": 0.754,
      "step": 8929
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.380709270853941,
      "learning_rate": 6.844719075416293e-06,
      "loss": 0.7578,
      "step": 8930
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.779486751349994,
      "learning_rate": 6.844047646961895e-06,
      "loss": 0.7115,
      "step": 8931
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.271453024753318,
      "learning_rate": 6.843376180017045e-06,
      "loss": 0.7239,
      "step": 8932
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.333861804517126,
      "learning_rate": 6.8427046745957525e-06,
      "loss": 0.738,
      "step": 8933
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.991637429665713,
      "learning_rate": 6.842033130712033e-06,
      "loss": 0.8104,
      "step": 8934
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.033951851124145,
      "learning_rate": 6.841361548379909e-06,
      "loss": 0.8347,
      "step": 8935
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.71164174866826,
      "learning_rate": 6.840689927613394e-06,
      "loss": 0.7626,
      "step": 8936
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.772787002906953,
      "learning_rate": 6.840018268426507e-06,
      "loss": 0.7536,
      "step": 8937
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.74288147652492,
      "learning_rate": 6.8393465708332696e-06,
      "loss": 0.7833,
      "step": 8938
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.881635084840276,
      "learning_rate": 6.8386748348477e-06,
      "loss": 0.7373,
      "step": 8939
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.357382489159556,
      "learning_rate": 6.83800306048382e-06,
      "loss": 0.7695,
      "step": 8940
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.031769554534384,
      "learning_rate": 6.837331247755651e-06,
      "loss": 0.7945,
      "step": 8941
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.788737115865319,
      "learning_rate": 6.836659396677216e-06,
      "loss": 0.7719,
      "step": 8942
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.245812328279025,
      "learning_rate": 6.835987507262538e-06,
      "loss": 0.7974,
      "step": 8943
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.498715404712324,
      "learning_rate": 6.835315579525642e-06,
      "loss": 0.8269,
      "step": 8944
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.427676297581858,
      "learning_rate": 6.834643613480553e-06,
      "loss": 0.7642,
      "step": 8945
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.352290356651545,
      "learning_rate": 6.833971609141294e-06,
      "loss": 0.8166,
      "step": 8946
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.967278826634626,
      "learning_rate": 6.8332995665218974e-06,
      "loss": 0.7049,
      "step": 8947
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.706706214341631,
      "learning_rate": 6.832627485636386e-06,
      "loss": 0.7647,
      "step": 8948
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.151264270034733,
      "learning_rate": 6.831955366498788e-06,
      "loss": 0.7187,
      "step": 8949
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.901451604733674,
      "learning_rate": 6.831283209123134e-06,
      "loss": 0.7887,
      "step": 8950
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.8069065238291735,
      "learning_rate": 6.830611013523455e-06,
      "loss": 0.7949,
      "step": 8951
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.777627024748749,
      "learning_rate": 6.829938779713778e-06,
      "loss": 0.7731,
      "step": 8952
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.2930136064236635,
      "learning_rate": 6.829266507708138e-06,
      "loss": 0.8153,
      "step": 8953
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.323762556715136,
      "learning_rate": 6.828594197520565e-06,
      "loss": 0.7372,
      "step": 8954
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.042257852614123,
      "learning_rate": 6.827921849165094e-06,
      "loss": 0.743,
      "step": 8955
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.532124582968514,
      "learning_rate": 6.827249462655755e-06,
      "loss": 0.8231,
      "step": 8956
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.7640365898871035,
      "learning_rate": 6.8265770380065865e-06,
      "loss": 0.788,
      "step": 8957
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.814610220235087,
      "learning_rate": 6.825904575231623e-06,
      "loss": 0.8216,
      "step": 8958
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.818041294787947,
      "learning_rate": 6.825232074344899e-06,
      "loss": 0.8016,
      "step": 8959
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.579175247459985,
      "learning_rate": 6.824559535360453e-06,
      "loss": 0.8041,
      "step": 8960
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.544542893134898,
      "learning_rate": 6.823886958292322e-06,
      "loss": 0.8165,
      "step": 8961
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.3734226503405225,
      "learning_rate": 6.8232143431545465e-06,
      "loss": 0.7454,
      "step": 8962
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.122217069429538,
      "learning_rate": 6.822541689961163e-06,
      "loss": 0.767,
      "step": 8963
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.27368178911124,
      "learning_rate": 6.821868998726213e-06,
      "loss": 0.7987,
      "step": 8964
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.323843161914525,
      "learning_rate": 6.821196269463738e-06,
      "loss": 0.7273,
      "step": 8965
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.454473026252531,
      "learning_rate": 6.820523502187779e-06,
      "loss": 0.7604,
      "step": 8966
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.6886932728790685,
      "learning_rate": 6.819850696912379e-06,
      "loss": 0.7297,
      "step": 8967
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.200467942075487,
      "learning_rate": 6.819177853651582e-06,
      "loss": 0.7838,
      "step": 8968
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.6732345759023435,
      "learning_rate": 6.818504972419429e-06,
      "loss": 0.8396,
      "step": 8969
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.948356621737164,
      "learning_rate": 6.817832053229969e-06,
      "loss": 0.8414,
      "step": 8970
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.1215228264217,
      "learning_rate": 6.817159096097246e-06,
      "loss": 0.8159,
      "step": 8971
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.600232289202041,
      "learning_rate": 6.8164861010353045e-06,
      "loss": 0.7897,
      "step": 8972
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.445965194437669,
      "learning_rate": 6.815813068058196e-06,
      "loss": 0.7861,
      "step": 8973
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.5400118123772355,
      "learning_rate": 6.8151399971799635e-06,
      "loss": 0.6926,
      "step": 8974
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.338548754085299,
      "learning_rate": 6.814466888414661e-06,
      "loss": 0.8646,
      "step": 8975
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.391516056292958,
      "learning_rate": 6.813793741776335e-06,
      "loss": 0.7706,
      "step": 8976
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.050186717071382,
      "learning_rate": 6.813120557279037e-06,
      "loss": 0.8605,
      "step": 8977
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.661275537208663,
      "learning_rate": 6.812447334936818e-06,
      "loss": 0.8271,
      "step": 8978
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.812292946319696,
      "learning_rate": 6.811774074763729e-06,
      "loss": 0.7575,
      "step": 8979
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.920845824356115,
      "learning_rate": 6.811100776773826e-06,
      "loss": 0.7731,
      "step": 8980
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.698305877414938,
      "learning_rate": 6.8104274409811586e-06,
      "loss": 0.8471,
      "step": 8981
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.91481553155522,
      "learning_rate": 6.809754067399783e-06,
      "loss": 0.789,
      "step": 8982
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.492634929410201,
      "learning_rate": 6.809080656043758e-06,
      "loss": 0.7458,
      "step": 8983
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.224994620398197,
      "learning_rate": 6.808407206927133e-06,
      "loss": 0.7859,
      "step": 8984
    },
    {
      "epoch": 0.8,
      "grad_norm": 9.829323056238552,
      "learning_rate": 6.807733720063967e-06,
      "loss": 0.8009,
      "step": 8985
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.395512871014527,
      "learning_rate": 6.807060195468321e-06,
      "loss": 0.8261,
      "step": 8986
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.226849611082985,
      "learning_rate": 6.80638663315425e-06,
      "loss": 0.783,
      "step": 8987
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.216187731707186,
      "learning_rate": 6.8057130331358125e-06,
      "loss": 0.7911,
      "step": 8988
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.814493244353561,
      "learning_rate": 6.805039395427071e-06,
      "loss": 0.8861,
      "step": 8989
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.7408557759747065,
      "learning_rate": 6.804365720042086e-06,
      "loss": 0.7942,
      "step": 8990
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.621859480269982,
      "learning_rate": 6.803692006994918e-06,
      "loss": 0.7776,
      "step": 8991
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.604723875696861,
      "learning_rate": 6.803018256299627e-06,
      "loss": 0.7836,
      "step": 8992
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.210415287070769,
      "learning_rate": 6.802344467970281e-06,
      "loss": 0.7352,
      "step": 8993
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.820463901836649,
      "learning_rate": 6.80167064202094e-06,
      "loss": 0.8188,
      "step": 8994
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.117751216031344,
      "learning_rate": 6.80099677846567e-06,
      "loss": 0.794,
      "step": 8995
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.036376109089421,
      "learning_rate": 6.800322877318537e-06,
      "loss": 0.7841,
      "step": 8996
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.6564262592407,
      "learning_rate": 6.799648938593606e-06,
      "loss": 0.7198,
      "step": 8997
    },
    {
      "epoch": 0.8,
      "grad_norm": 8.707786483463643,
      "learning_rate": 6.798974962304945e-06,
      "loss": 0.7589,
      "step": 8998
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.738571033737522,
      "learning_rate": 6.798300948466622e-06,
      "loss": 0.824,
      "step": 8999
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.094796114802285,
      "learning_rate": 6.797626897092704e-06,
      "loss": 0.7398,
      "step": 9000
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.788521713733064,
      "learning_rate": 6.79695280819726e-06,
      "loss": 0.8178,
      "step": 9001
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.280779490782635,
      "learning_rate": 6.796278681794364e-06,
      "loss": 0.7448,
      "step": 9002
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.172165665807148,
      "learning_rate": 6.795604517898082e-06,
      "loss": 0.77,
      "step": 9003
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.48492146257616,
      "learning_rate": 6.794930316522489e-06,
      "loss": 0.8206,
      "step": 9004
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.791921575871331,
      "learning_rate": 6.794256077681657e-06,
      "loss": 0.729,
      "step": 9005
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.690671746661968,
      "learning_rate": 6.793581801389658e-06,
      "loss": 0.7626,
      "step": 9006
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.8631866546735316,
      "learning_rate": 6.7929074876605675e-06,
      "loss": 0.8093,
      "step": 9007
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.868744491043089,
      "learning_rate": 6.792233136508459e-06,
      "loss": 0.7956,
      "step": 9008
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.324871311485903,
      "learning_rate": 6.791558747947409e-06,
      "loss": 0.7975,
      "step": 9009
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.301041022352684,
      "learning_rate": 6.7908843219914926e-06,
      "loss": 0.74,
      "step": 9010
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.337519238154856,
      "learning_rate": 6.790209858654789e-06,
      "loss": 0.858,
      "step": 9011
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.022151802886974,
      "learning_rate": 6.789535357951376e-06,
      "loss": 0.7667,
      "step": 9012
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.304019727707982,
      "learning_rate": 6.788860819895329e-06,
      "loss": 0.7888,
      "step": 9013
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.177946072194452,
      "learning_rate": 6.7881862445007305e-06,
      "loss": 0.7579,
      "step": 9014
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.132663641885758,
      "learning_rate": 6.7875116317816625e-06,
      "loss": 0.8391,
      "step": 9015
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.5107180970293586,
      "learning_rate": 6.786836981752201e-06,
      "loss": 0.7789,
      "step": 9016
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.586771278943711,
      "learning_rate": 6.786162294426431e-06,
      "loss": 0.7598,
      "step": 9017
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.152289000865504,
      "learning_rate": 6.785487569818436e-06,
      "loss": 0.74,
      "step": 9018
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.945654536699646,
      "learning_rate": 6.784812807942298e-06,
      "loss": 0.7692,
      "step": 9019
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.7083158575206525,
      "learning_rate": 6.784138008812101e-06,
      "loss": 0.7817,
      "step": 9020
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.300619429857944,
      "learning_rate": 6.783463172441931e-06,
      "loss": 0.8131,
      "step": 9021
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.727063648176615,
      "learning_rate": 6.782788298845872e-06,
      "loss": 0.7242,
      "step": 9022
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.903991124160926,
      "learning_rate": 6.782113388038011e-06,
      "loss": 0.7997,
      "step": 9023
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.532985294444383,
      "learning_rate": 6.781438440032438e-06,
      "loss": 0.7347,
      "step": 9024
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.448103758655382,
      "learning_rate": 6.780763454843236e-06,
      "loss": 0.824,
      "step": 9025
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.149875609784967,
      "learning_rate": 6.780088432484499e-06,
      "loss": 0.7501,
      "step": 9026
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.712536811692886,
      "learning_rate": 6.7794133729703115e-06,
      "loss": 0.7924,
      "step": 9027
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.968367709837906,
      "learning_rate": 6.778738276314769e-06,
      "loss": 0.7821,
      "step": 9028
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.348375415703808,
      "learning_rate": 6.7780631425319594e-06,
      "loss": 0.7691,
      "step": 9029
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.138032260380681,
      "learning_rate": 6.777387971635976e-06,
      "loss": 0.7618,
      "step": 9030
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.357861661550286,
      "learning_rate": 6.776712763640911e-06,
      "loss": 0.8459,
      "step": 9031
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.872447275163056,
      "learning_rate": 6.7760375185608586e-06,
      "loss": 0.7858,
      "step": 9032
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.995962011007614,
      "learning_rate": 6.7753622364099125e-06,
      "loss": 0.7358,
      "step": 9033
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.972094239405947,
      "learning_rate": 6.774686917202168e-06,
      "loss": 0.8198,
      "step": 9034
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.963086518202751,
      "learning_rate": 6.774011560951719e-06,
      "loss": 0.7381,
      "step": 9035
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.5637838252865,
      "learning_rate": 6.773336167672664e-06,
      "loss": 0.7284,
      "step": 9036
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.783214691661278,
      "learning_rate": 6.772660737379101e-06,
      "loss": 0.7685,
      "step": 9037
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.629247921909131,
      "learning_rate": 6.771985270085127e-06,
      "loss": 0.8087,
      "step": 9038
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.751093627423333,
      "learning_rate": 6.7713097658048385e-06,
      "loss": 0.7214,
      "step": 9039
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.244987664175362,
      "learning_rate": 6.770634224552341e-06,
      "loss": 0.7625,
      "step": 9040
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.920598199591266,
      "learning_rate": 6.769958646341733e-06,
      "loss": 0.7785,
      "step": 9041
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.1787303430185405,
      "learning_rate": 6.76928303118711e-06,
      "loss": 0.7179,
      "step": 9042
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.595860950818168,
      "learning_rate": 6.76860737910258e-06,
      "loss": 0.7805,
      "step": 9043
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.549026867023611,
      "learning_rate": 6.767931690102245e-06,
      "loss": 0.7547,
      "step": 9044
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.291792084506269,
      "learning_rate": 6.767255964200208e-06,
      "loss": 0.8019,
      "step": 9045
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.9059759855424465,
      "learning_rate": 6.766580201410572e-06,
      "loss": 0.7326,
      "step": 9046
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.109622057211347,
      "learning_rate": 6.765904401747443e-06,
      "loss": 0.774,
      "step": 9047
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.262291659582292,
      "learning_rate": 6.765228565224927e-06,
      "loss": 0.7447,
      "step": 9048
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.6380018170324115,
      "learning_rate": 6.764552691857131e-06,
      "loss": 0.7631,
      "step": 9049
    },
    {
      "epoch": 0.81,
      "grad_norm": 8.801002822451169,
      "learning_rate": 6.763876781658162e-06,
      "loss": 0.7989,
      "step": 9050
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.957229886900569,
      "learning_rate": 6.763200834642127e-06,
      "loss": 0.7605,
      "step": 9051
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.442124930835852,
      "learning_rate": 6.762524850823136e-06,
      "loss": 0.8174,
      "step": 9052
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.797481182071632,
      "learning_rate": 6.761848830215299e-06,
      "loss": 0.6879,
      "step": 9053
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.960131133290143,
      "learning_rate": 6.761172772832726e-06,
      "loss": 0.6961,
      "step": 9054
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.344293623655568,
      "learning_rate": 6.760496678689529e-06,
      "loss": 0.8731,
      "step": 9055
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.438291056581272,
      "learning_rate": 6.7598205477998195e-06,
      "loss": 0.7751,
      "step": 9056
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.533819513816837,
      "learning_rate": 6.759144380177708e-06,
      "loss": 0.7651,
      "step": 9057
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.9954378801150625,
      "learning_rate": 6.758468175837312e-06,
      "loss": 0.8857,
      "step": 9058
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.928905215288331,
      "learning_rate": 6.757791934792742e-06,
      "loss": 0.7626,
      "step": 9059
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.935427648879806,
      "learning_rate": 6.757115657058115e-06,
      "loss": 0.7234,
      "step": 9060
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.756285790191452,
      "learning_rate": 6.756439342647547e-06,
      "loss": 0.7324,
      "step": 9061
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.058768209600252,
      "learning_rate": 6.755762991575156e-06,
      "loss": 0.8128,
      "step": 9062
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.9806948488850775,
      "learning_rate": 6.755086603855057e-06,
      "loss": 0.7021,
      "step": 9063
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.125107251314303,
      "learning_rate": 6.754410179501367e-06,
      "loss": 0.8024,
      "step": 9064
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.980247183133425,
      "learning_rate": 6.753733718528208e-06,
      "loss": 0.8717,
      "step": 9065
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.4993379040274215,
      "learning_rate": 6.7530572209496984e-06,
      "loss": 0.7787,
      "step": 9066
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.082129497557983,
      "learning_rate": 6.752380686779958e-06,
      "loss": 0.7309,
      "step": 9067
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.9101174428366035,
      "learning_rate": 6.751704116033107e-06,
      "loss": 0.8154,
      "step": 9068
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.729077176347446,
      "learning_rate": 6.751027508723273e-06,
      "loss": 0.7544,
      "step": 9069
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.359909761582978,
      "learning_rate": 6.75035086486457e-06,
      "loss": 0.7983,
      "step": 9070
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.02936114910174,
      "learning_rate": 6.749674184471127e-06,
      "loss": 0.7381,
      "step": 9071
    },
    {
      "epoch": 0.81,
      "grad_norm": 9.374414053173028,
      "learning_rate": 6.748997467557068e-06,
      "loss": 0.7983,
      "step": 9072
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.490177724461621,
      "learning_rate": 6.748320714136516e-06,
      "loss": 0.8459,
      "step": 9073
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.178101908622659,
      "learning_rate": 6.747643924223599e-06,
      "loss": 0.8487,
      "step": 9074
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.062747464687736,
      "learning_rate": 6.746967097832442e-06,
      "loss": 0.7756,
      "step": 9075
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.699292418490548,
      "learning_rate": 6.746290234977172e-06,
      "loss": 0.7426,
      "step": 9076
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.203508252600028,
      "learning_rate": 6.745613335671919e-06,
      "loss": 0.7614,
      "step": 9077
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.218007974553774,
      "learning_rate": 6.744936399930808e-06,
      "loss": 0.7441,
      "step": 9078
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.897196098271374,
      "learning_rate": 6.744259427767974e-06,
      "loss": 0.7367,
      "step": 9079
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.631907894283829,
      "learning_rate": 6.743582419197543e-06,
      "loss": 0.7467,
      "step": 9080
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.445908822946712,
      "learning_rate": 6.742905374233646e-06,
      "loss": 0.7044,
      "step": 9081
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.057748582546555,
      "learning_rate": 6.742228292890418e-06,
      "loss": 0.7452,
      "step": 9082
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.458978352523111,
      "learning_rate": 6.74155117518199e-06,
      "loss": 0.7642,
      "step": 9083
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.962435922308668,
      "learning_rate": 6.740874021122494e-06,
      "loss": 0.7467,
      "step": 9084
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.840280922931275,
      "learning_rate": 6.7401968307260665e-06,
      "loss": 0.7412,
      "step": 9085
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.913396395277208,
      "learning_rate": 6.73951960400684e-06,
      "loss": 0.7788,
      "step": 9086
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.355900084606462,
      "learning_rate": 6.73884234097895e-06,
      "loss": 0.8072,
      "step": 9087
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.089418294402775,
      "learning_rate": 6.738165041656537e-06,
      "loss": 0.7935,
      "step": 9088
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.732213344814303,
      "learning_rate": 6.737487706053734e-06,
      "loss": 0.7977,
      "step": 9089
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.413970945933723,
      "learning_rate": 6.7368103341846795e-06,
      "loss": 0.768,
      "step": 9090
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.797082456670281,
      "learning_rate": 6.736132926063515e-06,
      "loss": 0.8431,
      "step": 9091
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.704713167315286,
      "learning_rate": 6.735455481704376e-06,
      "loss": 0.8042,
      "step": 9092
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.422378628106721,
      "learning_rate": 6.734778001121404e-06,
      "loss": 0.7406,
      "step": 9093
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.40349602728623,
      "learning_rate": 6.734100484328742e-06,
      "loss": 0.7788,
      "step": 9094
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.1377100530034445,
      "learning_rate": 6.733422931340528e-06,
      "loss": 0.776,
      "step": 9095
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.676639968708884,
      "learning_rate": 6.732745342170907e-06,
      "loss": 0.7294,
      "step": 9096
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.654546719789466,
      "learning_rate": 6.732067716834022e-06,
      "loss": 0.819,
      "step": 9097
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.82698152153159,
      "learning_rate": 6.731390055344018e-06,
      "loss": 0.7198,
      "step": 9098
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.9320144581148675,
      "learning_rate": 6.730712357715036e-06,
      "loss": 0.7552,
      "step": 9099
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.310956504574239,
      "learning_rate": 6.730034623961224e-06,
      "loss": 0.7767,
      "step": 9100
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.2993293852292584,
      "learning_rate": 6.729356854096729e-06,
      "loss": 0.772,
      "step": 9101
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.584913906610851,
      "learning_rate": 6.728679048135695e-06,
      "loss": 0.7429,
      "step": 9102
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.956494763984893,
      "learning_rate": 6.728001206092274e-06,
      "loss": 0.7821,
      "step": 9103
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.179743581810423,
      "learning_rate": 6.727323327980611e-06,
      "loss": 0.7167,
      "step": 9104
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.381994388276381,
      "learning_rate": 6.726645413814857e-06,
      "loss": 0.7494,
      "step": 9105
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.552847230149235,
      "learning_rate": 6.7259674636091606e-06,
      "loss": 0.7253,
      "step": 9106
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.019923454622278,
      "learning_rate": 6.725289477377675e-06,
      "loss": 0.8318,
      "step": 9107
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.554346855450403,
      "learning_rate": 6.724611455134547e-06,
      "loss": 0.7855,
      "step": 9108
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.892025837408739,
      "learning_rate": 6.723933396893932e-06,
      "loss": 0.746,
      "step": 9109
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.557491534199239,
      "learning_rate": 6.723255302669984e-06,
      "loss": 0.7375,
      "step": 9110
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.66373075473867,
      "learning_rate": 6.722577172476854e-06,
      "loss": 0.7693,
      "step": 9111
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.452442523353855,
      "learning_rate": 6.7218990063286996e-06,
      "loss": 0.8076,
      "step": 9112
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.8692719062231005,
      "learning_rate": 6.7212208042396745e-06,
      "loss": 0.724,
      "step": 9113
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.838215398036215,
      "learning_rate": 6.720542566223933e-06,
      "loss": 0.731,
      "step": 9114
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.796920545872942,
      "learning_rate": 6.7198642922956334e-06,
      "loss": 0.7356,
      "step": 9115
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.40226511278919,
      "learning_rate": 6.7191859824689345e-06,
      "loss": 0.7502,
      "step": 9116
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.7377975352727826,
      "learning_rate": 6.718507636757992e-06,
      "loss": 0.7706,
      "step": 9117
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.041443667044097,
      "learning_rate": 6.717829255176967e-06,
      "loss": 0.7095,
      "step": 9118
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.00743581573742,
      "learning_rate": 6.717150837740016e-06,
      "loss": 0.8115,
      "step": 9119
    },
    {
      "epoch": 0.81,
      "grad_norm": 8.499398604149865,
      "learning_rate": 6.716472384461306e-06,
      "loss": 0.7675,
      "step": 9120
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.057859442556996,
      "learning_rate": 6.71579389535499e-06,
      "loss": 0.8166,
      "step": 9121
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.840720890919522,
      "learning_rate": 6.715115370435236e-06,
      "loss": 0.813,
      "step": 9122
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.982257857465485,
      "learning_rate": 6.714436809716205e-06,
      "loss": 0.7604,
      "step": 9123
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.012551785073297,
      "learning_rate": 6.713758213212058e-06,
      "loss": 0.738,
      "step": 9124
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.924849023216129,
      "learning_rate": 6.7130795809369635e-06,
      "loss": 0.7821,
      "step": 9125
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.723099173898512,
      "learning_rate": 6.712400912905083e-06,
      "loss": 0.7518,
      "step": 9126
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.505998817158916,
      "learning_rate": 6.7117222091305846e-06,
      "loss": 0.8052,
      "step": 9127
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.127821701793819,
      "learning_rate": 6.711043469627633e-06,
      "loss": 0.7726,
      "step": 9128
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.518616696234241,
      "learning_rate": 6.710364694410396e-06,
      "loss": 0.7576,
      "step": 9129
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.7084337505403635,
      "learning_rate": 6.709685883493043e-06,
      "loss": 0.7415,
      "step": 9130
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.142218252330915,
      "learning_rate": 6.709007036889739e-06,
      "loss": 0.7803,
      "step": 9131
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.271720348328558,
      "learning_rate": 6.708328154614657e-06,
      "loss": 0.8149,
      "step": 9132
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.430171974276038,
      "learning_rate": 6.707649236681966e-06,
      "loss": 0.7979,
      "step": 9133
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.286023982013857,
      "learning_rate": 6.706970283105837e-06,
      "loss": 0.9138,
      "step": 9134
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.070866696522087,
      "learning_rate": 6.706291293900441e-06,
      "loss": 0.8188,
      "step": 9135
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.50575275280628,
      "learning_rate": 6.705612269079951e-06,
      "loss": 0.7231,
      "step": 9136
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.079711250521131,
      "learning_rate": 6.70493320865854e-06,
      "loss": 0.8098,
      "step": 9137
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.30638111095326,
      "learning_rate": 6.704254112650382e-06,
      "loss": 0.8653,
      "step": 9138
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.065920183485765,
      "learning_rate": 6.703574981069653e-06,
      "loss": 0.7989,
      "step": 9139
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.474011792552387,
      "learning_rate": 6.702895813930526e-06,
      "loss": 0.8721,
      "step": 9140
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.311899803455266,
      "learning_rate": 6.702216611247176e-06,
      "loss": 0.7301,
      "step": 9141
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.840893244326878,
      "learning_rate": 6.701537373033787e-06,
      "loss": 0.7566,
      "step": 9142
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.483266517957952,
      "learning_rate": 6.700858099304528e-06,
      "loss": 0.7957,
      "step": 9143
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.735192443882309,
      "learning_rate": 6.700178790073582e-06,
      "loss": 0.711,
      "step": 9144
    },
    {
      "epoch": 0.82,
      "grad_norm": 8.18337706786702,
      "learning_rate": 6.6994994453551274e-06,
      "loss": 0.8281,
      "step": 9145
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.923255960125858,
      "learning_rate": 6.698820065163343e-06,
      "loss": 0.7113,
      "step": 9146
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.907377275548001,
      "learning_rate": 6.69814064951241e-06,
      "loss": 0.7929,
      "step": 9147
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.90011308191444,
      "learning_rate": 6.6974611984165106e-06,
      "loss": 0.7288,
      "step": 9148
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.450382091149885,
      "learning_rate": 6.696781711889826e-06,
      "loss": 0.8296,
      "step": 9149
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.702912950847071,
      "learning_rate": 6.6961021899465385e-06,
      "loss": 0.7379,
      "step": 9150
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.34026374896719,
      "learning_rate": 6.695422632600832e-06,
      "loss": 0.6886,
      "step": 9151
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.411821829548468,
      "learning_rate": 6.694743039866891e-06,
      "loss": 0.8001,
      "step": 9152
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.265831106793937,
      "learning_rate": 6.694063411758901e-06,
      "loss": 0.7724,
      "step": 9153
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.630785396505117,
      "learning_rate": 6.6933837482910465e-06,
      "loss": 0.8042,
      "step": 9154
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.682882160925468,
      "learning_rate": 6.692704049477517e-06,
      "loss": 0.8075,
      "step": 9155
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.522978061173116,
      "learning_rate": 6.692024315332495e-06,
      "loss": 0.7901,
      "step": 9156
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.589824414535788,
      "learning_rate": 6.691344545870171e-06,
      "loss": 0.6906,
      "step": 9157
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.3963834554433046,
      "learning_rate": 6.690664741104736e-06,
      "loss": 0.7899,
      "step": 9158
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.04078981648403,
      "learning_rate": 6.6899849010503736e-06,
      "loss": 0.8428,
      "step": 9159
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.528349520005722,
      "learning_rate": 6.689305025721278e-06,
      "loss": 0.7501,
      "step": 9160
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.233638511071927,
      "learning_rate": 6.688625115131642e-06,
      "loss": 0.8235,
      "step": 9161
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.426153092618917,
      "learning_rate": 6.687945169295652e-06,
      "loss": 0.7069,
      "step": 9162
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.989908301336699,
      "learning_rate": 6.687265188227505e-06,
      "loss": 0.7495,
      "step": 9163
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.350896262188914,
      "learning_rate": 6.686585171941391e-06,
      "loss": 0.7377,
      "step": 9164
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.041948832057526,
      "learning_rate": 6.6859051204515056e-06,
      "loss": 0.7589,
      "step": 9165
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.670240678654193,
      "learning_rate": 6.685225033772042e-06,
      "loss": 0.7588,
      "step": 9166
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.589324936404713,
      "learning_rate": 6.684544911917199e-06,
      "loss": 0.7927,
      "step": 9167
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.124572799578937,
      "learning_rate": 6.683864754901168e-06,
      "loss": 0.7399,
      "step": 9168
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.605939273795954,
      "learning_rate": 6.683184562738147e-06,
      "loss": 0.7542,
      "step": 9169
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.236992036135941,
      "learning_rate": 6.682504335442337e-06,
      "loss": 0.7582,
      "step": 9170
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.2175149990144885,
      "learning_rate": 6.681824073027932e-06,
      "loss": 0.7505,
      "step": 9171
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.068630173613845,
      "learning_rate": 6.681143775509133e-06,
      "loss": 0.6977,
      "step": 9172
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.723855312064117,
      "learning_rate": 6.680463442900139e-06,
      "loss": 0.7783,
      "step": 9173
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.525590744171022,
      "learning_rate": 6.679783075215152e-06,
      "loss": 0.772,
      "step": 9174
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.712894486850788,
      "learning_rate": 6.679102672468369e-06,
      "loss": 0.7274,
      "step": 9175
    },
    {
      "epoch": 0.82,
      "grad_norm": 9.429183091489785,
      "learning_rate": 6.678422234673997e-06,
      "loss": 0.8055,
      "step": 9176
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.526893658980028,
      "learning_rate": 6.677741761846237e-06,
      "loss": 0.7863,
      "step": 9177
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.390006884137587,
      "learning_rate": 6.677061253999292e-06,
      "loss": 0.726,
      "step": 9178
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.799464412573038,
      "learning_rate": 6.676380711147364e-06,
      "loss": 0.8181,
      "step": 9179
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.927791013478489,
      "learning_rate": 6.675700133304662e-06,
      "loss": 0.8459,
      "step": 9180
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.595248942027759,
      "learning_rate": 6.675019520485388e-06,
      "loss": 0.7017,
      "step": 9181
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.5331004595809175,
      "learning_rate": 6.67433887270375e-06,
      "loss": 0.7471,
      "step": 9182
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.195866366210726,
      "learning_rate": 6.673658189973956e-06,
      "loss": 0.8198,
      "step": 9183
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.57307111183648,
      "learning_rate": 6.672977472310211e-06,
      "loss": 0.796,
      "step": 9184
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.958181446148806,
      "learning_rate": 6.6722967197267264e-06,
      "loss": 0.7821,
      "step": 9185
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.5746210909198455,
      "learning_rate": 6.671615932237708e-06,
      "loss": 0.7209,
      "step": 9186
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.015818896899873,
      "learning_rate": 6.67093510985737e-06,
      "loss": 0.8394,
      "step": 9187
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.935359750186638,
      "learning_rate": 6.6702542525999195e-06,
      "loss": 0.8084,
      "step": 9188
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.45763842725462,
      "learning_rate": 6.6695733604795684e-06,
      "loss": 0.7586,
      "step": 9189
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.156111708579572,
      "learning_rate": 6.668892433510532e-06,
      "loss": 0.8521,
      "step": 9190
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.796532449832661,
      "learning_rate": 6.66821147170702e-06,
      "loss": 0.7319,
      "step": 9191
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.511409845679262,
      "learning_rate": 6.667530475083247e-06,
      "loss": 0.8368,
      "step": 9192
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.141351822992895,
      "learning_rate": 6.666849443653427e-06,
      "loss": 0.7949,
      "step": 9193
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.617332008594722,
      "learning_rate": 6.666168377431776e-06,
      "loss": 0.7719,
      "step": 9194
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.6021662106766685,
      "learning_rate": 6.665487276432508e-06,
      "loss": 0.7921,
      "step": 9195
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.801139592040418,
      "learning_rate": 6.6648061406698395e-06,
      "loss": 0.8607,
      "step": 9196
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.195669710214925,
      "learning_rate": 6.66412497015799e-06,
      "loss": 0.7524,
      "step": 9197
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.822795667118691,
      "learning_rate": 6.663443764911175e-06,
      "loss": 0.7351,
      "step": 9198
    },
    {
      "epoch": 0.82,
      "grad_norm": 10.278779803199567,
      "learning_rate": 6.662762524943616e-06,
      "loss": 0.8917,
      "step": 9199
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.6289287513526025,
      "learning_rate": 6.66208125026953e-06,
      "loss": 0.7691,
      "step": 9200
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.219945086872416,
      "learning_rate": 6.6613999409031375e-06,
      "loss": 0.7636,
      "step": 9201
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.667169234508395,
      "learning_rate": 6.660718596858658e-06,
      "loss": 0.6813,
      "step": 9202
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.43393113116461,
      "learning_rate": 6.660037218150318e-06,
      "loss": 0.7676,
      "step": 9203
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.3479330224371875,
      "learning_rate": 6.6593558047923344e-06,
      "loss": 0.7496,
      "step": 9204
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.19315678162661,
      "learning_rate": 6.658674356798933e-06,
      "loss": 0.7622,
      "step": 9205
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.424329275290144,
      "learning_rate": 6.657992874184338e-06,
      "loss": 0.7872,
      "step": 9206
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.228312366368319,
      "learning_rate": 6.65731135696277e-06,
      "loss": 0.6999,
      "step": 9207
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.191926610008593,
      "learning_rate": 6.656629805148458e-06,
      "loss": 0.8397,
      "step": 9208
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.986232531620126,
      "learning_rate": 6.655948218755628e-06,
      "loss": 0.7973,
      "step": 9209
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.253408289901351,
      "learning_rate": 6.655266597798503e-06,
      "loss": 0.7334,
      "step": 9210
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.754746528021622,
      "learning_rate": 6.6545849422913145e-06,
      "loss": 0.7558,
      "step": 9211
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.1281475446936,
      "learning_rate": 6.6539032522482885e-06,
      "loss": 0.7732,
      "step": 9212
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.872874654863938,
      "learning_rate": 6.653221527683653e-06,
      "loss": 0.8353,
      "step": 9213
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.28655531581945,
      "learning_rate": 6.652539768611641e-06,
      "loss": 0.7297,
      "step": 9214
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.487427450203813,
      "learning_rate": 6.65185797504648e-06,
      "loss": 0.8404,
      "step": 9215
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.795706657007864,
      "learning_rate": 6.6511761470023996e-06,
      "loss": 0.769,
      "step": 9216
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.013014968625061,
      "learning_rate": 6.650494284493632e-06,
      "loss": 0.8346,
      "step": 9217
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.83971565902096,
      "learning_rate": 6.649812387534413e-06,
      "loss": 0.8,
      "step": 9218
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.04021493146678,
      "learning_rate": 6.6491304561389735e-06,
      "loss": 0.7875,
      "step": 9219
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.689449883124337,
      "learning_rate": 6.6484484903215465e-06,
      "loss": 0.8489,
      "step": 9220
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.797897028246807,
      "learning_rate": 6.647766490096368e-06,
      "loss": 0.7434,
      "step": 9221
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.023938330465407,
      "learning_rate": 6.647084455477673e-06,
      "loss": 0.7498,
      "step": 9222
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.166201376431791,
      "learning_rate": 6.6464023864796955e-06,
      "loss": 0.8184,
      "step": 9223
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.460568075953246,
      "learning_rate": 6.6457202831166734e-06,
      "loss": 0.7509,
      "step": 9224
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.247281875559364,
      "learning_rate": 6.645038145402847e-06,
      "loss": 0.7195,
      "step": 9225
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.6946803479857655,
      "learning_rate": 6.6443559733524484e-06,
      "loss": 0.7703,
      "step": 9226
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.893133654477158,
      "learning_rate": 6.643673766979721e-06,
      "loss": 0.7261,
      "step": 9227
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.819059854209314,
      "learning_rate": 6.642991526298907e-06,
      "loss": 0.7477,
      "step": 9228
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.314341487705419,
      "learning_rate": 6.64230925132424e-06,
      "loss": 0.783,
      "step": 9229
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.251618588622973,
      "learning_rate": 6.641626942069963e-06,
      "loss": 0.8815,
      "step": 9230
    },
    {
      "epoch": 0.82,
      "grad_norm": 8.019923728707433,
      "learning_rate": 6.64094459855032e-06,
      "loss": 0.8572,
      "step": 9231
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.675704132000863,
      "learning_rate": 6.640262220779552e-06,
      "loss": 0.8783,
      "step": 9232
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.627118667226743,
      "learning_rate": 6.639579808771901e-06,
      "loss": 0.8131,
      "step": 9233
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.781423011233196,
      "learning_rate": 6.6388973625416145e-06,
      "loss": 0.7774,
      "step": 9234
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.507598205656157,
      "learning_rate": 6.638214882102934e-06,
      "loss": 0.795,
      "step": 9235
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.549660281876527,
      "learning_rate": 6.637532367470104e-06,
      "loss": 0.8327,
      "step": 9236
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.181392326393004,
      "learning_rate": 6.636849818657373e-06,
      "loss": 0.8219,
      "step": 9237
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.053139779380874,
      "learning_rate": 6.6361672356789875e-06,
      "loss": 0.8178,
      "step": 9238
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.135555366803391,
      "learning_rate": 6.635484618549192e-06,
      "loss": 0.7056,
      "step": 9239
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.221347302960785,
      "learning_rate": 6.634801967282237e-06,
      "loss": 0.7841,
      "step": 9240
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.992837058641838,
      "learning_rate": 6.634119281892373e-06,
      "loss": 0.7767,
      "step": 9241
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.107546644312385,
      "learning_rate": 6.633436562393847e-06,
      "loss": 0.6797,
      "step": 9242
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.129832495415342,
      "learning_rate": 6.632753808800909e-06,
      "loss": 0.7883,
      "step": 9243
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.753666243037954,
      "learning_rate": 6.632071021127812e-06,
      "loss": 0.8234,
      "step": 9244
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.184328860541277,
      "learning_rate": 6.6313881993888066e-06,
      "loss": 0.7687,
      "step": 9245
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.837297877645717,
      "learning_rate": 6.630705343598144e-06,
      "loss": 0.7344,
      "step": 9246
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.528570368649965,
      "learning_rate": 6.63002245377008e-06,
      "loss": 0.7605,
      "step": 9247
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.373297466537811,
      "learning_rate": 6.6293395299188655e-06,
      "loss": 0.8562,
      "step": 9248
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.996260641964184,
      "learning_rate": 6.628656572058756e-06,
      "loss": 0.7499,
      "step": 9249
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.915657819380187,
      "learning_rate": 6.627973580204009e-06,
      "loss": 0.785,
      "step": 9250
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.968558225714612,
      "learning_rate": 6.627290554368879e-06,
      "loss": 0.7726,
      "step": 9251
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.029586113267461,
      "learning_rate": 6.626607494567621e-06,
      "loss": 0.7751,
      "step": 9252
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.470269862485086,
      "learning_rate": 6.625924400814495e-06,
      "loss": 0.7333,
      "step": 9253
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.079174350154124,
      "learning_rate": 6.625241273123757e-06,
      "loss": 0.7962,
      "step": 9254
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.2146186660828455,
      "learning_rate": 6.624558111509666e-06,
      "loss": 0.7464,
      "step": 9255
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.452822448698582,
      "learning_rate": 6.623874915986483e-06,
      "loss": 0.6866,
      "step": 9256
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.2479928081596,
      "learning_rate": 6.6231916865684685e-06,
      "loss": 0.7357,
      "step": 9257
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.211056826898012,
      "learning_rate": 6.62250842326988e-06,
      "loss": 0.8206,
      "step": 9258
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.457961012287179,
      "learning_rate": 6.621825126104983e-06,
      "loss": 0.8005,
      "step": 9259
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.912316596448745,
      "learning_rate": 6.621141795088037e-06,
      "loss": 0.7906,
      "step": 9260
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.342943523756849,
      "learning_rate": 6.620458430233307e-06,
      "loss": 0.8242,
      "step": 9261
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.370707624856441,
      "learning_rate": 6.619775031555056e-06,
      "loss": 0.7302,
      "step": 9262
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.144529284392769,
      "learning_rate": 6.619091599067549e-06,
      "loss": 0.8391,
      "step": 9263
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.7878541840736375,
      "learning_rate": 6.61840813278505e-06,
      "loss": 0.8514,
      "step": 9264
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.32004645889888,
      "learning_rate": 6.617724632721826e-06,
      "loss": 0.7752,
      "step": 9265
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.793122439130676,
      "learning_rate": 6.617041098892143e-06,
      "loss": 0.809,
      "step": 9266
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.5453441624994415,
      "learning_rate": 6.6163575313102666e-06,
      "loss": 0.7685,
      "step": 9267
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.69452457319781,
      "learning_rate": 6.6156739299904685e-06,
      "loss": 0.7395,
      "step": 9268
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.775445369297455,
      "learning_rate": 6.614990294947015e-06,
      "loss": 0.8155,
      "step": 9269
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.150895192936199,
      "learning_rate": 6.614306626194174e-06,
      "loss": 0.7413,
      "step": 9270
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.4181600956107605,
      "learning_rate": 6.61362292374622e-06,
      "loss": 0.8399,
      "step": 9271
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.0559154556494725,
      "learning_rate": 6.61293918761742e-06,
      "loss": 0.7604,
      "step": 9272
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.202471875614134,
      "learning_rate": 6.612255417822048e-06,
      "loss": 0.6871,
      "step": 9273
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.81408871772347,
      "learning_rate": 6.6115716143743726e-06,
      "loss": 0.7425,
      "step": 9274
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.789683027741783,
      "learning_rate": 6.61088777728867e-06,
      "loss": 0.7296,
      "step": 9275
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.364767618110003,
      "learning_rate": 6.610203906579213e-06,
      "loss": 0.8089,
      "step": 9276
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.880332672665388,
      "learning_rate": 6.609520002260277e-06,
      "loss": 0.8343,
      "step": 9277
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.906409923630362,
      "learning_rate": 6.608836064346135e-06,
      "loss": 0.7148,
      "step": 9278
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.812606070995352,
      "learning_rate": 6.608152092851063e-06,
      "loss": 0.8045,
      "step": 9279
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.290504129671277,
      "learning_rate": 6.607468087789338e-06,
      "loss": 0.8716,
      "step": 9280
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.211277178257234,
      "learning_rate": 6.606784049175237e-06,
      "loss": 0.6418,
      "step": 9281
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.716675392183513,
      "learning_rate": 6.606099977023037e-06,
      "loss": 0.8144,
      "step": 9282
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.714086465459584,
      "learning_rate": 6.605415871347018e-06,
      "loss": 0.7539,
      "step": 9283
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.833905534933337,
      "learning_rate": 6.604731732161458e-06,
      "loss": 0.8442,
      "step": 9284
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.376479273078114,
      "learning_rate": 6.604047559480639e-06,
      "loss": 0.8273,
      "step": 9285
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.220177948884824,
      "learning_rate": 6.603363353318839e-06,
      "loss": 0.7677,
      "step": 9286
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.528253187878936,
      "learning_rate": 6.60267911369034e-06,
      "loss": 0.7917,
      "step": 9287
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.7657088491680293,
      "learning_rate": 6.601994840609424e-06,
      "loss": 0.7602,
      "step": 9288
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.025348767678168,
      "learning_rate": 6.601310534090375e-06,
      "loss": 0.8643,
      "step": 9289
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.46474454884221,
      "learning_rate": 6.600626194147473e-06,
      "loss": 0.8021,
      "step": 9290
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.3419994414191585,
      "learning_rate": 6.599941820795008e-06,
      "loss": 0.7729,
      "step": 9291
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.923456966932649,
      "learning_rate": 6.59925741404726e-06,
      "loss": 0.7715,
      "step": 9292
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.734376007055663,
      "learning_rate": 6.598572973918516e-06,
      "loss": 0.7669,
      "step": 9293
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.187776980489579,
      "learning_rate": 6.59788850042306e-06,
      "loss": 0.8017,
      "step": 9294
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.314042967289795,
      "learning_rate": 6.597203993575183e-06,
      "loss": 0.7914,
      "step": 9295
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.144752312899453,
      "learning_rate": 6.59651945338917e-06,
      "loss": 0.8347,
      "step": 9296
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.7487264447185655,
      "learning_rate": 6.595834879879308e-06,
      "loss": 0.7473,
      "step": 9297
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.120026428640556,
      "learning_rate": 6.595150273059891e-06,
      "loss": 0.7516,
      "step": 9298
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.877638539082457,
      "learning_rate": 6.594465632945203e-06,
      "loss": 0.846,
      "step": 9299
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.5149391174032125,
      "learning_rate": 6.5937809595495376e-06,
      "loss": 0.8248,
      "step": 9300
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.254889248262842,
      "learning_rate": 6.593096252887186e-06,
      "loss": 0.6985,
      "step": 9301
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.946199252511024,
      "learning_rate": 6.592411512972436e-06,
      "loss": 0.7505,
      "step": 9302
    },
    {
      "epoch": 0.83,
      "grad_norm": 9.412465498012743,
      "learning_rate": 6.591726739819584e-06,
      "loss": 0.7772,
      "step": 9303
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.796554786525381,
      "learning_rate": 6.591041933442923e-06,
      "loss": 0.7768,
      "step": 9304
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.602723256971588,
      "learning_rate": 6.5903570938567464e-06,
      "loss": 0.7416,
      "step": 9305
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.098240338197,
      "learning_rate": 6.5896722210753475e-06,
      "loss": 0.8607,
      "step": 9306
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.740825197467731,
      "learning_rate": 6.588987315113023e-06,
      "loss": 0.8308,
      "step": 9307
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.66663838349238,
      "learning_rate": 6.588302375984067e-06,
      "loss": 0.7655,
      "step": 9308
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.60389603558631,
      "learning_rate": 6.587617403702777e-06,
      "loss": 0.8069,
      "step": 9309
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.08797233208417,
      "learning_rate": 6.586932398283452e-06,
      "loss": 0.7804,
      "step": 9310
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.099687459340384,
      "learning_rate": 6.586247359740387e-06,
      "loss": 0.7242,
      "step": 9311
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.479021003152057,
      "learning_rate": 6.585562288087884e-06,
      "loss": 0.7919,
      "step": 9312
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.859486004557756,
      "learning_rate": 6.584877183340238e-06,
      "loss": 0.8154,
      "step": 9313
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.476655648834488,
      "learning_rate": 6.584192045511755e-06,
      "loss": 0.7812,
      "step": 9314
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.325719671832365,
      "learning_rate": 6.58350687461673e-06,
      "loss": 0.7672,
      "step": 9315
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.499036442972537,
      "learning_rate": 6.582821670669467e-06,
      "loss": 0.8119,
      "step": 9316
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.046120769808518,
      "learning_rate": 6.58213643368427e-06,
      "loss": 0.8324,
      "step": 9317
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.4561092148943935,
      "learning_rate": 6.581451163675436e-06,
      "loss": 0.7093,
      "step": 9318
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.240683508485538,
      "learning_rate": 6.580765860657275e-06,
      "loss": 0.7479,
      "step": 9319
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.699579831547706,
      "learning_rate": 6.580080524644088e-06,
      "loss": 0.8072,
      "step": 9320
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.254794584097842,
      "learning_rate": 6.579395155650179e-06,
      "loss": 0.699,
      "step": 9321
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.20031573859095,
      "learning_rate": 6.578709753689857e-06,
      "loss": 0.7101,
      "step": 9322
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.5996288862491035,
      "learning_rate": 6.578024318777425e-06,
      "loss": 0.8205,
      "step": 9323
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.826077833052689,
      "learning_rate": 6.5773388509271926e-06,
      "loss": 0.7139,
      "step": 9324
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.535618711399195,
      "learning_rate": 6.576653350153463e-06,
      "loss": 0.7817,
      "step": 9325
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.347141038170473,
      "learning_rate": 6.57596781647055e-06,
      "loss": 0.7377,
      "step": 9326
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.5458820337352215,
      "learning_rate": 6.575282249892761e-06,
      "loss": 0.8087,
      "step": 9327
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.0411609367458174,
      "learning_rate": 6.574596650434402e-06,
      "loss": 0.8362,
      "step": 9328
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.787303473566768,
      "learning_rate": 6.573911018109789e-06,
      "loss": 0.7725,
      "step": 9329
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.5674753752555475,
      "learning_rate": 6.573225352933229e-06,
      "loss": 0.8364,
      "step": 9330
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.302168181009234,
      "learning_rate": 6.572539654919034e-06,
      "loss": 0.7477,
      "step": 9331
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.307333187435429,
      "learning_rate": 6.571853924081518e-06,
      "loss": 0.7542,
      "step": 9332
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.2774890246825645,
      "learning_rate": 6.571168160434995e-06,
      "loss": 0.8702,
      "step": 9333
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.102749680764938,
      "learning_rate": 6.570482363993777e-06,
      "loss": 0.7629,
      "step": 9334
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.393026536173051,
      "learning_rate": 6.569796534772177e-06,
      "loss": 0.722,
      "step": 9335
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.723600092369484,
      "learning_rate": 6.569110672784515e-06,
      "loss": 0.7441,
      "step": 9336
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.192978020884205,
      "learning_rate": 6.568424778045102e-06,
      "loss": 0.7619,
      "step": 9337
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.210185148054354,
      "learning_rate": 6.567738850568257e-06,
      "loss": 0.7036,
      "step": 9338
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.796429547702919,
      "learning_rate": 6.567052890368299e-06,
      "loss": 0.817,
      "step": 9339
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.398107835952964,
      "learning_rate": 6.566366897459541e-06,
      "loss": 0.7668,
      "step": 9340
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.012061984595038,
      "learning_rate": 6.565680871856304e-06,
      "loss": 0.8193,
      "step": 9341
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.143235548265437,
      "learning_rate": 6.564994813572909e-06,
      "loss": 0.7825,
      "step": 9342
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.620892845628608,
      "learning_rate": 6.564308722623675e-06,
      "loss": 0.8344,
      "step": 9343
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.241028274455972,
      "learning_rate": 6.56362259902292e-06,
      "loss": 0.7142,
      "step": 9344
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.625125090218979,
      "learning_rate": 6.5629364427849684e-06,
      "loss": 0.7598,
      "step": 9345
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.8539986274109035,
      "learning_rate": 6.562250253924143e-06,
      "loss": 0.7442,
      "step": 9346
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.074766112144884,
      "learning_rate": 6.5615640324547635e-06,
      "loss": 0.7802,
      "step": 9347
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.464471279938751,
      "learning_rate": 6.560877778391154e-06,
      "loss": 0.71,
      "step": 9348
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.1676974237056665,
      "learning_rate": 6.5601914917476405e-06,
      "loss": 0.72,
      "step": 9349
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.338109431074726,
      "learning_rate": 6.559505172538545e-06,
      "loss": 0.7845,
      "step": 9350
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.69799995446669,
      "learning_rate": 6.558818820778195e-06,
      "loss": 0.7059,
      "step": 9351
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.823735261747067,
      "learning_rate": 6.558132436480918e-06,
      "loss": 0.7604,
      "step": 9352
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.165204624555681,
      "learning_rate": 6.557446019661036e-06,
      "loss": 0.8237,
      "step": 9353
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.668712238592543,
      "learning_rate": 6.556759570332878e-06,
      "loss": 0.8277,
      "step": 9354
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.5360945942898905,
      "learning_rate": 6.556073088510776e-06,
      "loss": 0.8131,
      "step": 9355
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.235499108088945,
      "learning_rate": 6.555386574209054e-06,
      "loss": 0.8363,
      "step": 9356
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.1336195701134075,
      "learning_rate": 6.554700027442044e-06,
      "loss": 0.764,
      "step": 9357
    },
    {
      "epoch": 0.83,
      "grad_norm": 9.387252824520331,
      "learning_rate": 6.5540134482240766e-06,
      "loss": 0.7796,
      "step": 9358
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.202915048337988,
      "learning_rate": 6.553326836569481e-06,
      "loss": 0.7527,
      "step": 9359
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.918427239961853,
      "learning_rate": 6.55264019249259e-06,
      "loss": 0.7519,
      "step": 9360
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.9670865152616,
      "learning_rate": 6.551953516007734e-06,
      "loss": 0.7769,
      "step": 9361
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.3582982647282495,
      "learning_rate": 6.551266807129249e-06,
      "loss": 0.7535,
      "step": 9362
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.784733914049712,
      "learning_rate": 6.5505800658714635e-06,
      "loss": 0.785,
      "step": 9363
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.866126456100069,
      "learning_rate": 6.549893292248717e-06,
      "loss": 0.7668,
      "step": 9364
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.7093507920902455,
      "learning_rate": 6.549206486275343e-06,
      "loss": 0.8078,
      "step": 9365
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.392679800846632,
      "learning_rate": 6.548519647965675e-06,
      "loss": 0.9365,
      "step": 9366
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.424821099396832,
      "learning_rate": 6.547832777334051e-06,
      "loss": 0.7181,
      "step": 9367
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.103780868493945,
      "learning_rate": 6.547145874394807e-06,
      "loss": 0.8768,
      "step": 9368
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.860755170616341,
      "learning_rate": 6.5464589391622815e-06,
      "loss": 0.7193,
      "step": 9369
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.099389586548273,
      "learning_rate": 6.545771971650812e-06,
      "loss": 0.7725,
      "step": 9370
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.698505409116462,
      "learning_rate": 6.545084971874738e-06,
      "loss": 0.8299,
      "step": 9371
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.149108424180016,
      "learning_rate": 6.544397939848399e-06,
      "loss": 0.7978,
      "step": 9372
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.382091344588842,
      "learning_rate": 6.543710875586134e-06,
      "loss": 0.8239,
      "step": 9373
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.057556909934731,
      "learning_rate": 6.5430237791022875e-06,
      "loss": 0.7429,
      "step": 9374
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.506799028951691,
      "learning_rate": 6.542336650411197e-06,
      "loss": 0.7948,
      "step": 9375
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.625574976807894,
      "learning_rate": 6.5416494895272065e-06,
      "loss": 0.8271,
      "step": 9376
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.448150484599392,
      "learning_rate": 6.54096229646466e-06,
      "loss": 0.7868,
      "step": 9377
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.3058796499741945,
      "learning_rate": 6.540275071237899e-06,
      "loss": 0.696,
      "step": 9378
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.604166138188836,
      "learning_rate": 6.539587813861271e-06,
      "loss": 0.7196,
      "step": 9379
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.768993685714996,
      "learning_rate": 6.538900524349117e-06,
      "loss": 0.8068,
      "step": 9380
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.30224679324688,
      "learning_rate": 6.538213202715787e-06,
      "loss": 0.6955,
      "step": 9381
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.562395721617138,
      "learning_rate": 6.537525848975623e-06,
      "loss": 0.767,
      "step": 9382
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.980475884469279,
      "learning_rate": 6.536838463142973e-06,
      "loss": 0.6983,
      "step": 9383
    },
    {
      "epoch": 0.84,
      "grad_norm": 7.1649780130294145,
      "learning_rate": 6.536151045232188e-06,
      "loss": 0.6985,
      "step": 9384
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.765909171535991,
      "learning_rate": 6.535463595257614e-06,
      "loss": 0.7363,
      "step": 9385
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.726795864361377,
      "learning_rate": 6.534776113233599e-06,
      "loss": 0.7618,
      "step": 9386
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.988626692830045,
      "learning_rate": 6.534088599174495e-06,
      "loss": 0.7534,
      "step": 9387
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.183295830765261,
      "learning_rate": 6.53340105309465e-06,
      "loss": 0.8292,
      "step": 9388
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.32899631133422,
      "learning_rate": 6.532713475008416e-06,
      "loss": 0.7988,
      "step": 9389
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.62868168421306,
      "learning_rate": 6.532025864930145e-06,
      "loss": 0.8586,
      "step": 9390
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.282997199850087,
      "learning_rate": 6.531338222874189e-06,
      "loss": 0.7982,
      "step": 9391
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.061076536348874,
      "learning_rate": 6.530650548854901e-06,
      "loss": 0.8003,
      "step": 9392
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.237239668460433,
      "learning_rate": 6.529962842886637e-06,
      "loss": 0.7582,
      "step": 9393
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.741380066204617,
      "learning_rate": 6.529275104983748e-06,
      "loss": 0.7669,
      "step": 9394
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.938825262478454,
      "learning_rate": 6.528587335160589e-06,
      "loss": 0.7561,
      "step": 9395
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.751444779612163,
      "learning_rate": 6.527899533431516e-06,
      "loss": 0.7518,
      "step": 9396
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.736820732310771,
      "learning_rate": 6.527211699810889e-06,
      "loss": 0.7923,
      "step": 9397
    },
    {
      "epoch": 0.84,
      "grad_norm": 7.474309280000098,
      "learning_rate": 6.526523834313061e-06,
      "loss": 0.7761,
      "step": 9398
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.220839179871017,
      "learning_rate": 6.525835936952391e-06,
      "loss": 0.7618,
      "step": 9399
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.931245666645234,
      "learning_rate": 6.525148007743238e-06,
      "loss": 0.825,
      "step": 9400
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.897219521044759,
      "learning_rate": 6.524460046699961e-06,
      "loss": 0.7739,
      "step": 9401
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.373075572151494,
      "learning_rate": 6.523772053836918e-06,
      "loss": 0.7382,
      "step": 9402
    },
    {
      "epoch": 0.84,
      "grad_norm": 7.041556308792199,
      "learning_rate": 6.5230840291684705e-06,
      "loss": 0.7266,
      "step": 9403
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.7045779837449375,
      "learning_rate": 6.522395972708978e-06,
      "loss": 0.7802,
      "step": 9404
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.6407704344506895,
      "learning_rate": 6.521707884472805e-06,
      "loss": 0.8273,
      "step": 9405
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.889276153940064,
      "learning_rate": 6.521019764474312e-06,
      "loss": 0.8027,
      "step": 9406
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.678790910575011,
      "learning_rate": 6.520331612727863e-06,
      "loss": 0.8067,
      "step": 9407
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.273262913229003,
      "learning_rate": 6.519643429247821e-06,
      "loss": 0.7727,
      "step": 9408
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.979193303493956,
      "learning_rate": 6.51895521404855e-06,
      "loss": 0.8126,
      "step": 9409
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.044569055114587,
      "learning_rate": 6.518266967144416e-06,
      "loss": 0.7986,
      "step": 9410
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.774187163894605,
      "learning_rate": 6.517578688549783e-06,
      "loss": 0.7885,
      "step": 9411
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.9336235379802025,
      "learning_rate": 6.516890378279019e-06,
      "loss": 0.7662,
      "step": 9412
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.303508484621717,
      "learning_rate": 6.5162020363464896e-06,
      "loss": 0.7068,
      "step": 9413
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.603597362731738,
      "learning_rate": 6.515513662766563e-06,
      "loss": 0.8061,
      "step": 9414
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.9762408523613075,
      "learning_rate": 6.514825257553608e-06,
      "loss": 0.7211,
      "step": 9415
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.5628098732941025,
      "learning_rate": 6.514136820721995e-06,
      "loss": 0.7499,
      "step": 9416
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.189217828087101,
      "learning_rate": 6.513448352286089e-06,
      "loss": 0.7373,
      "step": 9417
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.627368002657835,
      "learning_rate": 6.512759852260264e-06,
      "loss": 0.725,
      "step": 9418
    },
    {
      "epoch": 0.84,
      "grad_norm": 7.012489715560954,
      "learning_rate": 6.51207132065889e-06,
      "loss": 0.7112,
      "step": 9419
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.454838362730212,
      "learning_rate": 6.5113827574963385e-06,
      "loss": 0.8101,
      "step": 9420
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.2168377924351335,
      "learning_rate": 6.510694162786982e-06,
      "loss": 0.7265,
      "step": 9421
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.693202391128046,
      "learning_rate": 6.510005536545193e-06,
      "loss": 0.7997,
      "step": 9422
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.595378103239235,
      "learning_rate": 6.509316878785345e-06,
      "loss": 0.8162,
      "step": 9423
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.239281820173799,
      "learning_rate": 6.508628189521814e-06,
      "loss": 0.7297,
      "step": 9424
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.452615824145354,
      "learning_rate": 6.5079394687689725e-06,
      "loss": 0.8155,
      "step": 9425
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.897663840133408,
      "learning_rate": 6.5072507165411955e-06,
      "loss": 0.8329,
      "step": 9426
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.910108177528046,
      "learning_rate": 6.5065619328528615e-06,
      "loss": 0.7672,
      "step": 9427
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.203014868969037,
      "learning_rate": 6.505873117718349e-06,
      "loss": 0.7812,
      "step": 9428
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.558775238792903,
      "learning_rate": 6.50518427115203e-06,
      "loss": 0.811,
      "step": 9429
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.478090989184489,
      "learning_rate": 6.504495393168287e-06,
      "loss": 0.7497,
      "step": 9430
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.296602875538748,
      "learning_rate": 6.5038064837814965e-06,
      "loss": 0.7583,
      "step": 9431
    },
    {
      "epoch": 0.84,
      "grad_norm": 7.309645838031168,
      "learning_rate": 6.503117543006039e-06,
      "loss": 0.7294,
      "step": 9432
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.659402894943269,
      "learning_rate": 6.502428570856295e-06,
      "loss": 0.7706,
      "step": 9433
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.936603196584094,
      "learning_rate": 6.501739567346644e-06,
      "loss": 0.9055,
      "step": 9434
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.284664429628865,
      "learning_rate": 6.50105053249147e-06,
      "loss": 0.8351,
      "step": 9435
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.260013905503574,
      "learning_rate": 6.5003614663051515e-06,
      "loss": 0.6766,
      "step": 9436
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.0709590724946665,
      "learning_rate": 6.4996723688020745e-06,
      "loss": 0.7964,
      "step": 9437
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.2697306878779475,
      "learning_rate": 6.498983239996621e-06,
      "loss": 0.7703,
      "step": 9438
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.645803437456238,
      "learning_rate": 6.498294079903175e-06,
      "loss": 0.7786,
      "step": 9439
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.842480487030007,
      "learning_rate": 6.4976048885361195e-06,
      "loss": 0.6554,
      "step": 9440
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.959448800868754,
      "learning_rate": 6.496915665909845e-06,
      "loss": 0.7559,
      "step": 9441
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.619694559534239,
      "learning_rate": 6.496226412038731e-06,
      "loss": 0.7706,
      "step": 9442
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.499554689360852,
      "learning_rate": 6.495537126937168e-06,
      "loss": 0.839,
      "step": 9443
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.780274543954469,
      "learning_rate": 6.494847810619544e-06,
      "loss": 0.794,
      "step": 9444
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.242326466764863,
      "learning_rate": 6.494158463100245e-06,
      "loss": 0.7019,
      "step": 9445
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.862876868982146,
      "learning_rate": 6.49346908439366e-06,
      "loss": 0.7714,
      "step": 9446
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.494067671312142,
      "learning_rate": 6.492779674514178e-06,
      "loss": 0.7424,
      "step": 9447
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.082272362418472,
      "learning_rate": 6.492090233476191e-06,
      "loss": 0.8138,
      "step": 9448
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.98451963811429,
      "learning_rate": 6.491400761294086e-06,
      "loss": 0.7302,
      "step": 9449
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.9084431710916085,
      "learning_rate": 6.490711257982258e-06,
      "loss": 0.7329,
      "step": 9450
    },
    {
      "epoch": 0.84,
      "grad_norm": 8.232235224728774,
      "learning_rate": 6.490021723555095e-06,
      "loss": 0.7817,
      "step": 9451
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.102128028204994,
      "learning_rate": 6.489332158026994e-06,
      "loss": 0.8103,
      "step": 9452
    },
    {
      "epoch": 0.84,
      "grad_norm": 7.54179087787043,
      "learning_rate": 6.488642561412344e-06,
      "loss": 0.8335,
      "step": 9453
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.889359196317132,
      "learning_rate": 6.487952933725542e-06,
      "loss": 0.8135,
      "step": 9454
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.821643885785941,
      "learning_rate": 6.48726327498098e-06,
      "loss": 0.8007,
      "step": 9455
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.114789199067065,
      "learning_rate": 6.486573585193054e-06,
      "loss": 0.7188,
      "step": 9456
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.071268844526726,
      "learning_rate": 6.485883864376161e-06,
      "loss": 0.8337,
      "step": 9457
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.684197524411109,
      "learning_rate": 6.485194112544696e-06,
      "loss": 0.778,
      "step": 9458
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.43065998690409,
      "learning_rate": 6.484504329713057e-06,
      "loss": 0.7983,
      "step": 9459
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.199418890114732,
      "learning_rate": 6.4838145158956415e-06,
      "loss": 0.7717,
      "step": 9460
    },
    {
      "epoch": 0.84,
      "grad_norm": 7.274225217919038,
      "learning_rate": 6.483124671106845e-06,
      "loss": 0.8048,
      "step": 9461
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.754643860973987,
      "learning_rate": 6.48243479536107e-06,
      "loss": 0.7138,
      "step": 9462
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.698038873216083,
      "learning_rate": 6.4817448886727164e-06,
      "loss": 0.8015,
      "step": 9463
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.77493837204741,
      "learning_rate": 6.481054951056181e-06,
      "loss": 0.7363,
      "step": 9464
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.789577075158253,
      "learning_rate": 6.480364982525869e-06,
      "loss": 0.7795,
      "step": 9465
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.930475254000997,
      "learning_rate": 6.479674983096179e-06,
      "loss": 0.7337,
      "step": 9466
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.510164872189907,
      "learning_rate": 6.4789849527815156e-06,
      "loss": 0.7667,
      "step": 9467
    },
    {
      "epoch": 0.84,
      "grad_norm": 7.084035067866466,
      "learning_rate": 6.478294891596279e-06,
      "loss": 0.8784,
      "step": 9468
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.401443727030793,
      "learning_rate": 6.477604799554873e-06,
      "loss": 0.7713,
      "step": 9469
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.429276089639123,
      "learning_rate": 6.476914676671704e-06,
      "loss": 0.8047,
      "step": 9470
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.0953449004326865,
      "learning_rate": 6.476224522961175e-06,
      "loss": 0.7899,
      "step": 9471
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.1193668121232045,
      "learning_rate": 6.4755343384376915e-06,
      "loss": 0.8291,
      "step": 9472
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.8491094297630895,
      "learning_rate": 6.474844123115662e-06,
      "loss": 0.7993,
      "step": 9473
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.382945887061991,
      "learning_rate": 6.474153877009489e-06,
      "loss": 0.8617,
      "step": 9474
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.3604486240660005,
      "learning_rate": 6.473463600133583e-06,
      "loss": 0.7681,
      "step": 9475
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.706659629893536,
      "learning_rate": 6.4727732925023526e-06,
      "loss": 0.768,
      "step": 9476
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.4809384329189985,
      "learning_rate": 6.472082954130203e-06,
      "loss": 0.8078,
      "step": 9477
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.725671252496022,
      "learning_rate": 6.471392585031545e-06,
      "loss": 0.8189,
      "step": 9478
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.044534122313022,
      "learning_rate": 6.470702185220792e-06,
      "loss": 0.7752,
      "step": 9479
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.062055885616479,
      "learning_rate": 6.47001175471235e-06,
      "loss": 0.786,
      "step": 9480
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.0646247179523245,
      "learning_rate": 6.4693212935206314e-06,
      "loss": 0.8016,
      "step": 9481
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.7348589853236955,
      "learning_rate": 6.468630801660048e-06,
      "loss": 0.7555,
      "step": 9482
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.861196209037025,
      "learning_rate": 6.467940279145015e-06,
      "loss": 0.7878,
      "step": 9483
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.817401010710393,
      "learning_rate": 6.467249725989941e-06,
      "loss": 0.76,
      "step": 9484
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.608166257131713,
      "learning_rate": 6.4665591422092445e-06,
      "loss": 0.8293,
      "step": 9485
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.096770724347656,
      "learning_rate": 6.465868527817337e-06,
      "loss": 0.7122,
      "step": 9486
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.639885021912108,
      "learning_rate": 6.465177882828632e-06,
      "loss": 0.7508,
      "step": 9487
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.628082305577703,
      "learning_rate": 6.46448720725755e-06,
      "loss": 0.7956,
      "step": 9488
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.579388018772583,
      "learning_rate": 6.463796501118506e-06,
      "loss": 0.7705,
      "step": 9489
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.020931787607997,
      "learning_rate": 6.463105764425912e-06,
      "loss": 0.7713,
      "step": 9490
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.038920757437064,
      "learning_rate": 6.462414997194191e-06,
      "loss": 0.7071,
      "step": 9491
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.744117498739114,
      "learning_rate": 6.46172419943776e-06,
      "loss": 0.803,
      "step": 9492
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.2790810665014405,
      "learning_rate": 6.461033371171037e-06,
      "loss": 0.8514,
      "step": 9493
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.529085062590118,
      "learning_rate": 6.460342512408442e-06,
      "loss": 0.8022,
      "step": 9494
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.537193277723638,
      "learning_rate": 6.459651623164394e-06,
      "loss": 0.7808,
      "step": 9495
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.1996721020317205,
      "learning_rate": 6.458960703453317e-06,
      "loss": 0.7426,
      "step": 9496
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.751659537916116,
      "learning_rate": 6.45826975328963e-06,
      "loss": 0.7583,
      "step": 9497
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.826849393663395,
      "learning_rate": 6.457578772687755e-06,
      "loss": 0.7606,
      "step": 9498
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.067813193084385,
      "learning_rate": 6.456887761662114e-06,
      "loss": 0.7334,
      "step": 9499
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.654639858166285,
      "learning_rate": 6.456196720227133e-06,
      "loss": 0.7529,
      "step": 9500
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.466501323227361,
      "learning_rate": 6.455505648397234e-06,
      "loss": 0.7689,
      "step": 9501
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.837091725243561,
      "learning_rate": 6.454814546186842e-06,
      "loss": 0.7422,
      "step": 9502
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.2314110244705025,
      "learning_rate": 6.454123413610383e-06,
      "loss": 0.774,
      "step": 9503
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.759604749124786,
      "learning_rate": 6.4534322506822814e-06,
      "loss": 0.7369,
      "step": 9504
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.405381900443038,
      "learning_rate": 6.452741057416965e-06,
      "loss": 0.7284,
      "step": 9505
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.4367036372457225,
      "learning_rate": 6.45204983382886e-06,
      "loss": 0.7382,
      "step": 9506
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.542157677493962,
      "learning_rate": 6.451358579932394e-06,
      "loss": 0.8405,
      "step": 9507
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.973240321821131,
      "learning_rate": 6.450667295741998e-06,
      "loss": 0.7874,
      "step": 9508
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.016150237838678,
      "learning_rate": 6.449975981272098e-06,
      "loss": 0.7691,
      "step": 9509
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.331024528250204,
      "learning_rate": 6.449284636537126e-06,
      "loss": 0.8072,
      "step": 9510
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.240646167379919,
      "learning_rate": 6.44859326155151e-06,
      "loss": 0.7033,
      "step": 9511
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.637427449026751,
      "learning_rate": 6.447901856329681e-06,
      "loss": 0.7685,
      "step": 9512
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.721048347316337,
      "learning_rate": 6.447210420886072e-06,
      "loss": 0.7902,
      "step": 9513
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.969690827665251,
      "learning_rate": 6.446518955235117e-06,
      "loss": 0.7581,
      "step": 9514
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.256913585650462,
      "learning_rate": 6.445827459391244e-06,
      "loss": 0.7854,
      "step": 9515
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.4876556749167715,
      "learning_rate": 6.4451359333688915e-06,
      "loss": 0.772,
      "step": 9516
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.33707408536969,
      "learning_rate": 6.444444377182488e-06,
      "loss": 0.7986,
      "step": 9517
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.731884410582803,
      "learning_rate": 6.443752790846475e-06,
      "loss": 0.7464,
      "step": 9518
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.912347865598169,
      "learning_rate": 6.4430611743752815e-06,
      "loss": 0.7782,
      "step": 9519
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.784455606447906,
      "learning_rate": 6.442369527783347e-06,
      "loss": 0.7287,
      "step": 9520
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.721981437503124,
      "learning_rate": 6.441677851085109e-06,
      "loss": 0.7469,
      "step": 9521
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.252140621174557,
      "learning_rate": 6.440986144295001e-06,
      "loss": 0.7972,
      "step": 9522
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.306176448046879,
      "learning_rate": 6.440294407427462e-06,
      "loss": 0.7982,
      "step": 9523
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.425681974727794,
      "learning_rate": 6.439602640496935e-06,
      "loss": 0.7702,
      "step": 9524
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.110392129159989,
      "learning_rate": 6.438910843517854e-06,
      "loss": 0.7755,
      "step": 9525
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.941812063145456,
      "learning_rate": 6.438219016504659e-06,
      "loss": 0.7008,
      "step": 9526
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.498904023881269,
      "learning_rate": 6.437527159471793e-06,
      "loss": 0.7006,
      "step": 9527
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.778417742680642,
      "learning_rate": 6.436835272433694e-06,
      "loss": 0.7732,
      "step": 9528
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.995832543564532,
      "learning_rate": 6.436143355404807e-06,
      "loss": 0.6732,
      "step": 9529
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.91037264830442,
      "learning_rate": 6.435451408399572e-06,
      "loss": 0.7684,
      "step": 9530
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.1141333054641205,
      "learning_rate": 6.434759431432434e-06,
      "loss": 0.755,
      "step": 9531
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.132325941619388,
      "learning_rate": 6.434067424517832e-06,
      "loss": 0.7182,
      "step": 9532
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.903674209103317,
      "learning_rate": 6.433375387670215e-06,
      "loss": 0.7784,
      "step": 9533
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.406649848598607,
      "learning_rate": 6.4326833209040255e-06,
      "loss": 0.6866,
      "step": 9534
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.742680650769454,
      "learning_rate": 6.431991224233709e-06,
      "loss": 0.7314,
      "step": 9535
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.499342225988305,
      "learning_rate": 6.431299097673712e-06,
      "loss": 0.815,
      "step": 9536
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.996583204365729,
      "learning_rate": 6.43060694123848e-06,
      "loss": 0.7459,
      "step": 9537
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.542806627885849,
      "learning_rate": 6.429914754942462e-06,
      "loss": 0.735,
      "step": 9538
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.106008251487081,
      "learning_rate": 6.4292225388001035e-06,
      "loss": 0.656,
      "step": 9539
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.698595446220986,
      "learning_rate": 6.428530292825856e-06,
      "loss": 0.7344,
      "step": 9540
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.458587189233385,
      "learning_rate": 6.427838017034167e-06,
      "loss": 0.6875,
      "step": 9541
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.374955447314323,
      "learning_rate": 6.427145711439485e-06,
      "loss": 0.8394,
      "step": 9542
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.038252284986005,
      "learning_rate": 6.426453376056264e-06,
      "loss": 0.7893,
      "step": 9543
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.525123537903482,
      "learning_rate": 6.425761010898951e-06,
      "loss": 0.7484,
      "step": 9544
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.377710573467817,
      "learning_rate": 6.425068615982e-06,
      "loss": 0.8033,
      "step": 9545
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.501009216616729,
      "learning_rate": 6.424376191319862e-06,
      "loss": 0.7719,
      "step": 9546
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.52391824500763,
      "learning_rate": 6.42368373692699e-06,
      "loss": 0.8396,
      "step": 9547
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.002609928414761,
      "learning_rate": 6.4229912528178385e-06,
      "loss": 0.7063,
      "step": 9548
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.617952095215382,
      "learning_rate": 6.42229873900686e-06,
      "loss": 0.8053,
      "step": 9549
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.236710405032114,
      "learning_rate": 6.4216061955085105e-06,
      "loss": 0.7575,
      "step": 9550
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.429146937323263,
      "learning_rate": 6.420913622337245e-06,
      "loss": 0.7842,
      "step": 9551
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.205821386155081,
      "learning_rate": 6.42022101950752e-06,
      "loss": 0.6841,
      "step": 9552
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.180702946289573,
      "learning_rate": 6.419528387033791e-06,
      "loss": 0.7964,
      "step": 9553
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.730448785339682,
      "learning_rate": 6.4188357249305155e-06,
      "loss": 0.8772,
      "step": 9554
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.56921917396458,
      "learning_rate": 6.4181430332121506e-06,
      "loss": 0.7916,
      "step": 9555
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.098898727370793,
      "learning_rate": 6.417450311893157e-06,
      "loss": 0.7349,
      "step": 9556
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.033889672998069,
      "learning_rate": 6.416757560987991e-06,
      "loss": 0.7626,
      "step": 9557
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.605419462432562,
      "learning_rate": 6.416064780511114e-06,
      "loss": 0.7203,
      "step": 9558
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.895344987713936,
      "learning_rate": 6.4153719704769865e-06,
      "loss": 0.8078,
      "step": 9559
    },
    {
      "epoch": 0.85,
      "grad_norm": 9.735344369692337,
      "learning_rate": 6.41467913090007e-06,
      "loss": 0.7712,
      "step": 9560
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.132816373187437,
      "learning_rate": 6.413986261794823e-06,
      "loss": 0.7255,
      "step": 9561
    },
    {
      "epoch": 0.85,
      "grad_norm": 8.176762313962836,
      "learning_rate": 6.41329336317571e-06,
      "loss": 0.7797,
      "step": 9562
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.553582987551326,
      "learning_rate": 6.412600435057191e-06,
      "loss": 0.7595,
      "step": 9563
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.157066524220565,
      "learning_rate": 6.411907477453734e-06,
      "loss": 0.7705,
      "step": 9564
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.701113316142415,
      "learning_rate": 6.411214490379799e-06,
      "loss": 0.7673,
      "step": 9565
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.599217784664563,
      "learning_rate": 6.410521473849852e-06,
      "loss": 0.7874,
      "step": 9566
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.570601781727519,
      "learning_rate": 6.4098284278783595e-06,
      "loss": 0.8087,
      "step": 9567
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.01071703376564,
      "learning_rate": 6.409135352479785e-06,
      "loss": 0.7366,
      "step": 9568
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.5568491721604465,
      "learning_rate": 6.408442247668597e-06,
      "loss": 0.7421,
      "step": 9569
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.987956524730392,
      "learning_rate": 6.407749113459261e-06,
      "loss": 0.7401,
      "step": 9570
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.423050384354918,
      "learning_rate": 6.407055949866243e-06,
      "loss": 0.6945,
      "step": 9571
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.070290825002088,
      "learning_rate": 6.4063627569040164e-06,
      "loss": 0.7409,
      "step": 9572
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.519554953437143,
      "learning_rate": 6.405669534587046e-06,
      "loss": 0.7601,
      "step": 9573
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.73981409898587,
      "learning_rate": 6.404976282929802e-06,
      "loss": 0.7829,
      "step": 9574
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.348567953293338,
      "learning_rate": 6.404283001946757e-06,
      "loss": 0.736,
      "step": 9575
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.794586639815102,
      "learning_rate": 6.403589691652378e-06,
      "loss": 0.7798,
      "step": 9576
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.1302917612063075,
      "learning_rate": 6.402896352061138e-06,
      "loss": 0.7947,
      "step": 9577
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.870441049642524,
      "learning_rate": 6.402202983187509e-06,
      "loss": 0.7753,
      "step": 9578
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.994016662555639,
      "learning_rate": 6.401509585045963e-06,
      "loss": 0.8068,
      "step": 9579
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.124560939881607,
      "learning_rate": 6.4008161576509734e-06,
      "loss": 0.809,
      "step": 9580
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.021929822062798,
      "learning_rate": 6.400122701017016e-06,
      "loss": 0.7422,
      "step": 9581
    },
    {
      "epoch": 0.85,
      "grad_norm": 8.439546053771394,
      "learning_rate": 6.399429215158564e-06,
      "loss": 0.8027,
      "step": 9582
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.34780093249162,
      "learning_rate": 6.3987357000900885e-06,
      "loss": 0.7673,
      "step": 9583
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.472288639123464,
      "learning_rate": 6.3980421558260715e-06,
      "loss": 0.7112,
      "step": 9584
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.859714145040179,
      "learning_rate": 6.397348582380984e-06,
      "loss": 0.8546,
      "step": 9585
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.940841819397808,
      "learning_rate": 6.396654979769305e-06,
      "loss": 0.7729,
      "step": 9586
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.63557414543912,
      "learning_rate": 6.3959613480055136e-06,
      "loss": 0.7755,
      "step": 9587
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.083855344299968,
      "learning_rate": 6.395267687104085e-06,
      "loss": 0.7654,
      "step": 9588
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.392333284566177,
      "learning_rate": 6.394573997079501e-06,
      "loss": 0.7206,
      "step": 9589
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.370705587135131,
      "learning_rate": 6.393880277946236e-06,
      "loss": 0.7868,
      "step": 9590
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.622597587951707,
      "learning_rate": 6.393186529718776e-06,
      "loss": 0.8311,
      "step": 9591
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.614489443883015,
      "learning_rate": 6.392492752411595e-06,
      "loss": 0.6962,
      "step": 9592
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.986926147842197,
      "learning_rate": 6.391798946039178e-06,
      "loss": 0.7714,
      "step": 9593
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.429776178278837,
      "learning_rate": 6.391105110616007e-06,
      "loss": 0.8937,
      "step": 9594
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.8234114069123235,
      "learning_rate": 6.390411246156562e-06,
      "loss": 0.7297,
      "step": 9595
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.658063226930865,
      "learning_rate": 6.389717352675329e-06,
      "loss": 0.7181,
      "step": 9596
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.601391365592395,
      "learning_rate": 6.389023430186789e-06,
      "loss": 0.7584,
      "step": 9597
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.7202670661460004,
      "learning_rate": 6.388329478705426e-06,
      "loss": 0.8319,
      "step": 9598
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.765891946618638,
      "learning_rate": 6.387635498245725e-06,
      "loss": 0.7802,
      "step": 9599
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.295515569683688,
      "learning_rate": 6.386941488822174e-06,
      "loss": 0.7482,
      "step": 9600
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.559529460626731,
      "learning_rate": 6.386247450449255e-06,
      "loss": 0.7164,
      "step": 9601
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.306185664172008,
      "learning_rate": 6.385553383141457e-06,
      "loss": 0.7249,
      "step": 9602
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.201778368943527,
      "learning_rate": 6.384859286913268e-06,
      "loss": 0.8223,
      "step": 9603
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.750108044956159,
      "learning_rate": 6.384165161779172e-06,
      "loss": 0.7877,
      "step": 9604
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.674629139721951,
      "learning_rate": 6.38347100775366e-06,
      "loss": 0.8479,
      "step": 9605
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.783015232008509,
      "learning_rate": 6.38277682485122e-06,
      "loss": 0.7722,
      "step": 9606
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.5484561865786235,
      "learning_rate": 6.382082613086344e-06,
      "loss": 0.7745,
      "step": 9607
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.3825820354104605,
      "learning_rate": 6.381388372473517e-06,
      "loss": 0.802,
      "step": 9608
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.81688022314568,
      "learning_rate": 6.380694103027234e-06,
      "loss": 0.7119,
      "step": 9609
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.2847306040091,
      "learning_rate": 6.379999804761988e-06,
      "loss": 0.7912,
      "step": 9610
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.384531631051542,
      "learning_rate": 6.379305477692266e-06,
      "loss": 0.8215,
      "step": 9611
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.3725644864050714,
      "learning_rate": 6.378611121832562e-06,
      "loss": 0.7914,
      "step": 9612
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.643052970131539,
      "learning_rate": 6.377916737197372e-06,
      "loss": 0.727,
      "step": 9613
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.510025912356734,
      "learning_rate": 6.377222323801185e-06,
      "loss": 0.7564,
      "step": 9614
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.827755218490466,
      "learning_rate": 6.3765278816585e-06,
      "loss": 0.7714,
      "step": 9615
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.308076665659627,
      "learning_rate": 6.3758334107838094e-06,
      "loss": 0.7974,
      "step": 9616
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.51238909891741,
      "learning_rate": 6.37513891119161e-06,
      "loss": 0.7911,
      "step": 9617
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.738011050335456,
      "learning_rate": 6.374444382896396e-06,
      "loss": 0.7646,
      "step": 9618
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.978078403967973,
      "learning_rate": 6.373749825912667e-06,
      "loss": 0.7717,
      "step": 9619
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.387442652963802,
      "learning_rate": 6.373055240254919e-06,
      "loss": 0.8127,
      "step": 9620
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.909185629345546,
      "learning_rate": 6.372360625937648e-06,
      "loss": 0.8193,
      "step": 9621
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.63492516804353,
      "learning_rate": 6.371665982975357e-06,
      "loss": 0.7915,
      "step": 9622
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.341202800122641,
      "learning_rate": 6.37097131138254e-06,
      "loss": 0.843,
      "step": 9623
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.255990001585032,
      "learning_rate": 6.3702766111737e-06,
      "loss": 0.663,
      "step": 9624
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.35900443954684,
      "learning_rate": 6.3695818823633374e-06,
      "loss": 0.7118,
      "step": 9625
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.740648511949091,
      "learning_rate": 6.3688871249659526e-06,
      "loss": 0.7908,
      "step": 9626
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.86790605496722,
      "learning_rate": 6.368192338996044e-06,
      "loss": 0.7617,
      "step": 9627
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.245811868476802,
      "learning_rate": 6.367497524468118e-06,
      "loss": 0.7641,
      "step": 9628
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.597582641964364,
      "learning_rate": 6.366802681396678e-06,
      "loss": 0.7041,
      "step": 9629
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.056824027272064,
      "learning_rate": 6.366107809796223e-06,
      "loss": 0.7632,
      "step": 9630
    },
    {
      "epoch": 0.86,
      "grad_norm": 8.653906417471454,
      "learning_rate": 6.36541290968126e-06,
      "loss": 0.7919,
      "step": 9631
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.0359267310783355,
      "learning_rate": 6.364717981066294e-06,
      "loss": 0.7773,
      "step": 9632
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.272816019358891,
      "learning_rate": 6.364023023965827e-06,
      "loss": 0.7004,
      "step": 9633
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.38899992508028,
      "learning_rate": 6.3633280383943675e-06,
      "loss": 0.7548,
      "step": 9634
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.951842709900067,
      "learning_rate": 6.36263302436642e-06,
      "loss": 0.7783,
      "step": 9635
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.309120425953189,
      "learning_rate": 6.3619379818964925e-06,
      "loss": 0.7596,
      "step": 9636
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.9253448801597606,
      "learning_rate": 6.361242910999092e-06,
      "loss": 0.7738,
      "step": 9637
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.149601982252259,
      "learning_rate": 6.360547811688728e-06,
      "loss": 0.764,
      "step": 9638
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.5309234127529034,
      "learning_rate": 6.359852683979908e-06,
      "loss": 0.7703,
      "step": 9639
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.981599484626527,
      "learning_rate": 6.359157527887139e-06,
      "loss": 0.7317,
      "step": 9640
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.809079650149916,
      "learning_rate": 6.358462343424934e-06,
      "loss": 0.771,
      "step": 9641
    },
    {
      "epoch": 0.86,
      "grad_norm": 8.102591800247954,
      "learning_rate": 6.357767130607804e-06,
      "loss": 0.81,
      "step": 9642
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.965538116913747,
      "learning_rate": 6.357071889450257e-06,
      "loss": 0.759,
      "step": 9643
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.910111440759103,
      "learning_rate": 6.356376619966807e-06,
      "loss": 0.7212,
      "step": 9644
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.319572592559808,
      "learning_rate": 6.355681322171965e-06,
      "loss": 0.8215,
      "step": 9645
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.968792559045306,
      "learning_rate": 6.354985996080245e-06,
      "loss": 0.7166,
      "step": 9646
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.421344843831731,
      "learning_rate": 6.35429064170616e-06,
      "loss": 0.727,
      "step": 9647
    },
    {
      "epoch": 0.86,
      "grad_norm": 9.497870017712973,
      "learning_rate": 6.353595259064223e-06,
      "loss": 0.7889,
      "step": 9648
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.480426679771631,
      "learning_rate": 6.352899848168949e-06,
      "loss": 0.786,
      "step": 9649
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.468689564823583,
      "learning_rate": 6.352204409034853e-06,
      "loss": 0.7909,
      "step": 9650
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.9841523550496385,
      "learning_rate": 6.3515089416764534e-06,
      "loss": 0.7635,
      "step": 9651
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.649429577759701,
      "learning_rate": 6.350813446108263e-06,
      "loss": 0.7562,
      "step": 9652
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.314009911187804,
      "learning_rate": 6.350117922344801e-06,
      "loss": 0.7712,
      "step": 9653
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.54972436487417,
      "learning_rate": 6.3494223704005816e-06,
      "loss": 0.7116,
      "step": 9654
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.930818601158032,
      "learning_rate": 6.348726790290128e-06,
      "loss": 0.7679,
      "step": 9655
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.648315657821372,
      "learning_rate": 6.348031182027955e-06,
      "loss": 0.8574,
      "step": 9656
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.826822324480468,
      "learning_rate": 6.347335545628584e-06,
      "loss": 0.7228,
      "step": 9657
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.45357739437551,
      "learning_rate": 6.3466398811065335e-06,
      "loss": 0.8511,
      "step": 9658
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.908836830808258,
      "learning_rate": 6.345944188476325e-06,
      "loss": 0.7354,
      "step": 9659
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.883851183934523,
      "learning_rate": 6.345248467752478e-06,
      "loss": 0.7541,
      "step": 9660
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.153830388505684,
      "learning_rate": 6.344552718949518e-06,
      "loss": 0.8227,
      "step": 9661
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.133076835914646,
      "learning_rate": 6.343856942081962e-06,
      "loss": 0.7598,
      "step": 9662
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.864469238982529,
      "learning_rate": 6.343161137164335e-06,
      "loss": 0.7438,
      "step": 9663
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.258163220243833,
      "learning_rate": 6.342465304211163e-06,
      "loss": 0.754,
      "step": 9664
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.463545279868044,
      "learning_rate": 6.341769443236966e-06,
      "loss": 0.7471,
      "step": 9665
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.292944923143582,
      "learning_rate": 6.34107355425627e-06,
      "loss": 0.7734,
      "step": 9666
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.072850028149302,
      "learning_rate": 6.340377637283602e-06,
      "loss": 0.7243,
      "step": 9667
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.750335166684634,
      "learning_rate": 6.339681692333486e-06,
      "loss": 0.7213,
      "step": 9668
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.929208381344161,
      "learning_rate": 6.338985719420447e-06,
      "loss": 0.6626,
      "step": 9669
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.614168437915558,
      "learning_rate": 6.3382897185590165e-06,
      "loss": 0.767,
      "step": 9670
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.6852280605651835,
      "learning_rate": 6.337593689763716e-06,
      "loss": 0.7302,
      "step": 9671
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.8516600725868475,
      "learning_rate": 6.3368976330490776e-06,
      "loss": 0.7153,
      "step": 9672
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.69805493874649,
      "learning_rate": 6.336201548429628e-06,
      "loss": 0.7213,
      "step": 9673
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.190546510302859,
      "learning_rate": 6.335505435919897e-06,
      "loss": 0.7779,
      "step": 9674
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.043502593244326,
      "learning_rate": 6.334809295534416e-06,
      "loss": 0.8064,
      "step": 9675
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.678446128257518,
      "learning_rate": 6.334113127287714e-06,
      "loss": 0.7962,
      "step": 9676
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.389223083147459,
      "learning_rate": 6.333416931194323e-06,
      "loss": 0.7437,
      "step": 9677
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.95557824679926,
      "learning_rate": 6.332720707268772e-06,
      "loss": 0.7406,
      "step": 9678
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.267830063979681,
      "learning_rate": 6.332024455525595e-06,
      "loss": 0.7905,
      "step": 9679
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.142082586469153,
      "learning_rate": 6.331328175979325e-06,
      "loss": 0.8081,
      "step": 9680
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.187690882985472,
      "learning_rate": 6.330631868644495e-06,
      "loss": 0.7281,
      "step": 9681
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.2007701721987765,
      "learning_rate": 6.329935533535639e-06,
      "loss": 0.8127,
      "step": 9682
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.970951646290749,
      "learning_rate": 6.329239170667292e-06,
      "loss": 0.7451,
      "step": 9683
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.789900846701699,
      "learning_rate": 6.328542780053987e-06,
      "loss": 0.7847,
      "step": 9684
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.882511808840508,
      "learning_rate": 6.3278463617102606e-06,
      "loss": 0.8155,
      "step": 9685
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.800216619352267,
      "learning_rate": 6.32714991565065e-06,
      "loss": 0.732,
      "step": 9686
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.60481732182165,
      "learning_rate": 6.326453441889691e-06,
      "loss": 0.7414,
      "step": 9687
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.777320389395315,
      "learning_rate": 6.3257569404419205e-06,
      "loss": 0.9373,
      "step": 9688
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.600170908611265,
      "learning_rate": 6.32506041132188e-06,
      "loss": 0.7491,
      "step": 9689
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.860002759638754,
      "learning_rate": 6.324363854544103e-06,
      "loss": 0.7906,
      "step": 9690
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.732816759013145,
      "learning_rate": 6.323667270123132e-06,
      "loss": 0.7343,
      "step": 9691
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.138212671728253,
      "learning_rate": 6.322970658073504e-06,
      "loss": 0.7533,
      "step": 9692
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.9304384114322435,
      "learning_rate": 6.322274018409762e-06,
      "loss": 0.7259,
      "step": 9693
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.538010567294612,
      "learning_rate": 6.3215773511464455e-06,
      "loss": 0.6934,
      "step": 9694
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.905715190860067,
      "learning_rate": 6.320880656298094e-06,
      "loss": 0.7655,
      "step": 9695
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.84397310816804,
      "learning_rate": 6.320183933879253e-06,
      "loss": 0.7907,
      "step": 9696
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.201392640522567,
      "learning_rate": 6.319487183904466e-06,
      "loss": 0.7421,
      "step": 9697
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.006391928661296,
      "learning_rate": 6.318790406388271e-06,
      "loss": 0.7895,
      "step": 9698
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.778412208700614,
      "learning_rate": 6.318093601345215e-06,
      "loss": 0.727,
      "step": 9699
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.19698117656942,
      "learning_rate": 6.3173967687898404e-06,
      "loss": 0.8328,
      "step": 9700
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.354407010929948,
      "learning_rate": 6.316699908736694e-06,
      "loss": 0.6964,
      "step": 9701
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.308359953055685,
      "learning_rate": 6.3160030212003225e-06,
      "loss": 0.6949,
      "step": 9702
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.732590648022395,
      "learning_rate": 6.315306106195267e-06,
      "loss": 0.751,
      "step": 9703
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.04920604548987,
      "learning_rate": 6.314609163736078e-06,
      "loss": 0.7146,
      "step": 9704
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.111519793738839,
      "learning_rate": 6.313912193837303e-06,
      "loss": 0.823,
      "step": 9705
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.340380772890688,
      "learning_rate": 6.313215196513487e-06,
      "loss": 0.7715,
      "step": 9706
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.17946212875513,
      "learning_rate": 6.31251817177918e-06,
      "loss": 0.7901,
      "step": 9707
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.8837403279075176,
      "learning_rate": 6.311821119648931e-06,
      "loss": 0.8199,
      "step": 9708
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.5144519839903685,
      "learning_rate": 6.3111240401372875e-06,
      "loss": 0.7939,
      "step": 9709
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.331943428639324,
      "learning_rate": 6.3104269332588015e-06,
      "loss": 0.7678,
      "step": 9710
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.645339658817871,
      "learning_rate": 6.309729799028023e-06,
      "loss": 0.7583,
      "step": 9711
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.897468541297062,
      "learning_rate": 6.309032637459506e-06,
      "loss": 0.8177,
      "step": 9712
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.746494859902457,
      "learning_rate": 6.308335448567796e-06,
      "loss": 0.7781,
      "step": 9713
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.401305494776823,
      "learning_rate": 6.30763823236745e-06,
      "loss": 0.7762,
      "step": 9714
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.109057651879473,
      "learning_rate": 6.30694098887302e-06,
      "loss": 0.767,
      "step": 9715
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.8431798601119915,
      "learning_rate": 6.306243718099057e-06,
      "loss": 0.7617,
      "step": 9716
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.412527459772984,
      "learning_rate": 6.305546420060118e-06,
      "loss": 0.7494,
      "step": 9717
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.180288025341259,
      "learning_rate": 6.30484909477076e-06,
      "loss": 0.811,
      "step": 9718
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.656689151863945,
      "learning_rate": 6.30415174224553e-06,
      "loss": 0.7106,
      "step": 9719
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.85171134944672,
      "learning_rate": 6.303454362498991e-06,
      "loss": 0.7983,
      "step": 9720
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.548652587058567,
      "learning_rate": 6.302756955545696e-06,
      "loss": 0.7616,
      "step": 9721
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.727495958499595,
      "learning_rate": 6.302059521400201e-06,
      "loss": 0.7887,
      "step": 9722
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.196384398140186,
      "learning_rate": 6.301362060077067e-06,
      "loss": 0.7641,
      "step": 9723
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.854533115196373,
      "learning_rate": 6.300664571590849e-06,
      "loss": 0.8086,
      "step": 9724
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.077895696447907,
      "learning_rate": 6.2999670559561065e-06,
      "loss": 0.7314,
      "step": 9725
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.2199288099545536,
      "learning_rate": 6.299269513187399e-06,
      "loss": 0.7802,
      "step": 9726
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.206560837203606,
      "learning_rate": 6.298571943299285e-06,
      "loss": 0.7969,
      "step": 9727
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.382693834873623,
      "learning_rate": 6.297874346306327e-06,
      "loss": 0.7595,
      "step": 9728
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.224437957624343,
      "learning_rate": 6.297176722223083e-06,
      "loss": 0.7414,
      "step": 9729
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.499876545896473,
      "learning_rate": 6.296479071064115e-06,
      "loss": 0.7434,
      "step": 9730
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.847516885582258,
      "learning_rate": 6.295781392843988e-06,
      "loss": 0.7547,
      "step": 9731
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.558753388591179,
      "learning_rate": 6.2950836875772605e-06,
      "loss": 0.7334,
      "step": 9732
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.736459166516716,
      "learning_rate": 6.294385955278498e-06,
      "loss": 0.8063,
      "step": 9733
    },
    {
      "epoch": 0.87,
      "grad_norm": 10.109051645774318,
      "learning_rate": 6.293688195962264e-06,
      "loss": 0.7178,
      "step": 9734
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.9361968608457065,
      "learning_rate": 6.292990409643121e-06,
      "loss": 0.7742,
      "step": 9735
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.460872153741046,
      "learning_rate": 6.292292596335634e-06,
      "loss": 0.8311,
      "step": 9736
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.433031431305875,
      "learning_rate": 6.29159475605437e-06,
      "loss": 0.7846,
      "step": 9737
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.857989579236105,
      "learning_rate": 6.2908968888138935e-06,
      "loss": 0.8038,
      "step": 9738
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.569643503778049,
      "learning_rate": 6.290198994628771e-06,
      "loss": 0.7248,
      "step": 9739
    },
    {
      "epoch": 0.87,
      "grad_norm": 9.571273350934446,
      "learning_rate": 6.289501073513572e-06,
      "loss": 0.7424,
      "step": 9740
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.425659957641195,
      "learning_rate": 6.2888031254828605e-06,
      "loss": 0.7492,
      "step": 9741
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.407756744256636,
      "learning_rate": 6.288105150551207e-06,
      "loss": 0.7759,
      "step": 9742
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.495054598506165,
      "learning_rate": 6.28740714873318e-06,
      "loss": 0.8099,
      "step": 9743
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.311643341693338,
      "learning_rate": 6.286709120043346e-06,
      "loss": 0.7989,
      "step": 9744
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.872150947401154,
      "learning_rate": 6.286011064496278e-06,
      "loss": 0.7214,
      "step": 9745
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.351364656475472,
      "learning_rate": 6.285312982106546e-06,
      "loss": 0.7623,
      "step": 9746
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.93795385029351,
      "learning_rate": 6.28461487288872e-06,
      "loss": 0.782,
      "step": 9747
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.7431909970469657,
      "learning_rate": 6.283916736857372e-06,
      "loss": 0.7588,
      "step": 9748
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.336567771686006,
      "learning_rate": 6.283218574027072e-06,
      "loss": 0.7275,
      "step": 9749
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.511240842298394,
      "learning_rate": 6.2825203844123964e-06,
      "loss": 0.8472,
      "step": 9750
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.463778359552672,
      "learning_rate": 6.281822168027915e-06,
      "loss": 0.7117,
      "step": 9751
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.789918912099097,
      "learning_rate": 6.281123924888204e-06,
      "loss": 0.8157,
      "step": 9752
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.321640373913825,
      "learning_rate": 6.280425655007837e-06,
      "loss": 0.7299,
      "step": 9753
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.877590118784282,
      "learning_rate": 6.279727358401388e-06,
      "loss": 0.7225,
      "step": 9754
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.8442221470586295,
      "learning_rate": 6.279029035083433e-06,
      "loss": 0.7851,
      "step": 9755
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.34055109166387,
      "learning_rate": 6.278330685068549e-06,
      "loss": 0.8127,
      "step": 9756
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.592131280234987,
      "learning_rate": 6.27763230837131e-06,
      "loss": 0.8634,
      "step": 9757
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.134920724723284,
      "learning_rate": 6.276933905006294e-06,
      "loss": 0.8235,
      "step": 9758
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.190500089275022,
      "learning_rate": 6.276235474988081e-06,
      "loss": 0.7727,
      "step": 9759
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.209558641718262,
      "learning_rate": 6.2755370183312455e-06,
      "loss": 0.7507,
      "step": 9760
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.230863139527202,
      "learning_rate": 6.274838535050368e-06,
      "loss": 0.7228,
      "step": 9761
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.101414379246327,
      "learning_rate": 6.27414002516003e-06,
      "loss": 0.782,
      "step": 9762
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.7469131883652285,
      "learning_rate": 6.27344148867481e-06,
      "loss": 0.8655,
      "step": 9763
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.995456967869784,
      "learning_rate": 6.272742925609284e-06,
      "loss": 0.7096,
      "step": 9764
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.0891935586739665,
      "learning_rate": 6.272044335978039e-06,
      "loss": 0.7578,
      "step": 9765
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.793230684641359,
      "learning_rate": 6.271345719795654e-06,
      "loss": 0.803,
      "step": 9766
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.641366637544686,
      "learning_rate": 6.270647077076712e-06,
      "loss": 0.7562,
      "step": 9767
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.480318732455397,
      "learning_rate": 6.269948407835793e-06,
      "loss": 0.7606,
      "step": 9768
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.943779578022765,
      "learning_rate": 6.269249712087484e-06,
      "loss": 0.8211,
      "step": 9769
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.659187168561962,
      "learning_rate": 6.268550989846366e-06,
      "loss": 0.7322,
      "step": 9770
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.138023512846651,
      "learning_rate": 6.267852241127024e-06,
      "loss": 0.8062,
      "step": 9771
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.26433237441788,
      "learning_rate": 6.267153465944044e-06,
      "loss": 0.7732,
      "step": 9772
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.130434754997048,
      "learning_rate": 6.26645466431201e-06,
      "loss": 0.8221,
      "step": 9773
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.464948469415333,
      "learning_rate": 6.265755836245506e-06,
      "loss": 0.7215,
      "step": 9774
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.390970233903345,
      "learning_rate": 6.2650569817591246e-06,
      "loss": 0.8311,
      "step": 9775
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.9748642101933696,
      "learning_rate": 6.264358100867448e-06,
      "loss": 0.8102,
      "step": 9776
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.396353472123679,
      "learning_rate": 6.263659193585063e-06,
      "loss": 0.7037,
      "step": 9777
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.641845806086245,
      "learning_rate": 6.262960259926562e-06,
      "loss": 0.7212,
      "step": 9778
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.073172200723437,
      "learning_rate": 6.2622612999065316e-06,
      "loss": 0.7877,
      "step": 9779
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.6781154246567365,
      "learning_rate": 6.261562313539559e-06,
      "loss": 0.796,
      "step": 9780
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.889775074326401,
      "learning_rate": 6.260863300840236e-06,
      "loss": 0.7303,
      "step": 9781
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.248690123335354,
      "learning_rate": 6.260164261823154e-06,
      "loss": 0.8359,
      "step": 9782
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.628509215390753,
      "learning_rate": 6.259465196502903e-06,
      "loss": 0.7448,
      "step": 9783
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.397228932471153,
      "learning_rate": 6.258766104894073e-06,
      "loss": 0.7903,
      "step": 9784
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.750413185837244,
      "learning_rate": 6.258066987011258e-06,
      "loss": 0.8418,
      "step": 9785
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.765393854953187,
      "learning_rate": 6.257367842869049e-06,
      "loss": 0.6927,
      "step": 9786
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.664058517144299,
      "learning_rate": 6.25666867248204e-06,
      "loss": 0.7535,
      "step": 9787
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.8413468096439685,
      "learning_rate": 6.255969475864825e-06,
      "loss": 0.7816,
      "step": 9788
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.939419908556113,
      "learning_rate": 6.255270253031997e-06,
      "loss": 0.7555,
      "step": 9789
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.675445841706478,
      "learning_rate": 6.254571003998153e-06,
      "loss": 0.7374,
      "step": 9790
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.258271396872061,
      "learning_rate": 6.253871728777887e-06,
      "loss": 0.8041,
      "step": 9791
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.237165624338756,
      "learning_rate": 6.253172427385792e-06,
      "loss": 0.7318,
      "step": 9792
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.97391673899181,
      "learning_rate": 6.252473099836468e-06,
      "loss": 0.7823,
      "step": 9793
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.62725143017116,
      "learning_rate": 6.251773746144512e-06,
      "loss": 0.7209,
      "step": 9794
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.227032551796015,
      "learning_rate": 6.251074366324519e-06,
      "loss": 0.7249,
      "step": 9795
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.47700944849814,
      "learning_rate": 6.250374960391089e-06,
      "loss": 0.775,
      "step": 9796
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.745307835948027,
      "learning_rate": 6.249675528358821e-06,
      "loss": 0.7855,
      "step": 9797
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.735720615309904,
      "learning_rate": 6.248976070242312e-06,
      "loss": 0.8509,
      "step": 9798
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.016392953332443,
      "learning_rate": 6.248276586056164e-06,
      "loss": 0.77,
      "step": 9799
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.910986800627424,
      "learning_rate": 6.247577075814974e-06,
      "loss": 0.7444,
      "step": 9800
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.530083286415009,
      "learning_rate": 6.246877539533345e-06,
      "loss": 0.8654,
      "step": 9801
    },
    {
      "epoch": 0.87,
      "grad_norm": 8.218722774095697,
      "learning_rate": 6.246177977225878e-06,
      "loss": 0.7235,
      "step": 9802
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.9857800587362044,
      "learning_rate": 6.245478388907175e-06,
      "loss": 0.7817,
      "step": 9803
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.413422600785966,
      "learning_rate": 6.24477877459184e-06,
      "loss": 0.829,
      "step": 9804
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.052256947450342,
      "learning_rate": 6.244079134294473e-06,
      "loss": 0.7134,
      "step": 9805
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.274180735199831,
      "learning_rate": 6.243379468029677e-06,
      "loss": 0.8488,
      "step": 9806
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.254375315197664,
      "learning_rate": 6.242679775812059e-06,
      "loss": 0.8056,
      "step": 9807
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.382070136401744,
      "learning_rate": 6.2419800576562215e-06,
      "loss": 0.7802,
      "step": 9808
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.0295743014485765,
      "learning_rate": 6.241280313576771e-06,
      "loss": 0.7505,
      "step": 9809
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.407889549345209,
      "learning_rate": 6.240580543588311e-06,
      "loss": 0.7725,
      "step": 9810
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.218968171172818,
      "learning_rate": 6.2398807477054504e-06,
      "loss": 0.7413,
      "step": 9811
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.060846227752618,
      "learning_rate": 6.239180925942794e-06,
      "loss": 0.7572,
      "step": 9812
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.8622348164976295,
      "learning_rate": 6.238481078314949e-06,
      "loss": 0.813,
      "step": 9813
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.24001652070445,
      "learning_rate": 6.237781204836525e-06,
      "loss": 0.7636,
      "step": 9814
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.55707352381635,
      "learning_rate": 6.237081305522128e-06,
      "loss": 0.837,
      "step": 9815
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.625670235453415,
      "learning_rate": 6.236381380386367e-06,
      "loss": 0.7711,
      "step": 9816
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.379156261099252,
      "learning_rate": 6.235681429443854e-06,
      "loss": 0.7697,
      "step": 9817
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.451108602408712,
      "learning_rate": 6.234981452709197e-06,
      "loss": 0.7571,
      "step": 9818
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.899880208049548,
      "learning_rate": 6.234281450197006e-06,
      "loss": 0.7011,
      "step": 9819
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.443604525105231,
      "learning_rate": 6.233581421921894e-06,
      "loss": 0.7502,
      "step": 9820
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.638481191283446,
      "learning_rate": 6.23288136789847e-06,
      "loss": 0.7472,
      "step": 9821
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.384864944442155,
      "learning_rate": 6.232181288141347e-06,
      "loss": 0.7378,
      "step": 9822
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.607874529528891,
      "learning_rate": 6.23148118266514e-06,
      "loss": 0.8002,
      "step": 9823
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.95233334445,
      "learning_rate": 6.230781051484458e-06,
      "loss": 0.8404,
      "step": 9824
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.5368550010793305,
      "learning_rate": 6.230080894613918e-06,
      "loss": 0.7554,
      "step": 9825
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.124440877781158,
      "learning_rate": 6.229380712068134e-06,
      "loss": 0.7439,
      "step": 9826
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.604613160183594,
      "learning_rate": 6.228680503861719e-06,
      "loss": 0.6907,
      "step": 9827
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.54661559806082,
      "learning_rate": 6.2279802700092895e-06,
      "loss": 0.7889,
      "step": 9828
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.5884258738460995,
      "learning_rate": 6.227280010525462e-06,
      "loss": 0.7331,
      "step": 9829
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.92609050916543,
      "learning_rate": 6.2265797254248496e-06,
      "loss": 0.8959,
      "step": 9830
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.721371076632458,
      "learning_rate": 6.225879414722073e-06,
      "loss": 0.7396,
      "step": 9831
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.675046520298822,
      "learning_rate": 6.225179078431749e-06,
      "loss": 0.7759,
      "step": 9832
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.734424718990778,
      "learning_rate": 6.224478716568493e-06,
      "loss": 0.8332,
      "step": 9833
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.813233278884142,
      "learning_rate": 6.223778329146926e-06,
      "loss": 0.7346,
      "step": 9834
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.533396986931334,
      "learning_rate": 6.223077916181668e-06,
      "loss": 0.7139,
      "step": 9835
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.824997391290138,
      "learning_rate": 6.222377477687336e-06,
      "loss": 0.7967,
      "step": 9836
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.435201711908418,
      "learning_rate": 6.22167701367855e-06,
      "loss": 0.8049,
      "step": 9837
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.84024165376351,
      "learning_rate": 6.220976524169933e-06,
      "loss": 0.7473,
      "step": 9838
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.292495970896735,
      "learning_rate": 6.220276009176105e-06,
      "loss": 0.7425,
      "step": 9839
    },
    {
      "epoch": 0.88,
      "grad_norm": 8.540935655678451,
      "learning_rate": 6.219575468711686e-06,
      "loss": 0.7082,
      "step": 9840
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.3673207731489425,
      "learning_rate": 6.218874902791302e-06,
      "loss": 0.7466,
      "step": 9841
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.471366730282035,
      "learning_rate": 6.218174311429573e-06,
      "loss": 0.709,
      "step": 9842
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.073493636428725,
      "learning_rate": 6.217473694641123e-06,
      "loss": 0.7796,
      "step": 9843
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.409516413250098,
      "learning_rate": 6.216773052440575e-06,
      "loss": 0.7717,
      "step": 9844
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.034277959291131,
      "learning_rate": 6.216072384842555e-06,
      "loss": 0.8032,
      "step": 9845
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.199711364132453,
      "learning_rate": 6.215371691861687e-06,
      "loss": 0.7656,
      "step": 9846
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.068191820197493,
      "learning_rate": 6.214670973512597e-06,
      "loss": 0.7784,
      "step": 9847
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.593190241652087,
      "learning_rate": 6.2139702298099105e-06,
      "loss": 0.7403,
      "step": 9848
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.273192634488201,
      "learning_rate": 6.213269460768254e-06,
      "loss": 0.7523,
      "step": 9849
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.3117759622183405,
      "learning_rate": 6.2125686664022535e-06,
      "loss": 0.8214,
      "step": 9850
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.4138003334489335,
      "learning_rate": 6.2118678467265385e-06,
      "loss": 0.777,
      "step": 9851
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.160305888734228,
      "learning_rate": 6.2111670017557355e-06,
      "loss": 0.7833,
      "step": 9852
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.665382394899319,
      "learning_rate": 6.2104661315044735e-06,
      "loss": 0.6944,
      "step": 9853
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.713495891370325,
      "learning_rate": 6.2097652359873825e-06,
      "loss": 0.7949,
      "step": 9854
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.265703804137476,
      "learning_rate": 6.209064315219093e-06,
      "loss": 0.8059,
      "step": 9855
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.858570053957441,
      "learning_rate": 6.208363369214232e-06,
      "loss": 0.7778,
      "step": 9856
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.781248456350245,
      "learning_rate": 6.207662397987432e-06,
      "loss": 0.7537,
      "step": 9857
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.601402282377116,
      "learning_rate": 6.206961401553324e-06,
      "loss": 0.7967,
      "step": 9858
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.957806345611381,
      "learning_rate": 6.206260379926541e-06,
      "loss": 0.729,
      "step": 9859
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.349063241000264,
      "learning_rate": 6.205559333121712e-06,
      "loss": 0.8146,
      "step": 9860
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.831368950822578,
      "learning_rate": 6.2048582611534745e-06,
      "loss": 0.7914,
      "step": 9861
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.857696199709002,
      "learning_rate": 6.2041571640364596e-06,
      "loss": 0.7346,
      "step": 9862
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.839866061630495,
      "learning_rate": 6.2034560417852996e-06,
      "loss": 0.7562,
      "step": 9863
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.072856223420023,
      "learning_rate": 6.20275489441463e-06,
      "loss": 0.8224,
      "step": 9864
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.085497580555607,
      "learning_rate": 6.202053721939086e-06,
      "loss": 0.6989,
      "step": 9865
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.724531931464815,
      "learning_rate": 6.201352524373304e-06,
      "loss": 0.7377,
      "step": 9866
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.536589617298664,
      "learning_rate": 6.2006513017319195e-06,
      "loss": 0.7293,
      "step": 9867
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.134751604167117,
      "learning_rate": 6.199950054029567e-06,
      "loss": 0.8487,
      "step": 9868
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.742377482116037,
      "learning_rate": 6.199248781280884e-06,
      "loss": 0.694,
      "step": 9869
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.797167232460141,
      "learning_rate": 6.198547483500511e-06,
      "loss": 0.762,
      "step": 9870
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.635590357252279,
      "learning_rate": 6.197846160703084e-06,
      "loss": 0.8541,
      "step": 9871
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.44350449516486,
      "learning_rate": 6.197144812903241e-06,
      "loss": 0.7159,
      "step": 9872
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.774322462155159,
      "learning_rate": 6.196443440115621e-06,
      "loss": 0.7571,
      "step": 9873
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.832464951253434,
      "learning_rate": 6.195742042354867e-06,
      "loss": 0.7492,
      "step": 9874
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.1981025830356975,
      "learning_rate": 6.195040619635613e-06,
      "loss": 0.6939,
      "step": 9875
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.216948315005162,
      "learning_rate": 6.194339171972504e-06,
      "loss": 0.7563,
      "step": 9876
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.494237381620149,
      "learning_rate": 6.193637699380183e-06,
      "loss": 0.8334,
      "step": 9877
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.748936491391593,
      "learning_rate": 6.1929362018732875e-06,
      "loss": 0.8176,
      "step": 9878
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.815403970786437,
      "learning_rate": 6.19223467946646e-06,
      "loss": 0.7331,
      "step": 9879
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.988813265145291,
      "learning_rate": 6.191533132174346e-06,
      "loss": 0.7949,
      "step": 9880
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.119290827075412,
      "learning_rate": 6.190831560011587e-06,
      "loss": 0.7462,
      "step": 9881
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.354332812928787,
      "learning_rate": 6.190129962992825e-06,
      "loss": 0.7294,
      "step": 9882
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.739268766731975,
      "learning_rate": 6.1894283411327095e-06,
      "loss": 0.7549,
      "step": 9883
    },
    {
      "epoch": 0.88,
      "grad_norm": 8.269171136319507,
      "learning_rate": 6.188726694445881e-06,
      "loss": 0.7387,
      "step": 9884
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.592298546498344,
      "learning_rate": 6.1880250229469864e-06,
      "loss": 0.7891,
      "step": 9885
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.649529788930542,
      "learning_rate": 6.18732332665067e-06,
      "loss": 0.7135,
      "step": 9886
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.487248039224758,
      "learning_rate": 6.18662160557158e-06,
      "loss": 0.8339,
      "step": 9887
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.191795688632441,
      "learning_rate": 6.185919859724363e-06,
      "loss": 0.7617,
      "step": 9888
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.536568846719598,
      "learning_rate": 6.185218089123665e-06,
      "loss": 0.7355,
      "step": 9889
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.457957308384774,
      "learning_rate": 6.184516293784136e-06,
      "loss": 0.7719,
      "step": 9890
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.1524146471923675,
      "learning_rate": 6.183814473720424e-06,
      "loss": 0.7661,
      "step": 9891
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.647911330361317,
      "learning_rate": 6.183112628947176e-06,
      "loss": 0.7436,
      "step": 9892
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.197298556638414,
      "learning_rate": 6.182410759479044e-06,
      "loss": 0.8343,
      "step": 9893
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.014020875297456,
      "learning_rate": 6.181708865330676e-06,
      "loss": 0.8063,
      "step": 9894
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.254773602607386,
      "learning_rate": 6.1810069465167225e-06,
      "loss": 0.7956,
      "step": 9895
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.570829429039235,
      "learning_rate": 6.180305003051836e-06,
      "loss": 0.857,
      "step": 9896
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.695056987900873,
      "learning_rate": 6.179603034950667e-06,
      "loss": 0.7609,
      "step": 9897
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.281681373151997,
      "learning_rate": 6.178901042227869e-06,
      "loss": 0.781,
      "step": 9898
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.913419575781941,
      "learning_rate": 6.178199024898094e-06,
      "loss": 0.7454,
      "step": 9899
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.811890768573447,
      "learning_rate": 6.1774969829759946e-06,
      "loss": 0.7161,
      "step": 9900
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.360663192042909,
      "learning_rate": 6.176794916476223e-06,
      "loss": 0.7234,
      "step": 9901
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.060096520468763,
      "learning_rate": 6.176092825413435e-06,
      "loss": 0.7674,
      "step": 9902
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.308445189994336,
      "learning_rate": 6.1753907098022845e-06,
      "loss": 0.7939,
      "step": 9903
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.366564507937351,
      "learning_rate": 6.174688569657428e-06,
      "loss": 0.8274,
      "step": 9904
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.318543323550969,
      "learning_rate": 6.1739864049935196e-06,
      "loss": 0.7273,
      "step": 9905
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.1867724327927975,
      "learning_rate": 6.1732842158252175e-06,
      "loss": 0.8589,
      "step": 9906
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.206113139862672,
      "learning_rate": 6.1725820021671735e-06,
      "loss": 0.7706,
      "step": 9907
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.1861446086850815,
      "learning_rate": 6.17187976403405e-06,
      "loss": 0.7603,
      "step": 9908
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.888317944665131,
      "learning_rate": 6.171177501440503e-06,
      "loss": 0.7572,
      "step": 9909
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.853713656033985,
      "learning_rate": 6.17047521440119e-06,
      "loss": 0.8133,
      "step": 9910
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.7852457081629325,
      "learning_rate": 6.16977290293077e-06,
      "loss": 0.7985,
      "step": 9911
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.973910445947179,
      "learning_rate": 6.169070567043903e-06,
      "loss": 0.7511,
      "step": 9912
    },
    {
      "epoch": 0.88,
      "grad_norm": 8.686211967465981,
      "learning_rate": 6.168368206755249e-06,
      "loss": 0.7867,
      "step": 9913
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.3604987410375875,
      "learning_rate": 6.167665822079466e-06,
      "loss": 0.7089,
      "step": 9914
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.704179874781647,
      "learning_rate": 6.166963413031217e-06,
      "loss": 0.6883,
      "step": 9915
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.199409241040635,
      "learning_rate": 6.166260979625162e-06,
      "loss": 0.7706,
      "step": 9916
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.293311353888366,
      "learning_rate": 6.165558521875961e-06,
      "loss": 0.7513,
      "step": 9917
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.615946942219975,
      "learning_rate": 6.1648560397982815e-06,
      "loss": 0.8617,
      "step": 9918
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.299422588502069,
      "learning_rate": 6.164153533406782e-06,
      "loss": 0.7774,
      "step": 9919
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.6395639156996635,
      "learning_rate": 6.163451002716128e-06,
      "loss": 0.7657,
      "step": 9920
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.558679436021888,
      "learning_rate": 6.162748447740981e-06,
      "loss": 0.8147,
      "step": 9921
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.70867369170834,
      "learning_rate": 6.162045868496006e-06,
      "loss": 0.7988,
      "step": 9922
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.652359245621111,
      "learning_rate": 6.161343264995869e-06,
      "loss": 0.7659,
      "step": 9923
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.432275932699342,
      "learning_rate": 6.160640637255234e-06,
      "loss": 0.7363,
      "step": 9924
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.8900369421039995,
      "learning_rate": 6.159937985288769e-06,
      "loss": 0.7123,
      "step": 9925
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.783861648226687,
      "learning_rate": 6.159235309111137e-06,
      "loss": 0.7926,
      "step": 9926
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.088594430153176,
      "learning_rate": 6.158532608737008e-06,
      "loss": 0.7676,
      "step": 9927
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.337841527623541,
      "learning_rate": 6.157829884181048e-06,
      "loss": 0.7408,
      "step": 9928
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.410127942879893,
      "learning_rate": 6.157127135457924e-06,
      "loss": 0.7401,
      "step": 9929
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.4489336181597094,
      "learning_rate": 6.156424362582305e-06,
      "loss": 0.7588,
      "step": 9930
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.097794303015817,
      "learning_rate": 6.155721565568861e-06,
      "loss": 0.8204,
      "step": 9931
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.800590315559956,
      "learning_rate": 6.155018744432259e-06,
      "loss": 0.7604,
      "step": 9932
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.871408568930423,
      "learning_rate": 6.1543158991871696e-06,
      "loss": 0.7552,
      "step": 9933
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.293399265363556,
      "learning_rate": 6.153613029848267e-06,
      "loss": 0.6806,
      "step": 9934
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.355082011566736,
      "learning_rate": 6.152910136430217e-06,
      "loss": 0.7717,
      "step": 9935
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.209321721744387,
      "learning_rate": 6.152207218947691e-06,
      "loss": 0.7195,
      "step": 9936
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.658416390741104,
      "learning_rate": 6.151504277415363e-06,
      "loss": 0.7812,
      "step": 9937
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.162827111585358,
      "learning_rate": 6.150801311847907e-06,
      "loss": 0.8035,
      "step": 9938
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.517315034367639,
      "learning_rate": 6.150098322259992e-06,
      "loss": 0.7499,
      "step": 9939
    },
    {
      "epoch": 0.89,
      "grad_norm": 12.376673801749577,
      "learning_rate": 6.149395308666294e-06,
      "loss": 0.722,
      "step": 9940
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.812406824099804,
      "learning_rate": 6.148692271081487e-06,
      "loss": 0.7427,
      "step": 9941
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.789268857645545,
      "learning_rate": 6.147989209520243e-06,
      "loss": 0.8063,
      "step": 9942
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.081661830991714,
      "learning_rate": 6.147286123997239e-06,
      "loss": 0.8247,
      "step": 9943
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.177937220663609,
      "learning_rate": 6.14658301452715e-06,
      "loss": 0.7775,
      "step": 9944
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.296086440226046,
      "learning_rate": 6.14587988112465e-06,
      "loss": 0.8018,
      "step": 9945
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.31034367047131,
      "learning_rate": 6.145176723804418e-06,
      "loss": 0.7613,
      "step": 9946
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.367915722935144,
      "learning_rate": 6.14447354258113e-06,
      "loss": 0.7323,
      "step": 9947
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.715031615547743,
      "learning_rate": 6.143770337469463e-06,
      "loss": 0.7277,
      "step": 9948
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.440645120633575,
      "learning_rate": 6.143067108484096e-06,
      "loss": 0.6938,
      "step": 9949
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.652528111811027,
      "learning_rate": 6.142363855639704e-06,
      "loss": 0.7634,
      "step": 9950
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.7452619743242535,
      "learning_rate": 6.14166057895097e-06,
      "loss": 0.7273,
      "step": 9951
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.348928122225856,
      "learning_rate": 6.14095727843257e-06,
      "loss": 0.7472,
      "step": 9952
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.003946677838617,
      "learning_rate": 6.1402539540991875e-06,
      "loss": 0.7568,
      "step": 9953
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.933015932333282,
      "learning_rate": 6.1395506059655e-06,
      "loss": 0.7719,
      "step": 9954
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.461634939290451,
      "learning_rate": 6.138847234046188e-06,
      "loss": 0.7956,
      "step": 9955
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.272669479192141,
      "learning_rate": 6.138143838355936e-06,
      "loss": 0.8185,
      "step": 9956
    },
    {
      "epoch": 0.89,
      "grad_norm": 8.217522583808606,
      "learning_rate": 6.137440418909423e-06,
      "loss": 0.766,
      "step": 9957
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.827210194161117,
      "learning_rate": 6.136736975721332e-06,
      "loss": 0.8066,
      "step": 9958
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.226471277970875,
      "learning_rate": 6.136033508806345e-06,
      "loss": 0.859,
      "step": 9959
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.188991729044971,
      "learning_rate": 6.135330018179149e-06,
      "loss": 0.7618,
      "step": 9960
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.83083007985564,
      "learning_rate": 6.134626503854423e-06,
      "loss": 0.8219,
      "step": 9961
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.960559076085251,
      "learning_rate": 6.133922965846853e-06,
      "loss": 0.734,
      "step": 9962
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.9525788599371285,
      "learning_rate": 6.133219404171128e-06,
      "loss": 0.7148,
      "step": 9963
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.87106501882532,
      "learning_rate": 6.132515818841927e-06,
      "loss": 0.795,
      "step": 9964
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.9746742327432125,
      "learning_rate": 6.131812209873937e-06,
      "loss": 0.8194,
      "step": 9965
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.467382899722966,
      "learning_rate": 6.131108577281849e-06,
      "loss": 0.7362,
      "step": 9966
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.278482148525703,
      "learning_rate": 6.130404921080344e-06,
      "loss": 0.7624,
      "step": 9967
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.151906703395872,
      "learning_rate": 6.129701241284113e-06,
      "loss": 0.7435,
      "step": 9968
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.611433006178244,
      "learning_rate": 6.128997537907842e-06,
      "loss": 0.6936,
      "step": 9969
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.569739234088302,
      "learning_rate": 6.12829381096622e-06,
      "loss": 0.8724,
      "step": 9970
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.279732669095729,
      "learning_rate": 6.127590060473937e-06,
      "loss": 0.7944,
      "step": 9971
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.060564500467176,
      "learning_rate": 6.126886286445679e-06,
      "loss": 0.7236,
      "step": 9972
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.9973168054322175,
      "learning_rate": 6.12618248889614e-06,
      "loss": 0.7693,
      "step": 9973
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.518326534640806,
      "learning_rate": 6.125478667840006e-06,
      "loss": 0.7054,
      "step": 9974
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.269288927047092,
      "learning_rate": 6.124774823291968e-06,
      "loss": 0.7579,
      "step": 9975
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.002691076097937,
      "learning_rate": 6.124070955266721e-06,
      "loss": 0.7676,
      "step": 9976
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.512147588469111,
      "learning_rate": 6.1233670637789545e-06,
      "loss": 0.6848,
      "step": 9977
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.519773017488289,
      "learning_rate": 6.122663148843361e-06,
      "loss": 0.7581,
      "step": 9978
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.308623433024361,
      "learning_rate": 6.121959210474632e-06,
      "loss": 0.73,
      "step": 9979
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.027926797192309,
      "learning_rate": 6.121255248687461e-06,
      "loss": 0.794,
      "step": 9980
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.089798457938509,
      "learning_rate": 6.120551263496543e-06,
      "loss": 0.7278,
      "step": 9981
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.242809207026988,
      "learning_rate": 6.119847254916571e-06,
      "loss": 0.7407,
      "step": 9982
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.073283783776094,
      "learning_rate": 6.1191432229622395e-06,
      "loss": 0.7847,
      "step": 9983
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.638679284923997,
      "learning_rate": 6.1184391676482445e-06,
      "loss": 0.7783,
      "step": 9984
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.922893123326059,
      "learning_rate": 6.117735088989282e-06,
      "loss": 0.7346,
      "step": 9985
    },
    {
      "epoch": 0.89,
      "grad_norm": 8.15366719924164,
      "learning_rate": 6.117030987000047e-06,
      "loss": 0.8147,
      "step": 9986
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.394937677931595,
      "learning_rate": 6.116326861695236e-06,
      "loss": 0.7214,
      "step": 9987
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.539290256828647,
      "learning_rate": 6.1156227130895475e-06,
      "loss": 0.7775,
      "step": 9988
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.891447296131274,
      "learning_rate": 6.114918541197676e-06,
      "loss": 0.794,
      "step": 9989
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.860324355783745,
      "learning_rate": 6.114214346034323e-06,
      "loss": 0.7441,
      "step": 9990
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.2798531898199474,
      "learning_rate": 6.113510127614186e-06,
      "loss": 0.7232,
      "step": 9991
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.6169567353272765,
      "learning_rate": 6.112805885951964e-06,
      "loss": 0.8362,
      "step": 9992
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.00586446045036,
      "learning_rate": 6.112101621062354e-06,
      "loss": 0.804,
      "step": 9993
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.276672972729107,
      "learning_rate": 6.111397332960058e-06,
      "loss": 0.7192,
      "step": 9994
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.8162262709805175,
      "learning_rate": 6.110693021659779e-06,
      "loss": 0.8544,
      "step": 9995
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.5965964650582185,
      "learning_rate": 6.1099886871762134e-06,
      "loss": 0.7483,
      "step": 9996
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.822203589771857,
      "learning_rate": 6.1092843295240655e-06,
      "loss": 0.7419,
      "step": 9997
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.042248790297229,
      "learning_rate": 6.108579948718037e-06,
      "loss": 0.7648,
      "step": 9998
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.830030508812699,
      "learning_rate": 6.107875544772831e-06,
      "loss": 0.838,
      "step": 9999
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.727371859945585,
      "learning_rate": 6.1071711177031465e-06,
      "loss": 0.7615,
      "step": 10000
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.735783227055713,
      "learning_rate": 6.106466667523692e-06,
      "loss": 0.7854,
      "step": 10001
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.304545210862161,
      "learning_rate": 6.105762194249167e-06,
      "loss": 0.7767,
      "step": 10002
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.99249786634484,
      "learning_rate": 6.105057697894279e-06,
      "loss": 0.8448,
      "step": 10003
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.387622246294665,
      "learning_rate": 6.104353178473732e-06,
      "loss": 0.7574,
      "step": 10004
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.083313528126039,
      "learning_rate": 6.103648636002229e-06,
      "loss": 0.7714,
      "step": 10005
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.983867207279958,
      "learning_rate": 6.102944070494479e-06,
      "loss": 0.7167,
      "step": 10006
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.394388653670665,
      "learning_rate": 6.102239481965187e-06,
      "loss": 0.7278,
      "step": 10007
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.108046015205648,
      "learning_rate": 6.10153487042906e-06,
      "loss": 0.733,
      "step": 10008
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.854866686076289,
      "learning_rate": 6.100830235900803e-06,
      "loss": 0.7207,
      "step": 10009
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.674054705932066,
      "learning_rate": 6.100125578395126e-06,
      "loss": 0.7674,
      "step": 10010
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.257686403511286,
      "learning_rate": 6.099420897926739e-06,
      "loss": 0.6923,
      "step": 10011
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.066290451115364,
      "learning_rate": 6.0987161945103465e-06,
      "loss": 0.7688,
      "step": 10012
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.187575334233862,
      "learning_rate": 6.09801146816066e-06,
      "loss": 0.7538,
      "step": 10013
    },
    {
      "epoch": 0.89,
      "grad_norm": 8.71687445044388,
      "learning_rate": 6.097306718892391e-06,
      "loss": 0.8285,
      "step": 10014
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.871311096069346,
      "learning_rate": 6.096601946720244e-06,
      "loss": 0.7834,
      "step": 10015
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.773263185873478,
      "learning_rate": 6.095897151658934e-06,
      "loss": 0.7738,
      "step": 10016
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.610930909008848,
      "learning_rate": 6.095192333723171e-06,
      "loss": 0.7324,
      "step": 10017
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.776279945821417,
      "learning_rate": 6.094487492927667e-06,
      "loss": 0.7449,
      "step": 10018
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.0715849201214445,
      "learning_rate": 6.093782629287132e-06,
      "loss": 0.7183,
      "step": 10019
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.482334863950214,
      "learning_rate": 6.0930777428162815e-06,
      "loss": 0.8058,
      "step": 10020
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.710100899183156,
      "learning_rate": 6.092372833529827e-06,
      "loss": 0.7524,
      "step": 10021
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.477060983194241,
      "learning_rate": 6.091667901442481e-06,
      "loss": 0.842,
      "step": 10022
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.368188635859025,
      "learning_rate": 6.090962946568957e-06,
      "loss": 0.7527,
      "step": 10023
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.009059587859002,
      "learning_rate": 6.090257968923972e-06,
      "loss": 0.795,
      "step": 10024
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.990158182959132,
      "learning_rate": 6.08955296852224e-06,
      "loss": 0.7738,
      "step": 10025
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.486058449075646,
      "learning_rate": 6.088847945378477e-06,
      "loss": 0.7258,
      "step": 10026
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.174720632378773,
      "learning_rate": 6.088142899507394e-06,
      "loss": 0.7741,
      "step": 10027
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.686638366020004,
      "learning_rate": 6.087437830923715e-06,
      "loss": 0.7409,
      "step": 10028
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.332738098931653,
      "learning_rate": 6.086732739642149e-06,
      "loss": 0.7563,
      "step": 10029
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.225133777397662,
      "learning_rate": 6.08602762567742e-06,
      "loss": 0.7404,
      "step": 10030
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.419400766468994,
      "learning_rate": 6.085322489044241e-06,
      "loss": 0.8044,
      "step": 10031
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.626995438997508,
      "learning_rate": 6.084617329757332e-06,
      "loss": 0.7529,
      "step": 10032
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.203065256208566,
      "learning_rate": 6.083912147831412e-06,
      "loss": 0.7891,
      "step": 10033
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.782370649653533,
      "learning_rate": 6.083206943281199e-06,
      "loss": 0.6985,
      "step": 10034
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.362973724616038,
      "learning_rate": 6.082501716121414e-06,
      "loss": 0.7801,
      "step": 10035
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.263373466010143,
      "learning_rate": 6.081796466366778e-06,
      "loss": 0.7308,
      "step": 10036
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.4120379684649045,
      "learning_rate": 6.081091194032007e-06,
      "loss": 0.7899,
      "step": 10037
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.829747760899516,
      "learning_rate": 6.0803858991318255e-06,
      "loss": 0.8223,
      "step": 10038
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.73773182302649,
      "learning_rate": 6.079680581680955e-06,
      "loss": 0.8008,
      "step": 10039
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.30934508061525,
      "learning_rate": 6.078975241694116e-06,
      "loss": 0.7545,
      "step": 10040
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.198314782251604,
      "learning_rate": 6.078269879186032e-06,
      "loss": 0.7324,
      "step": 10041
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.862874533805563,
      "learning_rate": 6.077564494171426e-06,
      "loss": 0.7197,
      "step": 10042
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.860026868749205,
      "learning_rate": 6.076859086665023e-06,
      "loss": 0.8082,
      "step": 10043
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.0187565234896345,
      "learning_rate": 6.076153656681541e-06,
      "loss": 0.7073,
      "step": 10044
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.301271178481069,
      "learning_rate": 6.07544820423571e-06,
      "loss": 0.8341,
      "step": 10045
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.354473595692216,
      "learning_rate": 6.074742729342254e-06,
      "loss": 0.7598,
      "step": 10046
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.178272236488297,
      "learning_rate": 6.074037232015895e-06,
      "loss": 0.7395,
      "step": 10047
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.905838041484551,
      "learning_rate": 6.073331712271361e-06,
      "loss": 0.7398,
      "step": 10048
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.502493954700063,
      "learning_rate": 6.072626170123379e-06,
      "loss": 0.7049,
      "step": 10049
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.917720054291424,
      "learning_rate": 6.071920605586676e-06,
      "loss": 0.7386,
      "step": 10050
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.264001219494541,
      "learning_rate": 6.071215018675975e-06,
      "loss": 0.7051,
      "step": 10051
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.441411026066102,
      "learning_rate": 6.0705094094060094e-06,
      "loss": 0.7068,
      "step": 10052
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.803366175328362,
      "learning_rate": 6.069803777791502e-06,
      "loss": 0.7277,
      "step": 10053
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.582451690344822,
      "learning_rate": 6.069098123847184e-06,
      "loss": 0.804,
      "step": 10054
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.067186353647767,
      "learning_rate": 6.0683924475877846e-06,
      "loss": 0.7401,
      "step": 10055
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.209597911305351,
      "learning_rate": 6.067686749028031e-06,
      "loss": 0.7778,
      "step": 10056
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.320557693865471,
      "learning_rate": 6.066981028182657e-06,
      "loss": 0.6987,
      "step": 10057
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.308111495323908,
      "learning_rate": 6.0662752850663895e-06,
      "loss": 0.721,
      "step": 10058
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.8280425356427985,
      "learning_rate": 6.0655695196939615e-06,
      "loss": 0.7324,
      "step": 10059
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.587302455941787,
      "learning_rate": 6.064863732080102e-06,
      "loss": 0.8344,
      "step": 10060
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.505566552592791,
      "learning_rate": 6.064157922239543e-06,
      "loss": 0.7773,
      "step": 10061
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.22513331125461,
      "learning_rate": 6.063452090187021e-06,
      "loss": 0.7809,
      "step": 10062
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.113180543984132,
      "learning_rate": 6.062746235937263e-06,
      "loss": 0.7741,
      "step": 10063
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.069250187426377,
      "learning_rate": 6.062040359505006e-06,
      "loss": 0.7135,
      "step": 10064
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.433675683390304,
      "learning_rate": 6.061334460904983e-06,
      "loss": 0.7618,
      "step": 10065
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.768941339917,
      "learning_rate": 6.060628540151926e-06,
      "loss": 0.7509,
      "step": 10066
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.150916963886978,
      "learning_rate": 6.05992259726057e-06,
      "loss": 0.755,
      "step": 10067
    },
    {
      "epoch": 0.9,
      "grad_norm": 8.215783877601135,
      "learning_rate": 6.059216632245654e-06,
      "loss": 0.8039,
      "step": 10068
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.499978799247384,
      "learning_rate": 6.058510645121908e-06,
      "loss": 0.8913,
      "step": 10069
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.9848393176263475,
      "learning_rate": 6.057804635904071e-06,
      "loss": 0.8056,
      "step": 10070
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.534488885163003,
      "learning_rate": 6.057098604606879e-06,
      "loss": 0.7829,
      "step": 10071
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.148841790775409,
      "learning_rate": 6.056392551245069e-06,
      "loss": 0.8182,
      "step": 10072
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.2844800462880475,
      "learning_rate": 6.0556864758333776e-06,
      "loss": 0.7231,
      "step": 10073
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.5931348799494085,
      "learning_rate": 6.054980378386544e-06,
      "loss": 0.7877,
      "step": 10074
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.627605700556318,
      "learning_rate": 6.054274258919303e-06,
      "loss": 0.7675,
      "step": 10075
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.930126989311527,
      "learning_rate": 6.053568117446397e-06,
      "loss": 0.7409,
      "step": 10076
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.781661996215399,
      "learning_rate": 6.052861953982564e-06,
      "loss": 0.7348,
      "step": 10077
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.482067947570703,
      "learning_rate": 6.052155768542542e-06,
      "loss": 0.7827,
      "step": 10078
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.747027824749389,
      "learning_rate": 6.051449561141074e-06,
      "loss": 0.7234,
      "step": 10079
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.948424052656772,
      "learning_rate": 6.050743331792898e-06,
      "loss": 0.7187,
      "step": 10080
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.755905530630202,
      "learning_rate": 6.0500370805127585e-06,
      "loss": 0.7673,
      "step": 10081
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.08734165175219,
      "learning_rate": 6.049330807315392e-06,
      "loss": 0.7175,
      "step": 10082
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.105004777273502,
      "learning_rate": 6.048624512215543e-06,
      "loss": 0.7786,
      "step": 10083
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.953459918845217,
      "learning_rate": 6.047918195227955e-06,
      "loss": 0.7721,
      "step": 10084
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.814288047102402,
      "learning_rate": 6.0472118563673685e-06,
      "loss": 0.7514,
      "step": 10085
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.560200332037439,
      "learning_rate": 6.046505495648529e-06,
      "loss": 0.8118,
      "step": 10086
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.69089871195553,
      "learning_rate": 6.045799113086179e-06,
      "loss": 0.8098,
      "step": 10087
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.671740650771017,
      "learning_rate": 6.045092708695062e-06,
      "loss": 0.7746,
      "step": 10088
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.356871446898439,
      "learning_rate": 6.044386282489922e-06,
      "loss": 0.7609,
      "step": 10089
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.351122634515333,
      "learning_rate": 6.043679834485509e-06,
      "loss": 0.8129,
      "step": 10090
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.199569568170179,
      "learning_rate": 6.0429733646965626e-06,
      "loss": 0.7708,
      "step": 10091
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.1130480560467335,
      "learning_rate": 6.042266873137831e-06,
      "loss": 0.7062,
      "step": 10092
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.068170327086019,
      "learning_rate": 6.041560359824062e-06,
      "loss": 0.8145,
      "step": 10093
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.146578788201856,
      "learning_rate": 6.040853824770002e-06,
      "loss": 0.6769,
      "step": 10094
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.769983583827237,
      "learning_rate": 6.040147267990397e-06,
      "loss": 0.7163,
      "step": 10095
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.923191879051833,
      "learning_rate": 6.039440689499996e-06,
      "loss": 0.761,
      "step": 10096
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.747811911644929,
      "learning_rate": 6.038734089313547e-06,
      "loss": 0.8004,
      "step": 10097
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.952685061465306,
      "learning_rate": 6.038027467445797e-06,
      "loss": 0.7147,
      "step": 10098
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.3895768598002585,
      "learning_rate": 6.037320823911498e-06,
      "loss": 0.7682,
      "step": 10099
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.789455252931676,
      "learning_rate": 6.0366141587254005e-06,
      "loss": 0.7475,
      "step": 10100
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.172091641545261,
      "learning_rate": 6.03590747190225e-06,
      "loss": 0.7418,
      "step": 10101
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.726220343871182,
      "learning_rate": 6.035200763456801e-06,
      "loss": 0.8353,
      "step": 10102
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.134183284480216,
      "learning_rate": 6.034494033403803e-06,
      "loss": 0.7338,
      "step": 10103
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.123594598953369,
      "learning_rate": 6.033787281758007e-06,
      "loss": 0.7079,
      "step": 10104
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.209517733680748,
      "learning_rate": 6.033080508534165e-06,
      "loss": 0.6924,
      "step": 10105
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.332955005841327,
      "learning_rate": 6.03237371374703e-06,
      "loss": 0.7834,
      "step": 10106
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.775242489923759,
      "learning_rate": 6.031666897411355e-06,
      "loss": 0.7341,
      "step": 10107
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.4851396099760334,
      "learning_rate": 6.030960059541894e-06,
      "loss": 0.7067,
      "step": 10108
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.097324463878127,
      "learning_rate": 6.030253200153398e-06,
      "loss": 0.7683,
      "step": 10109
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.273093521053382,
      "learning_rate": 6.029546319260623e-06,
      "loss": 0.7913,
      "step": 10110
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.821081694592558,
      "learning_rate": 6.028839416878323e-06,
      "loss": 0.769,
      "step": 10111
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.641786314944227,
      "learning_rate": 6.028132493021254e-06,
      "loss": 0.7347,
      "step": 10112
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.617135824502022,
      "learning_rate": 6.0274255477041695e-06,
      "loss": 0.687,
      "step": 10113
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.324514969329878,
      "learning_rate": 6.026718580941827e-06,
      "loss": 0.7408,
      "step": 10114
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.048135206116677,
      "learning_rate": 6.026011592748984e-06,
      "loss": 0.8175,
      "step": 10115
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.120532959706313,
      "learning_rate": 6.025304583140396e-06,
      "loss": 0.7004,
      "step": 10116
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.281929852680507,
      "learning_rate": 6.024597552130818e-06,
      "loss": 0.8099,
      "step": 10117
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.3943211602964,
      "learning_rate": 6.0238904997350116e-06,
      "loss": 0.7829,
      "step": 10118
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.920056221303454,
      "learning_rate": 6.023183425967732e-06,
      "loss": 0.8205,
      "step": 10119
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.54667145987195,
      "learning_rate": 6.02247633084374e-06,
      "loss": 0.7261,
      "step": 10120
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.845720934551642,
      "learning_rate": 6.021769214377794e-06,
      "loss": 0.771,
      "step": 10121
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.4836455224539264,
      "learning_rate": 6.021062076584654e-06,
      "loss": 0.7356,
      "step": 10122
    },
    {
      "epoch": 0.9,
      "grad_norm": 9.641724845102303,
      "learning_rate": 6.0203549174790774e-06,
      "loss": 0.79,
      "step": 10123
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.749782728503014,
      "learning_rate": 6.019647737075828e-06,
      "loss": 0.7644,
      "step": 10124
    },
    {
      "epoch": 0.9,
      "grad_norm": 8.008392028997855,
      "learning_rate": 6.018940535389664e-06,
      "loss": 0.8416,
      "step": 10125
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.698032796611498,
      "learning_rate": 6.018233312435349e-06,
      "loss": 0.7558,
      "step": 10126
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.312920975506399,
      "learning_rate": 6.0175260682276415e-06,
      "loss": 0.8431,
      "step": 10127
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.772987394100591,
      "learning_rate": 6.016818802781308e-06,
      "loss": 0.737,
      "step": 10128
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.973141706310906,
      "learning_rate": 6.016111516111108e-06,
      "loss": 0.71,
      "step": 10129
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.420661799350445,
      "learning_rate": 6.015404208231804e-06,
      "loss": 0.7971,
      "step": 10130
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.2295498432948975,
      "learning_rate": 6.014696879158162e-06,
      "loss": 0.7579,
      "step": 10131
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.018855134838867,
      "learning_rate": 6.013989528904944e-06,
      "loss": 0.7167,
      "step": 10132
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.110255160365863,
      "learning_rate": 6.013282157486916e-06,
      "loss": 0.7215,
      "step": 10133
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.166774050905314,
      "learning_rate": 6.012574764918841e-06,
      "loss": 0.7224,
      "step": 10134
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.281741455106094,
      "learning_rate": 6.011867351215487e-06,
      "loss": 0.7208,
      "step": 10135
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.903080868371061,
      "learning_rate": 6.011159916391618e-06,
      "loss": 0.7718,
      "step": 10136
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.43860259129649,
      "learning_rate": 6.010452460461998e-06,
      "loss": 0.8073,
      "step": 10137
    },
    {
      "epoch": 0.9,
      "grad_norm": 8.54332453093333,
      "learning_rate": 6.009744983441398e-06,
      "loss": 0.7278,
      "step": 10138
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.9317512724742105,
      "learning_rate": 6.009037485344582e-06,
      "loss": 0.7973,
      "step": 10139
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.898519067128321,
      "learning_rate": 6.008329966186317e-06,
      "loss": 0.7935,
      "step": 10140
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.841711738682992,
      "learning_rate": 6.007622425981374e-06,
      "loss": 0.785,
      "step": 10141
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.9822828056981296,
      "learning_rate": 6.006914864744517e-06,
      "loss": 0.7687,
      "step": 10142
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.019336005640037,
      "learning_rate": 6.0062072824905195e-06,
      "loss": 0.8059,
      "step": 10143
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.082063576322833,
      "learning_rate": 6.005499679234149e-06,
      "loss": 0.7934,
      "step": 10144
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.408641323200552,
      "learning_rate": 6.004792054990174e-06,
      "loss": 0.7944,
      "step": 10145
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.222675246254403,
      "learning_rate": 6.004084409773364e-06,
      "loss": 0.7249,
      "step": 10146
    },
    {
      "epoch": 0.91,
      "grad_norm": 8.601163465490684,
      "learning_rate": 6.003376743598492e-06,
      "loss": 0.7604,
      "step": 10147
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.6605740215032645,
      "learning_rate": 6.002669056480328e-06,
      "loss": 0.7687,
      "step": 10148
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.126747031031846,
      "learning_rate": 6.001961348433642e-06,
      "loss": 0.8262,
      "step": 10149
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.048416241417372,
      "learning_rate": 6.001253619473209e-06,
      "loss": 0.7466,
      "step": 10150
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.8441006645314575,
      "learning_rate": 6.0005458696138e-06,
      "loss": 0.7948,
      "step": 10151
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.184457871113128,
      "learning_rate": 5.9998380988701845e-06,
      "loss": 0.8017,
      "step": 10152
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.16397216425896,
      "learning_rate": 5.9991303072571395e-06,
      "loss": 0.7721,
      "step": 10153
    },
    {
      "epoch": 0.91,
      "grad_norm": 8.341066575556022,
      "learning_rate": 5.998422494789438e-06,
      "loss": 0.8287,
      "step": 10154
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.482470107070238,
      "learning_rate": 5.9977146614818515e-06,
      "loss": 0.7623,
      "step": 10155
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.344598088749769,
      "learning_rate": 5.997006807349157e-06,
      "loss": 0.7679,
      "step": 10156
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.635837145415517,
      "learning_rate": 5.996298932406131e-06,
      "loss": 0.7421,
      "step": 10157
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.3405981021877285,
      "learning_rate": 5.995591036667545e-06,
      "loss": 0.8404,
      "step": 10158
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.93853278804704,
      "learning_rate": 5.994883120148176e-06,
      "loss": 0.7526,
      "step": 10159
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.0336639916880825,
      "learning_rate": 5.994175182862802e-06,
      "loss": 0.7239,
      "step": 10160
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.984680383713284,
      "learning_rate": 5.993467224826197e-06,
      "loss": 0.8013,
      "step": 10161
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.8556186652738695,
      "learning_rate": 5.992759246053139e-06,
      "loss": 0.7894,
      "step": 10162
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.534400100311656,
      "learning_rate": 5.992051246558407e-06,
      "loss": 0.7504,
      "step": 10163
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.124377546536591,
      "learning_rate": 5.9913432263567775e-06,
      "loss": 0.7599,
      "step": 10164
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.065690904053198,
      "learning_rate": 5.99063518546303e-06,
      "loss": 0.7196,
      "step": 10165
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.67918373889295,
      "learning_rate": 5.9899271238919395e-06,
      "loss": 0.7879,
      "step": 10166
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.3733512130602765,
      "learning_rate": 5.989219041658292e-06,
      "loss": 0.7752,
      "step": 10167
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.793435821293688,
      "learning_rate": 5.9885109387768605e-06,
      "loss": 0.7392,
      "step": 10168
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.4255563103600104,
      "learning_rate": 5.987802815262429e-06,
      "loss": 0.7809,
      "step": 10169
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.544626023788492,
      "learning_rate": 5.987094671129777e-06,
      "loss": 0.7796,
      "step": 10170
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.250783963214054,
      "learning_rate": 5.986386506393686e-06,
      "loss": 0.7924,
      "step": 10171
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.401299321511758,
      "learning_rate": 5.9856783210689355e-06,
      "loss": 0.7409,
      "step": 10172
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.62978406185081,
      "learning_rate": 5.98497011517031e-06,
      "loss": 0.7235,
      "step": 10173
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.938909699381534,
      "learning_rate": 5.984261888712589e-06,
      "loss": 0.7152,
      "step": 10174
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.9423853534991755,
      "learning_rate": 5.983553641710558e-06,
      "loss": 0.8283,
      "step": 10175
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.625666816171898,
      "learning_rate": 5.982845374178999e-06,
      "loss": 0.7528,
      "step": 10176
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.849159932641711,
      "learning_rate": 5.9821370861326936e-06,
      "loss": 0.7741,
      "step": 10177
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.811759627358219,
      "learning_rate": 5.981428777586427e-06,
      "loss": 0.7736,
      "step": 10178
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.0867673195608605,
      "learning_rate": 5.980720448554985e-06,
      "loss": 0.8026,
      "step": 10179
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.871666937271671,
      "learning_rate": 5.980012099053152e-06,
      "loss": 0.7687,
      "step": 10180
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.550254771156421,
      "learning_rate": 5.979303729095711e-06,
      "loss": 0.67,
      "step": 10181
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.403168501484566,
      "learning_rate": 5.978595338697451e-06,
      "loss": 0.7745,
      "step": 10182
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.833699124739106,
      "learning_rate": 5.977886927873155e-06,
      "loss": 0.7829,
      "step": 10183
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.878447063134297,
      "learning_rate": 5.9771784966376126e-06,
      "loss": 0.8032,
      "step": 10184
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.964439574424301,
      "learning_rate": 5.9764700450056065e-06,
      "loss": 0.7753,
      "step": 10185
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.747248068486609,
      "learning_rate": 5.975761572991929e-06,
      "loss": 0.8838,
      "step": 10186
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.494419879989593,
      "learning_rate": 5.975053080611365e-06,
      "loss": 0.7595,
      "step": 10187
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.037945476084634,
      "learning_rate": 5.974344567878703e-06,
      "loss": 0.7639,
      "step": 10188
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.4617793844962454,
      "learning_rate": 5.973636034808732e-06,
      "loss": 0.7098,
      "step": 10189
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.9561752575729425,
      "learning_rate": 5.9729274814162405e-06,
      "loss": 0.7834,
      "step": 10190
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.5104831278262125,
      "learning_rate": 5.972218907716018e-06,
      "loss": 0.7463,
      "step": 10191
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.244261278841452,
      "learning_rate": 5.971510313722857e-06,
      "loss": 0.7724,
      "step": 10192
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.7579344364370275,
      "learning_rate": 5.970801699451544e-06,
      "loss": 0.7346,
      "step": 10193
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.147625337172605,
      "learning_rate": 5.970093064916873e-06,
      "loss": 0.7623,
      "step": 10194
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.815283855281739,
      "learning_rate": 5.969384410133632e-06,
      "loss": 0.7623,
      "step": 10195
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.979515260478697,
      "learning_rate": 5.968675735116616e-06,
      "loss": 0.734,
      "step": 10196
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.054534094317187,
      "learning_rate": 5.967967039880614e-06,
      "loss": 0.6948,
      "step": 10197
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.92188626744924,
      "learning_rate": 5.967258324440421e-06,
      "loss": 0.8388,
      "step": 10198
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.349287666226595,
      "learning_rate": 5.9665495888108285e-06,
      "loss": 0.7093,
      "step": 10199
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.999478698249883,
      "learning_rate": 5.965840833006629e-06,
      "loss": 0.72,
      "step": 10200
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.0080130748785034,
      "learning_rate": 5.965132057042619e-06,
      "loss": 0.7168,
      "step": 10201
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.213317350673265,
      "learning_rate": 5.9644232609335905e-06,
      "loss": 0.7541,
      "step": 10202
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.3889212633825885,
      "learning_rate": 5.963714444694337e-06,
      "loss": 0.7895,
      "step": 10203
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.769421033905793,
      "learning_rate": 5.963005608339656e-06,
      "loss": 0.831,
      "step": 10204
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.9197544387462315,
      "learning_rate": 5.9622967518843425e-06,
      "loss": 0.6931,
      "step": 10205
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.594457269245087,
      "learning_rate": 5.96158787534319e-06,
      "loss": 0.864,
      "step": 10206
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.875402121695587,
      "learning_rate": 5.960878978730998e-06,
      "loss": 0.8205,
      "step": 10207
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.233978370962491,
      "learning_rate": 5.96017006206256e-06,
      "loss": 0.7241,
      "step": 10208
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.839445752282132,
      "learning_rate": 5.959461125352677e-06,
      "loss": 0.7592,
      "step": 10209
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.600687005484644,
      "learning_rate": 5.958752168616142e-06,
      "loss": 0.7423,
      "step": 10210
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.248519781800196,
      "learning_rate": 5.958043191867757e-06,
      "loss": 0.8532,
      "step": 10211
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.440763487065326,
      "learning_rate": 5.957334195122316e-06,
      "loss": 0.8065,
      "step": 10212
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.273071068090524,
      "learning_rate": 5.956625178394622e-06,
      "loss": 0.8017,
      "step": 10213
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.3393760530749175,
      "learning_rate": 5.955916141699473e-06,
      "loss": 0.7366,
      "step": 10214
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.009460421139024,
      "learning_rate": 5.955207085051666e-06,
      "loss": 0.7754,
      "step": 10215
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.24644008568989,
      "learning_rate": 5.954498008466004e-06,
      "loss": 0.7384,
      "step": 10216
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.5411862043126785,
      "learning_rate": 5.953788911957287e-06,
      "loss": 0.7487,
      "step": 10217
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.044253759104281,
      "learning_rate": 5.953079795540314e-06,
      "loss": 0.8138,
      "step": 10218
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.822614510412595,
      "learning_rate": 5.952370659229888e-06,
      "loss": 0.7711,
      "step": 10219
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.139392977134978,
      "learning_rate": 5.95166150304081e-06,
      "loss": 0.7674,
      "step": 10220
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.716702171710714,
      "learning_rate": 5.950952326987883e-06,
      "loss": 0.7662,
      "step": 10221
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.607393314711142,
      "learning_rate": 5.950243131085908e-06,
      "loss": 0.713,
      "step": 10222
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.698212825370309,
      "learning_rate": 5.949533915349689e-06,
      "loss": 0.7416,
      "step": 10223
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.901374933639409,
      "learning_rate": 5.94882467979403e-06,
      "loss": 0.7657,
      "step": 10224
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.277682104149696,
      "learning_rate": 5.948115424433732e-06,
      "loss": 0.7382,
      "step": 10225
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.199364599579645,
      "learning_rate": 5.9474061492835995e-06,
      "loss": 0.7585,
      "step": 10226
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.743683224819009,
      "learning_rate": 5.9466968543584415e-06,
      "loss": 0.7377,
      "step": 10227
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.706162499848515,
      "learning_rate": 5.9459875396730584e-06,
      "loss": 0.8489,
      "step": 10228
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.93606856886517,
      "learning_rate": 5.945278205242257e-06,
      "loss": 0.7588,
      "step": 10229
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.667029038509037,
      "learning_rate": 5.944568851080843e-06,
      "loss": 0.7113,
      "step": 10230
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.681734489503555,
      "learning_rate": 5.9438594772036236e-06,
      "loss": 0.7126,
      "step": 10231
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.1548601584572475,
      "learning_rate": 5.943150083625404e-06,
      "loss": 0.733,
      "step": 10232
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.060351325497627,
      "learning_rate": 5.942440670360992e-06,
      "loss": 0.7322,
      "step": 10233
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.244535210468549,
      "learning_rate": 5.941731237425196e-06,
      "loss": 0.7439,
      "step": 10234
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.474595827267809,
      "learning_rate": 5.941021784832821e-06,
      "loss": 0.7821,
      "step": 10235
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.149365825171848,
      "learning_rate": 5.94031231259868e-06,
      "loss": 0.8077,
      "step": 10236
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.604554043797558,
      "learning_rate": 5.939602820737576e-06,
      "loss": 0.7793,
      "step": 10237
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.3463142247907935,
      "learning_rate": 5.938893309264321e-06,
      "loss": 0.7112,
      "step": 10238
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.279540793060562,
      "learning_rate": 5.9381837781937245e-06,
      "loss": 0.8195,
      "step": 10239
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.050913011049116,
      "learning_rate": 5.937474227540597e-06,
      "loss": 0.8252,
      "step": 10240
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.165954218964528,
      "learning_rate": 5.936764657319747e-06,
      "loss": 0.8221,
      "step": 10241
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.841602496829226,
      "learning_rate": 5.936055067545986e-06,
      "loss": 0.7006,
      "step": 10242
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.2423112063594655,
      "learning_rate": 5.935345458234128e-06,
      "loss": 0.7373,
      "step": 10243
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.393968280154461,
      "learning_rate": 5.934635829398979e-06,
      "loss": 0.7647,
      "step": 10244
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.733558035126318,
      "learning_rate": 5.933926181055355e-06,
      "loss": 0.7825,
      "step": 10245
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.698497496963701,
      "learning_rate": 5.933216513218069e-06,
      "loss": 0.7136,
      "step": 10246
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.165242828459354,
      "learning_rate": 5.932506825901929e-06,
      "loss": 0.7985,
      "step": 10247
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.5555363539972,
      "learning_rate": 5.931797119121752e-06,
      "loss": 0.8016,
      "step": 10248
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.121251574037313,
      "learning_rate": 5.931087392892352e-06,
      "loss": 0.7135,
      "step": 10249
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.5199037689258,
      "learning_rate": 5.93037764722854e-06,
      "loss": 0.7838,
      "step": 10250
    },
    {
      "epoch": 0.91,
      "grad_norm": 8.084462957977724,
      "learning_rate": 5.929667882145133e-06,
      "loss": 0.7718,
      "step": 10251
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.605646446882679,
      "learning_rate": 5.928958097656945e-06,
      "loss": 0.7214,
      "step": 10252
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.103123179590603,
      "learning_rate": 5.928248293778791e-06,
      "loss": 0.6932,
      "step": 10253
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.420309280541527,
      "learning_rate": 5.927538470525487e-06,
      "loss": 0.791,
      "step": 10254
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.433796437079402,
      "learning_rate": 5.926828627911847e-06,
      "loss": 0.7878,
      "step": 10255
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.614681843317519,
      "learning_rate": 5.926118765952691e-06,
      "loss": 0.7868,
      "step": 10256
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.721420884966442,
      "learning_rate": 5.9254088846628334e-06,
      "loss": 0.7851,
      "step": 10257
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.395996853961319,
      "learning_rate": 5.924698984057092e-06,
      "loss": 0.7292,
      "step": 10258
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.048717911283639,
      "learning_rate": 5.923989064150285e-06,
      "loss": 0.7391,
      "step": 10259
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.419366523033852,
      "learning_rate": 5.923279124957228e-06,
      "loss": 0.7213,
      "step": 10260
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.6388047724926,
      "learning_rate": 5.922569166492742e-06,
      "loss": 0.7582,
      "step": 10261
    },
    {
      "epoch": 0.92,
      "grad_norm": 11.252333126976932,
      "learning_rate": 5.921859188771647e-06,
      "loss": 0.8129,
      "step": 10262
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.1425472048752985,
      "learning_rate": 5.921149191808757e-06,
      "loss": 0.7949,
      "step": 10263
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.68021286840027,
      "learning_rate": 5.920439175618897e-06,
      "loss": 0.6935,
      "step": 10264
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.120487722410045,
      "learning_rate": 5.919729140216884e-06,
      "loss": 0.7978,
      "step": 10265
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.817397009794762,
      "learning_rate": 5.919019085617541e-06,
      "loss": 0.7118,
      "step": 10266
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.598504002989604,
      "learning_rate": 5.918309011835686e-06,
      "loss": 0.8079,
      "step": 10267
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.086787909908084,
      "learning_rate": 5.917598918886141e-06,
      "loss": 0.7574,
      "step": 10268
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.705525652651661,
      "learning_rate": 5.9168888067837286e-06,
      "loss": 0.6873,
      "step": 10269
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.1889257299656455,
      "learning_rate": 5.916178675543269e-06,
      "loss": 0.7859,
      "step": 10270
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.3052941778272125,
      "learning_rate": 5.915468525179588e-06,
      "loss": 0.733,
      "step": 10271
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.915900079386993,
      "learning_rate": 5.914758355707506e-06,
      "loss": 0.7145,
      "step": 10272
    },
    {
      "epoch": 0.92,
      "grad_norm": 7.680818432999958,
      "learning_rate": 5.914048167141846e-06,
      "loss": 0.7894,
      "step": 10273
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.0757167228610225,
      "learning_rate": 5.913337959497433e-06,
      "loss": 0.7202,
      "step": 10274
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.849375295542815,
      "learning_rate": 5.912627732789089e-06,
      "loss": 0.7753,
      "step": 10275
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.070136730501652,
      "learning_rate": 5.911917487031641e-06,
      "loss": 0.8111,
      "step": 10276
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.944433457796826,
      "learning_rate": 5.911207222239911e-06,
      "loss": 0.7748,
      "step": 10277
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.67943034409326,
      "learning_rate": 5.910496938428727e-06,
      "loss": 0.7819,
      "step": 10278
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.439583326947833,
      "learning_rate": 5.909786635612913e-06,
      "loss": 0.7274,
      "step": 10279
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.655363060879445,
      "learning_rate": 5.909076313807295e-06,
      "loss": 0.8355,
      "step": 10280
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.749378105623691,
      "learning_rate": 5.908365973026702e-06,
      "loss": 0.7852,
      "step": 10281
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.578349842518201,
      "learning_rate": 5.907655613285957e-06,
      "loss": 0.735,
      "step": 10282
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.94192012514465,
      "learning_rate": 5.906945234599887e-06,
      "loss": 0.7844,
      "step": 10283
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.744945539299703,
      "learning_rate": 5.906234836983323e-06,
      "loss": 0.8119,
      "step": 10284
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.687302574560532,
      "learning_rate": 5.905524420451092e-06,
      "loss": 0.7687,
      "step": 10285
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.429418575884567,
      "learning_rate": 5.904813985018021e-06,
      "loss": 0.7618,
      "step": 10286
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.366395482336362,
      "learning_rate": 5.904103530698939e-06,
      "loss": 0.786,
      "step": 10287
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.925739570850399,
      "learning_rate": 5.903393057508677e-06,
      "loss": 0.7678,
      "step": 10288
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.459284382438698,
      "learning_rate": 5.902682565462064e-06,
      "loss": 0.7215,
      "step": 10289
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.198719574093342,
      "learning_rate": 5.901972054573927e-06,
      "loss": 0.768,
      "step": 10290
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.199791313280977,
      "learning_rate": 5.9012615248591e-06,
      "loss": 0.8258,
      "step": 10291
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.7869140878290155,
      "learning_rate": 5.900550976332412e-06,
      "loss": 0.7307,
      "step": 10292
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.640913265627771,
      "learning_rate": 5.899840409008693e-06,
      "loss": 0.813,
      "step": 10293
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.434595488179786,
      "learning_rate": 5.8991298229027795e-06,
      "loss": 0.7795,
      "step": 10294
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.877006451601237,
      "learning_rate": 5.898419218029499e-06,
      "loss": 0.8245,
      "step": 10295
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.106576136554656,
      "learning_rate": 5.897708594403684e-06,
      "loss": 0.7578,
      "step": 10296
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.773713429721693,
      "learning_rate": 5.896997952040168e-06,
      "loss": 0.7496,
      "step": 10297
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.167837520183137,
      "learning_rate": 5.896287290953784e-06,
      "loss": 0.724,
      "step": 10298
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.750423640308995,
      "learning_rate": 5.895576611159365e-06,
      "loss": 0.7548,
      "step": 10299
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.6414621803823275,
      "learning_rate": 5.894865912671748e-06,
      "loss": 0.7867,
      "step": 10300
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.934873624403338,
      "learning_rate": 5.894155195505762e-06,
      "loss": 0.7436,
      "step": 10301
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.322468548155096,
      "learning_rate": 5.893444459676246e-06,
      "loss": 0.7069,
      "step": 10302
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.153724410587487,
      "learning_rate": 5.892733705198033e-06,
      "loss": 0.7747,
      "step": 10303
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.075851175729716,
      "learning_rate": 5.89202293208596e-06,
      "loss": 0.7383,
      "step": 10304
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.7816544760087005,
      "learning_rate": 5.8913121403548605e-06,
      "loss": 0.7307,
      "step": 10305
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.007082653587733,
      "learning_rate": 5.890601330019572e-06,
      "loss": 0.7555,
      "step": 10306
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.666168129069355,
      "learning_rate": 5.889890501094932e-06,
      "loss": 0.7232,
      "step": 10307
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.453098121153691,
      "learning_rate": 5.889179653595776e-06,
      "loss": 0.6662,
      "step": 10308
    },
    {
      "epoch": 0.92,
      "grad_norm": 8.686504383003271,
      "learning_rate": 5.888468787536943e-06,
      "loss": 0.7967,
      "step": 10309
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.367405327499712,
      "learning_rate": 5.887757902933269e-06,
      "loss": 0.7079,
      "step": 10310
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.755159719408923,
      "learning_rate": 5.887046999799593e-06,
      "loss": 0.8253,
      "step": 10311
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.897067130097319,
      "learning_rate": 5.886336078150753e-06,
      "loss": 0.7955,
      "step": 10312
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.654706902678017,
      "learning_rate": 5.885625138001589e-06,
      "loss": 0.7877,
      "step": 10313
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.383833474739498,
      "learning_rate": 5.88491417936694e-06,
      "loss": 0.7654,
      "step": 10314
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.075102691656701,
      "learning_rate": 5.884203202261643e-06,
      "loss": 0.797,
      "step": 10315
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.090295343243966,
      "learning_rate": 5.883492206700544e-06,
      "loss": 0.7241,
      "step": 10316
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.346853261343785,
      "learning_rate": 5.882781192698479e-06,
      "loss": 0.8664,
      "step": 10317
    },
    {
      "epoch": 0.92,
      "grad_norm": 7.910441799543443,
      "learning_rate": 5.882070160270289e-06,
      "loss": 0.8855,
      "step": 10318
    },
    {
      "epoch": 0.92,
      "grad_norm": 8.458027462217437,
      "learning_rate": 5.881359109430816e-06,
      "loss": 0.7845,
      "step": 10319
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.941103665928735,
      "learning_rate": 5.880648040194903e-06,
      "loss": 0.7134,
      "step": 10320
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.277356727217884,
      "learning_rate": 5.879936952577389e-06,
      "loss": 0.8185,
      "step": 10321
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.88626592785158,
      "learning_rate": 5.87922584659312e-06,
      "loss": 0.7747,
      "step": 10322
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.606761594443453,
      "learning_rate": 5.878514722256935e-06,
      "loss": 0.7821,
      "step": 10323
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.577610080341137,
      "learning_rate": 5.8778035795836805e-06,
      "loss": 0.8278,
      "step": 10324
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.787826382015443,
      "learning_rate": 5.8770924185881986e-06,
      "loss": 0.7681,
      "step": 10325
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.824005570876082,
      "learning_rate": 5.876381239285334e-06,
      "loss": 0.7989,
      "step": 10326
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.878096438500129,
      "learning_rate": 5.875670041689928e-06,
      "loss": 0.763,
      "step": 10327
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.568807387049608,
      "learning_rate": 5.8749588258168274e-06,
      "loss": 0.7465,
      "step": 10328
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.532008265704843,
      "learning_rate": 5.874247591680881e-06,
      "loss": 0.7503,
      "step": 10329
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.193955682249286,
      "learning_rate": 5.873536339296927e-06,
      "loss": 0.7827,
      "step": 10330
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.1295558430872426,
      "learning_rate": 5.872825068679817e-06,
      "loss": 0.7895,
      "step": 10331
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.9245282441763143,
      "learning_rate": 5.872113779844395e-06,
      "loss": 0.787,
      "step": 10332
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.129126951365239,
      "learning_rate": 5.871402472805507e-06,
      "loss": 0.7499,
      "step": 10333
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.728993177312453,
      "learning_rate": 5.870691147578001e-06,
      "loss": 0.7819,
      "step": 10334
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.197886358451952,
      "learning_rate": 5.869979804176724e-06,
      "loss": 0.7739,
      "step": 10335
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.3982869379694165,
      "learning_rate": 5.8692684426165235e-06,
      "loss": 0.8111,
      "step": 10336
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.148365410356932,
      "learning_rate": 5.868557062912248e-06,
      "loss": 0.7162,
      "step": 10337
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.114282040573901,
      "learning_rate": 5.867845665078746e-06,
      "loss": 0.7651,
      "step": 10338
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.402617446184281,
      "learning_rate": 5.8671342491308684e-06,
      "loss": 0.8049,
      "step": 10339
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.607720925592822,
      "learning_rate": 5.866422815083459e-06,
      "loss": 0.763,
      "step": 10340
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.5224531370962895,
      "learning_rate": 5.865711362951372e-06,
      "loss": 0.7817,
      "step": 10341
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.595438115496167,
      "learning_rate": 5.864999892749458e-06,
      "loss": 0.7785,
      "step": 10342
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.882187300794333,
      "learning_rate": 5.864288404492563e-06,
      "loss": 0.8399,
      "step": 10343
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.521319483182579,
      "learning_rate": 5.863576898195541e-06,
      "loss": 0.7402,
      "step": 10344
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.365883489702586,
      "learning_rate": 5.862865373873243e-06,
      "loss": 0.7596,
      "step": 10345
    },
    {
      "epoch": 0.92,
      "grad_norm": 9.4401196917089,
      "learning_rate": 5.86215383154052e-06,
      "loss": 0.7991,
      "step": 10346
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.932178646703814,
      "learning_rate": 5.861442271212223e-06,
      "loss": 0.8174,
      "step": 10347
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.907031448300558,
      "learning_rate": 5.860730692903207e-06,
      "loss": 0.8017,
      "step": 10348
    },
    {
      "epoch": 0.92,
      "grad_norm": 10.675850629778722,
      "learning_rate": 5.860019096628321e-06,
      "loss": 0.8396,
      "step": 10349
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.311565954651198,
      "learning_rate": 5.859307482402418e-06,
      "loss": 0.7717,
      "step": 10350
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.96821983252965,
      "learning_rate": 5.8585958502403576e-06,
      "loss": 0.7639,
      "step": 10351
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.241393107918069,
      "learning_rate": 5.857884200156985e-06,
      "loss": 0.8147,
      "step": 10352
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.586874388800151,
      "learning_rate": 5.857172532167162e-06,
      "loss": 0.7754,
      "step": 10353
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.719524147420923,
      "learning_rate": 5.8564608462857375e-06,
      "loss": 0.7503,
      "step": 10354
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.484857305083872,
      "learning_rate": 5.855749142527569e-06,
      "loss": 0.783,
      "step": 10355
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.2154446857985,
      "learning_rate": 5.855037420907511e-06,
      "loss": 0.7986,
      "step": 10356
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.8154011031568205,
      "learning_rate": 5.85432568144042e-06,
      "loss": 0.7204,
      "step": 10357
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.87853289994059,
      "learning_rate": 5.853613924141151e-06,
      "loss": 0.8086,
      "step": 10358
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.992968615896541,
      "learning_rate": 5.852902149024559e-06,
      "loss": 0.7373,
      "step": 10359
    },
    {
      "epoch": 0.92,
      "grad_norm": 7.200317375051821,
      "learning_rate": 5.8521903561055045e-06,
      "loss": 0.7661,
      "step": 10360
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.679027155574282,
      "learning_rate": 5.8514785453988435e-06,
      "loss": 0.7649,
      "step": 10361
    },
    {
      "epoch": 0.92,
      "grad_norm": 7.500005858373713,
      "learning_rate": 5.85076671691943e-06,
      "loss": 0.6932,
      "step": 10362
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.360151891061298,
      "learning_rate": 5.850054870682125e-06,
      "loss": 0.7837,
      "step": 10363
    },
    {
      "epoch": 0.92,
      "grad_norm": 8.225898483090898,
      "learning_rate": 5.849343006701787e-06,
      "loss": 0.8123,
      "step": 10364
    },
    {
      "epoch": 0.92,
      "grad_norm": 7.160769501992757,
      "learning_rate": 5.848631124993272e-06,
      "loss": 0.7747,
      "step": 10365
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.144609586022871,
      "learning_rate": 5.8479192255714424e-06,
      "loss": 0.7709,
      "step": 10366
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.605235779283128,
      "learning_rate": 5.847207308451157e-06,
      "loss": 0.7071,
      "step": 10367
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.887815086444653,
      "learning_rate": 5.846495373647273e-06,
      "loss": 0.7386,
      "step": 10368
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.785840600454714,
      "learning_rate": 5.845783421174651e-06,
      "loss": 0.785,
      "step": 10369
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.043759062248905,
      "learning_rate": 5.845071451048155e-06,
      "loss": 0.7723,
      "step": 10370
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.967429647991739,
      "learning_rate": 5.844359463282641e-06,
      "loss": 0.7819,
      "step": 10371
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.583230789410842,
      "learning_rate": 5.843647457892972e-06,
      "loss": 0.7868,
      "step": 10372
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.511121774507842,
      "learning_rate": 5.84293543489401e-06,
      "loss": 0.7167,
      "step": 10373
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.700681910852306,
      "learning_rate": 5.842223394300619e-06,
      "loss": 0.7354,
      "step": 10374
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.0420660209514425,
      "learning_rate": 5.841511336127657e-06,
      "loss": 0.7945,
      "step": 10375
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.746516795867465,
      "learning_rate": 5.840799260389988e-06,
      "loss": 0.758,
      "step": 10376
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.337950254921207,
      "learning_rate": 5.840087167102477e-06,
      "loss": 0.7016,
      "step": 10377
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.546522385332707,
      "learning_rate": 5.8393750562799856e-06,
      "loss": 0.7711,
      "step": 10378
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.988052959663414,
      "learning_rate": 5.838662927937379e-06,
      "loss": 0.7234,
      "step": 10379
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.630674204213422,
      "learning_rate": 5.83795078208952e-06,
      "loss": 0.7757,
      "step": 10380
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.362696298640356,
      "learning_rate": 5.837238618751273e-06,
      "loss": 0.7335,
      "step": 10381
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.003077476298915,
      "learning_rate": 5.8365264379375046e-06,
      "loss": 0.7131,
      "step": 10382
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.004186379932984,
      "learning_rate": 5.835814239663078e-06,
      "loss": 0.778,
      "step": 10383
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.532055525441049,
      "learning_rate": 5.83510202394286e-06,
      "loss": 0.7782,
      "step": 10384
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.905027598599512,
      "learning_rate": 5.834389790791715e-06,
      "loss": 0.7878,
      "step": 10385
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.072600511236851,
      "learning_rate": 5.833677540224512e-06,
      "loss": 0.6893,
      "step": 10386
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.679437753037188,
      "learning_rate": 5.832965272256113e-06,
      "loss": 0.7008,
      "step": 10387
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.198797708859641,
      "learning_rate": 5.832252986901391e-06,
      "loss": 0.7977,
      "step": 10388
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.296903660646669,
      "learning_rate": 5.83154068417521e-06,
      "loss": 0.7808,
      "step": 10389
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.249615796148007,
      "learning_rate": 5.830828364092439e-06,
      "loss": 0.7826,
      "step": 10390
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.491532383954514,
      "learning_rate": 5.830116026667944e-06,
      "loss": 0.7339,
      "step": 10391
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.614076751677338,
      "learning_rate": 5.829403671916595e-06,
      "loss": 0.8052,
      "step": 10392
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.352334955242177,
      "learning_rate": 5.828691299853261e-06,
      "loss": 0.7729,
      "step": 10393
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.400938430184248,
      "learning_rate": 5.8279789104928096e-06,
      "loss": 0.7433,
      "step": 10394
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.469780981972323,
      "learning_rate": 5.827266503850114e-06,
      "loss": 0.8114,
      "step": 10395
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.865486527662251,
      "learning_rate": 5.826554079940041e-06,
      "loss": 0.7663,
      "step": 10396
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.341372956952108,
      "learning_rate": 5.825841638777462e-06,
      "loss": 0.7748,
      "step": 10397
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.902952488388827,
      "learning_rate": 5.8251291803772455e-06,
      "loss": 0.7105,
      "step": 10398
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.9490741282528745,
      "learning_rate": 5.824416704754267e-06,
      "loss": 0.7878,
      "step": 10399
    },
    {
      "epoch": 0.93,
      "grad_norm": 8.250681407778234,
      "learning_rate": 5.823704211923393e-06,
      "loss": 0.7874,
      "step": 10400
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.356418186587142,
      "learning_rate": 5.822991701899498e-06,
      "loss": 0.732,
      "step": 10401
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.3271152827567745,
      "learning_rate": 5.822279174697454e-06,
      "loss": 0.7315,
      "step": 10402
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.646328934502949,
      "learning_rate": 5.821566630332134e-06,
      "loss": 0.7895,
      "step": 10403
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.210015052851324,
      "learning_rate": 5.820854068818408e-06,
      "loss": 0.7801,
      "step": 10404
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.840249579728816,
      "learning_rate": 5.820141490171153e-06,
      "loss": 0.7388,
      "step": 10405
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.815994395193933,
      "learning_rate": 5.819428894405238e-06,
      "loss": 0.7987,
      "step": 10406
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.696700030652165,
      "learning_rate": 5.81871628153554e-06,
      "loss": 0.7622,
      "step": 10407
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.078937007130367,
      "learning_rate": 5.8180036515769335e-06,
      "loss": 0.7205,
      "step": 10408
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.6266404057736725,
      "learning_rate": 5.817291004544291e-06,
      "loss": 0.7178,
      "step": 10409
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.675522912573433,
      "learning_rate": 5.81657834045249e-06,
      "loss": 0.6995,
      "step": 10410
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.737443268600538,
      "learning_rate": 5.815865659316403e-06,
      "loss": 0.7794,
      "step": 10411
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.963491265462475,
      "learning_rate": 5.81515296115091e-06,
      "loss": 0.712,
      "step": 10412
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.424731870726073,
      "learning_rate": 5.81444024597088e-06,
      "loss": 0.7485,
      "step": 10413
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.284566255419097,
      "learning_rate": 5.813727513791195e-06,
      "loss": 0.7887,
      "step": 10414
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.813756629636914,
      "learning_rate": 5.813014764626731e-06,
      "loss": 0.7415,
      "step": 10415
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.169518779441186,
      "learning_rate": 5.812301998492363e-06,
      "loss": 0.7024,
      "step": 10416
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.878204096309575,
      "learning_rate": 5.811589215402971e-06,
      "loss": 0.6919,
      "step": 10417
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.3674481267699745,
      "learning_rate": 5.810876415373432e-06,
      "loss": 0.7647,
      "step": 10418
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.26157827807808,
      "learning_rate": 5.8101635984186225e-06,
      "loss": 0.7848,
      "step": 10419
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.096138818245598,
      "learning_rate": 5.809450764553421e-06,
      "loss": 0.7042,
      "step": 10420
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.963258268220989,
      "learning_rate": 5.80873791379271e-06,
      "loss": 0.7555,
      "step": 10421
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.7714489449697535,
      "learning_rate": 5.8080250461513645e-06,
      "loss": 0.7995,
      "step": 10422
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.2466404249603755,
      "learning_rate": 5.807312161644265e-06,
      "loss": 0.7461,
      "step": 10423
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.824790290782588,
      "learning_rate": 5.8065992602862945e-06,
      "loss": 0.7765,
      "step": 10424
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.654907867118062,
      "learning_rate": 5.8058863420923295e-06,
      "loss": 0.7387,
      "step": 10425
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.185685958378814,
      "learning_rate": 5.805173407077252e-06,
      "loss": 0.7211,
      "step": 10426
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.466302318333001,
      "learning_rate": 5.804460455255942e-06,
      "loss": 0.7616,
      "step": 10427
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.432810267952034,
      "learning_rate": 5.803747486643283e-06,
      "loss": 0.758,
      "step": 10428
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.389295915621514,
      "learning_rate": 5.8030345012541555e-06,
      "loss": 0.7091,
      "step": 10429
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.205600665551974,
      "learning_rate": 5.80232149910344e-06,
      "loss": 0.8211,
      "step": 10430
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.737980131623514,
      "learning_rate": 5.801608480206022e-06,
      "loss": 0.7954,
      "step": 10431
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.6566184818689935,
      "learning_rate": 5.800895444576782e-06,
      "loss": 0.7735,
      "step": 10432
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.667167109611212,
      "learning_rate": 5.8001823922306035e-06,
      "loss": 0.7099,
      "step": 10433
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.342543516234341,
      "learning_rate": 5.7994693231823695e-06,
      "loss": 0.7219,
      "step": 10434
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.531444399605784,
      "learning_rate": 5.798756237446964e-06,
      "loss": 0.7519,
      "step": 10435
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.915885758139498,
      "learning_rate": 5.798043135039271e-06,
      "loss": 0.7483,
      "step": 10436
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.712418906822784,
      "learning_rate": 5.7973300159741765e-06,
      "loss": 0.7784,
      "step": 10437
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.614322535472201,
      "learning_rate": 5.796616880266564e-06,
      "loss": 0.7117,
      "step": 10438
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.469834109473515,
      "learning_rate": 5.795903727931318e-06,
      "loss": 0.6904,
      "step": 10439
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.729979880727099,
      "learning_rate": 5.795190558983324e-06,
      "loss": 0.7873,
      "step": 10440
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.331494419043117,
      "learning_rate": 5.794477373437469e-06,
      "loss": 0.7651,
      "step": 10441
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.22137797540707,
      "learning_rate": 5.793764171308639e-06,
      "loss": 0.8413,
      "step": 10442
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.120418604977993,
      "learning_rate": 5.793050952611721e-06,
      "loss": 0.7196,
      "step": 10443
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.475075280260225,
      "learning_rate": 5.7923377173616e-06,
      "loss": 0.7385,
      "step": 10444
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.426603852709179,
      "learning_rate": 5.791624465573164e-06,
      "loss": 0.7692,
      "step": 10445
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.417080677731547,
      "learning_rate": 5.790911197261302e-06,
      "loss": 0.7284,
      "step": 10446
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.698665207725203,
      "learning_rate": 5.790197912440901e-06,
      "loss": 0.7626,
      "step": 10447
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.391641658389334,
      "learning_rate": 5.789484611126848e-06,
      "loss": 0.7358,
      "step": 10448
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.269168455114803,
      "learning_rate": 5.788771293334032e-06,
      "loss": 0.7683,
      "step": 10449
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.558423382879241,
      "learning_rate": 5.788057959077345e-06,
      "loss": 0.7692,
      "step": 10450
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.051031672193766,
      "learning_rate": 5.787344608371673e-06,
      "loss": 0.7321,
      "step": 10451
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.6524552123877765,
      "learning_rate": 5.786631241231905e-06,
      "loss": 0.7861,
      "step": 10452
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.698684852641717,
      "learning_rate": 5.7859178576729345e-06,
      "loss": 0.8164,
      "step": 10453
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.115658972944113,
      "learning_rate": 5.78520445770965e-06,
      "loss": 0.7988,
      "step": 10454
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.386004473914074,
      "learning_rate": 5.784491041356941e-06,
      "loss": 0.7718,
      "step": 10455
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.764386847165355,
      "learning_rate": 5.7837776086297e-06,
      "loss": 0.7386,
      "step": 10456
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.432645313548101,
      "learning_rate": 5.7830641595428175e-06,
      "loss": 0.8342,
      "step": 10457
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.3431230473735996,
      "learning_rate": 5.782350694111185e-06,
      "loss": 0.7182,
      "step": 10458
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.1839031329482665,
      "learning_rate": 5.781637212349696e-06,
      "loss": 0.8192,
      "step": 10459
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.108840070426629,
      "learning_rate": 5.780923714273241e-06,
      "loss": 0.7384,
      "step": 10460
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.567668280231709,
      "learning_rate": 5.780210199896715e-06,
      "loss": 0.7837,
      "step": 10461
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.721910508866932,
      "learning_rate": 5.7794966692350084e-06,
      "loss": 0.767,
      "step": 10462
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.507984346704366,
      "learning_rate": 5.778783122303018e-06,
      "loss": 0.7035,
      "step": 10463
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.205967295892064,
      "learning_rate": 5.778069559115632e-06,
      "loss": 0.7672,
      "step": 10464
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.82910086983143,
      "learning_rate": 5.77735597968775e-06,
      "loss": 0.7335,
      "step": 10465
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.309133983890882,
      "learning_rate": 5.7766423840342645e-06,
      "loss": 0.6998,
      "step": 10466
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.192800778588366,
      "learning_rate": 5.77592877217007e-06,
      "loss": 0.8045,
      "step": 10467
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.317274183117621,
      "learning_rate": 5.775215144110061e-06,
      "loss": 0.7715,
      "step": 10468
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.045096418471563,
      "learning_rate": 5.774501499869135e-06,
      "loss": 0.7734,
      "step": 10469
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.388706476378973,
      "learning_rate": 5.773787839462184e-06,
      "loss": 0.7175,
      "step": 10470
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.228926406268309,
      "learning_rate": 5.773074162904108e-06,
      "loss": 0.68,
      "step": 10471
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.050877007613123,
      "learning_rate": 5.7723604702098015e-06,
      "loss": 0.737,
      "step": 10472
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.228962791106061,
      "learning_rate": 5.7716467613941605e-06,
      "loss": 0.7774,
      "step": 10473
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.279199897197876,
      "learning_rate": 5.770933036472084e-06,
      "loss": 0.7974,
      "step": 10474
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.022123019130507,
      "learning_rate": 5.77021929545847e-06,
      "loss": 0.775,
      "step": 10475
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.630805213328504,
      "learning_rate": 5.7695055383682145e-06,
      "loss": 0.7517,
      "step": 10476
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.371124191348364,
      "learning_rate": 5.768791765216215e-06,
      "loss": 0.7599,
      "step": 10477
    },
    {
      "epoch": 0.93,
      "grad_norm": 8.15312427864812,
      "learning_rate": 5.768077976017371e-06,
      "loss": 0.8376,
      "step": 10478
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.484360755592375,
      "learning_rate": 5.767364170786582e-06,
      "loss": 0.7217,
      "step": 10479
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.199025843030411,
      "learning_rate": 5.766650349538747e-06,
      "loss": 0.7595,
      "step": 10480
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.629998480747124,
      "learning_rate": 5.765936512288764e-06,
      "loss": 0.7992,
      "step": 10481
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.2042431878974815,
      "learning_rate": 5.765222659051534e-06,
      "loss": 0.7108,
      "step": 10482
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.784961416878586,
      "learning_rate": 5.7645087898419576e-06,
      "loss": 0.7429,
      "step": 10483
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.266198666586149,
      "learning_rate": 5.763794904674933e-06,
      "loss": 0.7676,
      "step": 10484
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.9603124583250775,
      "learning_rate": 5.763081003565363e-06,
      "loss": 0.793,
      "step": 10485
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.11653436756417,
      "learning_rate": 5.762367086528147e-06,
      "loss": 0.7242,
      "step": 10486
    },
    {
      "epoch": 0.94,
      "grad_norm": 22.369367807618673,
      "learning_rate": 5.76165315357819e-06,
      "loss": 0.7952,
      "step": 10487
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.818259833216247,
      "learning_rate": 5.76093920473039e-06,
      "loss": 0.8467,
      "step": 10488
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.391197941907862,
      "learning_rate": 5.760225239999651e-06,
      "loss": 0.7586,
      "step": 10489
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.580002687958734,
      "learning_rate": 5.759511259400874e-06,
      "loss": 0.7375,
      "step": 10490
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.239952941872122,
      "learning_rate": 5.758797262948964e-06,
      "loss": 0.7908,
      "step": 10491
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.525852003762492,
      "learning_rate": 5.758083250658822e-06,
      "loss": 0.6977,
      "step": 10492
    },
    {
      "epoch": 0.94,
      "grad_norm": 8.217196421494338,
      "learning_rate": 5.7573692225453525e-06,
      "loss": 0.7887,
      "step": 10493
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.319922048149324,
      "learning_rate": 5.75665517862346e-06,
      "loss": 0.8497,
      "step": 10494
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.742060979365445,
      "learning_rate": 5.755941118908046e-06,
      "loss": 0.6583,
      "step": 10495
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.92379202557174,
      "learning_rate": 5.755227043414018e-06,
      "loss": 0.7534,
      "step": 10496
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.772748880033975,
      "learning_rate": 5.754512952156279e-06,
      "loss": 0.7185,
      "step": 10497
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.381573660971025,
      "learning_rate": 5.753798845149735e-06,
      "loss": 0.6913,
      "step": 10498
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.2102226712117465,
      "learning_rate": 5.75308472240929e-06,
      "loss": 0.8069,
      "step": 10499
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.206287080927273,
      "learning_rate": 5.75237058394985e-06,
      "loss": 0.7217,
      "step": 10500
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.080906558801783,
      "learning_rate": 5.751656429786323e-06,
      "loss": 0.7787,
      "step": 10501
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.3863757203362495,
      "learning_rate": 5.750942259933614e-06,
      "loss": 0.7339,
      "step": 10502
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.872315370986401,
      "learning_rate": 5.750228074406629e-06,
      "loss": 0.7529,
      "step": 10503
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.709167223722633,
      "learning_rate": 5.749513873220278e-06,
      "loss": 0.8074,
      "step": 10504
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.800341003602659,
      "learning_rate": 5.748799656389464e-06,
      "loss": 0.7343,
      "step": 10505
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.3444239839321614,
      "learning_rate": 5.748085423929099e-06,
      "loss": 0.7344,
      "step": 10506
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.000115820171544,
      "learning_rate": 5.747371175854087e-06,
      "loss": 0.7568,
      "step": 10507
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.226268722913593,
      "learning_rate": 5.746656912179339e-06,
      "loss": 0.7195,
      "step": 10508
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.932709708163412,
      "learning_rate": 5.745942632919763e-06,
      "loss": 0.7649,
      "step": 10509
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.178137859378232,
      "learning_rate": 5.74522833809027e-06,
      "loss": 0.7484,
      "step": 10510
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.8050701394726456,
      "learning_rate": 5.7445140277057665e-06,
      "loss": 0.7192,
      "step": 10511
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.136706071830449,
      "learning_rate": 5.743799701781162e-06,
      "loss": 0.7847,
      "step": 10512
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.160541403908818,
      "learning_rate": 5.743085360331368e-06,
      "loss": 0.6795,
      "step": 10513
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.563082121575759,
      "learning_rate": 5.7423710033712954e-06,
      "loss": 0.7529,
      "step": 10514
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.5844129041905495,
      "learning_rate": 5.741656630915853e-06,
      "loss": 0.7439,
      "step": 10515
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.102688210838052,
      "learning_rate": 5.740942242979952e-06,
      "loss": 0.8093,
      "step": 10516
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.477879444858365,
      "learning_rate": 5.740227839578506e-06,
      "loss": 0.7103,
      "step": 10517
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.363358285552373,
      "learning_rate": 5.739513420726424e-06,
      "loss": 0.7279,
      "step": 10518
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.665614128777816,
      "learning_rate": 5.738798986438619e-06,
      "loss": 0.7784,
      "step": 10519
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.0594426197810725,
      "learning_rate": 5.738084536730004e-06,
      "loss": 0.6948,
      "step": 10520
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.280316813783569,
      "learning_rate": 5.737370071615489e-06,
      "loss": 0.7661,
      "step": 10521
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.523339458968434,
      "learning_rate": 5.736655591109989e-06,
      "loss": 0.6888,
      "step": 10522
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.3015696602877025,
      "learning_rate": 5.735941095228418e-06,
      "loss": 0.7647,
      "step": 10523
    },
    {
      "epoch": 0.94,
      "grad_norm": 10.444392859702623,
      "learning_rate": 5.735226583985687e-06,
      "loss": 0.776,
      "step": 10524
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.751741781566641,
      "learning_rate": 5.73451205739671e-06,
      "loss": 0.729,
      "step": 10525
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.234003258918477,
      "learning_rate": 5.733797515476406e-06,
      "loss": 0.8208,
      "step": 10526
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.335600175065065,
      "learning_rate": 5.733082958239683e-06,
      "loss": 0.6749,
      "step": 10527
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.3797299214855965,
      "learning_rate": 5.732368385701459e-06,
      "loss": 0.7032,
      "step": 10528
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.048224864775324,
      "learning_rate": 5.731653797876651e-06,
      "loss": 0.7911,
      "step": 10529
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.062843288885905,
      "learning_rate": 5.73093919478017e-06,
      "loss": 0.746,
      "step": 10530
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.857003108063851,
      "learning_rate": 5.730224576426933e-06,
      "loss": 0.784,
      "step": 10531
    },
    {
      "epoch": 0.94,
      "grad_norm": 8.482367323577703,
      "learning_rate": 5.729509942831859e-06,
      "loss": 0.8381,
      "step": 10532
    },
    {
      "epoch": 0.94,
      "grad_norm": 9.487671896020563,
      "learning_rate": 5.728795294009863e-06,
      "loss": 0.7908,
      "step": 10533
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.513544083711457,
      "learning_rate": 5.728080629975861e-06,
      "loss": 0.7387,
      "step": 10534
    },
    {
      "epoch": 0.94,
      "grad_norm": 8.663195932943987,
      "learning_rate": 5.727365950744769e-06,
      "loss": 0.7865,
      "step": 10535
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.039117849878281,
      "learning_rate": 5.726651256331507e-06,
      "loss": 0.7133,
      "step": 10536
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.17767092428578,
      "learning_rate": 5.7259365467509895e-06,
      "loss": 0.6903,
      "step": 10537
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.808951258299627,
      "learning_rate": 5.725221822018138e-06,
      "loss": 0.8201,
      "step": 10538
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.210869802194983,
      "learning_rate": 5.72450708214787e-06,
      "loss": 0.7746,
      "step": 10539
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.317890573755662,
      "learning_rate": 5.723792327155102e-06,
      "loss": 0.8016,
      "step": 10540
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.67476286701007,
      "learning_rate": 5.723077557054755e-06,
      "loss": 0.8007,
      "step": 10541
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.792551488976828,
      "learning_rate": 5.722362771861747e-06,
      "loss": 0.7748,
      "step": 10542
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.314475116960023,
      "learning_rate": 5.721647971590998e-06,
      "loss": 0.6972,
      "step": 10543
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.70319239117721,
      "learning_rate": 5.720933156257427e-06,
      "loss": 0.7381,
      "step": 10544
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.408937876887455,
      "learning_rate": 5.720218325875957e-06,
      "loss": 0.709,
      "step": 10545
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.293910358916889,
      "learning_rate": 5.719503480461506e-06,
      "loss": 0.8367,
      "step": 10546
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.853258674229003,
      "learning_rate": 5.718788620028996e-06,
      "loss": 0.725,
      "step": 10547
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.847736954241616,
      "learning_rate": 5.7180737445933466e-06,
      "loss": 0.8204,
      "step": 10548
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.394800935294825,
      "learning_rate": 5.717358854169482e-06,
      "loss": 0.7974,
      "step": 10549
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.696083837242662,
      "learning_rate": 5.716643948772319e-06,
      "loss": 0.6854,
      "step": 10550
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.157388859027308,
      "learning_rate": 5.715929028416785e-06,
      "loss": 0.7778,
      "step": 10551
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.772259387156887,
      "learning_rate": 5.715214093117801e-06,
      "loss": 0.7737,
      "step": 10552
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.806732296966844,
      "learning_rate": 5.714499142890288e-06,
      "loss": 0.7107,
      "step": 10553
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.674795966904009,
      "learning_rate": 5.713784177749171e-06,
      "loss": 0.7881,
      "step": 10554
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.957508401894398,
      "learning_rate": 5.7130691977093735e-06,
      "loss": 0.7355,
      "step": 10555
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.514264089060806,
      "learning_rate": 5.712354202785817e-06,
      "loss": 0.834,
      "step": 10556
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.9821461537906115,
      "learning_rate": 5.711639192993426e-06,
      "loss": 0.6489,
      "step": 10557
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.761887423669251,
      "learning_rate": 5.710924168347125e-06,
      "loss": 0.7079,
      "step": 10558
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.4834057370897265,
      "learning_rate": 5.71020912886184e-06,
      "loss": 0.744,
      "step": 10559
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.150207667374416,
      "learning_rate": 5.709494074552493e-06,
      "loss": 0.6911,
      "step": 10560
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.637005324682461,
      "learning_rate": 5.708779005434013e-06,
      "loss": 0.7538,
      "step": 10561
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.537409106199605,
      "learning_rate": 5.708063921521324e-06,
      "loss": 0.7976,
      "step": 10562
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.551660941669284,
      "learning_rate": 5.707348822829349e-06,
      "loss": 0.8054,
      "step": 10563
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.293912815037526,
      "learning_rate": 5.706633709373019e-06,
      "loss": 0.7274,
      "step": 10564
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.825203256092446,
      "learning_rate": 5.7059185811672545e-06,
      "loss": 0.7935,
      "step": 10565
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.266956992963532,
      "learning_rate": 5.705203438226987e-06,
      "loss": 0.7379,
      "step": 10566
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.319227347685753,
      "learning_rate": 5.704488280567143e-06,
      "loss": 0.7465,
      "step": 10567
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.6613945429889485,
      "learning_rate": 5.703773108202648e-06,
      "loss": 0.778,
      "step": 10568
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.601466357272401,
      "learning_rate": 5.703057921148431e-06,
      "loss": 0.8086,
      "step": 10569
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.922661586074919,
      "learning_rate": 5.7023427194194195e-06,
      "loss": 0.7085,
      "step": 10570
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.7602596428269757,
      "learning_rate": 5.701627503030542e-06,
      "loss": 0.7649,
      "step": 10571
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.153425383996962,
      "learning_rate": 5.700912271996727e-06,
      "loss": 0.7709,
      "step": 10572
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.24851704273224,
      "learning_rate": 5.700197026332902e-06,
      "loss": 0.761,
      "step": 10573
    },
    {
      "epoch": 0.94,
      "grad_norm": 9.183066525520982,
      "learning_rate": 5.6994817660539996e-06,
      "loss": 0.6776,
      "step": 10574
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.0555538943175256,
      "learning_rate": 5.698766491174945e-06,
      "loss": 0.7505,
      "step": 10575
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.5741315011178925,
      "learning_rate": 5.6980512017106725e-06,
      "loss": 0.7749,
      "step": 10576
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.285887509299367,
      "learning_rate": 5.697335897676108e-06,
      "loss": 0.7664,
      "step": 10577
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.67133417229361,
      "learning_rate": 5.696620579086185e-06,
      "loss": 0.7809,
      "step": 10578
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.606709836009858,
      "learning_rate": 5.695905245955832e-06,
      "loss": 0.7171,
      "step": 10579
    },
    {
      "epoch": 0.94,
      "grad_norm": 20.44308600593045,
      "learning_rate": 5.695189898299982e-06,
      "loss": 0.7642,
      "step": 10580
    },
    {
      "epoch": 0.94,
      "grad_norm": 8.125027478798033,
      "learning_rate": 5.694474536133564e-06,
      "loss": 0.7711,
      "step": 10581
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.928184912893852,
      "learning_rate": 5.69375915947151e-06,
      "loss": 0.7122,
      "step": 10582
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.79868389918157,
      "learning_rate": 5.693043768328756e-06,
      "loss": 0.6725,
      "step": 10583
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.056637229007533,
      "learning_rate": 5.6923283627202294e-06,
      "loss": 0.7245,
      "step": 10584
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.1340321551556825,
      "learning_rate": 5.691612942660865e-06,
      "loss": 0.8587,
      "step": 10585
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.6433615407927045,
      "learning_rate": 5.690897508165594e-06,
      "loss": 0.8216,
      "step": 10586
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.649372504963632,
      "learning_rate": 5.690182059249352e-06,
      "loss": 0.6997,
      "step": 10587
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.459925966193677,
      "learning_rate": 5.689466595927069e-06,
      "loss": 0.7302,
      "step": 10588
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.938044694976502,
      "learning_rate": 5.688751118213681e-06,
      "loss": 0.7801,
      "step": 10589
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.460728672945716,
      "learning_rate": 5.688035626124124e-06,
      "loss": 0.7501,
      "step": 10590
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.049578428352045,
      "learning_rate": 5.687320119673329e-06,
      "loss": 0.7499,
      "step": 10591
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.591645689084926,
      "learning_rate": 5.686604598876231e-06,
      "loss": 0.7803,
      "step": 10592
    },
    {
      "epoch": 0.95,
      "grad_norm": 8.138861938854694,
      "learning_rate": 5.685889063747767e-06,
      "loss": 0.7889,
      "step": 10593
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.8647485309529825,
      "learning_rate": 5.685173514302869e-06,
      "loss": 0.8415,
      "step": 10594
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.001790642676916,
      "learning_rate": 5.684457950556475e-06,
      "loss": 0.8442,
      "step": 10595
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.864996972665924,
      "learning_rate": 5.683742372523521e-06,
      "loss": 0.7822,
      "step": 10596
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.001483549320294,
      "learning_rate": 5.683026780218941e-06,
      "loss": 0.8154,
      "step": 10597
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.543924225639487,
      "learning_rate": 5.682311173657674e-06,
      "loss": 0.7392,
      "step": 10598
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.473060235350142,
      "learning_rate": 5.681595552854655e-06,
      "loss": 0.8487,
      "step": 10599
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.227635742922879,
      "learning_rate": 5.680879917824822e-06,
      "loss": 0.7601,
      "step": 10600
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.981401883064955,
      "learning_rate": 5.68016426858311e-06,
      "loss": 0.7585,
      "step": 10601
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.250713039130418,
      "learning_rate": 5.679448605144461e-06,
      "loss": 0.7582,
      "step": 10602
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.120912657845478,
      "learning_rate": 5.678732927523809e-06,
      "loss": 0.7773,
      "step": 10603
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.922008188992745,
      "learning_rate": 5.678017235736093e-06,
      "loss": 0.8072,
      "step": 10604
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.384303245746343,
      "learning_rate": 5.677301529796253e-06,
      "loss": 0.7226,
      "step": 10605
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.16553505462511,
      "learning_rate": 5.676585809719227e-06,
      "loss": 0.781,
      "step": 10606
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.257252476048734,
      "learning_rate": 5.675870075519954e-06,
      "loss": 0.7069,
      "step": 10607
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.061989437671772,
      "learning_rate": 5.675154327213372e-06,
      "loss": 0.7892,
      "step": 10608
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.941533213954334,
      "learning_rate": 5.674438564814423e-06,
      "loss": 0.7934,
      "step": 10609
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.422928954500159,
      "learning_rate": 5.673722788338045e-06,
      "loss": 0.7303,
      "step": 10610
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.561990431902749,
      "learning_rate": 5.673006997799179e-06,
      "loss": 0.7378,
      "step": 10611
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.713988289723681,
      "learning_rate": 5.672291193212769e-06,
      "loss": 0.7103,
      "step": 10612
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.426396705556261,
      "learning_rate": 5.6715753745937505e-06,
      "loss": 0.7244,
      "step": 10613
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.48007063886861,
      "learning_rate": 5.670859541957066e-06,
      "loss": 0.6828,
      "step": 10614
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.560177100987123,
      "learning_rate": 5.670143695317659e-06,
      "loss": 0.7498,
      "step": 10615
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.535038728425869,
      "learning_rate": 5.6694278346904685e-06,
      "loss": 0.8167,
      "step": 10616
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.811904873715546,
      "learning_rate": 5.668711960090438e-06,
      "loss": 0.7891,
      "step": 10617
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.0308424635672955,
      "learning_rate": 5.6679960715325114e-06,
      "loss": 0.7814,
      "step": 10618
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.432602147406498,
      "learning_rate": 5.6672801690316285e-06,
      "loss": 0.7286,
      "step": 10619
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.004723888851333,
      "learning_rate": 5.666564252602734e-06,
      "loss": 0.807,
      "step": 10620
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.414302087072265,
      "learning_rate": 5.66584832226077e-06,
      "loss": 0.7395,
      "step": 10621
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.590218691418691,
      "learning_rate": 5.665132378020682e-06,
      "loss": 0.7446,
      "step": 10622
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.4007665810745955,
      "learning_rate": 5.664416419897409e-06,
      "loss": 0.8377,
      "step": 10623
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.647582444977295,
      "learning_rate": 5.663700447905901e-06,
      "loss": 0.7309,
      "step": 10624
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.406149784073318,
      "learning_rate": 5.662984462061099e-06,
      "loss": 0.7058,
      "step": 10625
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.982429903941107,
      "learning_rate": 5.662268462377948e-06,
      "loss": 0.7667,
      "step": 10626
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.794985432104046,
      "learning_rate": 5.661552448871393e-06,
      "loss": 0.7255,
      "step": 10627
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.755769379565185,
      "learning_rate": 5.66083642155638e-06,
      "loss": 0.7405,
      "step": 10628
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.554138044239582,
      "learning_rate": 5.660120380447854e-06,
      "loss": 0.7398,
      "step": 10629
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.370821533988524,
      "learning_rate": 5.65940432556076e-06,
      "loss": 0.7442,
      "step": 10630
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.613554427284437,
      "learning_rate": 5.658688256910046e-06,
      "loss": 0.7922,
      "step": 10631
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.139966233718298,
      "learning_rate": 5.657972174510655e-06,
      "loss": 0.7637,
      "step": 10632
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.067191973318115,
      "learning_rate": 5.657256078377536e-06,
      "loss": 0.7067,
      "step": 10633
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.36515011619705,
      "learning_rate": 5.656539968525638e-06,
      "loss": 0.7377,
      "step": 10634
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.03777964560069,
      "learning_rate": 5.655823844969905e-06,
      "loss": 0.7087,
      "step": 10635
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.241257607384797,
      "learning_rate": 5.655107707725285e-06,
      "loss": 0.7427,
      "step": 10636
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.668711749566408,
      "learning_rate": 5.654391556806725e-06,
      "loss": 0.7771,
      "step": 10637
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.267537926873053,
      "learning_rate": 5.653675392229178e-06,
      "loss": 0.7858,
      "step": 10638
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.7236647959620015,
      "learning_rate": 5.652959214007585e-06,
      "loss": 0.7596,
      "step": 10639
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.8694732343863345,
      "learning_rate": 5.6522430221569004e-06,
      "loss": 0.7508,
      "step": 10640
    },
    {
      "epoch": 0.95,
      "grad_norm": 9.88193012467719,
      "learning_rate": 5.6515268166920715e-06,
      "loss": 0.754,
      "step": 10641
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.489735502573147,
      "learning_rate": 5.650810597628046e-06,
      "loss": 0.765,
      "step": 10642
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.0482321491459805,
      "learning_rate": 5.650094364979774e-06,
      "loss": 0.7655,
      "step": 10643
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.490434511148844,
      "learning_rate": 5.649378118762207e-06,
      "loss": 0.7927,
      "step": 10644
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.9672910682123605,
      "learning_rate": 5.6486618589902935e-06,
      "loss": 0.7778,
      "step": 10645
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.030370270274002,
      "learning_rate": 5.647945585678982e-06,
      "loss": 0.693,
      "step": 10646
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.150857129540686,
      "learning_rate": 5.647229298843229e-06,
      "loss": 0.7697,
      "step": 10647
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.84791791725582,
      "learning_rate": 5.646512998497981e-06,
      "loss": 0.7648,
      "step": 10648
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.520211133255416,
      "learning_rate": 5.645796684658189e-06,
      "loss": 0.7883,
      "step": 10649
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.391184800779915,
      "learning_rate": 5.645080357338806e-06,
      "loss": 0.788,
      "step": 10650
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.7424224870392075,
      "learning_rate": 5.644364016554782e-06,
      "loss": 0.7973,
      "step": 10651
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.046044369398744,
      "learning_rate": 5.643647662321071e-06,
      "loss": 0.7175,
      "step": 10652
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.882666461048117,
      "learning_rate": 5.642931294652626e-06,
      "loss": 0.8116,
      "step": 10653
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.457199183562852,
      "learning_rate": 5.642214913564397e-06,
      "loss": 0.7455,
      "step": 10654
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.097195040072289,
      "learning_rate": 5.641498519071337e-06,
      "loss": 0.7887,
      "step": 10655
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.863772194466594,
      "learning_rate": 5.6407821111884024e-06,
      "loss": 0.7393,
      "step": 10656
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.580266362356395,
      "learning_rate": 5.6400656899305436e-06,
      "loss": 0.7109,
      "step": 10657
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.110254962935882,
      "learning_rate": 5.639349255312715e-06,
      "loss": 0.9318,
      "step": 10658
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.3561319169399315,
      "learning_rate": 5.638632807349871e-06,
      "loss": 0.6812,
      "step": 10659
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.307685948451638,
      "learning_rate": 5.637916346056966e-06,
      "loss": 0.7,
      "step": 10660
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.913633347905796,
      "learning_rate": 5.637199871448954e-06,
      "loss": 0.8009,
      "step": 10661
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.704537892364227,
      "learning_rate": 5.63648338354079e-06,
      "loss": 0.8207,
      "step": 10662
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.17523944710243,
      "learning_rate": 5.635766882347431e-06,
      "loss": 0.7511,
      "step": 10663
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.932949810619823,
      "learning_rate": 5.635050367883829e-06,
      "loss": 0.7519,
      "step": 10664
    },
    {
      "epoch": 0.95,
      "grad_norm": 9.517703119593028,
      "learning_rate": 5.6343338401649405e-06,
      "loss": 0.7538,
      "step": 10665
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.86925452250156,
      "learning_rate": 5.633617299205724e-06,
      "loss": 0.7592,
      "step": 10666
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.5895185413363375,
      "learning_rate": 5.632900745021133e-06,
      "loss": 0.7818,
      "step": 10667
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.060077849033257,
      "learning_rate": 5.632184177626124e-06,
      "loss": 0.8032,
      "step": 10668
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.113490545332831,
      "learning_rate": 5.631467597035656e-06,
      "loss": 0.8013,
      "step": 10669
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.3128102952792755,
      "learning_rate": 5.6307510032646866e-06,
      "loss": 0.761,
      "step": 10670
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.4453703933024915,
      "learning_rate": 5.630034396328168e-06,
      "loss": 0.7921,
      "step": 10671
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.727918782878589,
      "learning_rate": 5.629317776241062e-06,
      "loss": 0.7443,
      "step": 10672
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.345823940572856,
      "learning_rate": 5.628601143018327e-06,
      "loss": 0.7632,
      "step": 10673
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.030813211507775,
      "learning_rate": 5.627884496674918e-06,
      "loss": 0.7669,
      "step": 10674
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.893063987113317,
      "learning_rate": 5.6271678372257955e-06,
      "loss": 0.7662,
      "step": 10675
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.340974433669948,
      "learning_rate": 5.62645116468592e-06,
      "loss": 0.7817,
      "step": 10676
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.784494013256472,
      "learning_rate": 5.625734479070247e-06,
      "loss": 0.7037,
      "step": 10677
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.9243659451852935,
      "learning_rate": 5.6250177803937365e-06,
      "loss": 0.8111,
      "step": 10678
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.089870308805469,
      "learning_rate": 5.6243010686713496e-06,
      "loss": 0.8342,
      "step": 10679
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.285006547919683,
      "learning_rate": 5.623584343918043e-06,
      "loss": 0.7553,
      "step": 10680
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.754903450597129,
      "learning_rate": 5.622867606148781e-06,
      "loss": 0.7926,
      "step": 10681
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.585319833935165,
      "learning_rate": 5.622150855378521e-06,
      "loss": 0.725,
      "step": 10682
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.514578778583079,
      "learning_rate": 5.621434091622225e-06,
      "loss": 0.759,
      "step": 10683
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.131466121453619,
      "learning_rate": 5.620717314894852e-06,
      "loss": 0.8013,
      "step": 10684
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.731256976412902,
      "learning_rate": 5.620000525211366e-06,
      "loss": 0.7002,
      "step": 10685
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.4858986793358575,
      "learning_rate": 5.619283722586726e-06,
      "loss": 0.7515,
      "step": 10686
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.125580683336622,
      "learning_rate": 5.618566907035894e-06,
      "loss": 0.7035,
      "step": 10687
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.976497982957543,
      "learning_rate": 5.617850078573833e-06,
      "loss": 0.7486,
      "step": 10688
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.521918463697802,
      "learning_rate": 5.617133237215504e-06,
      "loss": 0.8199,
      "step": 10689
    },
    {
      "epoch": 0.95,
      "grad_norm": 8.92038726397508,
      "learning_rate": 5.616416382975871e-06,
      "loss": 0.7749,
      "step": 10690
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.334149706984305,
      "learning_rate": 5.6156995158698955e-06,
      "loss": 0.7683,
      "step": 10691
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.108888792430661,
      "learning_rate": 5.614982635912543e-06,
      "loss": 0.7878,
      "step": 10692
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.440014651546038,
      "learning_rate": 5.614265743118772e-06,
      "loss": 0.78,
      "step": 10693
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.055910991030995,
      "learning_rate": 5.613548837503548e-06,
      "loss": 0.8293,
      "step": 10694
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.867393594276292,
      "learning_rate": 5.61283191908184e-06,
      "loss": 0.7484,
      "step": 10695
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.358813530741499,
      "learning_rate": 5.612114987868603e-06,
      "loss": 0.804,
      "step": 10696
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.382498873444532,
      "learning_rate": 5.6113980438788085e-06,
      "loss": 0.7935,
      "step": 10697
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.084619184208191,
      "learning_rate": 5.610681087127418e-06,
      "loss": 0.8601,
      "step": 10698
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.793705620537771,
      "learning_rate": 5.6099641176293995e-06,
      "loss": 0.7808,
      "step": 10699
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.091789453290367,
      "learning_rate": 5.609247135399713e-06,
      "loss": 0.7931,
      "step": 10700
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.595553673038061,
      "learning_rate": 5.608530140453329e-06,
      "loss": 0.7687,
      "step": 10701
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.571177159852979,
      "learning_rate": 5.6078131328052095e-06,
      "loss": 0.7446,
      "step": 10702
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.739434960997699,
      "learning_rate": 5.607096112470321e-06,
      "loss": 0.7637,
      "step": 10703
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.807746198722069,
      "learning_rate": 5.606379079463633e-06,
      "loss": 0.7835,
      "step": 10704
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.877265563074078,
      "learning_rate": 5.605662033800109e-06,
      "loss": 0.8166,
      "step": 10705
    },
    {
      "epoch": 0.96,
      "grad_norm": 10.936338041091139,
      "learning_rate": 5.604944975494716e-06,
      "loss": 0.7646,
      "step": 10706
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.506535647237497,
      "learning_rate": 5.6042279045624215e-06,
      "loss": 0.8112,
      "step": 10707
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.9536408837729065,
      "learning_rate": 5.603510821018193e-06,
      "loss": 0.7206,
      "step": 10708
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.368995281812021,
      "learning_rate": 5.602793724876997e-06,
      "loss": 0.7744,
      "step": 10709
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.775677622857346,
      "learning_rate": 5.602076616153802e-06,
      "loss": 0.6652,
      "step": 10710
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.368150406012541,
      "learning_rate": 5.601359494863577e-06,
      "loss": 0.7261,
      "step": 10711
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.083465653920019,
      "learning_rate": 5.600642361021289e-06,
      "loss": 0.7295,
      "step": 10712
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.5629695235647665,
      "learning_rate": 5.599925214641908e-06,
      "loss": 0.8112,
      "step": 10713
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.84280749341674,
      "learning_rate": 5.5992080557404015e-06,
      "loss": 0.777,
      "step": 10714
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.842479768087824,
      "learning_rate": 5.598490884331739e-06,
      "loss": 0.752,
      "step": 10715
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.998495916147981,
      "learning_rate": 5.59777370043089e-06,
      "loss": 0.7384,
      "step": 10716
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.616983184864074,
      "learning_rate": 5.5970565040528234e-06,
      "loss": 0.7267,
      "step": 10717
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.293705506394972,
      "learning_rate": 5.5963392952125104e-06,
      "loss": 0.7871,
      "step": 10718
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.839736128201519,
      "learning_rate": 5.59562207392492e-06,
      "loss": 0.7461,
      "step": 10719
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.640076475261465,
      "learning_rate": 5.5949048402050245e-06,
      "loss": 0.7795,
      "step": 10720
    },
    {
      "epoch": 0.96,
      "grad_norm": 8.20933092338473,
      "learning_rate": 5.5941875940677925e-06,
      "loss": 0.8238,
      "step": 10721
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.0923548930411355,
      "learning_rate": 5.593470335528196e-06,
      "loss": 0.7445,
      "step": 10722
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.307340813811372,
      "learning_rate": 5.592753064601205e-06,
      "loss": 0.7974,
      "step": 10723
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.555544714894877,
      "learning_rate": 5.592035781301794e-06,
      "loss": 0.6808,
      "step": 10724
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.8248089448274,
      "learning_rate": 5.59131848564493e-06,
      "loss": 0.8255,
      "step": 10725
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.0295610595861575,
      "learning_rate": 5.59060117764559e-06,
      "loss": 0.7614,
      "step": 10726
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.034031392543963,
      "learning_rate": 5.589883857318741e-06,
      "loss": 0.7721,
      "step": 10727
    },
    {
      "epoch": 0.96,
      "grad_norm": 9.122500617665851,
      "learning_rate": 5.58916652467936e-06,
      "loss": 0.7802,
      "step": 10728
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.324792360586684,
      "learning_rate": 5.588449179742418e-06,
      "loss": 0.7405,
      "step": 10729
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.339675133654454,
      "learning_rate": 5.587731822522888e-06,
      "loss": 0.7666,
      "step": 10730
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.757497106349291,
      "learning_rate": 5.587014453035743e-06,
      "loss": 0.7605,
      "step": 10731
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.289306820598123,
      "learning_rate": 5.586297071295956e-06,
      "loss": 0.7713,
      "step": 10732
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.755484171933607,
      "learning_rate": 5.585579677318504e-06,
      "loss": 0.7652,
      "step": 10733
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.42013905598207,
      "learning_rate": 5.584862271118357e-06,
      "loss": 0.8028,
      "step": 10734
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.435379640819886,
      "learning_rate": 5.584144852710491e-06,
      "loss": 0.8026,
      "step": 10735
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.9592135828320965,
      "learning_rate": 5.5834274221098825e-06,
      "loss": 0.6856,
      "step": 10736
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.350021084577158,
      "learning_rate": 5.582709979331502e-06,
      "loss": 0.7402,
      "step": 10737
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.9190001422942355,
      "learning_rate": 5.581992524390326e-06,
      "loss": 0.7675,
      "step": 10738
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.967910954540599,
      "learning_rate": 5.581275057301333e-06,
      "loss": 0.735,
      "step": 10739
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.8934848885417574,
      "learning_rate": 5.580557578079494e-06,
      "loss": 0.8004,
      "step": 10740
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.628239390229365,
      "learning_rate": 5.579840086739789e-06,
      "loss": 0.674,
      "step": 10741
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.448732014005137,
      "learning_rate": 5.579122583297191e-06,
      "loss": 0.7722,
      "step": 10742
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.824158500368036,
      "learning_rate": 5.578405067766677e-06,
      "loss": 0.7057,
      "step": 10743
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.924629681966388,
      "learning_rate": 5.577687540163223e-06,
      "loss": 0.7225,
      "step": 10744
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.763506936378303,
      "learning_rate": 5.576970000501809e-06,
      "loss": 0.6747,
      "step": 10745
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.104354558695034,
      "learning_rate": 5.576252448797409e-06,
      "loss": 0.7987,
      "step": 10746
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.808018785697197,
      "learning_rate": 5.575534885065e-06,
      "loss": 0.7984,
      "step": 10747
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.908433311103883,
      "learning_rate": 5.574817309319561e-06,
      "loss": 0.7605,
      "step": 10748
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.4284983130881255,
      "learning_rate": 5.574099721576072e-06,
      "loss": 0.7097,
      "step": 10749
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.807095208190377,
      "learning_rate": 5.573382121849505e-06,
      "loss": 0.764,
      "step": 10750
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.441683340938526,
      "learning_rate": 5.572664510154844e-06,
      "loss": 0.7918,
      "step": 10751
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.109456285242718,
      "learning_rate": 5.571946886507064e-06,
      "loss": 0.8281,
      "step": 10752
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.193676926402861,
      "learning_rate": 5.571229250921146e-06,
      "loss": 0.772,
      "step": 10753
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.982173230664179,
      "learning_rate": 5.570511603412069e-06,
      "loss": 0.6982,
      "step": 10754
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.357854473239579,
      "learning_rate": 5.569793943994812e-06,
      "loss": 0.6939,
      "step": 10755
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.713659414415944,
      "learning_rate": 5.569076272684353e-06,
      "loss": 0.7669,
      "step": 10756
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.317259298856979,
      "learning_rate": 5.568358589495674e-06,
      "loss": 0.7546,
      "step": 10757
    },
    {
      "epoch": 0.96,
      "grad_norm": 8.892567803185804,
      "learning_rate": 5.567640894443753e-06,
      "loss": 0.7802,
      "step": 10758
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.846412932848237,
      "learning_rate": 5.566923187543573e-06,
      "loss": 0.8154,
      "step": 10759
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.068471552640093,
      "learning_rate": 5.566205468810111e-06,
      "loss": 0.7065,
      "step": 10760
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.6999439367677835,
      "learning_rate": 5.565487738258351e-06,
      "loss": 0.7289,
      "step": 10761
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.474354562243727,
      "learning_rate": 5.564769995903273e-06,
      "loss": 0.7344,
      "step": 10762
    },
    {
      "epoch": 0.96,
      "grad_norm": 9.910927881187835,
      "learning_rate": 5.564052241759858e-06,
      "loss": 0.7058,
      "step": 10763
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.282517001681445,
      "learning_rate": 5.563334475843087e-06,
      "loss": 0.8072,
      "step": 10764
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.3570970787606695,
      "learning_rate": 5.562616698167944e-06,
      "loss": 0.7352,
      "step": 10765
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.863273078112554,
      "learning_rate": 5.561898908749409e-06,
      "loss": 0.7442,
      "step": 10766
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.098503267501091,
      "learning_rate": 5.561181107602464e-06,
      "loss": 0.7942,
      "step": 10767
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.0722410880532784,
      "learning_rate": 5.560463294742094e-06,
      "loss": 0.8654,
      "step": 10768
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.604544241358215,
      "learning_rate": 5.559745470183278e-06,
      "loss": 0.8158,
      "step": 10769
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.542121453878259,
      "learning_rate": 5.559027633941001e-06,
      "loss": 0.7228,
      "step": 10770
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.974083517509441,
      "learning_rate": 5.55830978603025e-06,
      "loss": 0.8172,
      "step": 10771
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.232486534983515,
      "learning_rate": 5.557591926466002e-06,
      "loss": 0.7789,
      "step": 10772
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.828476193029375,
      "learning_rate": 5.556874055263243e-06,
      "loss": 0.7887,
      "step": 10773
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.310737379762359,
      "learning_rate": 5.556156172436958e-06,
      "loss": 0.7281,
      "step": 10774
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.573933299145076,
      "learning_rate": 5.555438278002132e-06,
      "loss": 0.7326,
      "step": 10775
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.138146596991702,
      "learning_rate": 5.554720371973747e-06,
      "loss": 0.7809,
      "step": 10776
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.801364955093416,
      "learning_rate": 5.55400245436679e-06,
      "loss": 0.752,
      "step": 10777
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.6574481048814995,
      "learning_rate": 5.553284525196246e-06,
      "loss": 0.8301,
      "step": 10778
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.929194771244951,
      "learning_rate": 5.552566584477098e-06,
      "loss": 0.7973,
      "step": 10779
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.836063940641461,
      "learning_rate": 5.551848632224333e-06,
      "loss": 0.815,
      "step": 10780
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.435930726399479,
      "learning_rate": 5.551130668452936e-06,
      "loss": 0.7325,
      "step": 10781
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.523674390823477,
      "learning_rate": 5.550412693177893e-06,
      "loss": 0.7691,
      "step": 10782
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.6180469253151895,
      "learning_rate": 5.5496947064141894e-06,
      "loss": 0.697,
      "step": 10783
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.135105493775467,
      "learning_rate": 5.548976708176815e-06,
      "loss": 0.7325,
      "step": 10784
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.7600323438977155,
      "learning_rate": 5.548258698480754e-06,
      "loss": 0.7838,
      "step": 10785
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.343729945456398,
      "learning_rate": 5.547540677340992e-06,
      "loss": 0.6605,
      "step": 10786
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.92056030427151,
      "learning_rate": 5.546822644772518e-06,
      "loss": 0.8129,
      "step": 10787
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.31900559432893,
      "learning_rate": 5.546104600790317e-06,
      "loss": 0.6927,
      "step": 10788
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.093399631653884,
      "learning_rate": 5.5453865454093804e-06,
      "loss": 0.7638,
      "step": 10789
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.5354269520252855,
      "learning_rate": 5.544668478644694e-06,
      "loss": 0.7214,
      "step": 10790
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.602634929394719,
      "learning_rate": 5.543950400511246e-06,
      "loss": 0.7239,
      "step": 10791
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.14763620197193,
      "learning_rate": 5.543232311024022e-06,
      "loss": 0.7094,
      "step": 10792
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.186740873558471,
      "learning_rate": 5.542514210198016e-06,
      "loss": 0.7498,
      "step": 10793
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.625108897839705,
      "learning_rate": 5.541796098048216e-06,
      "loss": 0.7043,
      "step": 10794
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.040329259090135,
      "learning_rate": 5.5410779745896045e-06,
      "loss": 0.7452,
      "step": 10795
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.953571668079244,
      "learning_rate": 5.540359839837177e-06,
      "loss": 0.7202,
      "step": 10796
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.257510274340713,
      "learning_rate": 5.539641693805922e-06,
      "loss": 0.734,
      "step": 10797
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.807550674479319,
      "learning_rate": 5.538923536510826e-06,
      "loss": 0.7629,
      "step": 10798
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.393003910189959,
      "learning_rate": 5.538205367966883e-06,
      "loss": 0.7306,
      "step": 10799
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.467274352010674,
      "learning_rate": 5.537487188189084e-06,
      "loss": 0.7198,
      "step": 10800
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.908547468165139,
      "learning_rate": 5.536768997192413e-06,
      "loss": 0.7529,
      "step": 10801
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.0980365721146,
      "learning_rate": 5.5360507949918665e-06,
      "loss": 0.7915,
      "step": 10802
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.6406987634162675,
      "learning_rate": 5.535332581602434e-06,
      "loss": 0.7467,
      "step": 10803
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.520230773249258,
      "learning_rate": 5.534614357039106e-06,
      "loss": 0.7633,
      "step": 10804
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.149237622316492,
      "learning_rate": 5.533896121316873e-06,
      "loss": 0.7358,
      "step": 10805
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.56208937303041,
      "learning_rate": 5.533177874450728e-06,
      "loss": 0.7972,
      "step": 10806
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.723721530361703,
      "learning_rate": 5.532459616455664e-06,
      "loss": 0.7009,
      "step": 10807
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.907545575773026,
      "learning_rate": 5.5317413473466695e-06,
      "loss": 0.7363,
      "step": 10808
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.491298275741591,
      "learning_rate": 5.5310230671387414e-06,
      "loss": 0.8063,
      "step": 10809
    },
    {
      "epoch": 0.96,
      "grad_norm": 9.92361604961312,
      "learning_rate": 5.530304775846867e-06,
      "loss": 0.7895,
      "step": 10810
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.862532215228409,
      "learning_rate": 5.529586473486043e-06,
      "loss": 0.7405,
      "step": 10811
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.185312047580206,
      "learning_rate": 5.528868160071261e-06,
      "loss": 0.7803,
      "step": 10812
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.961223488597242,
      "learning_rate": 5.528149835617514e-06,
      "loss": 0.7018,
      "step": 10813
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.115070386481663,
      "learning_rate": 5.527431500139797e-06,
      "loss": 0.6888,
      "step": 10814
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.240889203508808,
      "learning_rate": 5.526713153653102e-06,
      "loss": 0.7588,
      "step": 10815
    },
    {
      "epoch": 0.96,
      "grad_norm": 8.409514123785138,
      "learning_rate": 5.525994796172422e-06,
      "loss": 0.7608,
      "step": 10816
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.847199709647511,
      "learning_rate": 5.525276427712755e-06,
      "loss": 0.8109,
      "step": 10817
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.0468838379030485,
      "learning_rate": 5.5245580482890905e-06,
      "loss": 0.7985,
      "step": 10818
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.057783443753944,
      "learning_rate": 5.523839657916427e-06,
      "loss": 0.8106,
      "step": 10819
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.395911694185834,
      "learning_rate": 5.523121256609757e-06,
      "loss": 0.8074,
      "step": 10820
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.6349244125114355,
      "learning_rate": 5.522402844384077e-06,
      "loss": 0.7294,
      "step": 10821
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.382522609416082,
      "learning_rate": 5.521684421254382e-06,
      "loss": 0.8,
      "step": 10822
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.836391953693463,
      "learning_rate": 5.5209659872356654e-06,
      "loss": 0.7634,
      "step": 10823
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.868128572568108,
      "learning_rate": 5.5202475423429255e-06,
      "loss": 0.7749,
      "step": 10824
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.076462503549313,
      "learning_rate": 5.519529086591159e-06,
      "loss": 0.754,
      "step": 10825
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.832570910090234,
      "learning_rate": 5.5188106199953595e-06,
      "loss": 0.7709,
      "step": 10826
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.41740428907209,
      "learning_rate": 5.5180921425705235e-06,
      "loss": 0.7645,
      "step": 10827
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.860705760960257,
      "learning_rate": 5.51737365433165e-06,
      "loss": 0.7704,
      "step": 10828
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.479478864224294,
      "learning_rate": 5.516655155293735e-06,
      "loss": 0.7229,
      "step": 10829
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.2566102146947555,
      "learning_rate": 5.5159366454717735e-06,
      "loss": 0.7339,
      "step": 10830
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.557257423336992,
      "learning_rate": 5.515218124880765e-06,
      "loss": 0.7171,
      "step": 10831
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.852578937963471,
      "learning_rate": 5.514499593535707e-06,
      "loss": 0.783,
      "step": 10832
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.506309609137088,
      "learning_rate": 5.513781051451595e-06,
      "loss": 0.6963,
      "step": 10833
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.770704000006851,
      "learning_rate": 5.51306249864343e-06,
      "loss": 0.7856,
      "step": 10834
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.687536992812467,
      "learning_rate": 5.51234393512621e-06,
      "loss": 0.7903,
      "step": 10835
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.14151277853263,
      "learning_rate": 5.5116253609149315e-06,
      "loss": 0.693,
      "step": 10836
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.397605921093894,
      "learning_rate": 5.510906776024594e-06,
      "loss": 0.696,
      "step": 10837
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.153173229645279,
      "learning_rate": 5.5101881804701966e-06,
      "loss": 0.7653,
      "step": 10838
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.09032519402522,
      "learning_rate": 5.509469574266737e-06,
      "loss": 0.7554,
      "step": 10839
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.200858368297088,
      "learning_rate": 5.5087509574292165e-06,
      "loss": 0.7658,
      "step": 10840
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.323588625124677,
      "learning_rate": 5.508032329972635e-06,
      "loss": 0.7119,
      "step": 10841
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.897500151511188,
      "learning_rate": 5.507313691911988e-06,
      "loss": 0.8124,
      "step": 10842
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.843540263562123,
      "learning_rate": 5.506595043262282e-06,
      "loss": 0.7367,
      "step": 10843
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.940924451862673,
      "learning_rate": 5.505876384038512e-06,
      "loss": 0.8115,
      "step": 10844
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.142674815756071,
      "learning_rate": 5.505157714255682e-06,
      "loss": 0.7766,
      "step": 10845
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.827504295541371,
      "learning_rate": 5.50443903392879e-06,
      "loss": 0.7613,
      "step": 10846
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.877686315815717,
      "learning_rate": 5.503720343072838e-06,
      "loss": 0.8235,
      "step": 10847
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.506388751840898,
      "learning_rate": 5.5030016417028264e-06,
      "loss": 0.6973,
      "step": 10848
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.080662334653133,
      "learning_rate": 5.5022829298337574e-06,
      "loss": 0.7377,
      "step": 10849
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.368887439964926,
      "learning_rate": 5.501564207480633e-06,
      "loss": 0.7257,
      "step": 10850
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.437736372977898,
      "learning_rate": 5.500845474658454e-06,
      "loss": 0.7668,
      "step": 10851
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.595274744317537,
      "learning_rate": 5.500126731382222e-06,
      "loss": 0.8354,
      "step": 10852
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.495852183228257,
      "learning_rate": 5.499407977666939e-06,
      "loss": 0.793,
      "step": 10853
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.398505559717923,
      "learning_rate": 5.49868921352761e-06,
      "loss": 0.832,
      "step": 10854
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.837122273337541,
      "learning_rate": 5.497970438979235e-06,
      "loss": 0.7368,
      "step": 10855
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.633314221489813,
      "learning_rate": 5.497251654036816e-06,
      "loss": 0.7704,
      "step": 10856
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.899177043702602,
      "learning_rate": 5.496532858715361e-06,
      "loss": 0.7967,
      "step": 10857
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.390812539567147,
      "learning_rate": 5.495814053029867e-06,
      "loss": 0.669,
      "step": 10858
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.152273159417039,
      "learning_rate": 5.495095236995341e-06,
      "loss": 0.7353,
      "step": 10859
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.228486809690229,
      "learning_rate": 5.494376410626788e-06,
      "loss": 0.7295,
      "step": 10860
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.6030421015126635,
      "learning_rate": 5.4936575739392085e-06,
      "loss": 0.7071,
      "step": 10861
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.282241887527273,
      "learning_rate": 5.4929387269476076e-06,
      "loss": 0.7878,
      "step": 10862
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.664744075912512,
      "learning_rate": 5.4922198696669915e-06,
      "loss": 0.7436,
      "step": 10863
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.379604978079328,
      "learning_rate": 5.491501002112363e-06,
      "loss": 0.7499,
      "step": 10864
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.711792151969145,
      "learning_rate": 5.490782124298729e-06,
      "loss": 0.7216,
      "step": 10865
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.257201442553998,
      "learning_rate": 5.490063236241091e-06,
      "loss": 0.733,
      "step": 10866
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.698936147947885,
      "learning_rate": 5.489344337954457e-06,
      "loss": 0.7341,
      "step": 10867
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.262660272297028,
      "learning_rate": 5.488625429453831e-06,
      "loss": 0.709,
      "step": 10868
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.091025138917357,
      "learning_rate": 5.48790651075422e-06,
      "loss": 0.7364,
      "step": 10869
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.030706704439861,
      "learning_rate": 5.487187581870629e-06,
      "loss": 0.7644,
      "step": 10870
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.396493711182646,
      "learning_rate": 5.486468642818064e-06,
      "loss": 0.7689,
      "step": 10871
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.521708301865422,
      "learning_rate": 5.48574969361153e-06,
      "loss": 0.8386,
      "step": 10872
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.531950468405093,
      "learning_rate": 5.485030734266038e-06,
      "loss": 0.7249,
      "step": 10873
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.055894533866756,
      "learning_rate": 5.48431176479659e-06,
      "loss": 0.7865,
      "step": 10874
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.994955839876602,
      "learning_rate": 5.483592785218194e-06,
      "loss": 0.7545,
      "step": 10875
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.8720281031726715,
      "learning_rate": 5.482873795545858e-06,
      "loss": 0.8024,
      "step": 10876
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.09959164851131,
      "learning_rate": 5.482154795794588e-06,
      "loss": 0.7807,
      "step": 10877
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.498183640465294,
      "learning_rate": 5.481435785979394e-06,
      "loss": 0.7144,
      "step": 10878
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.570621921208779,
      "learning_rate": 5.480716766115283e-06,
      "loss": 0.6933,
      "step": 10879
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.809443695996478,
      "learning_rate": 5.479997736217261e-06,
      "loss": 0.82,
      "step": 10880
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.91908343473005,
      "learning_rate": 5.4792786963003376e-06,
      "loss": 0.7902,
      "step": 10881
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.911956032275399,
      "learning_rate": 5.47855964637952e-06,
      "loss": 0.7451,
      "step": 10882
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.292592110278131,
      "learning_rate": 5.4778405864698195e-06,
      "loss": 0.7956,
      "step": 10883
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.040662578521395,
      "learning_rate": 5.477121516586243e-06,
      "loss": 0.8381,
      "step": 10884
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.991990085727443,
      "learning_rate": 5.476402436743799e-06,
      "loss": 0.7801,
      "step": 10885
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.050521211636354,
      "learning_rate": 5.475683346957497e-06,
      "loss": 0.7659,
      "step": 10886
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.286247708002262,
      "learning_rate": 5.4749642472423485e-06,
      "loss": 0.6867,
      "step": 10887
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.781821682144777,
      "learning_rate": 5.47424513761336e-06,
      "loss": 0.829,
      "step": 10888
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.25061283162398,
      "learning_rate": 5.473526018085544e-06,
      "loss": 0.6939,
      "step": 10889
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.566651308412545,
      "learning_rate": 5.472806888673909e-06,
      "loss": 0.828,
      "step": 10890
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.085724297098351,
      "learning_rate": 5.472087749393465e-06,
      "loss": 0.7881,
      "step": 10891
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.166223927755239,
      "learning_rate": 5.471368600259223e-06,
      "loss": 0.8002,
      "step": 10892
    },
    {
      "epoch": 0.97,
      "grad_norm": 8.17966785273844,
      "learning_rate": 5.470649441286194e-06,
      "loss": 0.799,
      "step": 10893
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.181368284013485,
      "learning_rate": 5.469930272489389e-06,
      "loss": 0.7201,
      "step": 10894
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.750473902174519,
      "learning_rate": 5.469211093883819e-06,
      "loss": 0.7895,
      "step": 10895
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.409850444267522,
      "learning_rate": 5.468491905484494e-06,
      "loss": 0.68,
      "step": 10896
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.412174205988963,
      "learning_rate": 5.467772707306427e-06,
      "loss": 0.7322,
      "step": 10897
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.133378927836763,
      "learning_rate": 5.46705349936463e-06,
      "loss": 0.7358,
      "step": 10898
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.91671555772656,
      "learning_rate": 5.466334281674112e-06,
      "loss": 0.7659,
      "step": 10899
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.9215159926489624,
      "learning_rate": 5.465615054249888e-06,
      "loss": 0.6961,
      "step": 10900
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.2816955387597595,
      "learning_rate": 5.46489581710697e-06,
      "loss": 0.7533,
      "step": 10901
    },
    {
      "epoch": 0.97,
      "grad_norm": 9.882621229467905,
      "learning_rate": 5.46417657026037e-06,
      "loss": 0.8756,
      "step": 10902
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.708762956651207,
      "learning_rate": 5.4634573137251e-06,
      "loss": 0.8209,
      "step": 10903
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.345043098191782,
      "learning_rate": 5.462738047516173e-06,
      "loss": 0.8502,
      "step": 10904
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.252415881748944,
      "learning_rate": 5.462018771648604e-06,
      "loss": 0.7915,
      "step": 10905
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.984023331478982,
      "learning_rate": 5.461299486137404e-06,
      "loss": 0.7453,
      "step": 10906
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.321797581294392,
      "learning_rate": 5.460580190997587e-06,
      "loss": 0.7968,
      "step": 10907
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.670609197430789,
      "learning_rate": 5.459860886244171e-06,
      "loss": 0.743,
      "step": 10908
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.284889902389466,
      "learning_rate": 5.459141571892161e-06,
      "loss": 0.7583,
      "step": 10909
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.635824894749308,
      "learning_rate": 5.458422247956579e-06,
      "loss": 0.7977,
      "step": 10910
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.600541405788429,
      "learning_rate": 5.4577029144524355e-06,
      "loss": 0.7082,
      "step": 10911
    },
    {
      "epoch": 0.97,
      "grad_norm": 9.067959538016334,
      "learning_rate": 5.456983571394746e-06,
      "loss": 0.7831,
      "step": 10912
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.3605245078371695,
      "learning_rate": 5.456264218798525e-06,
      "loss": 0.788,
      "step": 10913
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.59972656581848,
      "learning_rate": 5.455544856678788e-06,
      "loss": 0.7643,
      "step": 10914
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.276688621234027,
      "learning_rate": 5.454825485050551e-06,
      "loss": 0.7258,
      "step": 10915
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.902128502988064,
      "learning_rate": 5.4541061039288265e-06,
      "loss": 0.7898,
      "step": 10916
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.074511796033352,
      "learning_rate": 5.453386713328631e-06,
      "loss": 0.7483,
      "step": 10917
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.2963896584454675,
      "learning_rate": 5.452667313264982e-06,
      "loss": 0.8307,
      "step": 10918
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.788046973127715,
      "learning_rate": 5.451947903752895e-06,
      "loss": 0.82,
      "step": 10919
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.991260409354889,
      "learning_rate": 5.4512284848073825e-06,
      "loss": 0.7517,
      "step": 10920
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.896170925086359,
      "learning_rate": 5.450509056443466e-06,
      "loss": 0.7919,
      "step": 10921
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.298583060987662,
      "learning_rate": 5.449789618676159e-06,
      "loss": 0.8471,
      "step": 10922
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.153903798141602,
      "learning_rate": 5.449070171520478e-06,
      "loss": 0.6738,
      "step": 10923
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.611644670033036,
      "learning_rate": 5.448350714991442e-06,
      "loss": 0.8042,
      "step": 10924
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.423400673310906,
      "learning_rate": 5.447631249104065e-06,
      "loss": 0.7167,
      "step": 10925
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.964824962476968,
      "learning_rate": 5.446911773873367e-06,
      "loss": 0.8151,
      "step": 10926
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.533647884779483,
      "learning_rate": 5.4461922893143635e-06,
      "loss": 0.7879,
      "step": 10927
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.5947120585872,
      "learning_rate": 5.445472795442074e-06,
      "loss": 0.7243,
      "step": 10928
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.4628424952932098,
      "learning_rate": 5.444753292271514e-06,
      "loss": 0.8739,
      "step": 10929
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.821373319907367,
      "learning_rate": 5.444033779817706e-06,
      "loss": 0.7448,
      "step": 10930
    },
    {
      "epoch": 0.98,
      "grad_norm": 8.54558232942016,
      "learning_rate": 5.443314258095663e-06,
      "loss": 0.7809,
      "step": 10931
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.183940544667082,
      "learning_rate": 5.442594727120406e-06,
      "loss": 0.7706,
      "step": 10932
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.160339184049071,
      "learning_rate": 5.4418751869069545e-06,
      "loss": 0.7798,
      "step": 10933
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.790313198179425,
      "learning_rate": 5.441155637470325e-06,
      "loss": 0.7999,
      "step": 10934
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.22865400726616,
      "learning_rate": 5.440436078825537e-06,
      "loss": 0.7152,
      "step": 10935
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.520589550067894,
      "learning_rate": 5.439716510987611e-06,
      "loss": 0.7224,
      "step": 10936
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.253168197368117,
      "learning_rate": 5.438996933971568e-06,
      "loss": 0.8284,
      "step": 10937
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.00109018867336,
      "learning_rate": 5.438277347792422e-06,
      "loss": 0.7435,
      "step": 10938
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.984674075021841,
      "learning_rate": 5.437557752465198e-06,
      "loss": 0.6891,
      "step": 10939
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.626382171257207,
      "learning_rate": 5.4368381480049146e-06,
      "loss": 0.7453,
      "step": 10940
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.188686310681479,
      "learning_rate": 5.4361185344265895e-06,
      "loss": 0.7425,
      "step": 10941
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.71696672107443,
      "learning_rate": 5.435398911745246e-06,
      "loss": 0.7494,
      "step": 10942
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.761570090407094,
      "learning_rate": 5.434679279975905e-06,
      "loss": 0.7178,
      "step": 10943
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.542822081852594,
      "learning_rate": 5.433959639133586e-06,
      "loss": 0.8043,
      "step": 10944
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.309655126576206,
      "learning_rate": 5.433239989233308e-06,
      "loss": 0.7624,
      "step": 10945
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.3565105270089814,
      "learning_rate": 5.432520330290095e-06,
      "loss": 0.7251,
      "step": 10946
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.064157990392767,
      "learning_rate": 5.431800662318967e-06,
      "loss": 0.7245,
      "step": 10947
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.107981509897668,
      "learning_rate": 5.431080985334945e-06,
      "loss": 0.7374,
      "step": 10948
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.839918550509735,
      "learning_rate": 5.430361299353054e-06,
      "loss": 0.7491,
      "step": 10949
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.249140372363708,
      "learning_rate": 5.4296416043883115e-06,
      "loss": 0.7865,
      "step": 10950
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.5981414290553335,
      "learning_rate": 5.428921900455741e-06,
      "loss": 0.8204,
      "step": 10951
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.868684612888946,
      "learning_rate": 5.428202187570366e-06,
      "loss": 0.719,
      "step": 10952
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.66179288022737,
      "learning_rate": 5.427482465747209e-06,
      "loss": 0.6998,
      "step": 10953
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.203700427451355,
      "learning_rate": 5.4267627350012895e-06,
      "loss": 0.7647,
      "step": 10954
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.703821703943381,
      "learning_rate": 5.426042995347634e-06,
      "loss": 0.7659,
      "step": 10955
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.810313708812205,
      "learning_rate": 5.425323246801264e-06,
      "loss": 0.6959,
      "step": 10956
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.124878584922569,
      "learning_rate": 5.4246034893772015e-06,
      "loss": 0.7705,
      "step": 10957
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.782854270363214,
      "learning_rate": 5.423883723090473e-06,
      "loss": 0.6905,
      "step": 10958
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.106513041089919,
      "learning_rate": 5.423163947956099e-06,
      "loss": 0.7214,
      "step": 10959
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.036914215044615,
      "learning_rate": 5.422444163989104e-06,
      "loss": 0.6942,
      "step": 10960
    },
    {
      "epoch": 0.98,
      "grad_norm": 9.218680754965723,
      "learning_rate": 5.421724371204512e-06,
      "loss": 0.8339,
      "step": 10961
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.829682076883969,
      "learning_rate": 5.421004569617348e-06,
      "loss": 0.7902,
      "step": 10962
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.996729331965973,
      "learning_rate": 5.420284759242636e-06,
      "loss": 0.7246,
      "step": 10963
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.784089209312177,
      "learning_rate": 5.419564940095398e-06,
      "loss": 0.75,
      "step": 10964
    },
    {
      "epoch": 0.98,
      "grad_norm": 8.472496721757127,
      "learning_rate": 5.418845112190663e-06,
      "loss": 0.7639,
      "step": 10965
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.622684587540335,
      "learning_rate": 5.418125275543453e-06,
      "loss": 0.7343,
      "step": 10966
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.792942923617605,
      "learning_rate": 5.417405430168793e-06,
      "loss": 0.6959,
      "step": 10967
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.012266940813307,
      "learning_rate": 5.4166855760817085e-06,
      "loss": 0.7104,
      "step": 10968
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.123479517454359,
      "learning_rate": 5.4159657132972255e-06,
      "loss": 0.7414,
      "step": 10969
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.059458844612427,
      "learning_rate": 5.415245841830368e-06,
      "loss": 0.7404,
      "step": 10970
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.884663455045666,
      "learning_rate": 5.414525961696165e-06,
      "loss": 0.7327,
      "step": 10971
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.4885457002636375,
      "learning_rate": 5.4138060729096394e-06,
      "loss": 0.7794,
      "step": 10972
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.132055494004316,
      "learning_rate": 5.41308617548582e-06,
      "loss": 0.7308,
      "step": 10973
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.731627618879089,
      "learning_rate": 5.412366269439728e-06,
      "loss": 0.7224,
      "step": 10974
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.246486169524868,
      "learning_rate": 5.411646354786397e-06,
      "loss": 0.782,
      "step": 10975
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.924513941975421,
      "learning_rate": 5.410926431540846e-06,
      "loss": 0.7091,
      "step": 10976
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.758059422257872,
      "learning_rate": 5.410206499718107e-06,
      "loss": 0.8375,
      "step": 10977
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.767472676778761,
      "learning_rate": 5.409486559333207e-06,
      "loss": 0.7906,
      "step": 10978
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.811519428828896,
      "learning_rate": 5.4087666104011705e-06,
      "loss": 0.7512,
      "step": 10979
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.769157249553341,
      "learning_rate": 5.408046652937027e-06,
      "loss": 0.7469,
      "step": 10980
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.234970326389391,
      "learning_rate": 5.407326686955804e-06,
      "loss": 0.8275,
      "step": 10981
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.517935192664958,
      "learning_rate": 5.4066067124725265e-06,
      "loss": 0.7943,
      "step": 10982
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.4386855719959115,
      "learning_rate": 5.405886729502224e-06,
      "loss": 0.815,
      "step": 10983
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.429937211251179,
      "learning_rate": 5.405166738059927e-06,
      "loss": 0.7624,
      "step": 10984
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.800785655966383,
      "learning_rate": 5.40444673816066e-06,
      "loss": 0.8036,
      "step": 10985
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.839310660289208,
      "learning_rate": 5.403726729819453e-06,
      "loss": 0.7926,
      "step": 10986
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.9388657699629075,
      "learning_rate": 5.403006713051336e-06,
      "loss": 0.7846,
      "step": 10987
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.101990328000364,
      "learning_rate": 5.4022866878713365e-06,
      "loss": 0.7198,
      "step": 10988
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.1357160780207725,
      "learning_rate": 5.4015666542944825e-06,
      "loss": 0.6939,
      "step": 10989
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.3646086531637724,
      "learning_rate": 5.400846612335803e-06,
      "loss": 0.6795,
      "step": 10990
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.693596334146372,
      "learning_rate": 5.40012656201033e-06,
      "loss": 0.6909,
      "step": 10991
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.769017161535011,
      "learning_rate": 5.39940650333309e-06,
      "loss": 0.8169,
      "step": 10992
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.399699925285489,
      "learning_rate": 5.398686436319115e-06,
      "loss": 0.7661,
      "step": 10993
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.90561808801369,
      "learning_rate": 5.3979663609834355e-06,
      "loss": 0.7491,
      "step": 10994
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.211008260075684,
      "learning_rate": 5.397246277341076e-06,
      "loss": 0.7199,
      "step": 10995
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.215269306267232,
      "learning_rate": 5.396526185407072e-06,
      "loss": 0.7719,
      "step": 10996
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.8419117740802555,
      "learning_rate": 5.395806085196453e-06,
      "loss": 0.8293,
      "step": 10997
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.922609747711222,
      "learning_rate": 5.395085976724248e-06,
      "loss": 0.7584,
      "step": 10998
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.770890993048914,
      "learning_rate": 5.3943658600054886e-06,
      "loss": 0.812,
      "step": 10999
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.252765837460645,
      "learning_rate": 5.393645735055205e-06,
      "loss": 0.7558,
      "step": 11000
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.1003718615310785,
      "learning_rate": 5.392925601888429e-06,
      "loss": 0.8175,
      "step": 11001
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.127926939834495,
      "learning_rate": 5.392205460520192e-06,
      "loss": 0.7528,
      "step": 11002
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.837037990237241,
      "learning_rate": 5.391485310965524e-06,
      "loss": 0.7439,
      "step": 11003
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.350434987831685,
      "learning_rate": 5.390765153239459e-06,
      "loss": 0.7745,
      "step": 11004
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.898594735039204,
      "learning_rate": 5.390044987357026e-06,
      "loss": 0.756,
      "step": 11005
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.2749602056664235,
      "learning_rate": 5.389324813333259e-06,
      "loss": 0.7679,
      "step": 11006
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.065526230824823,
      "learning_rate": 5.388604631183189e-06,
      "loss": 0.733,
      "step": 11007
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.971647564906095,
      "learning_rate": 5.387884440921846e-06,
      "loss": 0.6954,
      "step": 11008
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.654992520017964,
      "learning_rate": 5.3871642425642675e-06,
      "loss": 0.7647,
      "step": 11009
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.299184502539172,
      "learning_rate": 5.3864440361254836e-06,
      "loss": 0.7783,
      "step": 11010
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.626667894031781,
      "learning_rate": 5.385723821620525e-06,
      "loss": 0.7197,
      "step": 11011
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.02334049614243,
      "learning_rate": 5.385003599064426e-06,
      "loss": 0.7564,
      "step": 11012
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.862395593727241,
      "learning_rate": 5.384283368472223e-06,
      "loss": 0.8131,
      "step": 11013
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.618701518775597,
      "learning_rate": 5.383563129858943e-06,
      "loss": 0.7718,
      "step": 11014
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.356015050610118,
      "learning_rate": 5.382842883239624e-06,
      "loss": 0.7434,
      "step": 11015
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.342844156947658,
      "learning_rate": 5.382122628629298e-06,
      "loss": 0.7659,
      "step": 11016
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.35263850274779,
      "learning_rate": 5.381402366043001e-06,
      "loss": 0.8834,
      "step": 11017
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.087155007188754,
      "learning_rate": 5.3806820954957616e-06,
      "loss": 0.8347,
      "step": 11018
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.62816028355693,
      "learning_rate": 5.379961817002619e-06,
      "loss": 0.7922,
      "step": 11019
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.327301937430217,
      "learning_rate": 5.379241530578605e-06,
      "loss": 0.7781,
      "step": 11020
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.5433780651070474,
      "learning_rate": 5.378521236238754e-06,
      "loss": 0.7786,
      "step": 11021
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.531973255540814,
      "learning_rate": 5.377800933998102e-06,
      "loss": 0.6978,
      "step": 11022
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.530768261184267,
      "learning_rate": 5.3770806238716836e-06,
      "loss": 0.8534,
      "step": 11023
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.558703111807603,
      "learning_rate": 5.37636030587453e-06,
      "loss": 0.7905,
      "step": 11024
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.964483616505667,
      "learning_rate": 5.375639980021682e-06,
      "loss": 0.7456,
      "step": 11025
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.574431588536491,
      "learning_rate": 5.374919646328171e-06,
      "loss": 0.8059,
      "step": 11026
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.476183949429598,
      "learning_rate": 5.3741993048090335e-06,
      "loss": 0.7621,
      "step": 11027
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.570655043610124,
      "learning_rate": 5.373478955479306e-06,
      "loss": 0.6816,
      "step": 11028
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.537724318700964,
      "learning_rate": 5.372758598354022e-06,
      "loss": 0.79,
      "step": 11029
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.989313968311101,
      "learning_rate": 5.37203823344822e-06,
      "loss": 0.6836,
      "step": 11030
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.7452197611324065,
      "learning_rate": 5.371317860776933e-06,
      "loss": 0.7953,
      "step": 11031
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.589394212759536,
      "learning_rate": 5.3705974803552e-06,
      "loss": 0.7946,
      "step": 11032
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.8469425344130372,
      "learning_rate": 5.369877092198056e-06,
      "loss": 0.7421,
      "step": 11033
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.100493372293296,
      "learning_rate": 5.3691566963205366e-06,
      "loss": 0.7678,
      "step": 11034
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.967245258979909,
      "learning_rate": 5.368436292737681e-06,
      "loss": 0.7573,
      "step": 11035
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.5789971328599846,
      "learning_rate": 5.367715881464524e-06,
      "loss": 0.712,
      "step": 11036
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.136789739081765,
      "learning_rate": 5.3669954625161026e-06,
      "loss": 0.7196,
      "step": 11037
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.376699761360246,
      "learning_rate": 5.366275035907456e-06,
      "loss": 0.7486,
      "step": 11038
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.053732917571989,
      "learning_rate": 5.36555460165362e-06,
      "loss": 0.7453,
      "step": 11039
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.585110943815493,
      "learning_rate": 5.364834159769631e-06,
      "loss": 0.7496,
      "step": 11040
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.734102219818554,
      "learning_rate": 5.364113710270528e-06,
      "loss": 0.718,
      "step": 11041
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.565223050700412,
      "learning_rate": 5.36339325317135e-06,
      "loss": 0.6957,
      "step": 11042
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.010018216746038,
      "learning_rate": 5.362672788487132e-06,
      "loss": 0.738,
      "step": 11043
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.180877472304474,
      "learning_rate": 5.3619523162329145e-06,
      "loss": 0.7165,
      "step": 11044
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.934431818293741,
      "learning_rate": 5.361231836423737e-06,
      "loss": 0.8024,
      "step": 11045
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.79695194664858,
      "learning_rate": 5.360511349074634e-06,
      "loss": 0.7465,
      "step": 11046
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.871524056988522,
      "learning_rate": 5.359790854200646e-06,
      "loss": 0.7095,
      "step": 11047
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.517145947357526,
      "learning_rate": 5.359070351816813e-06,
      "loss": 0.7326,
      "step": 11048
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.573777801895005,
      "learning_rate": 5.358349841938171e-06,
      "loss": 0.7818,
      "step": 11049
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.494414394083082,
      "learning_rate": 5.357629324579762e-06,
      "loss": 0.709,
      "step": 11050
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.574138597080131,
      "learning_rate": 5.356908799756624e-06,
      "loss": 0.7052,
      "step": 11051
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.746537583242266,
      "learning_rate": 5.356188267483798e-06,
      "loss": 0.7231,
      "step": 11052
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.91583823624544,
      "learning_rate": 5.3554677277763186e-06,
      "loss": 0.7404,
      "step": 11053
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.367306997187707,
      "learning_rate": 5.35474718064923e-06,
      "loss": 0.8321,
      "step": 11054
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.7238502956997275,
      "learning_rate": 5.354026626117572e-06,
      "loss": 0.6996,
      "step": 11055
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.5815414442975175,
      "learning_rate": 5.353306064196382e-06,
      "loss": 0.7724,
      "step": 11056
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.6305107650338515,
      "learning_rate": 5.352585494900704e-06,
      "loss": 0.7703,
      "step": 11057
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.232995229234033,
      "learning_rate": 5.351864918245574e-06,
      "loss": 0.711,
      "step": 11058
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.217542122287174,
      "learning_rate": 5.351144334246035e-06,
      "loss": 0.7428,
      "step": 11059
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.188560929620792,
      "learning_rate": 5.350423742917125e-06,
      "loss": 0.7879,
      "step": 11060
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.214590432888082,
      "learning_rate": 5.34970314427389e-06,
      "loss": 0.7533,
      "step": 11061
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.707725072022295,
      "learning_rate": 5.348982538331367e-06,
      "loss": 0.7073,
      "step": 11062
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.603418241949082,
      "learning_rate": 5.348261925104597e-06,
      "loss": 0.7645,
      "step": 11063
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.536911551471322,
      "learning_rate": 5.347541304608622e-06,
      "loss": 0.7006,
      "step": 11064
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.667297216952301,
      "learning_rate": 5.3468206768584825e-06,
      "loss": 0.8391,
      "step": 11065
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.3362277448489985,
      "learning_rate": 5.346100041869222e-06,
      "loss": 0.7298,
      "step": 11066
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.836542275288702,
      "learning_rate": 5.345379399655882e-06,
      "loss": 0.698,
      "step": 11067
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.020966609390918,
      "learning_rate": 5.344658750233503e-06,
      "loss": 0.7337,
      "step": 11068
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.437600069422203,
      "learning_rate": 5.343938093617126e-06,
      "loss": 0.7535,
      "step": 11069
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.41711096062655,
      "learning_rate": 5.343217429821795e-06,
      "loss": 0.7028,
      "step": 11070
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.875467033523789,
      "learning_rate": 5.342496758862552e-06,
      "loss": 0.7131,
      "step": 11071
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.596263239440667,
      "learning_rate": 5.3417760807544386e-06,
      "loss": 0.7458,
      "step": 11072
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.090267105980639,
      "learning_rate": 5.341055395512499e-06,
      "loss": 0.685,
      "step": 11073
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.369899438742318,
      "learning_rate": 5.340334703151775e-06,
      "loss": 0.7364,
      "step": 11074
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.908078120353521,
      "learning_rate": 5.339614003687309e-06,
      "loss": 0.741,
      "step": 11075
    },
    {
      "epoch": 0.99,
      "grad_norm": 8.698311181713988,
      "learning_rate": 5.338893297134143e-06,
      "loss": 0.693,
      "step": 11076
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.333893968370529,
      "learning_rate": 5.338172583507324e-06,
      "loss": 0.744,
      "step": 11077
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.076349717455652,
      "learning_rate": 5.3374518628218915e-06,
      "loss": 0.7614,
      "step": 11078
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.688289627842928,
      "learning_rate": 5.33673113509289e-06,
      "loss": 0.7697,
      "step": 11079
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.001034098495634,
      "learning_rate": 5.336010400335366e-06,
      "loss": 0.7479,
      "step": 11080
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.293700404860387,
      "learning_rate": 5.3352896585643585e-06,
      "loss": 0.7127,
      "step": 11081
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.320618590815339,
      "learning_rate": 5.334568909794914e-06,
      "loss": 0.7957,
      "step": 11082
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.36013408435718,
      "learning_rate": 5.333848154042076e-06,
      "loss": 0.7685,
      "step": 11083
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.425353350697938,
      "learning_rate": 5.333127391320889e-06,
      "loss": 0.7163,
      "step": 11084
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.500995439363926,
      "learning_rate": 5.332406621646397e-06,
      "loss": 0.7429,
      "step": 11085
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.8887881187478297,
      "learning_rate": 5.331685845033646e-06,
      "loss": 0.7087,
      "step": 11086
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.995144707314016,
      "learning_rate": 5.330965061497678e-06,
      "loss": 0.7824,
      "step": 11087
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.272135191288286,
      "learning_rate": 5.3302442710535405e-06,
      "loss": 0.7469,
      "step": 11088
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.007328459995023,
      "learning_rate": 5.329523473716276e-06,
      "loss": 0.7622,
      "step": 11089
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.44990972062913,
      "learning_rate": 5.328802669500931e-06,
      "loss": 0.7835,
      "step": 11090
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.503068382090159,
      "learning_rate": 5.328081858422549e-06,
      "loss": 0.8321,
      "step": 11091
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.735618697969628,
      "learning_rate": 5.3273610404961785e-06,
      "loss": 0.7674,
      "step": 11092
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.625291475298846,
      "learning_rate": 5.326640215736863e-06,
      "loss": 0.7432,
      "step": 11093
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.90159579411522,
      "learning_rate": 5.325919384159647e-06,
      "loss": 0.7451,
      "step": 11094
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.450107463345857,
      "learning_rate": 5.32519854577958e-06,
      "loss": 0.7488,
      "step": 11095
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.948842562469714,
      "learning_rate": 5.324477700611704e-06,
      "loss": 0.7821,
      "step": 11096
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.441562555587381,
      "learning_rate": 5.323756848671067e-06,
      "loss": 0.6984,
      "step": 11097
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.941430913409872,
      "learning_rate": 5.323035989972714e-06,
      "loss": 0.7725,
      "step": 11098
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.152779379546134,
      "learning_rate": 5.322315124531692e-06,
      "loss": 0.7781,
      "step": 11099
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.455084110953734,
      "learning_rate": 5.321594252363048e-06,
      "loss": 0.7319,
      "step": 11100
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.58101134498056,
      "learning_rate": 5.320873373481827e-06,
      "loss": 0.7814,
      "step": 11101
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.804214450777967,
      "learning_rate": 5.32015248790308e-06,
      "loss": 0.7617,
      "step": 11102
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.496609120332069,
      "learning_rate": 5.31943159564185e-06,
      "loss": 0.7617,
      "step": 11103
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.822165199977477,
      "learning_rate": 5.318710696713184e-06,
      "loss": 0.7368,
      "step": 11104
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.3913241971728745,
      "learning_rate": 5.317989791132131e-06,
      "loss": 0.7767,
      "step": 11105
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.966836248529469,
      "learning_rate": 5.3172688789137375e-06,
      "loss": 0.7985,
      "step": 11106
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.569565415828155,
      "learning_rate": 5.31654796007305e-06,
      "loss": 0.7807,
      "step": 11107
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.855007805784122,
      "learning_rate": 5.3158270346251175e-06,
      "loss": 0.6857,
      "step": 11108
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.7023718898429605,
      "learning_rate": 5.315106102584987e-06,
      "loss": 0.7814,
      "step": 11109
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.9456521524255335,
      "learning_rate": 5.314385163967708e-06,
      "loss": 0.7301,
      "step": 11110
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.674611081080166,
      "learning_rate": 5.313664218788326e-06,
      "loss": 0.7883,
      "step": 11111
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.467399671637663,
      "learning_rate": 5.312943267061891e-06,
      "loss": 0.7245,
      "step": 11112
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.001527424091409,
      "learning_rate": 5.3122223088034495e-06,
      "loss": 0.8032,
      "step": 11113
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.129455901720221,
      "learning_rate": 5.311501344028052e-06,
      "loss": 0.6819,
      "step": 11114
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.97272682739805,
      "learning_rate": 5.310780372750746e-06,
      "loss": 0.7385,
      "step": 11115
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.0324857106532175,
      "learning_rate": 5.31005939498658e-06,
      "loss": 0.7178,
      "step": 11116
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.1917020583694145,
      "learning_rate": 5.309338410750604e-06,
      "loss": 0.7157,
      "step": 11117
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.216383665502591,
      "learning_rate": 5.3086174200578654e-06,
      "loss": 0.6789,
      "step": 11118
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.03592545385452,
      "learning_rate": 5.3078964229234145e-06,
      "loss": 0.6922,
      "step": 11119
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.4534259871065744,
      "learning_rate": 5.307175419362299e-06,
      "loss": 0.7069,
      "step": 11120
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.417561982450537,
      "learning_rate": 5.306454409389571e-06,
      "loss": 0.752,
      "step": 11121
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.202252689188949,
      "learning_rate": 5.305733393020276e-06,
      "loss": 0.7489,
      "step": 11122
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.229794599311198,
      "learning_rate": 5.305012370269466e-06,
      "loss": 0.8313,
      "step": 11123
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.846325936212159,
      "learning_rate": 5.304291341152193e-06,
      "loss": 0.779,
      "step": 11124
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.243116482925004,
      "learning_rate": 5.303570305683505e-06,
      "loss": 0.772,
      "step": 11125
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.784615047089913,
      "learning_rate": 5.30284926387845e-06,
      "loss": 0.6988,
      "step": 11126
    },
    {
      "epoch": 0.99,
      "grad_norm": 9.51757655345791,
      "learning_rate": 5.3021282157520805e-06,
      "loss": 0.7724,
      "step": 11127
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.675569148559299,
      "learning_rate": 5.301407161319446e-06,
      "loss": 0.754,
      "step": 11128
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.335646812850775,
      "learning_rate": 5.300686100595597e-06,
      "loss": 0.7535,
      "step": 11129
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.457738368359142,
      "learning_rate": 5.299965033595583e-06,
      "loss": 0.7597,
      "step": 11130
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.664368820911588,
      "learning_rate": 5.299243960334458e-06,
      "loss": 0.6787,
      "step": 11131
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.508968935688197,
      "learning_rate": 5.298522880827269e-06,
      "loss": 0.7746,
      "step": 11132
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.6085156884599785,
      "learning_rate": 5.297801795089069e-06,
      "loss": 0.7747,
      "step": 11133
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.3766850913658,
      "learning_rate": 5.297080703134909e-06,
      "loss": 0.67,
      "step": 11134
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.349451374275716,
      "learning_rate": 5.29635960497984e-06,
      "loss": 0.7498,
      "step": 11135
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.819946108006879,
      "learning_rate": 5.295638500638912e-06,
      "loss": 0.7019,
      "step": 11136
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.988975181112236,
      "learning_rate": 5.29491739012718e-06,
      "loss": 0.8002,
      "step": 11137
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.425962746352961,
      "learning_rate": 5.294196273459692e-06,
      "loss": 0.6879,
      "step": 11138
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.509832263465465,
      "learning_rate": 5.293475150651501e-06,
      "loss": 0.8263,
      "step": 11139
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.163186818760884,
      "learning_rate": 5.29275402171766e-06,
      "loss": 0.7276,
      "step": 11140
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.2366056447604015,
      "learning_rate": 5.292032886673219e-06,
      "loss": 0.6903,
      "step": 11141
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.024996295916544,
      "learning_rate": 5.291311745533231e-06,
      "loss": 0.7363,
      "step": 11142
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.420552446417939,
      "learning_rate": 5.290590598312748e-06,
      "loss": 0.7357,
      "step": 11143
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.416391416934465,
      "learning_rate": 5.289869445026823e-06,
      "loss": 0.8028,
      "step": 11144
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.83724320236594,
      "learning_rate": 5.289148285690508e-06,
      "loss": 0.7402,
      "step": 11145
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.7418543464112113,
      "learning_rate": 5.288427120318856e-06,
      "loss": 0.7111,
      "step": 11146
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.967061001565642,
      "learning_rate": 5.28770594892692e-06,
      "loss": 0.7769,
      "step": 11147
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.656524098120945,
      "learning_rate": 5.286984771529752e-06,
      "loss": 0.8369,
      "step": 11148
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.496758741244326,
      "learning_rate": 5.286263588142404e-06,
      "loss": 0.7336,
      "step": 11149
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.80131776733186,
      "learning_rate": 5.285542398779932e-06,
      "loss": 0.7299,
      "step": 11150
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.255954373363686,
      "learning_rate": 5.284821203457387e-06,
      "loss": 0.7618,
      "step": 11151
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.6399007238879575,
      "learning_rate": 5.284100002189824e-06,
      "loss": 0.7557,
      "step": 11152
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.646126179423611,
      "learning_rate": 5.283378794992297e-06,
      "loss": 0.7998,
      "step": 11153
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.551077985229171,
      "learning_rate": 5.282657581879855e-06,
      "loss": 0.7077,
      "step": 11154
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.887965551964678,
      "learning_rate": 5.281936362867557e-06,
      "loss": 0.784,
      "step": 11155
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.062014855435554,
      "learning_rate": 5.281215137970455e-06,
      "loss": 0.7241,
      "step": 11156
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.917226639235365,
      "learning_rate": 5.280493907203601e-06,
      "loss": 0.7768,
      "step": 11157
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.889973499642333,
      "learning_rate": 5.279772670582051e-06,
      "loss": 0.7557,
      "step": 11158
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.6137971153273805,
      "learning_rate": 5.279051428120862e-06,
      "loss": 0.7933,
      "step": 11159
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.380429543675351,
      "learning_rate": 5.278330179835084e-06,
      "loss": 0.7302,
      "step": 11160
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.799749573624482,
      "learning_rate": 5.2776089257397725e-06,
      "loss": 0.7327,
      "step": 11161
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.868394833331379,
      "learning_rate": 5.276887665849982e-06,
      "loss": 0.8077,
      "step": 11162
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.4912062246692335,
      "learning_rate": 5.27616640018077e-06,
      "loss": 0.7417,
      "step": 11163
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.834906321563252,
      "learning_rate": 5.275445128747188e-06,
      "loss": 0.7388,
      "step": 11164
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.655867120785422,
      "learning_rate": 5.274723851564291e-06,
      "loss": 0.7743,
      "step": 11165
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.291826937812054,
      "learning_rate": 5.274002568647136e-06,
      "loss": 0.7713,
      "step": 11166
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.568515272696508,
      "learning_rate": 5.273281280010779e-06,
      "loss": 0.8011,
      "step": 11167
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.121924475269006,
      "learning_rate": 5.2725599856702726e-06,
      "loss": 0.7356,
      "step": 11168
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.699011438143868,
      "learning_rate": 5.271838685640673e-06,
      "loss": 0.7521,
      "step": 11169
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.980007856629774,
      "learning_rate": 5.271117379937036e-06,
      "loss": 0.7627,
      "step": 11170
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.66795948985015,
      "learning_rate": 5.270396068574418e-06,
      "loss": 0.823,
      "step": 11171
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.194293718619441,
      "learning_rate": 5.269674751567875e-06,
      "loss": 0.718,
      "step": 11172
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.425510881431933,
      "learning_rate": 5.26895342893246e-06,
      "loss": 0.7227,
      "step": 11173
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.969507683872716,
      "learning_rate": 5.268232100683233e-06,
      "loss": 0.75,
      "step": 11174
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.810486147772585,
      "learning_rate": 5.2675107668352484e-06,
      "loss": 0.7721,
      "step": 11175
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.6796142905639115,
      "learning_rate": 5.266789427403562e-06,
      "loss": 0.6626,
      "step": 11176
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.250851617819861,
      "learning_rate": 5.26606808240323e-06,
      "loss": 0.7819,
      "step": 11177
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.237223959406041,
      "learning_rate": 5.265346731849311e-06,
      "loss": 0.6466,
      "step": 11178
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.409894501679666,
      "learning_rate": 5.264625375756858e-06,
      "loss": 0.7991,
      "step": 11179
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.5321797721443895,
      "learning_rate": 5.26390401414093e-06,
      "loss": 0.7389,
      "step": 11180
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.927430400664432,
      "learning_rate": 5.263182647016585e-06,
      "loss": 0.7599,
      "step": 11181
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.367990838837778,
      "learning_rate": 5.262461274398879e-06,
      "loss": 0.7427,
      "step": 11182
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.673366907806207,
      "learning_rate": 5.261739896302867e-06,
      "loss": 0.7264,
      "step": 11183
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.659779488131939,
      "learning_rate": 5.261018512743607e-06,
      "loss": 0.7585,
      "step": 11184
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.399644321678648,
      "learning_rate": 5.26029712373616e-06,
      "loss": 0.7786,
      "step": 11185
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.3341215573417236,
      "learning_rate": 5.259575729295579e-06,
      "loss": 0.6887,
      "step": 11186
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.153505715991838,
      "learning_rate": 5.258854329436923e-06,
      "loss": 0.7236,
      "step": 11187
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.464931435434329,
      "learning_rate": 5.258132924175251e-06,
      "loss": 0.8749,
      "step": 11188
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.683147710879489,
      "learning_rate": 5.257411513525619e-06,
      "loss": 0.7581,
      "step": 11189
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.181117449786323,
      "learning_rate": 5.256690097503084e-06,
      "loss": 0.7166,
      "step": 11190
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.351594563678765,
      "learning_rate": 5.255968676122707e-06,
      "loss": 0.7148,
      "step": 11191
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.3658055852887445,
      "learning_rate": 5.255247249399544e-06,
      "loss": 0.719,
      "step": 11192
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.754852526676091,
      "learning_rate": 5.2545258173486525e-06,
      "loss": 0.7795,
      "step": 11193
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.84950436200714,
      "learning_rate": 5.253804379985094e-06,
      "loss": 0.7463,
      "step": 11194
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.323464547394593,
      "learning_rate": 5.253082937323923e-06,
      "loss": 0.702,
      "step": 11195
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.255173612874839,
      "learning_rate": 5.252361489380201e-06,
      "loss": 0.759,
      "step": 11196
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.651375860622127,
      "learning_rate": 5.2516400361689846e-06,
      "loss": 0.7888,
      "step": 11197
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.691488313503528,
      "learning_rate": 5.250918577705335e-06,
      "loss": 0.7758,
      "step": 11198
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.339017942335485,
      "learning_rate": 5.2501971140043085e-06,
      "loss": 0.8201,
      "step": 11199
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.588977533193598,
      "learning_rate": 5.2494756450809644e-06,
      "loss": 0.7573,
      "step": 11200
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.49654528048211,
      "learning_rate": 5.248754170950365e-06,
      "loss": 0.7584,
      "step": 11201
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.55074267439653,
      "learning_rate": 5.248032691627565e-06,
      "loss": 0.75,
      "step": 11202
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.4186145457978085,
      "learning_rate": 5.247311207127625e-06,
      "loss": 0.8155,
      "step": 11203
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.817904208077326,
      "learning_rate": 5.246589717465609e-06,
      "loss": 0.7919,
      "step": 11204
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.612027768834955,
      "learning_rate": 5.24586822265657e-06,
      "loss": 0.7815,
      "step": 11205
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.736399329693415,
      "learning_rate": 5.245146722715571e-06,
      "loss": 0.8042,
      "step": 11206
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.609914591005141,
      "learning_rate": 5.244425217657671e-06,
      "loss": 0.7256,
      "step": 11207
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.833565041493093,
      "learning_rate": 5.243703707497928e-06,
      "loss": 0.7549,
      "step": 11208
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.1984447723944385,
      "learning_rate": 5.242982192251405e-06,
      "loss": 0.7457,
      "step": 11209
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.597889396423226,
      "learning_rate": 5.242260671933162e-06,
      "loss": 0.7081,
      "step": 11210
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.748625060787291,
      "learning_rate": 5.2415391465582574e-06,
      "loss": 0.6447,
      "step": 11211
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.062348679614206,
      "learning_rate": 5.240817616141751e-06,
      "loss": 0.6636,
      "step": 11212
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.3993536315792925,
      "learning_rate": 5.240096080698704e-06,
      "loss": 0.5966,
      "step": 11213
    },
    {
      "epoch": 1.0,
      "grad_norm": 8.930451492117777,
      "learning_rate": 5.23937454024418e-06,
      "loss": 0.6387,
      "step": 11214
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.238289855437982,
      "learning_rate": 5.238652994793234e-06,
      "loss": 0.653,
      "step": 11215
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.247595818884971,
      "learning_rate": 5.2379314443609305e-06,
      "loss": 0.6438,
      "step": 11216
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.314688446725929,
      "learning_rate": 5.237209888962329e-06,
      "loss": 0.6966,
      "step": 11217
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.5633923668492025,
      "learning_rate": 5.236488328612491e-06,
      "loss": 0.6336,
      "step": 11218
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.268957245928129,
      "learning_rate": 5.235766763326475e-06,
      "loss": 0.6396,
      "step": 11219
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.611266944803511,
      "learning_rate": 5.235045193119347e-06,
      "loss": 0.5682,
      "step": 11220
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.540538641915241,
      "learning_rate": 5.234323618006163e-06,
      "loss": 0.6084,
      "step": 11221
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.108061926770547,
      "learning_rate": 5.233602038001988e-06,
      "loss": 0.6118,
      "step": 11222
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.024405065718537,
      "learning_rate": 5.232880453121881e-06,
      "loss": 0.6483,
      "step": 11223
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.859788820733571,
      "learning_rate": 5.232158863380905e-06,
      "loss": 0.5988,
      "step": 11224
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.543978678386403,
      "learning_rate": 5.231437268794123e-06,
      "loss": 0.628,
      "step": 11225
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.696374092180809,
      "learning_rate": 5.230715669376594e-06,
      "loss": 0.6055,
      "step": 11226
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.2664878234742565,
      "learning_rate": 5.229994065143379e-06,
      "loss": 0.6701,
      "step": 11227
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.030958593596821,
      "learning_rate": 5.229272456109541e-06,
      "loss": 0.6336,
      "step": 11228
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.497627142481068,
      "learning_rate": 5.228550842290146e-06,
      "loss": 0.6671,
      "step": 11229
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.709481196881171,
      "learning_rate": 5.227829223700249e-06,
      "loss": 0.7214,
      "step": 11230
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.36234947262567,
      "learning_rate": 5.227107600354917e-06,
      "loss": 0.5956,
      "step": 11231
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.723871360136007,
      "learning_rate": 5.2263859722692125e-06,
      "loss": 0.6789,
      "step": 11232
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.284392976485072,
      "learning_rate": 5.225664339458197e-06,
      "loss": 0.6497,
      "step": 11233
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.0378673568909695,
      "learning_rate": 5.224942701936931e-06,
      "loss": 0.5965,
      "step": 11234
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.414616625115518,
      "learning_rate": 5.224221059720478e-06,
      "loss": 0.5578,
      "step": 11235
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.877565732033814,
      "learning_rate": 5.223499412823904e-06,
      "loss": 0.5942,
      "step": 11236
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.309384903287145,
      "learning_rate": 5.222777761262266e-06,
      "loss": 0.7337,
      "step": 11237
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.343336920215003,
      "learning_rate": 5.222056105050632e-06,
      "loss": 0.6545,
      "step": 11238
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.351688957636159,
      "learning_rate": 5.221334444204063e-06,
      "loss": 0.6215,
      "step": 11239
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.007742946798826,
      "learning_rate": 5.220612778737623e-06,
      "loss": 0.6234,
      "step": 11240
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.505872130936188,
      "learning_rate": 5.219891108666374e-06,
      "loss": 0.6134,
      "step": 11241
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.262369688442417,
      "learning_rate": 5.219169434005379e-06,
      "loss": 0.6429,
      "step": 11242
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.6723811399728215,
      "learning_rate": 5.218447754769702e-06,
      "loss": 0.5991,
      "step": 11243
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.563464628838593,
      "learning_rate": 5.217726070974406e-06,
      "loss": 0.616,
      "step": 11244
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.987681432158209,
      "learning_rate": 5.2170043826345575e-06,
      "loss": 0.6316,
      "step": 11245
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.753575118632956,
      "learning_rate": 5.216282689765215e-06,
      "loss": 0.6656,
      "step": 11246
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.573892396098607,
      "learning_rate": 5.2155609923814465e-06,
      "loss": 0.6179,
      "step": 11247
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.381419439276616,
      "learning_rate": 5.214839290498313e-06,
      "loss": 0.6519,
      "step": 11248
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.067877224474092,
      "learning_rate": 5.214117584130882e-06,
      "loss": 0.5643,
      "step": 11249
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.557330105462775,
      "learning_rate": 5.213395873294213e-06,
      "loss": 0.6092,
      "step": 11250
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.034332328434427,
      "learning_rate": 5.2126741580033735e-06,
      "loss": 0.6496,
      "step": 11251
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.355680397065146,
      "learning_rate": 5.211952438273428e-06,
      "loss": 0.6821,
      "step": 11252
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.123372540034472,
      "learning_rate": 5.211230714119437e-06,
      "loss": 0.628,
      "step": 11253
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.029464815059172,
      "learning_rate": 5.2105089855564695e-06,
      "loss": 0.6615,
      "step": 11254
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.764609167748253,
      "learning_rate": 5.209787252599589e-06,
      "loss": 0.5431,
      "step": 11255
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.288690533502914,
      "learning_rate": 5.209065515263856e-06,
      "loss": 0.6318,
      "step": 11256
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.509925678063468,
      "learning_rate": 5.208343773564338e-06,
      "loss": 0.6449,
      "step": 11257
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.33513383864488,
      "learning_rate": 5.2076220275161026e-06,
      "loss": 0.5876,
      "step": 11258
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.979975109683134,
      "learning_rate": 5.20690027713421e-06,
      "loss": 0.5983,
      "step": 11259
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.315631701488811,
      "learning_rate": 5.206178522433728e-06,
      "loss": 0.6531,
      "step": 11260
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.604982178387352,
      "learning_rate": 5.205456763429721e-06,
      "loss": 0.6059,
      "step": 11261
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.986883389002421,
      "learning_rate": 5.204735000137255e-06,
      "loss": 0.59,
      "step": 11262
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.736180465463961,
      "learning_rate": 5.204013232571393e-06,
      "loss": 0.7132,
      "step": 11263
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.712472845251881,
      "learning_rate": 5.203291460747202e-06,
      "loss": 0.5942,
      "step": 11264
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.084005484462115,
      "learning_rate": 5.202569684679747e-06,
      "loss": 0.6562,
      "step": 11265
    },
    {
      "epoch": 1.01,
      "grad_norm": 7.45185463871948,
      "learning_rate": 5.201847904384092e-06,
      "loss": 0.6085,
      "step": 11266
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.676436496535014,
      "learning_rate": 5.201126119875306e-06,
      "loss": 0.5681,
      "step": 11267
    },
    {
      "epoch": 1.01,
      "grad_norm": 8.133483041093939,
      "learning_rate": 5.200404331168453e-06,
      "loss": 0.6041,
      "step": 11268
    },
    {
      "epoch": 1.01,
      "grad_norm": 6.873792717964926,
      "learning_rate": 5.199682538278596e-06,
      "loss": 0.6291,
      "step": 11269
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.15650307295505,
      "learning_rate": 5.198960741220804e-06,
      "loss": 0.6082,
      "step": 11270
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.694406860820382,
      "learning_rate": 5.198238940010144e-06,
      "loss": 0.5986,
      "step": 11271
    },
    {
      "epoch": 1.01,
      "grad_norm": 7.496649946112167,
      "learning_rate": 5.197517134661678e-06,
      "loss": 0.5993,
      "step": 11272
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.718472022820167,
      "learning_rate": 5.1967953251904755e-06,
      "loss": 0.6241,
      "step": 11273
    },
    {
      "epoch": 1.01,
      "grad_norm": 7.20808586463834,
      "learning_rate": 5.196073511611602e-06,
      "loss": 0.6282,
      "step": 11274
    },
    {
      "epoch": 1.01,
      "grad_norm": 7.288236761807121,
      "learning_rate": 5.195351693940123e-06,
      "loss": 0.6424,
      "step": 11275
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.255930015952369,
      "learning_rate": 5.194629872191106e-06,
      "loss": 0.6187,
      "step": 11276
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.469966388699414,
      "learning_rate": 5.193908046379616e-06,
      "loss": 0.6474,
      "step": 11277
    },
    {
      "epoch": 1.01,
      "grad_norm": 6.076805105153063,
      "learning_rate": 5.193186216520719e-06,
      "loss": 0.6385,
      "step": 11278
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.07908578794181,
      "learning_rate": 5.192464382629484e-06,
      "loss": 0.6007,
      "step": 11279
    },
    {
      "epoch": 1.01,
      "grad_norm": 4.811831026601546,
      "learning_rate": 5.191742544720977e-06,
      "loss": 0.6071,
      "step": 11280
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.508781304423288,
      "learning_rate": 5.191020702810262e-06,
      "loss": 0.6485,
      "step": 11281
    },
    {
      "epoch": 1.01,
      "grad_norm": 4.899720311240473,
      "learning_rate": 5.1902988569124105e-06,
      "loss": 0.6207,
      "step": 11282
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.039429607760431,
      "learning_rate": 5.189577007042487e-06,
      "loss": 0.5611,
      "step": 11283
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.954435534763285,
      "learning_rate": 5.188855153215557e-06,
      "loss": 0.5884,
      "step": 11284
    },
    {
      "epoch": 1.01,
      "grad_norm": 6.033904674662508,
      "learning_rate": 5.18813329544669e-06,
      "loss": 0.5819,
      "step": 11285
    },
    {
      "epoch": 1.01,
      "grad_norm": 4.774354861066126,
      "learning_rate": 5.187411433750953e-06,
      "loss": 0.6301,
      "step": 11286
    },
    {
      "epoch": 1.01,
      "grad_norm": 7.414579316868729,
      "learning_rate": 5.186689568143413e-06,
      "loss": 0.5818,
      "step": 11287
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.586596101825275,
      "learning_rate": 5.185967698639136e-06,
      "loss": 0.6479,
      "step": 11288
    },
    {
      "epoch": 1.01,
      "grad_norm": 6.54473033147053,
      "learning_rate": 5.1852458252531905e-06,
      "loss": 0.5921,
      "step": 11289
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.636608948701973,
      "learning_rate": 5.1845239480006456e-06,
      "loss": 0.6375,
      "step": 11290
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.738915246304771,
      "learning_rate": 5.183802066896567e-06,
      "loss": 0.6651,
      "step": 11291
    },
    {
      "epoch": 1.01,
      "grad_norm": 6.1904329214343,
      "learning_rate": 5.183080181956022e-06,
      "loss": 0.5846,
      "step": 11292
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.824754184344088,
      "learning_rate": 5.18235829319408e-06,
      "loss": 0.6611,
      "step": 11293
    },
    {
      "epoch": 1.01,
      "grad_norm": 7.014528694596895,
      "learning_rate": 5.181636400625807e-06,
      "loss": 0.6666,
      "step": 11294
    },
    {
      "epoch": 1.01,
      "grad_norm": 4.571633673316373,
      "learning_rate": 5.1809145042662715e-06,
      "loss": 0.5989,
      "step": 11295
    },
    {
      "epoch": 1.01,
      "grad_norm": 7.318148196236044,
      "learning_rate": 5.180192604130543e-06,
      "loss": 0.6382,
      "step": 11296
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.133725040963876,
      "learning_rate": 5.179470700233689e-06,
      "loss": 0.5716,
      "step": 11297
    },
    {
      "epoch": 1.01,
      "grad_norm": 6.0688069296680425,
      "learning_rate": 5.178748792590775e-06,
      "loss": 0.6643,
      "step": 11298
    },
    {
      "epoch": 1.01,
      "grad_norm": 6.923699030955574,
      "learning_rate": 5.178026881216873e-06,
      "loss": 0.6344,
      "step": 11299
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.84547869303841,
      "learning_rate": 5.17730496612705e-06,
      "loss": 0.6338,
      "step": 11300
    },
    {
      "epoch": 1.01,
      "grad_norm": 8.434718993274839,
      "learning_rate": 5.1765830473363734e-06,
      "loss": 0.5738,
      "step": 11301
    },
    {
      "epoch": 1.01,
      "grad_norm": 6.423533308120808,
      "learning_rate": 5.175861124859914e-06,
      "loss": 0.6633,
      "step": 11302
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.697385366193451,
      "learning_rate": 5.175139198712737e-06,
      "loss": 0.5872,
      "step": 11303
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.891422887680415,
      "learning_rate": 5.174417268909913e-06,
      "loss": 0.5506,
      "step": 11304
    },
    {
      "epoch": 1.01,
      "grad_norm": 7.322485467255316,
      "learning_rate": 5.17369533546651e-06,
      "loss": 0.6341,
      "step": 11305
    },
    {
      "epoch": 1.01,
      "grad_norm": 6.355660539530603,
      "learning_rate": 5.172973398397599e-06,
      "loss": 0.6686,
      "step": 11306
    },
    {
      "epoch": 1.01,
      "grad_norm": 6.871315290234635,
      "learning_rate": 5.172251457718245e-06,
      "loss": 0.6038,
      "step": 11307
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.184692412395197,
      "learning_rate": 5.171529513443519e-06,
      "loss": 0.5729,
      "step": 11308
    },
    {
      "epoch": 1.01,
      "grad_norm": 6.969204663915408,
      "learning_rate": 5.170807565588493e-06,
      "loss": 0.6186,
      "step": 11309
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.569334678967067,
      "learning_rate": 5.17008561416823e-06,
      "loss": 0.6076,
      "step": 11310
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.523142912042274,
      "learning_rate": 5.169363659197802e-06,
      "loss": 0.6096,
      "step": 11311
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.73092529076704,
      "learning_rate": 5.168641700692283e-06,
      "loss": 0.5501,
      "step": 11312
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.8714193386360805,
      "learning_rate": 5.167919738666735e-06,
      "loss": 0.6165,
      "step": 11313
    },
    {
      "epoch": 1.01,
      "grad_norm": 7.258713871809443,
      "learning_rate": 5.167197773136229e-06,
      "loss": 0.598,
      "step": 11314
    },
    {
      "epoch": 1.01,
      "grad_norm": 6.947388269975274,
      "learning_rate": 5.166475804115837e-06,
      "loss": 0.5948,
      "step": 11315
    },
    {
      "epoch": 1.01,
      "grad_norm": 6.736932915945489,
      "learning_rate": 5.165753831620626e-06,
      "loss": 0.6708,
      "step": 11316
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.153098203638437,
      "learning_rate": 5.165031855665668e-06,
      "loss": 0.5855,
      "step": 11317
    },
    {
      "epoch": 1.01,
      "grad_norm": 4.562760381941685,
      "learning_rate": 5.1643098762660315e-06,
      "loss": 0.5721,
      "step": 11318
    },
    {
      "epoch": 1.01,
      "grad_norm": 8.506232744561556,
      "learning_rate": 5.163587893436785e-06,
      "loss": 0.6036,
      "step": 11319
    },
    {
      "epoch": 1.01,
      "grad_norm": 9.770030512475426,
      "learning_rate": 5.162865907193001e-06,
      "loss": 0.626,
      "step": 11320
    },
    {
      "epoch": 1.01,
      "grad_norm": 4.3090432061831825,
      "learning_rate": 5.162143917549746e-06,
      "loss": 0.6524,
      "step": 11321
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.779302475295813,
      "learning_rate": 5.161421924522093e-06,
      "loss": 0.5916,
      "step": 11322
    },
    {
      "epoch": 1.01,
      "grad_norm": 8.833145520561269,
      "learning_rate": 5.16069992812511e-06,
      "loss": 0.5944,
      "step": 11323
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.011355850799018,
      "learning_rate": 5.1599779283738695e-06,
      "loss": 0.6226,
      "step": 11324
    },
    {
      "epoch": 1.01,
      "grad_norm": 7.183863226093159,
      "learning_rate": 5.159255925283439e-06,
      "loss": 0.605,
      "step": 11325
    },
    {
      "epoch": 1.01,
      "grad_norm": 6.0740766451245385,
      "learning_rate": 5.158533918868891e-06,
      "loss": 0.5745,
      "step": 11326
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.878321986070727,
      "learning_rate": 5.157811909145293e-06,
      "loss": 0.5789,
      "step": 11327
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.282948427254995,
      "learning_rate": 5.157089896127718e-06,
      "loss": 0.6432,
      "step": 11328
    },
    {
      "epoch": 1.01,
      "grad_norm": 4.564376770122776,
      "learning_rate": 5.1563678798312344e-06,
      "loss": 0.5911,
      "step": 11329
    },
    {
      "epoch": 1.01,
      "grad_norm": 4.838746964893542,
      "learning_rate": 5.155645860270914e-06,
      "loss": 0.6105,
      "step": 11330
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.7145180033186485,
      "learning_rate": 5.154923837461828e-06,
      "loss": 0.5988,
      "step": 11331
    },
    {
      "epoch": 1.01,
      "grad_norm": 8.442825008992337,
      "learning_rate": 5.154201811419046e-06,
      "loss": 0.6685,
      "step": 11332
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.595385631103522,
      "learning_rate": 5.153479782157639e-06,
      "loss": 0.6457,
      "step": 11333
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.34937284454787,
      "learning_rate": 5.152757749692676e-06,
      "loss": 0.5737,
      "step": 11334
    },
    {
      "epoch": 1.01,
      "grad_norm": 6.591145974024,
      "learning_rate": 5.152035714039232e-06,
      "loss": 0.6489,
      "step": 11335
    },
    {
      "epoch": 1.01,
      "grad_norm": 7.402016566084002,
      "learning_rate": 5.151313675212374e-06,
      "loss": 0.6102,
      "step": 11336
    },
    {
      "epoch": 1.01,
      "grad_norm": 6.1229472353528624,
      "learning_rate": 5.150591633227174e-06,
      "loss": 0.6238,
      "step": 11337
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.231695503645207,
      "learning_rate": 5.1498695880987046e-06,
      "loss": 0.5882,
      "step": 11338
    },
    {
      "epoch": 1.01,
      "grad_norm": 7.151166310107379,
      "learning_rate": 5.149147539842034e-06,
      "loss": 0.591,
      "step": 11339
    },
    {
      "epoch": 1.01,
      "grad_norm": 7.861912067328402,
      "learning_rate": 5.148425488472238e-06,
      "loss": 0.6251,
      "step": 11340
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.348021636205391,
      "learning_rate": 5.147703434004383e-06,
      "loss": 0.6583,
      "step": 11341
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.4612281392657325,
      "learning_rate": 5.146981376453542e-06,
      "loss": 0.6122,
      "step": 11342
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.41400060830754,
      "learning_rate": 5.1462593158347875e-06,
      "loss": 0.5711,
      "step": 11343
    },
    {
      "epoch": 1.01,
      "grad_norm": 7.327502499633114,
      "learning_rate": 5.14553725216319e-06,
      "loss": 0.6151,
      "step": 11344
    },
    {
      "epoch": 1.01,
      "grad_norm": 4.75494020510985,
      "learning_rate": 5.1448151854538205e-06,
      "loss": 0.642,
      "step": 11345
    },
    {
      "epoch": 1.01,
      "grad_norm": 7.484040052999811,
      "learning_rate": 5.144093115721751e-06,
      "loss": 0.634,
      "step": 11346
    },
    {
      "epoch": 1.01,
      "grad_norm": 6.086948507403673,
      "learning_rate": 5.143371042982055e-06,
      "loss": 0.634,
      "step": 11347
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.0331070225793235,
      "learning_rate": 5.142648967249802e-06,
      "loss": 0.6161,
      "step": 11348
    },
    {
      "epoch": 1.01,
      "grad_norm": 7.315633092783068,
      "learning_rate": 5.141926888540063e-06,
      "loss": 0.6645,
      "step": 11349
    },
    {
      "epoch": 1.01,
      "grad_norm": 4.4106214879668295,
      "learning_rate": 5.141204806867912e-06,
      "loss": 0.5429,
      "step": 11350
    },
    {
      "epoch": 1.01,
      "grad_norm": 8.334803539631114,
      "learning_rate": 5.140482722248419e-06,
      "loss": 0.6336,
      "step": 11351
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.394827787517125,
      "learning_rate": 5.139760634696656e-06,
      "loss": 0.6273,
      "step": 11352
    },
    {
      "epoch": 1.01,
      "grad_norm": 6.905956357096545,
      "learning_rate": 5.139038544227698e-06,
      "loss": 0.6243,
      "step": 11353
    },
    {
      "epoch": 1.01,
      "grad_norm": 8.014263228755812,
      "learning_rate": 5.138316450856613e-06,
      "loss": 0.6195,
      "step": 11354
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.836077695173028,
      "learning_rate": 5.137594354598477e-06,
      "loss": 0.6169,
      "step": 11355
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.1367932210217955,
      "learning_rate": 5.136872255468358e-06,
      "loss": 0.5456,
      "step": 11356
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.5981287360020495,
      "learning_rate": 5.136150153481333e-06,
      "loss": 0.6165,
      "step": 11357
    },
    {
      "epoch": 1.01,
      "grad_norm": 7.085463309280587,
      "learning_rate": 5.1354280486524696e-06,
      "loss": 0.6242,
      "step": 11358
    },
    {
      "epoch": 1.01,
      "grad_norm": 6.110363330752762,
      "learning_rate": 5.134705940996843e-06,
      "loss": 0.6686,
      "step": 11359
    },
    {
      "epoch": 1.01,
      "grad_norm": 7.299276794193033,
      "learning_rate": 5.133983830529524e-06,
      "loss": 0.6537,
      "step": 11360
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.325401268716434,
      "learning_rate": 5.133261717265586e-06,
      "loss": 0.5777,
      "step": 11361
    },
    {
      "epoch": 1.01,
      "grad_norm": 9.333297579746171,
      "learning_rate": 5.132539601220104e-06,
      "loss": 0.6107,
      "step": 11362
    },
    {
      "epoch": 1.01,
      "grad_norm": 7.3752934644884505,
      "learning_rate": 5.1318174824081465e-06,
      "loss": 0.6551,
      "step": 11363
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.351167156154887,
      "learning_rate": 5.131095360844787e-06,
      "loss": 0.6338,
      "step": 11364
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.636800416642211,
      "learning_rate": 5.130373236545099e-06,
      "loss": 0.6864,
      "step": 11365
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.51638844958901,
      "learning_rate": 5.1296511095241565e-06,
      "loss": 0.6207,
      "step": 11366
    },
    {
      "epoch": 1.01,
      "grad_norm": 8.746745259314302,
      "learning_rate": 5.1289289797970295e-06,
      "loss": 0.6467,
      "step": 11367
    },
    {
      "epoch": 1.01,
      "grad_norm": 6.854676511438797,
      "learning_rate": 5.1282068473787925e-06,
      "loss": 0.7029,
      "step": 11368
    },
    {
      "epoch": 1.01,
      "grad_norm": 6.69968387974251,
      "learning_rate": 5.127484712284519e-06,
      "loss": 0.5428,
      "step": 11369
    },
    {
      "epoch": 1.01,
      "grad_norm": 8.134382169637025,
      "learning_rate": 5.126762574529283e-06,
      "loss": 0.6022,
      "step": 11370
    },
    {
      "epoch": 1.01,
      "grad_norm": 7.759521435004208,
      "learning_rate": 5.126040434128153e-06,
      "loss": 0.6066,
      "step": 11371
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.393437943141763,
      "learning_rate": 5.125318291096205e-06,
      "loss": 0.6939,
      "step": 11372
    },
    {
      "epoch": 1.01,
      "grad_norm": 6.8047975171466835,
      "learning_rate": 5.124596145448514e-06,
      "loss": 0.6618,
      "step": 11373
    },
    {
      "epoch": 1.01,
      "grad_norm": 6.247199510972823,
      "learning_rate": 5.12387399720015e-06,
      "loss": 0.6096,
      "step": 11374
    },
    {
      "epoch": 1.01,
      "grad_norm": 6.5524838837178825,
      "learning_rate": 5.123151846366188e-06,
      "loss": 0.6278,
      "step": 11375
    },
    {
      "epoch": 1.01,
      "grad_norm": 6.665490253629418,
      "learning_rate": 5.122429692961701e-06,
      "loss": 0.6218,
      "step": 11376
    },
    {
      "epoch": 1.01,
      "grad_norm": 4.194413799344783,
      "learning_rate": 5.121707537001764e-06,
      "loss": 0.5666,
      "step": 11377
    },
    {
      "epoch": 1.02,
      "grad_norm": 7.300511763302451,
      "learning_rate": 5.120985378501445e-06,
      "loss": 0.6404,
      "step": 11378
    },
    {
      "epoch": 1.02,
      "grad_norm": 8.216023387768901,
      "learning_rate": 5.120263217475825e-06,
      "loss": 0.6223,
      "step": 11379
    },
    {
      "epoch": 1.02,
      "grad_norm": 10.961711852198805,
      "learning_rate": 5.119541053939971e-06,
      "loss": 0.6037,
      "step": 11380
    },
    {
      "epoch": 1.02,
      "grad_norm": 7.318922298290036,
      "learning_rate": 5.118818887908959e-06,
      "loss": 0.5804,
      "step": 11381
    },
    {
      "epoch": 1.02,
      "grad_norm": 7.836134337484413,
      "learning_rate": 5.118096719397865e-06,
      "loss": 0.5829,
      "step": 11382
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.489477004067507,
      "learning_rate": 5.11737454842176e-06,
      "loss": 0.5761,
      "step": 11383
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.021797842592924,
      "learning_rate": 5.116652374995718e-06,
      "loss": 0.6966,
      "step": 11384
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.825999992030382,
      "learning_rate": 5.115930199134813e-06,
      "loss": 0.6721,
      "step": 11385
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.123574970274254,
      "learning_rate": 5.115208020854119e-06,
      "loss": 0.6216,
      "step": 11386
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.992222818863251,
      "learning_rate": 5.114485840168709e-06,
      "loss": 0.6024,
      "step": 11387
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.084441262004154,
      "learning_rate": 5.11376365709366e-06,
      "loss": 0.5806,
      "step": 11388
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.849311915104958,
      "learning_rate": 5.1130414716440425e-06,
      "loss": 0.609,
      "step": 11389
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.602094818139735,
      "learning_rate": 5.112319283834931e-06,
      "loss": 0.6055,
      "step": 11390
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.54100686171314,
      "learning_rate": 5.111597093681402e-06,
      "loss": 0.6078,
      "step": 11391
    },
    {
      "epoch": 1.02,
      "grad_norm": 11.611014210205434,
      "learning_rate": 5.110874901198528e-06,
      "loss": 0.6585,
      "step": 11392
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.216896573717639,
      "learning_rate": 5.110152706401383e-06,
      "loss": 0.5489,
      "step": 11393
    },
    {
      "epoch": 1.02,
      "grad_norm": 10.463082972611163,
      "learning_rate": 5.10943050930504e-06,
      "loss": 0.6212,
      "step": 11394
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.602228594882577,
      "learning_rate": 5.108708309924576e-06,
      "loss": 0.6237,
      "step": 11395
    },
    {
      "epoch": 1.02,
      "grad_norm": 7.5767981249845535,
      "learning_rate": 5.107986108275064e-06,
      "loss": 0.6055,
      "step": 11396
    },
    {
      "epoch": 1.02,
      "grad_norm": 8.111452852558513,
      "learning_rate": 5.107263904371578e-06,
      "loss": 0.5733,
      "step": 11397
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.63531002547118,
      "learning_rate": 5.106541698229195e-06,
      "loss": 0.633,
      "step": 11398
    },
    {
      "epoch": 1.02,
      "grad_norm": 8.978541470179241,
      "learning_rate": 5.105819489862985e-06,
      "loss": 0.5728,
      "step": 11399
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.029244991781461,
      "learning_rate": 5.105097279288025e-06,
      "loss": 0.6509,
      "step": 11400
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.529001024073215,
      "learning_rate": 5.104375066519391e-06,
      "loss": 0.5786,
      "step": 11401
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.948032124036982,
      "learning_rate": 5.103652851572155e-06,
      "loss": 0.6255,
      "step": 11402
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.975780918465593,
      "learning_rate": 5.102930634461393e-06,
      "loss": 0.6729,
      "step": 11403
    },
    {
      "epoch": 1.02,
      "grad_norm": 7.316359102916173,
      "learning_rate": 5.102208415202179e-06,
      "loss": 0.6679,
      "step": 11404
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.958832126274536,
      "learning_rate": 5.101486193809589e-06,
      "loss": 0.6289,
      "step": 11405
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.477458264689184,
      "learning_rate": 5.100763970298697e-06,
      "loss": 0.6739,
      "step": 11406
    },
    {
      "epoch": 1.02,
      "grad_norm": 4.656585999986154,
      "learning_rate": 5.100041744684576e-06,
      "loss": 0.6204,
      "step": 11407
    },
    {
      "epoch": 1.02,
      "grad_norm": 4.544282710556619,
      "learning_rate": 5.099319516982304e-06,
      "loss": 0.6007,
      "step": 11408
    },
    {
      "epoch": 1.02,
      "grad_norm": 4.982818665043,
      "learning_rate": 5.0985972872069545e-06,
      "loss": 0.6201,
      "step": 11409
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.001293771451889,
      "learning_rate": 5.097875055373602e-06,
      "loss": 0.5226,
      "step": 11410
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.611818856797479,
      "learning_rate": 5.0971528214973235e-06,
      "loss": 0.6213,
      "step": 11411
    },
    {
      "epoch": 1.02,
      "grad_norm": 7.6068912902888375,
      "learning_rate": 5.09643058559319e-06,
      "loss": 0.6869,
      "step": 11412
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.867969643151915,
      "learning_rate": 5.095708347676282e-06,
      "loss": 0.6663,
      "step": 11413
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.728334604144465,
      "learning_rate": 5.0949861077616715e-06,
      "loss": 0.534,
      "step": 11414
    },
    {
      "epoch": 1.02,
      "grad_norm": 8.294957119915058,
      "learning_rate": 5.094263865864433e-06,
      "loss": 0.6413,
      "step": 11415
    },
    {
      "epoch": 1.02,
      "grad_norm": 7.746751795752345,
      "learning_rate": 5.0935416219996426e-06,
      "loss": 0.6369,
      "step": 11416
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.030213528351936,
      "learning_rate": 5.092819376182376e-06,
      "loss": 0.5969,
      "step": 11417
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.405853745610788,
      "learning_rate": 5.092097128427709e-06,
      "loss": 0.6051,
      "step": 11418
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.235119267744857,
      "learning_rate": 5.091374878750714e-06,
      "loss": 0.5606,
      "step": 11419
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.632487551018236,
      "learning_rate": 5.09065262716647e-06,
      "loss": 0.6539,
      "step": 11420
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.995152608978691,
      "learning_rate": 5.089930373690052e-06,
      "loss": 0.6431,
      "step": 11421
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.768388364564545,
      "learning_rate": 5.0892081183365325e-06,
      "loss": 0.6064,
      "step": 11422
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.3847951495610245,
      "learning_rate": 5.088485861120991e-06,
      "loss": 0.6294,
      "step": 11423
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.314571781682193,
      "learning_rate": 5.087763602058498e-06,
      "loss": 0.5334,
      "step": 11424
    },
    {
      "epoch": 1.02,
      "grad_norm": 8.137603820930277,
      "learning_rate": 5.087041341164134e-06,
      "loss": 0.6365,
      "step": 11425
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.676398265810293,
      "learning_rate": 5.086319078452973e-06,
      "loss": 0.6493,
      "step": 11426
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.531140024359236,
      "learning_rate": 5.08559681394009e-06,
      "loss": 0.6098,
      "step": 11427
    },
    {
      "epoch": 1.02,
      "grad_norm": 4.795685367607865,
      "learning_rate": 5.084874547640559e-06,
      "loss": 0.6085,
      "step": 11428
    },
    {
      "epoch": 1.02,
      "grad_norm": 4.977715780749774,
      "learning_rate": 5.08415227956946e-06,
      "loss": 0.5523,
      "step": 11429
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.701954916975521,
      "learning_rate": 5.0834300097418656e-06,
      "loss": 0.6568,
      "step": 11430
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.825649859262631,
      "learning_rate": 5.082707738172852e-06,
      "loss": 0.6338,
      "step": 11431
    },
    {
      "epoch": 1.02,
      "grad_norm": 7.424576363078336,
      "learning_rate": 5.081985464877496e-06,
      "loss": 0.6646,
      "step": 11432
    },
    {
      "epoch": 1.02,
      "grad_norm": 7.204824288897726,
      "learning_rate": 5.081263189870873e-06,
      "loss": 0.6183,
      "step": 11433
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.462092594874722,
      "learning_rate": 5.080540913168059e-06,
      "loss": 0.5577,
      "step": 11434
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.28645645414889,
      "learning_rate": 5.079818634784129e-06,
      "loss": 0.5972,
      "step": 11435
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.631719639151833,
      "learning_rate": 5.079096354734161e-06,
      "loss": 0.6259,
      "step": 11436
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.667478946581205,
      "learning_rate": 5.078374073033229e-06,
      "loss": 0.6103,
      "step": 11437
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.7740575441181665,
      "learning_rate": 5.077651789696409e-06,
      "loss": 0.5899,
      "step": 11438
    },
    {
      "epoch": 1.02,
      "grad_norm": 7.32525287815923,
      "learning_rate": 5.07692950473878e-06,
      "loss": 0.6151,
      "step": 11439
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.281783797969762,
      "learning_rate": 5.076207218175415e-06,
      "loss": 0.5403,
      "step": 11440
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.774511354060808,
      "learning_rate": 5.075484930021392e-06,
      "loss": 0.6538,
      "step": 11441
    },
    {
      "epoch": 1.02,
      "grad_norm": 8.662274836108477,
      "learning_rate": 5.074762640291785e-06,
      "loss": 0.6752,
      "step": 11442
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.661672031506345,
      "learning_rate": 5.0740403490016735e-06,
      "loss": 0.618,
      "step": 11443
    },
    {
      "epoch": 1.02,
      "grad_norm": 8.245042037497141,
      "learning_rate": 5.073318056166129e-06,
      "loss": 0.6574,
      "step": 11444
    },
    {
      "epoch": 1.02,
      "grad_norm": 8.052691900434517,
      "learning_rate": 5.0725957618002315e-06,
      "loss": 0.6663,
      "step": 11445
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.157634901345381,
      "learning_rate": 5.071873465919058e-06,
      "loss": 0.6099,
      "step": 11446
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.932033893410408,
      "learning_rate": 5.071151168537683e-06,
      "loss": 0.6296,
      "step": 11447
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.973200997771974,
      "learning_rate": 5.070428869671182e-06,
      "loss": 0.5977,
      "step": 11448
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.833855465900155,
      "learning_rate": 5.069706569334635e-06,
      "loss": 0.6604,
      "step": 11449
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.882596259893021,
      "learning_rate": 5.068984267543113e-06,
      "loss": 0.6768,
      "step": 11450
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.770479898384032,
      "learning_rate": 5.068261964311697e-06,
      "loss": 0.6783,
      "step": 11451
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.907892101400199,
      "learning_rate": 5.067539659655463e-06,
      "loss": 0.6376,
      "step": 11452
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.734263194858119,
      "learning_rate": 5.066817353589484e-06,
      "loss": 0.6173,
      "step": 11453
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.398599587220075,
      "learning_rate": 5.06609504612884e-06,
      "loss": 0.5926,
      "step": 11454
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.70067804449616,
      "learning_rate": 5.065372737288607e-06,
      "loss": 0.6311,
      "step": 11455
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.4150553119325755,
      "learning_rate": 5.064650427083863e-06,
      "loss": 0.639,
      "step": 11456
    },
    {
      "epoch": 1.02,
      "grad_norm": 8.442340269120825,
      "learning_rate": 5.0639281155296806e-06,
      "loss": 0.687,
      "step": 11457
    },
    {
      "epoch": 1.02,
      "grad_norm": 8.798305589387832,
      "learning_rate": 5.063205802641139e-06,
      "loss": 0.6037,
      "step": 11458
    },
    {
      "epoch": 1.02,
      "grad_norm": 4.673649266090739,
      "learning_rate": 5.0624834884333154e-06,
      "loss": 0.6838,
      "step": 11459
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.250625613610803,
      "learning_rate": 5.0617611729212844e-06,
      "loss": 0.6067,
      "step": 11460
    },
    {
      "epoch": 1.02,
      "grad_norm": 7.772826373606097,
      "learning_rate": 5.061038856120126e-06,
      "loss": 0.6212,
      "step": 11461
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.594908919090863,
      "learning_rate": 5.060316538044915e-06,
      "loss": 0.6241,
      "step": 11462
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.854159010753713,
      "learning_rate": 5.059594218710728e-06,
      "loss": 0.5981,
      "step": 11463
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.500377314854002,
      "learning_rate": 5.0588718981326414e-06,
      "loss": 0.6425,
      "step": 11464
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.80562941535456,
      "learning_rate": 5.0581495763257335e-06,
      "loss": 0.6323,
      "step": 11465
    },
    {
      "epoch": 1.02,
      "grad_norm": 7.571074823588971,
      "learning_rate": 5.057427253305081e-06,
      "loss": 0.6346,
      "step": 11466
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.722749322848404,
      "learning_rate": 5.0567049290857585e-06,
      "loss": 0.5943,
      "step": 11467
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.481658607449244,
      "learning_rate": 5.0559826036828475e-06,
      "loss": 0.5727,
      "step": 11468
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.321420713106929,
      "learning_rate": 5.05526027711142e-06,
      "loss": 0.6046,
      "step": 11469
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.69915151786212,
      "learning_rate": 5.054537949386556e-06,
      "loss": 0.5813,
      "step": 11470
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.993833275767788,
      "learning_rate": 5.0538156205233335e-06,
      "loss": 0.608,
      "step": 11471
    },
    {
      "epoch": 1.02,
      "grad_norm": 8.420672562897401,
      "learning_rate": 5.053093290536826e-06,
      "loss": 0.5879,
      "step": 11472
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.493765763491909,
      "learning_rate": 5.052370959442111e-06,
      "loss": 0.6876,
      "step": 11473
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.507581170107481,
      "learning_rate": 5.051648627254269e-06,
      "loss": 0.63,
      "step": 11474
    },
    {
      "epoch": 1.02,
      "grad_norm": 8.403338322853598,
      "learning_rate": 5.050926293988374e-06,
      "loss": 0.6017,
      "step": 11475
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.313249061872875,
      "learning_rate": 5.050203959659504e-06,
      "loss": 0.5424,
      "step": 11476
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.5872753999466305,
      "learning_rate": 5.049481624282737e-06,
      "loss": 0.5898,
      "step": 11477
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.183280672574792,
      "learning_rate": 5.0487592878731504e-06,
      "loss": 0.6401,
      "step": 11478
    },
    {
      "epoch": 1.02,
      "grad_norm": 8.261467317979394,
      "learning_rate": 5.048036950445819e-06,
      "loss": 0.6742,
      "step": 11479
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.76108353026768,
      "learning_rate": 5.047314612015821e-06,
      "loss": 0.6188,
      "step": 11480
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.913390375652617,
      "learning_rate": 5.046592272598236e-06,
      "loss": 0.5899,
      "step": 11481
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.628231017864223,
      "learning_rate": 5.045869932208137e-06,
      "loss": 0.5729,
      "step": 11482
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.07457548432494,
      "learning_rate": 5.045147590860604e-06,
      "loss": 0.6035,
      "step": 11483
    },
    {
      "epoch": 1.02,
      "grad_norm": 9.284740504858679,
      "learning_rate": 5.044425248570716e-06,
      "loss": 0.559,
      "step": 11484
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.907460166476627,
      "learning_rate": 5.043702905353548e-06,
      "loss": 0.6275,
      "step": 11485
    },
    {
      "epoch": 1.02,
      "grad_norm": 8.848072912635923,
      "learning_rate": 5.042980561224176e-06,
      "loss": 0.6687,
      "step": 11486
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.621763236752186,
      "learning_rate": 5.04225821619768e-06,
      "loss": 0.5943,
      "step": 11487
    },
    {
      "epoch": 1.02,
      "grad_norm": 7.46178128264674,
      "learning_rate": 5.0415358702891345e-06,
      "loss": 0.6173,
      "step": 11488
    },
    {
      "epoch": 1.02,
      "grad_norm": 7.005711819240018,
      "learning_rate": 5.0408135235136206e-06,
      "loss": 0.6857,
      "step": 11489
    },
    {
      "epoch": 1.03,
      "grad_norm": 7.297334044548831,
      "learning_rate": 5.0400911758862126e-06,
      "loss": 0.5889,
      "step": 11490
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.591909453487586,
      "learning_rate": 5.0393688274219885e-06,
      "loss": 0.5977,
      "step": 11491
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.061440818436598,
      "learning_rate": 5.038646478136029e-06,
      "loss": 0.6168,
      "step": 11492
    },
    {
      "epoch": 1.03,
      "grad_norm": 7.460224554245447,
      "learning_rate": 5.037924128043407e-06,
      "loss": 0.5664,
      "step": 11493
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.60422334316145,
      "learning_rate": 5.037201777159202e-06,
      "loss": 0.7056,
      "step": 11494
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.8994025750970716,
      "learning_rate": 5.036479425498492e-06,
      "loss": 0.5328,
      "step": 11495
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.5350167946378095,
      "learning_rate": 5.035757073076353e-06,
      "loss": 0.6421,
      "step": 11496
    },
    {
      "epoch": 1.03,
      "grad_norm": 8.927885163593757,
      "learning_rate": 5.035034719907864e-06,
      "loss": 0.6146,
      "step": 11497
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.294837036449359,
      "learning_rate": 5.034312366008101e-06,
      "loss": 0.6389,
      "step": 11498
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.900942595591899,
      "learning_rate": 5.033590011392144e-06,
      "loss": 0.6887,
      "step": 11499
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.690394694873084,
      "learning_rate": 5.032867656075069e-06,
      "loss": 0.6386,
      "step": 11500
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.283241499234376,
      "learning_rate": 5.032145300071952e-06,
      "loss": 0.665,
      "step": 11501
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.271389471213352,
      "learning_rate": 5.031422943397873e-06,
      "loss": 0.6237,
      "step": 11502
    },
    {
      "epoch": 1.03,
      "grad_norm": 8.125916459041276,
      "learning_rate": 5.03070058606791e-06,
      "loss": 0.6037,
      "step": 11503
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.703090067136761,
      "learning_rate": 5.029978228097139e-06,
      "loss": 0.6328,
      "step": 11504
    },
    {
      "epoch": 1.03,
      "grad_norm": 4.842789822427345,
      "learning_rate": 5.0292558695006365e-06,
      "loss": 0.6048,
      "step": 11505
    },
    {
      "epoch": 1.03,
      "grad_norm": 9.212363566239267,
      "learning_rate": 5.028533510293484e-06,
      "loss": 0.6041,
      "step": 11506
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.503166832871615,
      "learning_rate": 5.027811150490757e-06,
      "loss": 0.655,
      "step": 11507
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.59167392833282,
      "learning_rate": 5.027088790107532e-06,
      "loss": 0.6066,
      "step": 11508
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.474418554538446,
      "learning_rate": 5.026366429158887e-06,
      "loss": 0.5633,
      "step": 11509
    },
    {
      "epoch": 1.03,
      "grad_norm": 4.172980239802118,
      "learning_rate": 5.025644067659902e-06,
      "loss": 0.5761,
      "step": 11510
    },
    {
      "epoch": 1.03,
      "grad_norm": 7.488017010827383,
      "learning_rate": 5.024921705625652e-06,
      "loss": 0.6102,
      "step": 11511
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.446714887549067,
      "learning_rate": 5.024199343071218e-06,
      "loss": 0.5879,
      "step": 11512
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.789440163970139,
      "learning_rate": 5.023476980011673e-06,
      "loss": 0.5838,
      "step": 11513
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.354716789168824,
      "learning_rate": 5.0227546164621e-06,
      "loss": 0.6582,
      "step": 11514
    },
    {
      "epoch": 1.03,
      "grad_norm": 7.188615956744129,
      "learning_rate": 5.022032252437573e-06,
      "loss": 0.6195,
      "step": 11515
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.089118996545337,
      "learning_rate": 5.021309887953172e-06,
      "loss": 0.5577,
      "step": 11516
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.545948300449825,
      "learning_rate": 5.020587523023972e-06,
      "loss": 0.6115,
      "step": 11517
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.415812148689991,
      "learning_rate": 5.019865157665053e-06,
      "loss": 0.5819,
      "step": 11518
    },
    {
      "epoch": 1.03,
      "grad_norm": 7.439016580300034,
      "learning_rate": 5.019142791891493e-06,
      "loss": 0.6101,
      "step": 11519
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.911579774811184,
      "learning_rate": 5.018420425718369e-06,
      "loss": 0.6327,
      "step": 11520
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.230702755594394,
      "learning_rate": 5.017698059160759e-06,
      "loss": 0.648,
      "step": 11521
    },
    {
      "epoch": 1.03,
      "grad_norm": 7.964705443057674,
      "learning_rate": 5.016975692233741e-06,
      "loss": 0.6498,
      "step": 11522
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.529977629322342,
      "learning_rate": 5.016253324952391e-06,
      "loss": 0.6297,
      "step": 11523
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.572412333393989,
      "learning_rate": 5.0155309573317905e-06,
      "loss": 0.5947,
      "step": 11524
    },
    {
      "epoch": 1.03,
      "grad_norm": 7.187373485687559,
      "learning_rate": 5.014808589387015e-06,
      "loss": 0.6067,
      "step": 11525
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.471144636984977,
      "learning_rate": 5.014086221133141e-06,
      "loss": 0.6742,
      "step": 11526
    },
    {
      "epoch": 1.03,
      "grad_norm": 8.224234039349197,
      "learning_rate": 5.013363852585249e-06,
      "loss": 0.6082,
      "step": 11527
    },
    {
      "epoch": 1.03,
      "grad_norm": 4.0713006743089135,
      "learning_rate": 5.012641483758416e-06,
      "loss": 0.552,
      "step": 11528
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.570666742010075,
      "learning_rate": 5.01191911466772e-06,
      "loss": 0.6227,
      "step": 11529
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.981175160822479,
      "learning_rate": 5.011196745328239e-06,
      "loss": 0.6282,
      "step": 11530
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.271769193810429,
      "learning_rate": 5.010474375755049e-06,
      "loss": 0.59,
      "step": 11531
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.185141740809473,
      "learning_rate": 5.00975200596323e-06,
      "loss": 0.6131,
      "step": 11532
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.643681328053057,
      "learning_rate": 5.0090296359678595e-06,
      "loss": 0.6189,
      "step": 11533
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.128185880585026,
      "learning_rate": 5.008307265784014e-06,
      "loss": 0.6229,
      "step": 11534
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.182380969139973,
      "learning_rate": 5.007584895426775e-06,
      "loss": 0.5952,
      "step": 11535
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.8331041246897035,
      "learning_rate": 5.0068625249112155e-06,
      "loss": 0.6158,
      "step": 11536
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.2907842416158735,
      "learning_rate": 5.006140154252416e-06,
      "loss": 0.602,
      "step": 11537
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.292279184733193,
      "learning_rate": 5.005417783465456e-06,
      "loss": 0.5754,
      "step": 11538
    },
    {
      "epoch": 1.03,
      "grad_norm": 7.59613597818785,
      "learning_rate": 5.004695412565409e-06,
      "loss": 0.6877,
      "step": 11539
    },
    {
      "epoch": 1.03,
      "grad_norm": 8.42116736347381,
      "learning_rate": 5.003973041567357e-06,
      "loss": 0.6058,
      "step": 11540
    },
    {
      "epoch": 1.03,
      "grad_norm": 4.729166899613549,
      "learning_rate": 5.003250670486377e-06,
      "loss": 0.5908,
      "step": 11541
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.799978681526645,
      "learning_rate": 5.002528299337547e-06,
      "loss": 0.6326,
      "step": 11542
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.807465680695557,
      "learning_rate": 5.0018059281359425e-06,
      "loss": 0.6224,
      "step": 11543
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.108132583987435,
      "learning_rate": 5.001083556896643e-06,
      "loss": 0.6444,
      "step": 11544
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.20729466929514,
      "learning_rate": 5.000361185634728e-06,
      "loss": 0.6548,
      "step": 11545
    },
    {
      "epoch": 1.03,
      "grad_norm": 8.10717752010198,
      "learning_rate": 4.999638814365273e-06,
      "loss": 0.6755,
      "step": 11546
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.10324742911467,
      "learning_rate": 4.9989164431033585e-06,
      "loss": 0.6205,
      "step": 11547
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.243421256885166,
      "learning_rate": 4.998194071864059e-06,
      "loss": 0.5662,
      "step": 11548
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.34246403140865,
      "learning_rate": 4.9974717006624546e-06,
      "loss": 0.6997,
      "step": 11549
    },
    {
      "epoch": 1.03,
      "grad_norm": 8.113219008995605,
      "learning_rate": 4.996749329513624e-06,
      "loss": 0.6086,
      "step": 11550
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.707493788583779,
      "learning_rate": 4.996026958432643e-06,
      "loss": 0.6095,
      "step": 11551
    },
    {
      "epoch": 1.03,
      "grad_norm": 4.93636205535711,
      "learning_rate": 4.995304587434591e-06,
      "loss": 0.6413,
      "step": 11552
    },
    {
      "epoch": 1.03,
      "grad_norm": 7.062170286042081,
      "learning_rate": 4.994582216534547e-06,
      "loss": 0.6345,
      "step": 11553
    },
    {
      "epoch": 1.03,
      "grad_norm": 7.102956607353067,
      "learning_rate": 4.993859845747585e-06,
      "loss": 0.656,
      "step": 11554
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.958866486720514,
      "learning_rate": 4.993137475088787e-06,
      "loss": 0.6348,
      "step": 11555
    },
    {
      "epoch": 1.03,
      "grad_norm": 11.820802122633879,
      "learning_rate": 4.9924151045732274e-06,
      "loss": 0.572,
      "step": 11556
    },
    {
      "epoch": 1.03,
      "grad_norm": 8.392307876567111,
      "learning_rate": 4.991692734215988e-06,
      "loss": 0.643,
      "step": 11557
    },
    {
      "epoch": 1.03,
      "grad_norm": 7.145269653077524,
      "learning_rate": 4.990970364032143e-06,
      "loss": 0.5575,
      "step": 11558
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.046670788164138,
      "learning_rate": 4.990247994036772e-06,
      "loss": 0.5744,
      "step": 11559
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.9347660465118866,
      "learning_rate": 4.989525624244953e-06,
      "loss": 0.5537,
      "step": 11560
    },
    {
      "epoch": 1.03,
      "grad_norm": 7.769164539044649,
      "learning_rate": 4.988803254671763e-06,
      "loss": 0.6361,
      "step": 11561
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.40482128982242,
      "learning_rate": 4.988080885332281e-06,
      "loss": 0.6284,
      "step": 11562
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.9590693196092515,
      "learning_rate": 4.9873585162415846e-06,
      "loss": 0.6175,
      "step": 11563
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.751836284169477,
      "learning_rate": 4.986636147414752e-06,
      "loss": 0.6143,
      "step": 11564
    },
    {
      "epoch": 1.03,
      "grad_norm": 4.6238152204696465,
      "learning_rate": 4.985913778866859e-06,
      "loss": 0.5998,
      "step": 11565
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.4365254207761975,
      "learning_rate": 4.985191410612987e-06,
      "loss": 0.6015,
      "step": 11566
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.738063856239536,
      "learning_rate": 4.98446904266821e-06,
      "loss": 0.5378,
      "step": 11567
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.7074920040055535,
      "learning_rate": 4.98374667504761e-06,
      "loss": 0.6107,
      "step": 11568
    },
    {
      "epoch": 1.03,
      "grad_norm": 7.856502618767377,
      "learning_rate": 4.983024307766261e-06,
      "loss": 0.6593,
      "step": 11569
    },
    {
      "epoch": 1.03,
      "grad_norm": 8.979495406386846,
      "learning_rate": 4.982301940839244e-06,
      "loss": 0.62,
      "step": 11570
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.191810845899528,
      "learning_rate": 4.981579574281633e-06,
      "loss": 0.5745,
      "step": 11571
    },
    {
      "epoch": 1.03,
      "grad_norm": 4.752206465400866,
      "learning_rate": 4.980857208108508e-06,
      "loss": 0.6443,
      "step": 11572
    },
    {
      "epoch": 1.03,
      "grad_norm": 7.156432038916066,
      "learning_rate": 4.980134842334949e-06,
      "loss": 0.5946,
      "step": 11573
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.968628794496162,
      "learning_rate": 4.97941247697603e-06,
      "loss": 0.6263,
      "step": 11574
    },
    {
      "epoch": 1.03,
      "grad_norm": 4.558121292003021,
      "learning_rate": 4.97869011204683e-06,
      "loss": 0.5709,
      "step": 11575
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.8931835047507635,
      "learning_rate": 4.977967747562429e-06,
      "loss": 0.6132,
      "step": 11576
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.909841470891709,
      "learning_rate": 4.977245383537901e-06,
      "loss": 0.6207,
      "step": 11577
    },
    {
      "epoch": 1.03,
      "grad_norm": 8.025363744166805,
      "learning_rate": 4.976523019988327e-06,
      "loss": 0.6055,
      "step": 11578
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.645062046137986,
      "learning_rate": 4.975800656928784e-06,
      "loss": 0.6273,
      "step": 11579
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.044601132057996,
      "learning_rate": 4.975078294374348e-06,
      "loss": 0.5457,
      "step": 11580
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.970371495622717,
      "learning_rate": 4.974355932340099e-06,
      "loss": 0.6618,
      "step": 11581
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.174100868524889,
      "learning_rate": 4.973633570841115e-06,
      "loss": 0.6487,
      "step": 11582
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.13765437106652,
      "learning_rate": 4.972911209892472e-06,
      "loss": 0.6248,
      "step": 11583
    },
    {
      "epoch": 1.03,
      "grad_norm": 4.980104733223577,
      "learning_rate": 4.9721888495092465e-06,
      "loss": 0.5503,
      "step": 11584
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.950058306115187,
      "learning_rate": 4.971466489706518e-06,
      "loss": 0.6037,
      "step": 11585
    },
    {
      "epoch": 1.03,
      "grad_norm": 7.033951743911191,
      "learning_rate": 4.970744130499364e-06,
      "loss": 0.6054,
      "step": 11586
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.175963293025318,
      "learning_rate": 4.970021771902864e-06,
      "loss": 0.5927,
      "step": 11587
    },
    {
      "epoch": 1.03,
      "grad_norm": 7.206327114341463,
      "learning_rate": 4.9692994139320915e-06,
      "loss": 0.6588,
      "step": 11588
    },
    {
      "epoch": 1.03,
      "grad_norm": 9.496243306088898,
      "learning_rate": 4.968577056602128e-06,
      "loss": 0.6352,
      "step": 11589
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.609569710816613,
      "learning_rate": 4.9678546999280495e-06,
      "loss": 0.5895,
      "step": 11590
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.681324355487443,
      "learning_rate": 4.967132343924932e-06,
      "loss": 0.6258,
      "step": 11591
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.662207063779923,
      "learning_rate": 4.966409988607858e-06,
      "loss": 0.5889,
      "step": 11592
    },
    {
      "epoch": 1.03,
      "grad_norm": 7.192711181551438,
      "learning_rate": 4.965687633991899e-06,
      "loss": 0.6443,
      "step": 11593
    },
    {
      "epoch": 1.03,
      "grad_norm": 4.63277396377212,
      "learning_rate": 4.964965280092137e-06,
      "loss": 0.6249,
      "step": 11594
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.913799268249351,
      "learning_rate": 4.964242926923648e-06,
      "loss": 0.5912,
      "step": 11595
    },
    {
      "epoch": 1.03,
      "grad_norm": 7.036281536971418,
      "learning_rate": 4.9635205745015116e-06,
      "loss": 0.6056,
      "step": 11596
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.724884028849377,
      "learning_rate": 4.9627982228408e-06,
      "loss": 0.6211,
      "step": 11597
    },
    {
      "epoch": 1.03,
      "grad_norm": 8.415578932219454,
      "learning_rate": 4.962075871956595e-06,
      "loss": 0.5708,
      "step": 11598
    },
    {
      "epoch": 1.03,
      "grad_norm": 7.066689069973688,
      "learning_rate": 4.961353521863974e-06,
      "loss": 0.5954,
      "step": 11599
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.0070667699931155,
      "learning_rate": 4.960631172578012e-06,
      "loss": 0.5999,
      "step": 11600
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.908002225563063,
      "learning_rate": 4.959908824113789e-06,
      "loss": 0.6236,
      "step": 11601
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.746544597030718,
      "learning_rate": 4.959186476486382e-06,
      "loss": 0.589,
      "step": 11602
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.097525999553976,
      "learning_rate": 4.958464129710866e-06,
      "loss": 0.6051,
      "step": 11603
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.713746553368952,
      "learning_rate": 4.957741783802322e-06,
      "loss": 0.6213,
      "step": 11604
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.310710845586217,
      "learning_rate": 4.957019438775826e-06,
      "loss": 0.672,
      "step": 11605
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.450203043718635,
      "learning_rate": 4.956297094646454e-06,
      "loss": 0.6049,
      "step": 11606
    },
    {
      "epoch": 1.04,
      "grad_norm": 7.433116611329401,
      "learning_rate": 4.955574751429285e-06,
      "loss": 0.7148,
      "step": 11607
    },
    {
      "epoch": 1.04,
      "grad_norm": 4.612580659791629,
      "learning_rate": 4.954852409139396e-06,
      "loss": 0.5536,
      "step": 11608
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.545808617002117,
      "learning_rate": 4.954130067791863e-06,
      "loss": 0.6656,
      "step": 11609
    },
    {
      "epoch": 1.04,
      "grad_norm": 7.148509559005961,
      "learning_rate": 4.953407727401765e-06,
      "loss": 0.62,
      "step": 11610
    },
    {
      "epoch": 1.04,
      "grad_norm": 7.3410735902542585,
      "learning_rate": 4.95268538798418e-06,
      "loss": 0.6306,
      "step": 11611
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.706582097823274,
      "learning_rate": 4.9519630495541835e-06,
      "loss": 0.545,
      "step": 11612
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.173032299194765,
      "learning_rate": 4.951240712126852e-06,
      "loss": 0.5924,
      "step": 11613
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.548073661849845,
      "learning_rate": 4.950518375717264e-06,
      "loss": 0.5956,
      "step": 11614
    },
    {
      "epoch": 1.04,
      "grad_norm": 7.888479027226866,
      "learning_rate": 4.949796040340497e-06,
      "loss": 0.5896,
      "step": 11615
    },
    {
      "epoch": 1.04,
      "grad_norm": 8.433440162374461,
      "learning_rate": 4.949073706011627e-06,
      "loss": 0.6063,
      "step": 11616
    },
    {
      "epoch": 1.04,
      "grad_norm": 4.4849077217338165,
      "learning_rate": 4.948351372745732e-06,
      "loss": 0.6368,
      "step": 11617
    },
    {
      "epoch": 1.04,
      "grad_norm": 4.725372815886445,
      "learning_rate": 4.94762904055789e-06,
      "loss": 0.5773,
      "step": 11618
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.038803687674468,
      "learning_rate": 4.946906709463176e-06,
      "loss": 0.595,
      "step": 11619
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.705144894344487,
      "learning_rate": 4.946184379476668e-06,
      "loss": 0.5703,
      "step": 11620
    },
    {
      "epoch": 1.04,
      "grad_norm": 4.878014472896639,
      "learning_rate": 4.945462050613445e-06,
      "loss": 0.5613,
      "step": 11621
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.614552743859249,
      "learning_rate": 4.94473972288858e-06,
      "loss": 0.5928,
      "step": 11622
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.75923764594968,
      "learning_rate": 4.944017396317154e-06,
      "loss": 0.6249,
      "step": 11623
    },
    {
      "epoch": 1.04,
      "grad_norm": 7.006578633704678,
      "learning_rate": 4.9432950709142415e-06,
      "loss": 0.6367,
      "step": 11624
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.589058482318481,
      "learning_rate": 4.942572746694922e-06,
      "loss": 0.6534,
      "step": 11625
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.001916045976451,
      "learning_rate": 4.941850423674269e-06,
      "loss": 0.6191,
      "step": 11626
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.755334636529886,
      "learning_rate": 4.94112810186736e-06,
      "loss": 0.5983,
      "step": 11627
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.081945456418661,
      "learning_rate": 4.940405781289274e-06,
      "loss": 0.6493,
      "step": 11628
    },
    {
      "epoch": 1.04,
      "grad_norm": 8.957893307739983,
      "learning_rate": 4.939683461955087e-06,
      "loss": 0.6173,
      "step": 11629
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.663882749701962,
      "learning_rate": 4.938961143879875e-06,
      "loss": 0.6191,
      "step": 11630
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.139162442312647,
      "learning_rate": 4.938238827078716e-06,
      "loss": 0.6602,
      "step": 11631
    },
    {
      "epoch": 1.04,
      "grad_norm": 4.584409858489053,
      "learning_rate": 4.937516511566686e-06,
      "loss": 0.6692,
      "step": 11632
    },
    {
      "epoch": 1.04,
      "grad_norm": 4.228828919104155,
      "learning_rate": 4.936794197358863e-06,
      "loss": 0.6134,
      "step": 11633
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.684611024638545,
      "learning_rate": 4.936071884470321e-06,
      "loss": 0.5829,
      "step": 11634
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.8859301482553565,
      "learning_rate": 4.935349572916139e-06,
      "loss": 0.6087,
      "step": 11635
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.826469923314895,
      "learning_rate": 4.934627262711394e-06,
      "loss": 0.6042,
      "step": 11636
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.598903196079001,
      "learning_rate": 4.933904953871161e-06,
      "loss": 0.6358,
      "step": 11637
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.250299612901703,
      "learning_rate": 4.933182646410516e-06,
      "loss": 0.667,
      "step": 11638
    },
    {
      "epoch": 1.04,
      "grad_norm": 4.649298014739443,
      "learning_rate": 4.9324603403445395e-06,
      "loss": 0.6194,
      "step": 11639
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.003116751576399,
      "learning_rate": 4.9317380356883055e-06,
      "loss": 0.6118,
      "step": 11640
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.583683552386966,
      "learning_rate": 4.931015732456889e-06,
      "loss": 0.5302,
      "step": 11641
    },
    {
      "epoch": 1.04,
      "grad_norm": 7.669728210104926,
      "learning_rate": 4.930293430665367e-06,
      "loss": 0.6306,
      "step": 11642
    },
    {
      "epoch": 1.04,
      "grad_norm": 10.094003165916055,
      "learning_rate": 4.929571130328819e-06,
      "loss": 0.6901,
      "step": 11643
    },
    {
      "epoch": 1.04,
      "grad_norm": 4.87248056962988,
      "learning_rate": 4.928848831462319e-06,
      "loss": 0.577,
      "step": 11644
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.6630507968400075,
      "learning_rate": 4.9281265340809435e-06,
      "loss": 0.5678,
      "step": 11645
    },
    {
      "epoch": 1.04,
      "grad_norm": 7.88756660955437,
      "learning_rate": 4.927404238199769e-06,
      "loss": 0.6071,
      "step": 11646
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.455358973522802,
      "learning_rate": 4.926681943833872e-06,
      "loss": 0.6555,
      "step": 11647
    },
    {
      "epoch": 1.04,
      "grad_norm": 7.333423319955297,
      "learning_rate": 4.925959650998329e-06,
      "loss": 0.6352,
      "step": 11648
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.718901828710988,
      "learning_rate": 4.9252373597082165e-06,
      "loss": 0.5879,
      "step": 11649
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.374517572077803,
      "learning_rate": 4.92451506997861e-06,
      "loss": 0.609,
      "step": 11650
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.35732517137119,
      "learning_rate": 4.923792781824586e-06,
      "loss": 0.5759,
      "step": 11651
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.229527602133704,
      "learning_rate": 4.923070495261221e-06,
      "loss": 0.6201,
      "step": 11652
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.623533674160671,
      "learning_rate": 4.922348210303591e-06,
      "loss": 0.5854,
      "step": 11653
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.750875083333629,
      "learning_rate": 4.921625926966774e-06,
      "loss": 0.5757,
      "step": 11654
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.754604673351946,
      "learning_rate": 4.9209036452658415e-06,
      "loss": 0.5867,
      "step": 11655
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.631581296535587,
      "learning_rate": 4.920181365215873e-06,
      "loss": 0.6255,
      "step": 11656
    },
    {
      "epoch": 1.04,
      "grad_norm": 7.197460075668641,
      "learning_rate": 4.919459086831944e-06,
      "loss": 0.5927,
      "step": 11657
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.635072325130017,
      "learning_rate": 4.918736810129129e-06,
      "loss": 0.6422,
      "step": 11658
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.162098069424316,
      "learning_rate": 4.9180145351225064e-06,
      "loss": 0.6225,
      "step": 11659
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.101462297015165,
      "learning_rate": 4.91729226182715e-06,
      "loss": 0.6434,
      "step": 11660
    },
    {
      "epoch": 1.04,
      "grad_norm": 7.415433458426369,
      "learning_rate": 4.916569990258136e-06,
      "loss": 0.6466,
      "step": 11661
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.9637416447519085,
      "learning_rate": 4.9158477204305425e-06,
      "loss": 0.6318,
      "step": 11662
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.634312965043066,
      "learning_rate": 4.915125452359442e-06,
      "loss": 0.5878,
      "step": 11663
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.053452036631529,
      "learning_rate": 4.914403186059912e-06,
      "loss": 0.6623,
      "step": 11664
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.216970010796135,
      "learning_rate": 4.913680921547029e-06,
      "loss": 0.6405,
      "step": 11665
    },
    {
      "epoch": 1.04,
      "grad_norm": 7.462414792690141,
      "learning_rate": 4.9129586588358665e-06,
      "loss": 0.623,
      "step": 11666
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.97304119637614,
      "learning_rate": 4.912236397941502e-06,
      "loss": 0.6026,
      "step": 11667
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.8731889667952775,
      "learning_rate": 4.911514138879011e-06,
      "loss": 0.6311,
      "step": 11668
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.354703066798265,
      "learning_rate": 4.91079188166347e-06,
      "loss": 0.6191,
      "step": 11669
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.976543533995944,
      "learning_rate": 4.91006962630995e-06,
      "loss": 0.6492,
      "step": 11670
    },
    {
      "epoch": 1.04,
      "grad_norm": 4.474194872308304,
      "learning_rate": 4.9093473728335304e-06,
      "loss": 0.6665,
      "step": 11671
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.60806012304972,
      "learning_rate": 4.908625121249288e-06,
      "loss": 0.6702,
      "step": 11672
    },
    {
      "epoch": 1.04,
      "grad_norm": 7.698428746998802,
      "learning_rate": 4.9079028715722936e-06,
      "loss": 0.5867,
      "step": 11673
    },
    {
      "epoch": 1.04,
      "grad_norm": 4.888983564660257,
      "learning_rate": 4.907180623817625e-06,
      "loss": 0.5738,
      "step": 11674
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.994362240517239,
      "learning_rate": 4.906458378000359e-06,
      "loss": 0.6589,
      "step": 11675
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.750232005680473,
      "learning_rate": 4.905736134135569e-06,
      "loss": 0.6971,
      "step": 11676
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.079879369956708,
      "learning_rate": 4.90501389223833e-06,
      "loss": 0.5927,
      "step": 11677
    },
    {
      "epoch": 1.04,
      "grad_norm": 4.209765952711047,
      "learning_rate": 4.9042916523237195e-06,
      "loss": 0.5919,
      "step": 11678
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.446984685313848,
      "learning_rate": 4.90356941440681e-06,
      "loss": 0.5978,
      "step": 11679
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.681224008448138,
      "learning_rate": 4.902847178502677e-06,
      "loss": 0.6047,
      "step": 11680
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.384887739390651,
      "learning_rate": 4.9021249446263985e-06,
      "loss": 0.5902,
      "step": 11681
    },
    {
      "epoch": 1.04,
      "grad_norm": 8.243363080994994,
      "learning_rate": 4.9014027127930455e-06,
      "loss": 0.6559,
      "step": 11682
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.405491882290054,
      "learning_rate": 4.900680483017698e-06,
      "loss": 0.6092,
      "step": 11683
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.670552355639677,
      "learning_rate": 4.899958255315425e-06,
      "loss": 0.5982,
      "step": 11684
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.933360717651437,
      "learning_rate": 4.899236029701307e-06,
      "loss": 0.6296,
      "step": 11685
    },
    {
      "epoch": 1.04,
      "grad_norm": 7.71192642301557,
      "learning_rate": 4.898513806190414e-06,
      "loss": 0.5617,
      "step": 11686
    },
    {
      "epoch": 1.04,
      "grad_norm": 7.80469820721532,
      "learning_rate": 4.897791584797823e-06,
      "loss": 0.6257,
      "step": 11687
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.030858516977971,
      "learning_rate": 4.8970693655386094e-06,
      "loss": 0.6234,
      "step": 11688
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.118435147009873,
      "learning_rate": 4.896347148427847e-06,
      "loss": 0.5931,
      "step": 11689
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.20871929849968,
      "learning_rate": 4.895624933480611e-06,
      "loss": 0.5933,
      "step": 11690
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.526167848672538,
      "learning_rate": 4.894902720711976e-06,
      "loss": 0.6217,
      "step": 11691
    },
    {
      "epoch": 1.04,
      "grad_norm": 7.23542568669468,
      "learning_rate": 4.894180510137016e-06,
      "loss": 0.6284,
      "step": 11692
    },
    {
      "epoch": 1.04,
      "grad_norm": 10.75115138509132,
      "learning_rate": 4.893458301770807e-06,
      "loss": 0.6082,
      "step": 11693
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.1761951481333766,
      "learning_rate": 4.892736095628422e-06,
      "loss": 0.6078,
      "step": 11694
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.609762877523205,
      "learning_rate": 4.892013891724937e-06,
      "loss": 0.6178,
      "step": 11695
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.742834671067485,
      "learning_rate": 4.891291690075423e-06,
      "loss": 0.5613,
      "step": 11696
    },
    {
      "epoch": 1.04,
      "grad_norm": 4.481944019115577,
      "learning_rate": 4.890569490694961e-06,
      "loss": 0.6127,
      "step": 11697
    },
    {
      "epoch": 1.04,
      "grad_norm": 11.02079964139869,
      "learning_rate": 4.88984729359862e-06,
      "loss": 0.6473,
      "step": 11698
    },
    {
      "epoch": 1.04,
      "grad_norm": 8.767967686760194,
      "learning_rate": 4.889125098801474e-06,
      "loss": 0.6641,
      "step": 11699
    },
    {
      "epoch": 1.04,
      "grad_norm": 7.643585819764252,
      "learning_rate": 4.888402906318599e-06,
      "loss": 0.6108,
      "step": 11700
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.282890903374749,
      "learning_rate": 4.88768071616507e-06,
      "loss": 0.627,
      "step": 11701
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.615156392001707,
      "learning_rate": 4.886958528355959e-06,
      "loss": 0.5702,
      "step": 11702
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.298007604964225,
      "learning_rate": 4.886236342906341e-06,
      "loss": 0.6455,
      "step": 11703
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.191555906326668,
      "learning_rate": 4.885514159831292e-06,
      "loss": 0.5851,
      "step": 11704
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.2765043292873735,
      "learning_rate": 4.884791979145882e-06,
      "loss": 0.5403,
      "step": 11705
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.066795794683276,
      "learning_rate": 4.8840698008651875e-06,
      "loss": 0.5788,
      "step": 11706
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.681452545157281,
      "learning_rate": 4.883347625004284e-06,
      "loss": 0.6565,
      "step": 11707
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.170712738911591,
      "learning_rate": 4.882625451578241e-06,
      "loss": 0.6254,
      "step": 11708
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.813393733000676,
      "learning_rate": 4.881903280602137e-06,
      "loss": 0.6093,
      "step": 11709
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.975799472219051,
      "learning_rate": 4.881181112091041e-06,
      "loss": 0.6506,
      "step": 11710
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.496539030260688,
      "learning_rate": 4.880458946060029e-06,
      "loss": 0.6256,
      "step": 11711
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.402567592852545,
      "learning_rate": 4.879736782524178e-06,
      "loss": 0.6341,
      "step": 11712
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.015999049182045,
      "learning_rate": 4.879014621498556e-06,
      "loss": 0.5773,
      "step": 11713
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.670364941604027,
      "learning_rate": 4.878292462998239e-06,
      "loss": 0.7058,
      "step": 11714
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.557323492124184,
      "learning_rate": 4.8775703070383e-06,
      "loss": 0.6206,
      "step": 11715
    },
    {
      "epoch": 1.05,
      "grad_norm": 7.0107194420018715,
      "learning_rate": 4.876848153633814e-06,
      "loss": 0.6021,
      "step": 11716
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.040577756339709,
      "learning_rate": 4.876126002799851e-06,
      "loss": 0.671,
      "step": 11717
    },
    {
      "epoch": 1.05,
      "grad_norm": 7.632509162373895,
      "learning_rate": 4.875403854551487e-06,
      "loss": 0.6404,
      "step": 11718
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.286507927051945,
      "learning_rate": 4.874681708903796e-06,
      "loss": 0.595,
      "step": 11719
    },
    {
      "epoch": 1.05,
      "grad_norm": 7.05808090596921,
      "learning_rate": 4.8739595658718485e-06,
      "loss": 0.6502,
      "step": 11720
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.4026720830152,
      "learning_rate": 4.873237425470719e-06,
      "loss": 0.6475,
      "step": 11721
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.3336186252394215,
      "learning_rate": 4.8725152877154815e-06,
      "loss": 0.5929,
      "step": 11722
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.88418267580631,
      "learning_rate": 4.871793152621208e-06,
      "loss": 0.6251,
      "step": 11723
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.9624828129040575,
      "learning_rate": 4.871071020202971e-06,
      "loss": 0.6182,
      "step": 11724
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.804611481216006,
      "learning_rate": 4.870348890475845e-06,
      "loss": 0.6636,
      "step": 11725
    },
    {
      "epoch": 1.05,
      "grad_norm": 7.320686859210225,
      "learning_rate": 4.869626763454904e-06,
      "loss": 0.6106,
      "step": 11726
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.793488497837014,
      "learning_rate": 4.868904639155215e-06,
      "loss": 0.5963,
      "step": 11727
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.022841638407189,
      "learning_rate": 4.868182517591856e-06,
      "loss": 0.5903,
      "step": 11728
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.468389857595162,
      "learning_rate": 4.867460398779899e-06,
      "loss": 0.5934,
      "step": 11729
    },
    {
      "epoch": 1.05,
      "grad_norm": 4.547977899656438,
      "learning_rate": 4.866738282734415e-06,
      "loss": 0.6292,
      "step": 11730
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.274633832878775,
      "learning_rate": 4.866016169470477e-06,
      "loss": 0.6275,
      "step": 11731
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.952128760952303,
      "learning_rate": 4.86529405900316e-06,
      "loss": 0.6026,
      "step": 11732
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.359678376122914,
      "learning_rate": 4.864571951347532e-06,
      "loss": 0.6436,
      "step": 11733
    },
    {
      "epoch": 1.05,
      "grad_norm": 7.307390315880051,
      "learning_rate": 4.863849846518669e-06,
      "loss": 0.5927,
      "step": 11734
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.530438789690461,
      "learning_rate": 4.863127744531643e-06,
      "loss": 0.6116,
      "step": 11735
    },
    {
      "epoch": 1.05,
      "grad_norm": 8.413606672276698,
      "learning_rate": 4.862405645401524e-06,
      "loss": 0.6383,
      "step": 11736
    },
    {
      "epoch": 1.05,
      "grad_norm": 4.857676247374448,
      "learning_rate": 4.8616835491433865e-06,
      "loss": 0.6541,
      "step": 11737
    },
    {
      "epoch": 1.05,
      "grad_norm": 7.364604698564956,
      "learning_rate": 4.860961455772304e-06,
      "loss": 0.6116,
      "step": 11738
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.414958290502444,
      "learning_rate": 4.8602393653033446e-06,
      "loss": 0.598,
      "step": 11739
    },
    {
      "epoch": 1.05,
      "grad_norm": 8.926654397711882,
      "learning_rate": 4.859517277751581e-06,
      "loss": 0.6255,
      "step": 11740
    },
    {
      "epoch": 1.05,
      "grad_norm": 4.841402046486643,
      "learning_rate": 4.858795193132091e-06,
      "loss": 0.611,
      "step": 11741
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.185621988285458,
      "learning_rate": 4.85807311145994e-06,
      "loss": 0.5298,
      "step": 11742
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.757299462400628,
      "learning_rate": 4.857351032750201e-06,
      "loss": 0.5998,
      "step": 11743
    },
    {
      "epoch": 1.05,
      "grad_norm": 7.266290111563711,
      "learning_rate": 4.856628957017946e-06,
      "loss": 0.5789,
      "step": 11744
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.344375036575781,
      "learning_rate": 4.855906884278251e-06,
      "loss": 0.5669,
      "step": 11745
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.053005413706421,
      "learning_rate": 4.85518481454618e-06,
      "loss": 0.6027,
      "step": 11746
    },
    {
      "epoch": 1.05,
      "grad_norm": 7.564944720703027,
      "learning_rate": 4.854462747836811e-06,
      "loss": 0.587,
      "step": 11747
    },
    {
      "epoch": 1.05,
      "grad_norm": 7.1601398947009205,
      "learning_rate": 4.853740684165214e-06,
      "loss": 0.6055,
      "step": 11748
    },
    {
      "epoch": 1.05,
      "grad_norm": 8.50900557851205,
      "learning_rate": 4.8530186235464585e-06,
      "loss": 0.6081,
      "step": 11749
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.5734988469081665,
      "learning_rate": 4.852296565995618e-06,
      "loss": 0.6256,
      "step": 11750
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.85584956246809,
      "learning_rate": 4.851574511527764e-06,
      "loss": 0.5939,
      "step": 11751
    },
    {
      "epoch": 1.05,
      "grad_norm": 4.919077774199888,
      "learning_rate": 4.850852460157965e-06,
      "loss": 0.6636,
      "step": 11752
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.864336067456951,
      "learning_rate": 4.850130411901296e-06,
      "loss": 0.6385,
      "step": 11753
    },
    {
      "epoch": 1.05,
      "grad_norm": 8.15442365548519,
      "learning_rate": 4.849408366772827e-06,
      "loss": 0.6357,
      "step": 11754
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.838671404881356,
      "learning_rate": 4.848686324787629e-06,
      "loss": 0.6337,
      "step": 11755
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.509834530433153,
      "learning_rate": 4.847964285960771e-06,
      "loss": 0.6563,
      "step": 11756
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.368825778401634,
      "learning_rate": 4.847242250307324e-06,
      "loss": 0.6343,
      "step": 11757
    },
    {
      "epoch": 1.05,
      "grad_norm": 4.936999963245161,
      "learning_rate": 4.846520217842363e-06,
      "loss": 0.627,
      "step": 11758
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.7913211824017266,
      "learning_rate": 4.845798188580955e-06,
      "loss": 0.5478,
      "step": 11759
    },
    {
      "epoch": 1.05,
      "grad_norm": 4.632696324831424,
      "learning_rate": 4.8450761625381735e-06,
      "loss": 0.6042,
      "step": 11760
    },
    {
      "epoch": 1.05,
      "grad_norm": 7.6521022926579985,
      "learning_rate": 4.844354139729087e-06,
      "loss": 0.6315,
      "step": 11761
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.3153221456926945,
      "learning_rate": 4.843632120168767e-06,
      "loss": 0.6665,
      "step": 11762
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.074895151985437,
      "learning_rate": 4.842910103872283e-06,
      "loss": 0.6203,
      "step": 11763
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.331818924280271,
      "learning_rate": 4.8421880908547085e-06,
      "loss": 0.5445,
      "step": 11764
    },
    {
      "epoch": 1.05,
      "grad_norm": 8.368291181363428,
      "learning_rate": 4.841466081131111e-06,
      "loss": 0.6069,
      "step": 11765
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.442164322778927,
      "learning_rate": 4.840744074716562e-06,
      "loss": 0.6443,
      "step": 11766
    },
    {
      "epoch": 1.05,
      "grad_norm": 7.401104469258698,
      "learning_rate": 4.840022071626131e-06,
      "loss": 0.5857,
      "step": 11767
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.07560241036891,
      "learning_rate": 4.83930007187489e-06,
      "loss": 0.6009,
      "step": 11768
    },
    {
      "epoch": 1.05,
      "grad_norm": 8.475487922194178,
      "learning_rate": 4.838578075477907e-06,
      "loss": 0.5784,
      "step": 11769
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.352570880922076,
      "learning_rate": 4.837856082450255e-06,
      "loss": 0.5647,
      "step": 11770
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.367518751697547,
      "learning_rate": 4.837134092807002e-06,
      "loss": 0.584,
      "step": 11771
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.734740283243662,
      "learning_rate": 4.836412106563217e-06,
      "loss": 0.6306,
      "step": 11772
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.350988927168498,
      "learning_rate": 4.83569012373397e-06,
      "loss": 0.6096,
      "step": 11773
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.202142261853159,
      "learning_rate": 4.834968144334334e-06,
      "loss": 0.5993,
      "step": 11774
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.336177699501183,
      "learning_rate": 4.834246168379375e-06,
      "loss": 0.5938,
      "step": 11775
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.469469410509182,
      "learning_rate": 4.8335241958841645e-06,
      "loss": 0.701,
      "step": 11776
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.935301079037251,
      "learning_rate": 4.832802226863773e-06,
      "loss": 0.599,
      "step": 11777
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.049831617540954,
      "learning_rate": 4.832080261333267e-06,
      "loss": 0.6467,
      "step": 11778
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.864238274597081,
      "learning_rate": 4.831358299307718e-06,
      "loss": 0.5365,
      "step": 11779
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.338069643283755,
      "learning_rate": 4.830636340802197e-06,
      "loss": 0.6272,
      "step": 11780
    },
    {
      "epoch": 1.05,
      "grad_norm": 7.27295084210903,
      "learning_rate": 4.82991438583177e-06,
      "loss": 0.5687,
      "step": 11781
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.846659664084782,
      "learning_rate": 4.829192434411508e-06,
      "loss": 0.6196,
      "step": 11782
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.668511507025311,
      "learning_rate": 4.828470486556481e-06,
      "loss": 0.628,
      "step": 11783
    },
    {
      "epoch": 1.05,
      "grad_norm": 8.249759398659982,
      "learning_rate": 4.8277485422817575e-06,
      "loss": 0.6225,
      "step": 11784
    },
    {
      "epoch": 1.05,
      "grad_norm": 7.480873922819516,
      "learning_rate": 4.827026601602404e-06,
      "loss": 0.6246,
      "step": 11785
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.076827900538968,
      "learning_rate": 4.826304664533492e-06,
      "loss": 0.5912,
      "step": 11786
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.688156711095468,
      "learning_rate": 4.82558273109009e-06,
      "loss": 0.5545,
      "step": 11787
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.50881643696435,
      "learning_rate": 4.824860801287266e-06,
      "loss": 0.5811,
      "step": 11788
    },
    {
      "epoch": 1.05,
      "grad_norm": 7.582430602939867,
      "learning_rate": 4.824138875140088e-06,
      "loss": 0.6281,
      "step": 11789
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.401939204529119,
      "learning_rate": 4.823416952663628e-06,
      "loss": 0.5781,
      "step": 11790
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.629525836852875,
      "learning_rate": 4.822695033872951e-06,
      "loss": 0.6378,
      "step": 11791
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.996041153766011,
      "learning_rate": 4.821973118783128e-06,
      "loss": 0.5829,
      "step": 11792
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.203014446511726,
      "learning_rate": 4.821251207409226e-06,
      "loss": 0.6067,
      "step": 11793
    },
    {
      "epoch": 1.05,
      "grad_norm": 4.917285312091409,
      "learning_rate": 4.820529299766313e-06,
      "loss": 0.5868,
      "step": 11794
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.455050663404512,
      "learning_rate": 4.819807395869459e-06,
      "loss": 0.6627,
      "step": 11795
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.445390904022384,
      "learning_rate": 4.819085495733729e-06,
      "loss": 0.6293,
      "step": 11796
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.923701501862212,
      "learning_rate": 4.818363599374194e-06,
      "loss": 0.647,
      "step": 11797
    },
    {
      "epoch": 1.05,
      "grad_norm": 7.17997364272325,
      "learning_rate": 4.817641706805923e-06,
      "loss": 0.5981,
      "step": 11798
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.141109792885425,
      "learning_rate": 4.81691981804398e-06,
      "loss": 0.6331,
      "step": 11799
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.254763140363157,
      "learning_rate": 4.816197933103436e-06,
      "loss": 0.6249,
      "step": 11800
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.824578256902001,
      "learning_rate": 4.815476051999356e-06,
      "loss": 0.6569,
      "step": 11801
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.830397383273105,
      "learning_rate": 4.81475417474681e-06,
      "loss": 0.6132,
      "step": 11802
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.6828532902062205,
      "learning_rate": 4.814032301360865e-06,
      "loss": 0.6157,
      "step": 11803
    },
    {
      "epoch": 1.05,
      "grad_norm": 7.39721586283343,
      "learning_rate": 4.813310431856589e-06,
      "loss": 0.5733,
      "step": 11804
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.050004135871844,
      "learning_rate": 4.8125885662490486e-06,
      "loss": 0.6101,
      "step": 11805
    },
    {
      "epoch": 1.05,
      "grad_norm": 4.677923509319573,
      "learning_rate": 4.811866704553311e-06,
      "loss": 0.6194,
      "step": 11806
    },
    {
      "epoch": 1.05,
      "grad_norm": 7.667607783776006,
      "learning_rate": 4.811144846784444e-06,
      "loss": 0.6277,
      "step": 11807
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.402030388676414,
      "learning_rate": 4.810422992957515e-06,
      "loss": 0.6421,
      "step": 11808
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.49257382296834,
      "learning_rate": 4.80970114308759e-06,
      "loss": 0.6322,
      "step": 11809
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.363581181471328,
      "learning_rate": 4.808979297189738e-06,
      "loss": 0.6238,
      "step": 11810
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.81253949584474,
      "learning_rate": 4.808257455279025e-06,
      "loss": 0.5717,
      "step": 11811
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.7011622042862244,
      "learning_rate": 4.807535617370517e-06,
      "loss": 0.612,
      "step": 11812
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.276964419364145,
      "learning_rate": 4.8068137834792834e-06,
      "loss": 0.5343,
      "step": 11813
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.742077638653688,
      "learning_rate": 4.806091953620387e-06,
      "loss": 0.5519,
      "step": 11814
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.85163615848631,
      "learning_rate": 4.805370127808897e-06,
      "loss": 0.571,
      "step": 11815
    },
    {
      "epoch": 1.05,
      "grad_norm": 7.29849572895556,
      "learning_rate": 4.804648306059879e-06,
      "loss": 0.6105,
      "step": 11816
    },
    {
      "epoch": 1.05,
      "grad_norm": 7.2822846444648865,
      "learning_rate": 4.803926488388399e-06,
      "loss": 0.6074,
      "step": 11817
    },
    {
      "epoch": 1.05,
      "grad_norm": 7.515333532615236,
      "learning_rate": 4.803204674809525e-06,
      "loss": 0.5521,
      "step": 11818
    },
    {
      "epoch": 1.05,
      "grad_norm": 7.006201338075945,
      "learning_rate": 4.802482865338323e-06,
      "loss": 0.5984,
      "step": 11819
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.61751221215997,
      "learning_rate": 4.801761059989858e-06,
      "loss": 0.5718,
      "step": 11820
    },
    {
      "epoch": 1.05,
      "grad_norm": 9.082838722199938,
      "learning_rate": 4.801039258779197e-06,
      "loss": 0.5864,
      "step": 11821
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.427134211738039,
      "learning_rate": 4.800317461721405e-06,
      "loss": 0.5786,
      "step": 11822
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.369585005626349,
      "learning_rate": 4.799595668831549e-06,
      "loss": 0.6513,
      "step": 11823
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.166857242038706,
      "learning_rate": 4.798873880124696e-06,
      "loss": 0.6642,
      "step": 11824
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.4010376318464655,
      "learning_rate": 4.798152095615909e-06,
      "loss": 0.613,
      "step": 11825
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.502569614241491,
      "learning_rate": 4.797430315320254e-06,
      "loss": 0.6458,
      "step": 11826
    },
    {
      "epoch": 1.06,
      "grad_norm": 8.002812342884619,
      "learning_rate": 4.7967085392528e-06,
      "loss": 0.6212,
      "step": 11827
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.86572233503092,
      "learning_rate": 4.79598676742861e-06,
      "loss": 0.6678,
      "step": 11828
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.845179507658644,
      "learning_rate": 4.795264999862747e-06,
      "loss": 0.6288,
      "step": 11829
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.642335694734807,
      "learning_rate": 4.7945432365702795e-06,
      "loss": 0.5974,
      "step": 11830
    },
    {
      "epoch": 1.06,
      "grad_norm": 8.08874135073581,
      "learning_rate": 4.793821477566273e-06,
      "loss": 0.6038,
      "step": 11831
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.960528435970326,
      "learning_rate": 4.793099722865791e-06,
      "loss": 0.6272,
      "step": 11832
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.4316293474010315,
      "learning_rate": 4.792377972483899e-06,
      "loss": 0.6801,
      "step": 11833
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.732853082181118,
      "learning_rate": 4.7916562264356625e-06,
      "loss": 0.641,
      "step": 11834
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.455132652691217,
      "learning_rate": 4.790934484736146e-06,
      "loss": 0.6481,
      "step": 11835
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.445517598424585,
      "learning_rate": 4.790212747400414e-06,
      "loss": 0.5952,
      "step": 11836
    },
    {
      "epoch": 1.06,
      "grad_norm": 4.154493168262916,
      "learning_rate": 4.789491014443531e-06,
      "loss": 0.6622,
      "step": 11837
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.88869627008,
      "learning_rate": 4.788769285880563e-06,
      "loss": 0.6504,
      "step": 11838
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.388500239288493,
      "learning_rate": 4.788047561726573e-06,
      "loss": 0.6478,
      "step": 11839
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.861530116780925,
      "learning_rate": 4.7873258419966265e-06,
      "loss": 0.6467,
      "step": 11840
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.726625825366673,
      "learning_rate": 4.786604126705787e-06,
      "loss": 0.6143,
      "step": 11841
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.000359057397192,
      "learning_rate": 4.785882415869121e-06,
      "loss": 0.5954,
      "step": 11842
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.514104502732833,
      "learning_rate": 4.7851607095016875e-06,
      "loss": 0.6164,
      "step": 11843
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.314265722874526,
      "learning_rate": 4.784439007618556e-06,
      "loss": 0.6044,
      "step": 11844
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.6285984236473725,
      "learning_rate": 4.7837173102347865e-06,
      "loss": 0.5964,
      "step": 11845
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.7563929275340096,
      "learning_rate": 4.782995617365444e-06,
      "loss": 0.587,
      "step": 11846
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.1149452965973135,
      "learning_rate": 4.7822739290255945e-06,
      "loss": 0.6978,
      "step": 11847
    },
    {
      "epoch": 1.06,
      "grad_norm": 4.636319833060464,
      "learning_rate": 4.781552245230299e-06,
      "loss": 0.5569,
      "step": 11848
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.943608140169962,
      "learning_rate": 4.7808305659946215e-06,
      "loss": 0.5653,
      "step": 11849
    },
    {
      "epoch": 1.06,
      "grad_norm": 8.329407536551669,
      "learning_rate": 4.780108891333628e-06,
      "loss": 0.6082,
      "step": 11850
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.396040370931967,
      "learning_rate": 4.779387221262378e-06,
      "loss": 0.6275,
      "step": 11851
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.448320071335851,
      "learning_rate": 4.778665555795937e-06,
      "loss": 0.5853,
      "step": 11852
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.3428474972492355,
      "learning_rate": 4.777943894949369e-06,
      "loss": 0.5831,
      "step": 11853
    },
    {
      "epoch": 1.06,
      "grad_norm": 4.89987424343236,
      "learning_rate": 4.777222238737734e-06,
      "loss": 0.6278,
      "step": 11854
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.85224129101676,
      "learning_rate": 4.776500587176097e-06,
      "loss": 0.6732,
      "step": 11855
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.75314435969929,
      "learning_rate": 4.775778940279524e-06,
      "loss": 0.6944,
      "step": 11856
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.233119273525305,
      "learning_rate": 4.775057298063072e-06,
      "loss": 0.6348,
      "step": 11857
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.126632780411375,
      "learning_rate": 4.774335660541806e-06,
      "loss": 0.6362,
      "step": 11858
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.91072813493523,
      "learning_rate": 4.773614027730789e-06,
      "loss": 0.5663,
      "step": 11859
    },
    {
      "epoch": 1.06,
      "grad_norm": 9.123015786199197,
      "learning_rate": 4.772892399645084e-06,
      "loss": 0.6333,
      "step": 11860
    },
    {
      "epoch": 1.06,
      "grad_norm": 10.494177290373202,
      "learning_rate": 4.772170776299752e-06,
      "loss": 0.6567,
      "step": 11861
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.560968873110473,
      "learning_rate": 4.771449157709857e-06,
      "loss": 0.6268,
      "step": 11862
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.186380082156406,
      "learning_rate": 4.77072754389046e-06,
      "loss": 0.633,
      "step": 11863
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.559468478979155,
      "learning_rate": 4.770005934856623e-06,
      "loss": 0.5925,
      "step": 11864
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.2576950730321315,
      "learning_rate": 4.7692843306234086e-06,
      "loss": 0.6461,
      "step": 11865
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.508809671195857,
      "learning_rate": 4.768562731205879e-06,
      "loss": 0.6593,
      "step": 11866
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.196084687330982,
      "learning_rate": 4.767841136619094e-06,
      "loss": 0.5616,
      "step": 11867
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.165914950035639,
      "learning_rate": 4.76711954687812e-06,
      "loss": 0.5994,
      "step": 11868
    },
    {
      "epoch": 1.06,
      "grad_norm": 8.338427517383668,
      "learning_rate": 4.766397961998013e-06,
      "loss": 0.6474,
      "step": 11869
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.291790467353545,
      "learning_rate": 4.765676381993839e-06,
      "loss": 0.6168,
      "step": 11870
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.247941280259271,
      "learning_rate": 4.764954806880656e-06,
      "loss": 0.5581,
      "step": 11871
    },
    {
      "epoch": 1.06,
      "grad_norm": 8.266017154126727,
      "learning_rate": 4.764233236673526e-06,
      "loss": 0.6495,
      "step": 11872
    },
    {
      "epoch": 1.06,
      "grad_norm": 4.162647284284624,
      "learning_rate": 4.7635116713875115e-06,
      "loss": 0.6009,
      "step": 11873
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.009618857339073,
      "learning_rate": 4.762790111037673e-06,
      "loss": 0.6226,
      "step": 11874
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.973823875207583,
      "learning_rate": 4.76206855563907e-06,
      "loss": 0.652,
      "step": 11875
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.754235530560493,
      "learning_rate": 4.761347005206767e-06,
      "loss": 0.6005,
      "step": 11876
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.045635320784792,
      "learning_rate": 4.760625459755822e-06,
      "loss": 0.6177,
      "step": 11877
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.150281436532465,
      "learning_rate": 4.759903919301297e-06,
      "loss": 0.5876,
      "step": 11878
    },
    {
      "epoch": 1.06,
      "grad_norm": 8.518743556133074,
      "learning_rate": 4.7591823838582505e-06,
      "loss": 0.6168,
      "step": 11879
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.892860003121722,
      "learning_rate": 4.758460853441743e-06,
      "loss": 0.6755,
      "step": 11880
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.0778341762537345,
      "learning_rate": 4.75773932806684e-06,
      "loss": 0.5735,
      "step": 11881
    },
    {
      "epoch": 1.06,
      "grad_norm": 8.112329020544912,
      "learning_rate": 4.757017807748595e-06,
      "loss": 0.6288,
      "step": 11882
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.811071585328688,
      "learning_rate": 4.756296292502072e-06,
      "loss": 0.6424,
      "step": 11883
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.897903004446372,
      "learning_rate": 4.755574782342331e-06,
      "loss": 0.6563,
      "step": 11884
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.372758675233637,
      "learning_rate": 4.754853277284433e-06,
      "loss": 0.6261,
      "step": 11885
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.875711513889177,
      "learning_rate": 4.754131777343432e-06,
      "loss": 0.6851,
      "step": 11886
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.64910225667467,
      "learning_rate": 4.753410282534393e-06,
      "loss": 0.6213,
      "step": 11887
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.849204229988389,
      "learning_rate": 4.752688792872376e-06,
      "loss": 0.5927,
      "step": 11888
    },
    {
      "epoch": 1.06,
      "grad_norm": 4.831495853063509,
      "learning_rate": 4.751967308372436e-06,
      "loss": 0.5844,
      "step": 11889
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.207408142039933,
      "learning_rate": 4.751245829049637e-06,
      "loss": 0.5605,
      "step": 11890
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.810504368951274,
      "learning_rate": 4.750524354919036e-06,
      "loss": 0.6245,
      "step": 11891
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.99641686247119,
      "learning_rate": 4.749802885995693e-06,
      "loss": 0.6413,
      "step": 11892
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.546083541445912,
      "learning_rate": 4.749081422294666e-06,
      "loss": 0.6235,
      "step": 11893
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.174597369933894,
      "learning_rate": 4.748359963831016e-06,
      "loss": 0.5794,
      "step": 11894
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.000604768863522,
      "learning_rate": 4.7476385106198005e-06,
      "loss": 0.6354,
      "step": 11895
    },
    {
      "epoch": 1.06,
      "grad_norm": 4.851895648051762,
      "learning_rate": 4.746917062676078e-06,
      "loss": 0.6269,
      "step": 11896
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.139964652166451,
      "learning_rate": 4.746195620014908e-06,
      "loss": 0.6353,
      "step": 11897
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.396252719761897,
      "learning_rate": 4.7454741826513475e-06,
      "loss": 0.601,
      "step": 11898
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.777922978533689,
      "learning_rate": 4.744752750600459e-06,
      "loss": 0.611,
      "step": 11899
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.813611550679431,
      "learning_rate": 4.744031323877295e-06,
      "loss": 0.6517,
      "step": 11900
    },
    {
      "epoch": 1.06,
      "grad_norm": 4.87158070864444,
      "learning_rate": 4.743309902496918e-06,
      "loss": 0.6536,
      "step": 11901
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.978289460246686,
      "learning_rate": 4.742588486474383e-06,
      "loss": 0.6274,
      "step": 11902
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.317852709458828,
      "learning_rate": 4.741867075824751e-06,
      "loss": 0.613,
      "step": 11903
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.622324237581697,
      "learning_rate": 4.741145670563079e-06,
      "loss": 0.5851,
      "step": 11904
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.131163318314807,
      "learning_rate": 4.740424270704422e-06,
      "loss": 0.6736,
      "step": 11905
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.473256692765911,
      "learning_rate": 4.739702876263841e-06,
      "loss": 0.6117,
      "step": 11906
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.3117328819429686,
      "learning_rate": 4.738981487256394e-06,
      "loss": 0.6813,
      "step": 11907
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.839169450647451,
      "learning_rate": 4.738260103697134e-06,
      "loss": 0.6269,
      "step": 11908
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.381631031290111,
      "learning_rate": 4.737538725601122e-06,
      "loss": 0.665,
      "step": 11909
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.281182177184,
      "learning_rate": 4.736817352983416e-06,
      "loss": 0.6542,
      "step": 11910
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.199938607010687,
      "learning_rate": 4.73609598585907e-06,
      "loss": 0.6402,
      "step": 11911
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.331771006385832,
      "learning_rate": 4.735374624243142e-06,
      "loss": 0.6492,
      "step": 11912
    },
    {
      "epoch": 1.06,
      "grad_norm": 8.940182995941031,
      "learning_rate": 4.734653268150691e-06,
      "loss": 0.5808,
      "step": 11913
    },
    {
      "epoch": 1.06,
      "grad_norm": 10.054085035155888,
      "learning_rate": 4.733931917596772e-06,
      "loss": 0.635,
      "step": 11914
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.183574051727937,
      "learning_rate": 4.73321057259644e-06,
      "loss": 0.6629,
      "step": 11915
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.792533890532135,
      "learning_rate": 4.732489233164753e-06,
      "loss": 0.545,
      "step": 11916
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.396796725271783,
      "learning_rate": 4.731767899316769e-06,
      "loss": 0.5808,
      "step": 11917
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.097392051451859,
      "learning_rate": 4.7310465710675405e-06,
      "loss": 0.6788,
      "step": 11918
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.23340952897792,
      "learning_rate": 4.730325248432127e-06,
      "loss": 0.6729,
      "step": 11919
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.876469733306199,
      "learning_rate": 4.729603931425583e-06,
      "loss": 0.6362,
      "step": 11920
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.721322871942779,
      "learning_rate": 4.728882620062965e-06,
      "loss": 0.5914,
      "step": 11921
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.620378790164176,
      "learning_rate": 4.7281613143593274e-06,
      "loss": 0.5667,
      "step": 11922
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.2410195089417275,
      "learning_rate": 4.727440014329729e-06,
      "loss": 0.6081,
      "step": 11923
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.82248528179745,
      "learning_rate": 4.7267187199892225e-06,
      "loss": 0.623,
      "step": 11924
    },
    {
      "epoch": 1.06,
      "grad_norm": 4.945231296291067,
      "learning_rate": 4.725997431352864e-06,
      "loss": 0.5736,
      "step": 11925
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.427891630166787,
      "learning_rate": 4.725276148435709e-06,
      "loss": 0.6401,
      "step": 11926
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.003752325857021,
      "learning_rate": 4.724554871252813e-06,
      "loss": 0.5686,
      "step": 11927
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.555226987020725,
      "learning_rate": 4.723833599819233e-06,
      "loss": 0.6172,
      "step": 11928
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.5114240565046275,
      "learning_rate": 4.7231123341500195e-06,
      "loss": 0.6015,
      "step": 11929
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.785665601692966,
      "learning_rate": 4.72239107426023e-06,
      "loss": 0.5882,
      "step": 11930
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.691055845420415,
      "learning_rate": 4.721669820164918e-06,
      "loss": 0.6564,
      "step": 11931
    },
    {
      "epoch": 1.06,
      "grad_norm": 8.165784557960034,
      "learning_rate": 4.720948571879141e-06,
      "loss": 0.6232,
      "step": 11932
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.5009923279236155,
      "learning_rate": 4.72022732941795e-06,
      "loss": 0.6251,
      "step": 11933
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.776074208951971,
      "learning_rate": 4.719506092796401e-06,
      "loss": 0.5962,
      "step": 11934
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.641402152127873,
      "learning_rate": 4.718784862029547e-06,
      "loss": 0.5992,
      "step": 11935
    },
    {
      "epoch": 1.06,
      "grad_norm": 4.549928946557599,
      "learning_rate": 4.718063637132445e-06,
      "loss": 0.655,
      "step": 11936
    },
    {
      "epoch": 1.06,
      "grad_norm": 8.362204629935281,
      "learning_rate": 4.717342418120146e-06,
      "loss": 0.6465,
      "step": 11937
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.900429216286781,
      "learning_rate": 4.716621205007705e-06,
      "loss": 0.6326,
      "step": 11938
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.901279048389313,
      "learning_rate": 4.715899997810177e-06,
      "loss": 0.5977,
      "step": 11939
    },
    {
      "epoch": 1.07,
      "grad_norm": 7.588033708835188,
      "learning_rate": 4.715178796542613e-06,
      "loss": 0.6055,
      "step": 11940
    },
    {
      "epoch": 1.07,
      "grad_norm": 7.664661152385146,
      "learning_rate": 4.714457601220068e-06,
      "loss": 0.6116,
      "step": 11941
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.115518544087896,
      "learning_rate": 4.713736411857597e-06,
      "loss": 0.5694,
      "step": 11942
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.0910672246913755,
      "learning_rate": 4.713015228470251e-06,
      "loss": 0.6039,
      "step": 11943
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.812483987768947,
      "learning_rate": 4.712294051073082e-06,
      "loss": 0.6456,
      "step": 11944
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.934580418338028,
      "learning_rate": 4.711572879681145e-06,
      "loss": 0.6702,
      "step": 11945
    },
    {
      "epoch": 1.07,
      "grad_norm": 4.325186950965554,
      "learning_rate": 4.710851714309494e-06,
      "loss": 0.5976,
      "step": 11946
    },
    {
      "epoch": 1.07,
      "grad_norm": 4.368591349085769,
      "learning_rate": 4.710130554973179e-06,
      "loss": 0.5621,
      "step": 11947
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.333223989048208,
      "learning_rate": 4.709409401687253e-06,
      "loss": 0.6736,
      "step": 11948
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.370523091843543,
      "learning_rate": 4.708688254466771e-06,
      "loss": 0.6514,
      "step": 11949
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.394385442368738,
      "learning_rate": 4.707967113326782e-06,
      "loss": 0.6171,
      "step": 11950
    },
    {
      "epoch": 1.07,
      "grad_norm": 7.229174179025565,
      "learning_rate": 4.707245978282341e-06,
      "loss": 0.6465,
      "step": 11951
    },
    {
      "epoch": 1.07,
      "grad_norm": 7.3671810740176005,
      "learning_rate": 4.7065248493485e-06,
      "loss": 0.6482,
      "step": 11952
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.403478234840632,
      "learning_rate": 4.705803726540308e-06,
      "loss": 0.6063,
      "step": 11953
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.684188497482172,
      "learning_rate": 4.705082609872822e-06,
      "loss": 0.578,
      "step": 11954
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.092798056155695,
      "learning_rate": 4.704361499361087e-06,
      "loss": 0.6437,
      "step": 11955
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.462642036755844,
      "learning_rate": 4.7036403950201605e-06,
      "loss": 0.689,
      "step": 11956
    },
    {
      "epoch": 1.07,
      "grad_norm": 7.448240886193026,
      "learning_rate": 4.7029192968650926e-06,
      "loss": 0.538,
      "step": 11957
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.104863429324404,
      "learning_rate": 4.702198204910932e-06,
      "loss": 0.6002,
      "step": 11958
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.030524612116136,
      "learning_rate": 4.701477119172733e-06,
      "loss": 0.5963,
      "step": 11959
    },
    {
      "epoch": 1.07,
      "grad_norm": 7.072310648305771,
      "learning_rate": 4.700756039665544e-06,
      "loss": 0.6052,
      "step": 11960
    },
    {
      "epoch": 1.07,
      "grad_norm": 10.157621447263258,
      "learning_rate": 4.700034966404418e-06,
      "loss": 0.614,
      "step": 11961
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.49994185495951,
      "learning_rate": 4.6993138994044055e-06,
      "loss": 0.6113,
      "step": 11962
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.400381930664456,
      "learning_rate": 4.698592838680556e-06,
      "loss": 0.5192,
      "step": 11963
    },
    {
      "epoch": 1.07,
      "grad_norm": 7.310687894089106,
      "learning_rate": 4.697871784247922e-06,
      "loss": 0.6134,
      "step": 11964
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.742090377180881,
      "learning_rate": 4.6971507361215515e-06,
      "loss": 0.6166,
      "step": 11965
    },
    {
      "epoch": 1.07,
      "grad_norm": 8.950683256385949,
      "learning_rate": 4.696429694316496e-06,
      "loss": 0.6781,
      "step": 11966
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.588721792258092,
      "learning_rate": 4.695708658847808e-06,
      "loss": 0.6797,
      "step": 11967
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.15859957091745,
      "learning_rate": 4.694987629730533e-06,
      "loss": 0.6667,
      "step": 11968
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.601892782632707,
      "learning_rate": 4.6942666069797235e-06,
      "loss": 0.6441,
      "step": 11969
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.166572157612888,
      "learning_rate": 4.693545590610431e-06,
      "loss": 0.5799,
      "step": 11970
    },
    {
      "epoch": 1.07,
      "grad_norm": 8.01804035131906,
      "learning_rate": 4.692824580637704e-06,
      "loss": 0.7079,
      "step": 11971
    },
    {
      "epoch": 1.07,
      "grad_norm": 7.277492839877546,
      "learning_rate": 4.692103577076588e-06,
      "loss": 0.5903,
      "step": 11972
    },
    {
      "epoch": 1.07,
      "grad_norm": 7.014439720204969,
      "learning_rate": 4.691382579942136e-06,
      "loss": 0.6059,
      "step": 11973
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.876319343459802,
      "learning_rate": 4.690661589249399e-06,
      "loss": 0.609,
      "step": 11974
    },
    {
      "epoch": 1.07,
      "grad_norm": 8.672792765858931,
      "learning_rate": 4.689940605013422e-06,
      "loss": 0.6103,
      "step": 11975
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.824626051923619,
      "learning_rate": 4.689219627249255e-06,
      "loss": 0.654,
      "step": 11976
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.8295520323059025,
      "learning_rate": 4.68849865597195e-06,
      "loss": 0.5675,
      "step": 11977
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.326408509111778,
      "learning_rate": 4.687777691196551e-06,
      "loss": 0.6153,
      "step": 11978
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.481471536771306,
      "learning_rate": 4.687056732938111e-06,
      "loss": 0.6034,
      "step": 11979
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.2155806010816335,
      "learning_rate": 4.6863357812116765e-06,
      "loss": 0.6456,
      "step": 11980
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.744998030095122,
      "learning_rate": 4.685614836032294e-06,
      "loss": 0.59,
      "step": 11981
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.908646758444147,
      "learning_rate": 4.684893897415013e-06,
      "loss": 0.6127,
      "step": 11982
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.84652498101671,
      "learning_rate": 4.684172965374883e-06,
      "loss": 0.585,
      "step": 11983
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.2165932581120185,
      "learning_rate": 4.683452039926951e-06,
      "loss": 0.5844,
      "step": 11984
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.329453446115876,
      "learning_rate": 4.682731121086263e-06,
      "loss": 0.6003,
      "step": 11985
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.277953752776681,
      "learning_rate": 4.682010208867872e-06,
      "loss": 0.5871,
      "step": 11986
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.83383052371514,
      "learning_rate": 4.681289303286818e-06,
      "loss": 0.5969,
      "step": 11987
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.666121518606537,
      "learning_rate": 4.6805684043581525e-06,
      "loss": 0.575,
      "step": 11988
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.915760763976355,
      "learning_rate": 4.679847512096921e-06,
      "loss": 0.5542,
      "step": 11989
    },
    {
      "epoch": 1.07,
      "grad_norm": 7.5848834101012965,
      "learning_rate": 4.679126626518174e-06,
      "loss": 0.6229,
      "step": 11990
    },
    {
      "epoch": 1.07,
      "grad_norm": 7.1866452223808315,
      "learning_rate": 4.678405747636954e-06,
      "loss": 0.5676,
      "step": 11991
    },
    {
      "epoch": 1.07,
      "grad_norm": 7.0879100419871355,
      "learning_rate": 4.677684875468309e-06,
      "loss": 0.5744,
      "step": 11992
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.127791495541252,
      "learning_rate": 4.676964010027288e-06,
      "loss": 0.6335,
      "step": 11993
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.569321927022114,
      "learning_rate": 4.676243151328936e-06,
      "loss": 0.6087,
      "step": 11994
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.304489267013175,
      "learning_rate": 4.675522299388298e-06,
      "loss": 0.603,
      "step": 11995
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.3111917061422895,
      "learning_rate": 4.674801454220423e-06,
      "loss": 0.6719,
      "step": 11996
    },
    {
      "epoch": 1.07,
      "grad_norm": 7.824395466447082,
      "learning_rate": 4.674080615840353e-06,
      "loss": 0.604,
      "step": 11997
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.538552550467098,
      "learning_rate": 4.673359784263137e-06,
      "loss": 0.6104,
      "step": 11998
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.2883614185236345,
      "learning_rate": 4.672638959503822e-06,
      "loss": 0.6115,
      "step": 11999
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.984832050114474,
      "learning_rate": 4.671918141577453e-06,
      "loss": 0.5792,
      "step": 12000
    },
    {
      "epoch": 1.07,
      "grad_norm": 7.4149865572395655,
      "learning_rate": 4.6711973304990715e-06,
      "loss": 0.5777,
      "step": 12001
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.393975561739815,
      "learning_rate": 4.670476526283726e-06,
      "loss": 0.5811,
      "step": 12002
    },
    {
      "epoch": 1.07,
      "grad_norm": 7.058309309427976,
      "learning_rate": 4.669755728946462e-06,
      "loss": 0.6278,
      "step": 12003
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.237332933206511,
      "learning_rate": 4.6690349385023236e-06,
      "loss": 0.6641,
      "step": 12004
    },
    {
      "epoch": 1.07,
      "grad_norm": 7.595909316875707,
      "learning_rate": 4.6683141549663555e-06,
      "loss": 0.5979,
      "step": 12005
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.8455491370505674,
      "learning_rate": 4.667593378353604e-06,
      "loss": 0.5988,
      "step": 12006
    },
    {
      "epoch": 1.07,
      "grad_norm": 8.44536537519205,
      "learning_rate": 4.666872608679112e-06,
      "loss": 0.5795,
      "step": 12007
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.162995820974451,
      "learning_rate": 4.666151845957925e-06,
      "loss": 0.6203,
      "step": 12008
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.9392579739231515,
      "learning_rate": 4.665431090205088e-06,
      "loss": 0.6572,
      "step": 12009
    },
    {
      "epoch": 1.07,
      "grad_norm": 7.540427555049084,
      "learning_rate": 4.664710341435643e-06,
      "loss": 0.6462,
      "step": 12010
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.0895665213782495,
      "learning_rate": 4.663989599664635e-06,
      "loss": 0.5421,
      "step": 12011
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.9701478819068665,
      "learning_rate": 4.66326886490711e-06,
      "loss": 0.5854,
      "step": 12012
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.825920288168263,
      "learning_rate": 4.662548137178109e-06,
      "loss": 0.5791,
      "step": 12013
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.457803568835402,
      "learning_rate": 4.661827416492676e-06,
      "loss": 0.5463,
      "step": 12014
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.096793792702986,
      "learning_rate": 4.661106702865858e-06,
      "loss": 0.6094,
      "step": 12015
    },
    {
      "epoch": 1.07,
      "grad_norm": 4.8748151496922585,
      "learning_rate": 4.660385996312694e-06,
      "loss": 0.6325,
      "step": 12016
    },
    {
      "epoch": 1.07,
      "grad_norm": 4.868297345700508,
      "learning_rate": 4.659665296848227e-06,
      "loss": 0.5743,
      "step": 12017
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.914928349924997,
      "learning_rate": 4.658944604487502e-06,
      "loss": 0.6589,
      "step": 12018
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.607176621052412,
      "learning_rate": 4.658223919245563e-06,
      "loss": 0.6068,
      "step": 12019
    },
    {
      "epoch": 1.07,
      "grad_norm": 7.598039529525154,
      "learning_rate": 4.65750324113745e-06,
      "loss": 0.6663,
      "step": 12020
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.921123864964928,
      "learning_rate": 4.656782570178206e-06,
      "loss": 0.5816,
      "step": 12021
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.619711791958025,
      "learning_rate": 4.656061906382876e-06,
      "loss": 0.6152,
      "step": 12022
    },
    {
      "epoch": 1.07,
      "grad_norm": 4.9808794448671785,
      "learning_rate": 4.655341249766499e-06,
      "loss": 0.6362,
      "step": 12023
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.660330849402079,
      "learning_rate": 4.6546206003441195e-06,
      "loss": 0.6472,
      "step": 12024
    },
    {
      "epoch": 1.07,
      "grad_norm": 9.36586807806784,
      "learning_rate": 4.653899958130779e-06,
      "loss": 0.635,
      "step": 12025
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.833453452794807,
      "learning_rate": 4.6531793231415174e-06,
      "loss": 0.5625,
      "step": 12026
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.78080109161468,
      "learning_rate": 4.652458695391379e-06,
      "loss": 0.677,
      "step": 12027
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.111010727221864,
      "learning_rate": 4.651738074895404e-06,
      "loss": 0.5562,
      "step": 12028
    },
    {
      "epoch": 1.07,
      "grad_norm": 7.363112814335502,
      "learning_rate": 4.651017461668636e-06,
      "loss": 0.5856,
      "step": 12029
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.653322963230796,
      "learning_rate": 4.650296855726112e-06,
      "loss": 0.6553,
      "step": 12030
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.697609573549606,
      "learning_rate": 4.6495762570828755e-06,
      "loss": 0.5515,
      "step": 12031
    },
    {
      "epoch": 1.07,
      "grad_norm": 9.313159098032665,
      "learning_rate": 4.648855665753968e-06,
      "loss": 0.6253,
      "step": 12032
    },
    {
      "epoch": 1.07,
      "grad_norm": 9.231770424228843,
      "learning_rate": 4.648135081754428e-06,
      "loss": 0.6165,
      "step": 12033
    },
    {
      "epoch": 1.07,
      "grad_norm": 10.392286096453311,
      "learning_rate": 4.647414505099298e-06,
      "loss": 0.6418,
      "step": 12034
    },
    {
      "epoch": 1.07,
      "grad_norm": 8.068453116538683,
      "learning_rate": 4.646693935803619e-06,
      "loss": 0.642,
      "step": 12035
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.029806263517552,
      "learning_rate": 4.6459733738824295e-06,
      "loss": 0.6183,
      "step": 12036
    },
    {
      "epoch": 1.07,
      "grad_norm": 4.756783589564232,
      "learning_rate": 4.645252819350771e-06,
      "loss": 0.5826,
      "step": 12037
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.494720613584122,
      "learning_rate": 4.644532272223683e-06,
      "loss": 0.5881,
      "step": 12038
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.065618648146596,
      "learning_rate": 4.643811732516205e-06,
      "loss": 0.6171,
      "step": 12039
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.146332572856123,
      "learning_rate": 4.643091200243378e-06,
      "loss": 0.6171,
      "step": 12040
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.6333547328821485,
      "learning_rate": 4.642370675420239e-06,
      "loss": 0.6158,
      "step": 12041
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.5961002297373765,
      "learning_rate": 4.641650158061829e-06,
      "loss": 0.6489,
      "step": 12042
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.647150772168137,
      "learning_rate": 4.64092964818319e-06,
      "loss": 0.5965,
      "step": 12043
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.653058234789503,
      "learning_rate": 4.640209145799356e-06,
      "loss": 0.6157,
      "step": 12044
    },
    {
      "epoch": 1.07,
      "grad_norm": 7.4688926001170985,
      "learning_rate": 4.639488650925369e-06,
      "loss": 0.6581,
      "step": 12045
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.558730494480607,
      "learning_rate": 4.6387681635762655e-06,
      "loss": 0.6058,
      "step": 12046
    },
    {
      "epoch": 1.07,
      "grad_norm": 7.334274446442271,
      "learning_rate": 4.638047683767087e-06,
      "loss": 0.636,
      "step": 12047
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.731253096415271,
      "learning_rate": 4.637327211512869e-06,
      "loss": 0.5802,
      "step": 12048
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.41220219880376,
      "learning_rate": 4.636606746828651e-06,
      "loss": 0.6648,
      "step": 12049
    },
    {
      "epoch": 1.07,
      "grad_norm": 7.096633479035297,
      "learning_rate": 4.635886289729474e-06,
      "loss": 0.5981,
      "step": 12050
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.72637533998417,
      "learning_rate": 4.635165840230371e-06,
      "loss": 0.5717,
      "step": 12051
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.496541841386946,
      "learning_rate": 4.634445398346382e-06,
      "loss": 0.622,
      "step": 12052
    },
    {
      "epoch": 1.08,
      "grad_norm": 4.347419132950107,
      "learning_rate": 4.633724964092546e-06,
      "loss": 0.624,
      "step": 12053
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.15919707006026,
      "learning_rate": 4.633004537483898e-06,
      "loss": 0.6193,
      "step": 12054
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.383270221134732,
      "learning_rate": 4.632284118535477e-06,
      "loss": 0.6165,
      "step": 12055
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.795948583368702,
      "learning_rate": 4.631563707262321e-06,
      "loss": 0.596,
      "step": 12056
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.391270540430426,
      "learning_rate": 4.630843303679464e-06,
      "loss": 0.6642,
      "step": 12057
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.002638898551668,
      "learning_rate": 4.630122907801947e-06,
      "loss": 0.5754,
      "step": 12058
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.4147721428471245,
      "learning_rate": 4.629402519644803e-06,
      "loss": 0.6472,
      "step": 12059
    },
    {
      "epoch": 1.08,
      "grad_norm": 7.3653384128528545,
      "learning_rate": 4.6286821392230695e-06,
      "loss": 0.5081,
      "step": 12060
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.095476539842993,
      "learning_rate": 4.627961766551783e-06,
      "loss": 0.6619,
      "step": 12061
    },
    {
      "epoch": 1.08,
      "grad_norm": 4.946561731677316,
      "learning_rate": 4.627241401645979e-06,
      "loss": 0.6149,
      "step": 12062
    },
    {
      "epoch": 1.08,
      "grad_norm": 8.291245431005313,
      "learning_rate": 4.626521044520696e-06,
      "loss": 0.5789,
      "step": 12063
    },
    {
      "epoch": 1.08,
      "grad_norm": 4.96777229375153,
      "learning_rate": 4.625800695190967e-06,
      "loss": 0.656,
      "step": 12064
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.031272621498209,
      "learning_rate": 4.6250803536718294e-06,
      "loss": 0.5864,
      "step": 12065
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.785223336403202,
      "learning_rate": 4.62436001997832e-06,
      "loss": 0.5914,
      "step": 12066
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.414874728157363,
      "learning_rate": 4.6236396941254704e-06,
      "loss": 0.6072,
      "step": 12067
    },
    {
      "epoch": 1.08,
      "grad_norm": 4.699618381456815,
      "learning_rate": 4.622919376128318e-06,
      "loss": 0.6361,
      "step": 12068
    },
    {
      "epoch": 1.08,
      "grad_norm": 7.4551168138029436,
      "learning_rate": 4.622199066001899e-06,
      "loss": 0.6375,
      "step": 12069
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.768849779864432,
      "learning_rate": 4.621478763761247e-06,
      "loss": 0.5957,
      "step": 12070
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.224053711821635,
      "learning_rate": 4.620758469421396e-06,
      "loss": 0.6289,
      "step": 12071
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.745967093039336,
      "learning_rate": 4.620038182997384e-06,
      "loss": 0.6312,
      "step": 12072
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.539237950229194,
      "learning_rate": 4.619317904504241e-06,
      "loss": 0.6018,
      "step": 12073
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.790032094884742,
      "learning_rate": 4.618597633957003e-06,
      "loss": 0.6136,
      "step": 12074
    },
    {
      "epoch": 1.08,
      "grad_norm": 7.340390396466589,
      "learning_rate": 4.617877371370703e-06,
      "loss": 0.6594,
      "step": 12075
    },
    {
      "epoch": 1.08,
      "grad_norm": 8.871324533243085,
      "learning_rate": 4.6171571167603776e-06,
      "loss": 0.6071,
      "step": 12076
    },
    {
      "epoch": 1.08,
      "grad_norm": 7.983024231892863,
      "learning_rate": 4.616436870141059e-06,
      "loss": 0.5835,
      "step": 12077
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.223636743431161,
      "learning_rate": 4.615716631527779e-06,
      "loss": 0.5653,
      "step": 12078
    },
    {
      "epoch": 1.08,
      "grad_norm": 7.996878332110762,
      "learning_rate": 4.6149964009355745e-06,
      "loss": 0.6191,
      "step": 12079
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.851596321400062,
      "learning_rate": 4.614276178379476e-06,
      "loss": 0.5474,
      "step": 12080
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.694528085403803,
      "learning_rate": 4.613555963874517e-06,
      "loss": 0.6374,
      "step": 12081
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.710077790560148,
      "learning_rate": 4.612835757435733e-06,
      "loss": 0.6029,
      "step": 12082
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.032301855615928,
      "learning_rate": 4.612115559078154e-06,
      "loss": 0.6318,
      "step": 12083
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.114690617818625,
      "learning_rate": 4.611395368816811e-06,
      "loss": 0.5928,
      "step": 12084
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.058917552582104,
      "learning_rate": 4.610675186666742e-06,
      "loss": 0.6296,
      "step": 12085
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.951238036531276,
      "learning_rate": 4.609955012642974e-06,
      "loss": 0.5849,
      "step": 12086
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.185561162875943,
      "learning_rate": 4.609234846760543e-06,
      "loss": 0.5993,
      "step": 12087
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.813865926986282,
      "learning_rate": 4.608514689034477e-06,
      "loss": 0.5751,
      "step": 12088
    },
    {
      "epoch": 1.08,
      "grad_norm": 7.206999276422698,
      "learning_rate": 4.60779453947981e-06,
      "loss": 0.637,
      "step": 12089
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.215976501075295,
      "learning_rate": 4.607074398111572e-06,
      "loss": 0.5922,
      "step": 12090
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.887408112606375,
      "learning_rate": 4.606354264944796e-06,
      "loss": 0.6145,
      "step": 12091
    },
    {
      "epoch": 1.08,
      "grad_norm": 7.543329313808365,
      "learning_rate": 4.605634139994514e-06,
      "loss": 0.6157,
      "step": 12092
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.554584925375946,
      "learning_rate": 4.604914023275754e-06,
      "loss": 0.6379,
      "step": 12093
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.453022567297067,
      "learning_rate": 4.6041939148035484e-06,
      "loss": 0.6074,
      "step": 12094
    },
    {
      "epoch": 1.08,
      "grad_norm": 7.323731010552585,
      "learning_rate": 4.603473814592929e-06,
      "loss": 0.6222,
      "step": 12095
    },
    {
      "epoch": 1.08,
      "grad_norm": 7.636118104825007,
      "learning_rate": 4.6027537226589246e-06,
      "loss": 0.6914,
      "step": 12096
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.080530567479221,
      "learning_rate": 4.602033639016566e-06,
      "loss": 0.6055,
      "step": 12097
    },
    {
      "epoch": 1.08,
      "grad_norm": 7.671317788299645,
      "learning_rate": 4.601313563680886e-06,
      "loss": 0.5837,
      "step": 12098
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.583909578517009,
      "learning_rate": 4.6005934966669094e-06,
      "loss": 0.5916,
      "step": 12099
    },
    {
      "epoch": 1.08,
      "grad_norm": 7.981229278583459,
      "learning_rate": 4.599873437989671e-06,
      "loss": 0.6388,
      "step": 12100
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.224771115811171,
      "learning_rate": 4.599153387664198e-06,
      "loss": 0.6077,
      "step": 12101
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.8750182066549,
      "learning_rate": 4.598433345705521e-06,
      "loss": 0.6137,
      "step": 12102
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.248827198351331,
      "learning_rate": 4.597713312128666e-06,
      "loss": 0.6419,
      "step": 12103
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.1422709658653405,
      "learning_rate": 4.596993286948666e-06,
      "loss": 0.577,
      "step": 12104
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.566292236254606,
      "learning_rate": 4.596273270180549e-06,
      "loss": 0.6171,
      "step": 12105
    },
    {
      "epoch": 1.08,
      "grad_norm": 7.428699674564249,
      "learning_rate": 4.595553261839341e-06,
      "loss": 0.7029,
      "step": 12106
    },
    {
      "epoch": 1.08,
      "grad_norm": 7.527981573468961,
      "learning_rate": 4.594833261940075e-06,
      "loss": 0.621,
      "step": 12107
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.978220492474521,
      "learning_rate": 4.594113270497777e-06,
      "loss": 0.6035,
      "step": 12108
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.111645892068383,
      "learning_rate": 4.593393287527475e-06,
      "loss": 0.6482,
      "step": 12109
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.415084100878169,
      "learning_rate": 4.592673313044198e-06,
      "loss": 0.6085,
      "step": 12110
    },
    {
      "epoch": 1.08,
      "grad_norm": 7.583565054801552,
      "learning_rate": 4.591953347062975e-06,
      "loss": 0.6171,
      "step": 12111
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.488551658051941,
      "learning_rate": 4.5912333895988295e-06,
      "loss": 0.6039,
      "step": 12112
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.726993501881336,
      "learning_rate": 4.5905134406667945e-06,
      "loss": 0.5843,
      "step": 12113
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.513465220979658,
      "learning_rate": 4.589793500281893e-06,
      "loss": 0.5505,
      "step": 12114
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.31402293533623,
      "learning_rate": 4.589073568459153e-06,
      "loss": 0.6167,
      "step": 12115
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.184833878208935,
      "learning_rate": 4.588353645213606e-06,
      "loss": 0.6717,
      "step": 12116
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.9358038801035455,
      "learning_rate": 4.5876337305602726e-06,
      "loss": 0.6221,
      "step": 12117
    },
    {
      "epoch": 1.08,
      "grad_norm": 7.133085220955131,
      "learning_rate": 4.586913824514183e-06,
      "loss": 0.5934,
      "step": 12118
    },
    {
      "epoch": 1.08,
      "grad_norm": 8.945281789441415,
      "learning_rate": 4.586193927090361e-06,
      "loss": 0.6165,
      "step": 12119
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.508938074865498,
      "learning_rate": 4.585474038303835e-06,
      "loss": 0.6159,
      "step": 12120
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.449734299850156,
      "learning_rate": 4.584754158169632e-06,
      "loss": 0.6682,
      "step": 12121
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.03471414845222,
      "learning_rate": 4.584034286702775e-06,
      "loss": 0.5437,
      "step": 12122
    },
    {
      "epoch": 1.08,
      "grad_norm": 4.923315564435925,
      "learning_rate": 4.583314423918293e-06,
      "loss": 0.5731,
      "step": 12123
    },
    {
      "epoch": 1.08,
      "grad_norm": 8.767391181473451,
      "learning_rate": 4.582594569831209e-06,
      "loss": 0.6819,
      "step": 12124
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.614699790474262,
      "learning_rate": 4.5818747244565486e-06,
      "loss": 0.5999,
      "step": 12125
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.075181609663835,
      "learning_rate": 4.581154887809339e-06,
      "loss": 0.7166,
      "step": 12126
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.07651540682775,
      "learning_rate": 4.580435059904602e-06,
      "loss": 0.5571,
      "step": 12127
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.394427333263704,
      "learning_rate": 4.579715240757365e-06,
      "loss": 0.5815,
      "step": 12128
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.845883085898629,
      "learning_rate": 4.578995430382653e-06,
      "loss": 0.5452,
      "step": 12129
    },
    {
      "epoch": 1.08,
      "grad_norm": 7.1896648307884625,
      "learning_rate": 4.578275628795489e-06,
      "loss": 0.6524,
      "step": 12130
    },
    {
      "epoch": 1.08,
      "grad_norm": 7.8357344493109515,
      "learning_rate": 4.577555836010898e-06,
      "loss": 0.5971,
      "step": 12131
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.679404785601094,
      "learning_rate": 4.576836052043903e-06,
      "loss": 0.5557,
      "step": 12132
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.708870363318491,
      "learning_rate": 4.57611627690953e-06,
      "loss": 0.6347,
      "step": 12133
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.309345281386245,
      "learning_rate": 4.5753965106228e-06,
      "loss": 0.6963,
      "step": 12134
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.331666062172255,
      "learning_rate": 4.574676753198737e-06,
      "loss": 0.6839,
      "step": 12135
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.2793206163002475,
      "learning_rate": 4.573957004652368e-06,
      "loss": 0.7098,
      "step": 12136
    },
    {
      "epoch": 1.08,
      "grad_norm": 8.603115721946924,
      "learning_rate": 4.573237264998711e-06,
      "loss": 0.6989,
      "step": 12137
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.794590622181864,
      "learning_rate": 4.572517534252793e-06,
      "loss": 0.5644,
      "step": 12138
    },
    {
      "epoch": 1.08,
      "grad_norm": 7.158617160545257,
      "learning_rate": 4.5717978124296356e-06,
      "loss": 0.61,
      "step": 12139
    },
    {
      "epoch": 1.08,
      "grad_norm": 8.536619640600705,
      "learning_rate": 4.57107809954426e-06,
      "loss": 0.6448,
      "step": 12140
    },
    {
      "epoch": 1.08,
      "grad_norm": 7.4351472978341615,
      "learning_rate": 4.570358395611689e-06,
      "loss": 0.6132,
      "step": 12141
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.091325851212444,
      "learning_rate": 4.569638700646948e-06,
      "loss": 0.6084,
      "step": 12142
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.482410429409856,
      "learning_rate": 4.568919014665055e-06,
      "loss": 0.6077,
      "step": 12143
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.254367473884244,
      "learning_rate": 4.568199337681035e-06,
      "loss": 0.6737,
      "step": 12144
    },
    {
      "epoch": 1.08,
      "grad_norm": 4.2087272698140605,
      "learning_rate": 4.567479669709907e-06,
      "loss": 0.6127,
      "step": 12145
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.649835923055252,
      "learning_rate": 4.566760010766695e-06,
      "loss": 0.6204,
      "step": 12146
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.936716943874863,
      "learning_rate": 4.566040360866417e-06,
      "loss": 0.5985,
      "step": 12147
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.183499230430857,
      "learning_rate": 4.565320720024097e-06,
      "loss": 0.577,
      "step": 12148
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.6376103150746495,
      "learning_rate": 4.564601088254755e-06,
      "loss": 0.6088,
      "step": 12149
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.96148309397227,
      "learning_rate": 4.563881465573411e-06,
      "loss": 0.6765,
      "step": 12150
    },
    {
      "epoch": 1.08,
      "grad_norm": 8.942533286014006,
      "learning_rate": 4.563161851995087e-06,
      "loss": 0.6497,
      "step": 12151
    },
    {
      "epoch": 1.08,
      "grad_norm": 7.004761348988787,
      "learning_rate": 4.562442247534803e-06,
      "loss": 0.5874,
      "step": 12152
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.800899992565953,
      "learning_rate": 4.5617226522075784e-06,
      "loss": 0.698,
      "step": 12153
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.2971476643812965,
      "learning_rate": 4.561003066028434e-06,
      "loss": 0.5868,
      "step": 12154
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.901897579845932,
      "learning_rate": 4.5602834890123895e-06,
      "loss": 0.6263,
      "step": 12155
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.35392700242602,
      "learning_rate": 4.559563921174463e-06,
      "loss": 0.6335,
      "step": 12156
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.669500923508975,
      "learning_rate": 4.558844362529676e-06,
      "loss": 0.605,
      "step": 12157
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.414039149801606,
      "learning_rate": 4.558124813093047e-06,
      "loss": 0.6027,
      "step": 12158
    },
    {
      "epoch": 1.08,
      "grad_norm": 7.305014443085208,
      "learning_rate": 4.5574052728795965e-06,
      "loss": 0.5806,
      "step": 12159
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.153035057501803,
      "learning_rate": 4.5566857419043396e-06,
      "loss": 0.5582,
      "step": 12160
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.431489669288865,
      "learning_rate": 4.555966220182297e-06,
      "loss": 0.604,
      "step": 12161
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.1931837147166435,
      "learning_rate": 4.555246707728487e-06,
      "loss": 0.6357,
      "step": 12162
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.760356090643258,
      "learning_rate": 4.554527204557927e-06,
      "loss": 0.6039,
      "step": 12163
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.519162273357288,
      "learning_rate": 4.553807710685637e-06,
      "loss": 0.5506,
      "step": 12164
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.860784977503885,
      "learning_rate": 4.553088226126635e-06,
      "loss": 0.6411,
      "step": 12165
    },
    {
      "epoch": 1.09,
      "grad_norm": 7.696702516199977,
      "learning_rate": 4.5523687508959366e-06,
      "loss": 0.5676,
      "step": 12166
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.2426825677561055,
      "learning_rate": 4.55164928500856e-06,
      "loss": 0.6065,
      "step": 12167
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.903898326270965,
      "learning_rate": 4.550929828479523e-06,
      "loss": 0.6366,
      "step": 12168
    },
    {
      "epoch": 1.09,
      "grad_norm": 4.594958249920871,
      "learning_rate": 4.550210381323842e-06,
      "loss": 0.628,
      "step": 12169
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.992505809305903,
      "learning_rate": 4.549490943556535e-06,
      "loss": 0.6768,
      "step": 12170
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.5161474155536725,
      "learning_rate": 4.5487715151926175e-06,
      "loss": 0.7219,
      "step": 12171
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.9983282134198515,
      "learning_rate": 4.548052096247106e-06,
      "loss": 0.6254,
      "step": 12172
    },
    {
      "epoch": 1.09,
      "grad_norm": 7.33702874784635,
      "learning_rate": 4.5473326867350195e-06,
      "loss": 0.6143,
      "step": 12173
    },
    {
      "epoch": 1.09,
      "grad_norm": 7.545803323322835,
      "learning_rate": 4.54661328667137e-06,
      "loss": 0.6498,
      "step": 12174
    },
    {
      "epoch": 1.09,
      "grad_norm": 8.030333955912607,
      "learning_rate": 4.545893896071176e-06,
      "loss": 0.5757,
      "step": 12175
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.900675270903722,
      "learning_rate": 4.545174514949452e-06,
      "loss": 0.6195,
      "step": 12176
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.433352342393618,
      "learning_rate": 4.544455143321213e-06,
      "loss": 0.644,
      "step": 12177
    },
    {
      "epoch": 1.09,
      "grad_norm": 8.582431569888172,
      "learning_rate": 4.543735781201476e-06,
      "loss": 0.5921,
      "step": 12178
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.856981569103109,
      "learning_rate": 4.543016428605256e-06,
      "loss": 0.6685,
      "step": 12179
    },
    {
      "epoch": 1.09,
      "grad_norm": 7.73768524827604,
      "learning_rate": 4.542297085547565e-06,
      "loss": 0.6089,
      "step": 12180
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.916786377788344,
      "learning_rate": 4.541577752043424e-06,
      "loss": 0.6298,
      "step": 12181
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.408303371534851,
      "learning_rate": 4.5408584281078396e-06,
      "loss": 0.6148,
      "step": 12182
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.953881074790135,
      "learning_rate": 4.540139113755832e-06,
      "loss": 0.5911,
      "step": 12183
    },
    {
      "epoch": 1.09,
      "grad_norm": 8.843315242956763,
      "learning_rate": 4.539419809002413e-06,
      "loss": 0.6487,
      "step": 12184
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.120903965464881,
      "learning_rate": 4.538700513862596e-06,
      "loss": 0.5954,
      "step": 12185
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.719693989812334,
      "learning_rate": 4.537981228351396e-06,
      "loss": 0.6533,
      "step": 12186
    },
    {
      "epoch": 1.09,
      "grad_norm": 7.731159396154621,
      "learning_rate": 4.537261952483827e-06,
      "loss": 0.649,
      "step": 12187
    },
    {
      "epoch": 1.09,
      "grad_norm": 8.2516752935545,
      "learning_rate": 4.536542686274903e-06,
      "loss": 0.6613,
      "step": 12188
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.158957241297204,
      "learning_rate": 4.535823429739632e-06,
      "loss": 0.6502,
      "step": 12189
    },
    {
      "epoch": 1.09,
      "grad_norm": 11.603742605243331,
      "learning_rate": 4.535104182893031e-06,
      "loss": 0.6026,
      "step": 12190
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.524118859322126,
      "learning_rate": 4.534384945750114e-06,
      "loss": 0.6513,
      "step": 12191
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.9518388553144,
      "learning_rate": 4.533665718325889e-06,
      "loss": 0.6093,
      "step": 12192
    },
    {
      "epoch": 1.09,
      "grad_norm": 7.6996137758412075,
      "learning_rate": 4.532946500635372e-06,
      "loss": 0.6289,
      "step": 12193
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.84983464964117,
      "learning_rate": 4.532227292693574e-06,
      "loss": 0.6107,
      "step": 12194
    },
    {
      "epoch": 1.09,
      "grad_norm": 4.844525075675774,
      "learning_rate": 4.531508094515507e-06,
      "loss": 0.691,
      "step": 12195
    },
    {
      "epoch": 1.09,
      "grad_norm": 8.012257968740206,
      "learning_rate": 4.530788906116182e-06,
      "loss": 0.6412,
      "step": 12196
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.061339579801128,
      "learning_rate": 4.530069727510612e-06,
      "loss": 0.6613,
      "step": 12197
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.294308450327937,
      "learning_rate": 4.529350558713806e-06,
      "loss": 0.5858,
      "step": 12198
    },
    {
      "epoch": 1.09,
      "grad_norm": 7.990128359437187,
      "learning_rate": 4.528631399740778e-06,
      "loss": 0.6471,
      "step": 12199
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.237255012733461,
      "learning_rate": 4.527912250606536e-06,
      "loss": 0.6424,
      "step": 12200
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.73558155247318,
      "learning_rate": 4.527193111326092e-06,
      "loss": 0.6517,
      "step": 12201
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.919372977202312,
      "learning_rate": 4.526473981914459e-06,
      "loss": 0.6824,
      "step": 12202
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.085467487156427,
      "learning_rate": 4.525754862386642e-06,
      "loss": 0.6169,
      "step": 12203
    },
    {
      "epoch": 1.09,
      "grad_norm": 7.468404036353225,
      "learning_rate": 4.525035752757654e-06,
      "loss": 0.6112,
      "step": 12204
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.909048819144996,
      "learning_rate": 4.524316653042504e-06,
      "loss": 0.5611,
      "step": 12205
    },
    {
      "epoch": 1.09,
      "grad_norm": 8.08497456737673,
      "learning_rate": 4.523597563256202e-06,
      "loss": 0.5609,
      "step": 12206
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.861875161230403,
      "learning_rate": 4.52287848341376e-06,
      "loss": 0.6631,
      "step": 12207
    },
    {
      "epoch": 1.09,
      "grad_norm": 7.223697605052951,
      "learning_rate": 4.522159413530182e-06,
      "loss": 0.586,
      "step": 12208
    },
    {
      "epoch": 1.09,
      "grad_norm": 8.243593262353837,
      "learning_rate": 4.521440353620482e-06,
      "loss": 0.6558,
      "step": 12209
    },
    {
      "epoch": 1.09,
      "grad_norm": 4.083872121420837,
      "learning_rate": 4.520721303699664e-06,
      "loss": 0.6394,
      "step": 12210
    },
    {
      "epoch": 1.09,
      "grad_norm": 8.480147156170185,
      "learning_rate": 4.52000226378274e-06,
      "loss": 0.6396,
      "step": 12211
    },
    {
      "epoch": 1.09,
      "grad_norm": 8.318491605712856,
      "learning_rate": 4.519283233884719e-06,
      "loss": 0.6993,
      "step": 12212
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.3119167580318765,
      "learning_rate": 4.518564214020607e-06,
      "loss": 0.6025,
      "step": 12213
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.3869704792366155,
      "learning_rate": 4.517845204205412e-06,
      "loss": 0.6419,
      "step": 12214
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.986770597097819,
      "learning_rate": 4.517126204454143e-06,
      "loss": 0.6238,
      "step": 12215
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.611565167168221,
      "learning_rate": 4.516407214781807e-06,
      "loss": 0.5898,
      "step": 12216
    },
    {
      "epoch": 1.09,
      "grad_norm": 7.7562642593691224,
      "learning_rate": 4.515688235203413e-06,
      "loss": 0.607,
      "step": 12217
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.317543305264216,
      "learning_rate": 4.514969265733965e-06,
      "loss": 0.5765,
      "step": 12218
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.69957638326467,
      "learning_rate": 4.514250306388471e-06,
      "loss": 0.6346,
      "step": 12219
    },
    {
      "epoch": 1.09,
      "grad_norm": 7.0719332943096305,
      "learning_rate": 4.513531357181939e-06,
      "loss": 0.6357,
      "step": 12220
    },
    {
      "epoch": 1.09,
      "grad_norm": 7.149432804305595,
      "learning_rate": 4.512812418129373e-06,
      "loss": 0.6051,
      "step": 12221
    },
    {
      "epoch": 1.09,
      "grad_norm": 7.337803659007509,
      "learning_rate": 4.512093489245782e-06,
      "loss": 0.642,
      "step": 12222
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.556745310241483,
      "learning_rate": 4.511374570546171e-06,
      "loss": 0.6299,
      "step": 12223
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.905008658589598,
      "learning_rate": 4.510655662045544e-06,
      "loss": 0.6225,
      "step": 12224
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.556754924980502,
      "learning_rate": 4.5099367637589105e-06,
      "loss": 0.6107,
      "step": 12225
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.310635189962499,
      "learning_rate": 4.509217875701273e-06,
      "loss": 0.628,
      "step": 12226
    },
    {
      "epoch": 1.09,
      "grad_norm": 7.753661544794619,
      "learning_rate": 4.508498997887637e-06,
      "loss": 0.5794,
      "step": 12227
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.384414306004821,
      "learning_rate": 4.507780130333009e-06,
      "loss": 0.6091,
      "step": 12228
    },
    {
      "epoch": 1.09,
      "grad_norm": 7.024731976827323,
      "learning_rate": 4.507061273052392e-06,
      "loss": 0.6219,
      "step": 12229
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.151495166009182,
      "learning_rate": 4.506342426060792e-06,
      "loss": 0.599,
      "step": 12230
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.806505085302793,
      "learning_rate": 4.505623589373215e-06,
      "loss": 0.6584,
      "step": 12231
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.139605767067767,
      "learning_rate": 4.504904763004661e-06,
      "loss": 0.6684,
      "step": 12232
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.766903352706043,
      "learning_rate": 4.504185946970135e-06,
      "loss": 0.6728,
      "step": 12233
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.05974002230052,
      "learning_rate": 4.5034671412846415e-06,
      "loss": 0.63,
      "step": 12234
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.268483846264399,
      "learning_rate": 4.502748345963185e-06,
      "loss": 0.6921,
      "step": 12235
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.5862168821144325,
      "learning_rate": 4.502029561020767e-06,
      "loss": 0.6072,
      "step": 12236
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.738767589100811,
      "learning_rate": 4.5013107864723915e-06,
      "loss": 0.5942,
      "step": 12237
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.420906089489877,
      "learning_rate": 4.5005920223330625e-06,
      "loss": 0.7222,
      "step": 12238
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.600011065174285,
      "learning_rate": 4.499873268617779e-06,
      "loss": 0.6387,
      "step": 12239
    },
    {
      "epoch": 1.09,
      "grad_norm": 7.421544454901337,
      "learning_rate": 4.499154525341547e-06,
      "loss": 0.6238,
      "step": 12240
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.494780592092498,
      "learning_rate": 4.498435792519369e-06,
      "loss": 0.621,
      "step": 12241
    },
    {
      "epoch": 1.09,
      "grad_norm": 8.79119495428578,
      "learning_rate": 4.497717070166243e-06,
      "loss": 0.6364,
      "step": 12242
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.021538092301,
      "learning_rate": 4.4969983582971735e-06,
      "loss": 0.611,
      "step": 12243
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.629380610857261,
      "learning_rate": 4.4962796569271635e-06,
      "loss": 0.5972,
      "step": 12244
    },
    {
      "epoch": 1.09,
      "grad_norm": 7.912074580524507,
      "learning_rate": 4.495560966071213e-06,
      "loss": 0.5961,
      "step": 12245
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.047839209518593,
      "learning_rate": 4.494842285744321e-06,
      "loss": 0.7044,
      "step": 12246
    },
    {
      "epoch": 1.09,
      "grad_norm": 8.166387592508032,
      "learning_rate": 4.494123615961489e-06,
      "loss": 0.635,
      "step": 12247
    },
    {
      "epoch": 1.09,
      "grad_norm": 7.2202462897732005,
      "learning_rate": 4.493404956737721e-06,
      "loss": 0.6449,
      "step": 12248
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.524004144374282,
      "learning_rate": 4.492686308088013e-06,
      "loss": 0.568,
      "step": 12249
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.810940822898996,
      "learning_rate": 4.491967670027367e-06,
      "loss": 0.5877,
      "step": 12250
    },
    {
      "epoch": 1.09,
      "grad_norm": 4.959927580661747,
      "learning_rate": 4.491249042570785e-06,
      "loss": 0.6984,
      "step": 12251
    },
    {
      "epoch": 1.09,
      "grad_norm": 7.7442878349345206,
      "learning_rate": 4.490530425733265e-06,
      "loss": 0.5569,
      "step": 12252
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.034256743866717,
      "learning_rate": 4.489811819529805e-06,
      "loss": 0.6391,
      "step": 12253
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.3128632710510955,
      "learning_rate": 4.489093223975408e-06,
      "loss": 0.6075,
      "step": 12254
    },
    {
      "epoch": 1.09,
      "grad_norm": 4.680369249550314,
      "learning_rate": 4.488374639085069e-06,
      "loss": 0.6064,
      "step": 12255
    },
    {
      "epoch": 1.09,
      "grad_norm": 7.490410664782851,
      "learning_rate": 4.487656064873791e-06,
      "loss": 0.5862,
      "step": 12256
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.408086465010091,
      "learning_rate": 4.48693750135657e-06,
      "loss": 0.6774,
      "step": 12257
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.584965074450606,
      "learning_rate": 4.486218948548405e-06,
      "loss": 0.6759,
      "step": 12258
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.403060273272091,
      "learning_rate": 4.485500406464293e-06,
      "loss": 0.6782,
      "step": 12259
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.5644455433007565,
      "learning_rate": 4.484781875119237e-06,
      "loss": 0.5714,
      "step": 12260
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.897922949233469,
      "learning_rate": 4.484063354528228e-06,
      "loss": 0.5999,
      "step": 12261
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.868035626257705,
      "learning_rate": 4.4833448447062675e-06,
      "loss": 0.6042,
      "step": 12262
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.821417914057071,
      "learning_rate": 4.482626345668351e-06,
      "loss": 0.5426,
      "step": 12263
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.619706956751019,
      "learning_rate": 4.481907857429478e-06,
      "loss": 0.5775,
      "step": 12264
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.1393495085008825,
      "learning_rate": 4.481189380004642e-06,
      "loss": 0.6139,
      "step": 12265
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.726604928409019,
      "learning_rate": 4.480470913408842e-06,
      "loss": 0.6257,
      "step": 12266
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.34945879500513,
      "learning_rate": 4.479752457657075e-06,
      "loss": 0.5885,
      "step": 12267
    },
    {
      "epoch": 1.09,
      "grad_norm": 7.9158874829621215,
      "learning_rate": 4.479034012764335e-06,
      "loss": 0.6419,
      "step": 12268
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.872906185239631,
      "learning_rate": 4.47831557874562e-06,
      "loss": 0.549,
      "step": 12269
    },
    {
      "epoch": 1.09,
      "grad_norm": 7.166174919221646,
      "learning_rate": 4.477597155615924e-06,
      "loss": 0.5727,
      "step": 12270
    },
    {
      "epoch": 1.09,
      "grad_norm": 7.115910284592625,
      "learning_rate": 4.476878743390244e-06,
      "loss": 0.5946,
      "step": 12271
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.808928864151515,
      "learning_rate": 4.476160342083573e-06,
      "loss": 0.5792,
      "step": 12272
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.53654568764681,
      "learning_rate": 4.47544195171091e-06,
      "loss": 0.6614,
      "step": 12273
    },
    {
      "epoch": 1.09,
      "grad_norm": 7.959849949279879,
      "learning_rate": 4.4747235722872484e-06,
      "loss": 0.5878,
      "step": 12274
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.70533665586016,
      "learning_rate": 4.474005203827579e-06,
      "loss": 0.5805,
      "step": 12275
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.305688319614058,
      "learning_rate": 4.473286846346901e-06,
      "loss": 0.6237,
      "step": 12276
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.6664423211525685,
      "learning_rate": 4.4725684998602055e-06,
      "loss": 0.6477,
      "step": 12277
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.927597892470878,
      "learning_rate": 4.4718501643824874e-06,
      "loss": 0.6534,
      "step": 12278
    },
    {
      "epoch": 1.1,
      "grad_norm": 8.052070157125787,
      "learning_rate": 4.47113183992874e-06,
      "loss": 0.6414,
      "step": 12279
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.563652840711037,
      "learning_rate": 4.470413526513959e-06,
      "loss": 0.64,
      "step": 12280
    },
    {
      "epoch": 1.1,
      "grad_norm": 8.074049630588965,
      "learning_rate": 4.4696952241531344e-06,
      "loss": 0.6037,
      "step": 12281
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.896781707317337,
      "learning_rate": 4.46897693286126e-06,
      "loss": 0.5893,
      "step": 12282
    },
    {
      "epoch": 1.1,
      "grad_norm": 8.637428202503258,
      "learning_rate": 4.468258652653331e-06,
      "loss": 0.564,
      "step": 12283
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.727663766201061,
      "learning_rate": 4.467540383544338e-06,
      "loss": 0.5789,
      "step": 12284
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.484280948430802,
      "learning_rate": 4.466822125549273e-06,
      "loss": 0.5875,
      "step": 12285
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.571246326769667,
      "learning_rate": 4.466103878683128e-06,
      "loss": 0.7122,
      "step": 12286
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.293334115070593,
      "learning_rate": 4.465385642960895e-06,
      "loss": 0.6711,
      "step": 12287
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.651989369490887,
      "learning_rate": 4.464667418397568e-06,
      "loss": 0.6492,
      "step": 12288
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.403644646325852,
      "learning_rate": 4.463949205008134e-06,
      "loss": 0.6342,
      "step": 12289
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.080452113057424,
      "learning_rate": 4.463231002807589e-06,
      "loss": 0.5927,
      "step": 12290
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.5355422925202005,
      "learning_rate": 4.462512811810919e-06,
      "loss": 0.6339,
      "step": 12291
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.411445782785753,
      "learning_rate": 4.461794632033118e-06,
      "loss": 0.6392,
      "step": 12292
    },
    {
      "epoch": 1.1,
      "grad_norm": 11.073622358318127,
      "learning_rate": 4.4610764634891745e-06,
      "loss": 0.6507,
      "step": 12293
    },
    {
      "epoch": 1.1,
      "grad_norm": 8.278814292535646,
      "learning_rate": 4.46035830619408e-06,
      "loss": 0.6139,
      "step": 12294
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.547537672072074,
      "learning_rate": 4.459640160162825e-06,
      "loss": 0.5857,
      "step": 12295
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.208388523061698,
      "learning_rate": 4.458922025410396e-06,
      "loss": 0.6332,
      "step": 12296
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.750194341309088,
      "learning_rate": 4.458203901951787e-06,
      "loss": 0.6327,
      "step": 12297
    },
    {
      "epoch": 1.1,
      "grad_norm": 8.558155369040444,
      "learning_rate": 4.4574857898019845e-06,
      "loss": 0.6275,
      "step": 12298
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.65942440733257,
      "learning_rate": 4.456767688975977e-06,
      "loss": 0.6796,
      "step": 12299
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.4589926030428755,
      "learning_rate": 4.456049599488755e-06,
      "loss": 0.5629,
      "step": 12300
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.51413672002122,
      "learning_rate": 4.455331521355307e-06,
      "loss": 0.6239,
      "step": 12301
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.115248376621213,
      "learning_rate": 4.45461345459062e-06,
      "loss": 0.6354,
      "step": 12302
    },
    {
      "epoch": 1.1,
      "grad_norm": 8.002535954580285,
      "learning_rate": 4.4538953992096845e-06,
      "loss": 0.6429,
      "step": 12303
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.431163707315679,
      "learning_rate": 4.453177355227485e-06,
      "loss": 0.6138,
      "step": 12304
    },
    {
      "epoch": 1.1,
      "grad_norm": 4.91134669728509,
      "learning_rate": 4.452459322659011e-06,
      "loss": 0.576,
      "step": 12305
    },
    {
      "epoch": 1.1,
      "grad_norm": 8.351551576368372,
      "learning_rate": 4.451741301519249e-06,
      "loss": 0.5982,
      "step": 12306
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.707691539682638,
      "learning_rate": 4.4510232918231866e-06,
      "loss": 0.5871,
      "step": 12307
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.203384223131686,
      "learning_rate": 4.450305293585811e-06,
      "loss": 0.6248,
      "step": 12308
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.541717880530045,
      "learning_rate": 4.449587306822109e-06,
      "loss": 0.6363,
      "step": 12309
    },
    {
      "epoch": 1.1,
      "grad_norm": 12.296621688930657,
      "learning_rate": 4.448869331547065e-06,
      "loss": 0.6302,
      "step": 12310
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.641099279382813,
      "learning_rate": 4.448151367775669e-06,
      "loss": 0.6254,
      "step": 12311
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.952893194050032,
      "learning_rate": 4.447433415522904e-06,
      "loss": 0.6428,
      "step": 12312
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.003795634632259,
      "learning_rate": 4.446715474803756e-06,
      "loss": 0.5882,
      "step": 12313
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.040308905317815,
      "learning_rate": 4.445997545633211e-06,
      "loss": 0.5298,
      "step": 12314
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.926834470766514,
      "learning_rate": 4.445279628026254e-06,
      "loss": 0.6244,
      "step": 12315
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.559586506491983,
      "learning_rate": 4.4445617219978685e-06,
      "loss": 0.5755,
      "step": 12316
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.203405085628962,
      "learning_rate": 4.4438438275630435e-06,
      "loss": 0.6477,
      "step": 12317
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.251462698610916,
      "learning_rate": 4.443125944736759e-06,
      "loss": 0.5764,
      "step": 12318
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.348878059629113,
      "learning_rate": 4.442408073534002e-06,
      "loss": 0.6345,
      "step": 12319
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.090945574769848,
      "learning_rate": 4.441690213969753e-06,
      "loss": 0.5535,
      "step": 12320
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.202938594131278,
      "learning_rate": 4.440972366059e-06,
      "loss": 0.54,
      "step": 12321
    },
    {
      "epoch": 1.1,
      "grad_norm": 8.506620296000484,
      "learning_rate": 4.440254529816724e-06,
      "loss": 0.6361,
      "step": 12322
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.064689937246912,
      "learning_rate": 4.439536705257909e-06,
      "loss": 0.5722,
      "step": 12323
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.028254494834181,
      "learning_rate": 4.438818892397538e-06,
      "loss": 0.7365,
      "step": 12324
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.858246225220811,
      "learning_rate": 4.438101091250593e-06,
      "loss": 0.5999,
      "step": 12325
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.698644628943827,
      "learning_rate": 4.437383301832057e-06,
      "loss": 0.5651,
      "step": 12326
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.032581522567862,
      "learning_rate": 4.436665524156914e-06,
      "loss": 0.6306,
      "step": 12327
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.7791307370559,
      "learning_rate": 4.435947758240143e-06,
      "loss": 0.6821,
      "step": 12328
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.125175355200564,
      "learning_rate": 4.435230004096727e-06,
      "loss": 0.6201,
      "step": 12329
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.933177633314032,
      "learning_rate": 4.434512261741649e-06,
      "loss": 0.5902,
      "step": 12330
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.44008733821542,
      "learning_rate": 4.433794531189889e-06,
      "loss": 0.665,
      "step": 12331
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.432819785012518,
      "learning_rate": 4.43307681245643e-06,
      "loss": 0.6371,
      "step": 12332
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.584359129538197,
      "learning_rate": 4.4323591055562486e-06,
      "loss": 0.559,
      "step": 12333
    },
    {
      "epoch": 1.1,
      "grad_norm": 8.342848590895008,
      "learning_rate": 4.431641410504329e-06,
      "loss": 0.5731,
      "step": 12334
    },
    {
      "epoch": 1.1,
      "grad_norm": 4.670150368832689,
      "learning_rate": 4.430923727315649e-06,
      "loss": 0.5474,
      "step": 12335
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.614852778674106,
      "learning_rate": 4.430206056005189e-06,
      "loss": 0.6394,
      "step": 12336
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.897686349855462,
      "learning_rate": 4.429488396587933e-06,
      "loss": 0.5372,
      "step": 12337
    },
    {
      "epoch": 1.1,
      "grad_norm": 8.86987137800057,
      "learning_rate": 4.428770749078855e-06,
      "loss": 0.6493,
      "step": 12338
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.322595701834847,
      "learning_rate": 4.428053113492936e-06,
      "loss": 0.6548,
      "step": 12339
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.609767476672049,
      "learning_rate": 4.427335489845158e-06,
      "loss": 0.5989,
      "step": 12340
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.467316270168323,
      "learning_rate": 4.426617878150496e-06,
      "loss": 0.6862,
      "step": 12341
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.99362208302428,
      "learning_rate": 4.42590027842393e-06,
      "loss": 0.6091,
      "step": 12342
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.0991015502895705,
      "learning_rate": 4.4251826906804395e-06,
      "loss": 0.6264,
      "step": 12343
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.780292242503265,
      "learning_rate": 4.424465114935001e-06,
      "loss": 0.6215,
      "step": 12344
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.621525573859071,
      "learning_rate": 4.4237475512025915e-06,
      "loss": 0.628,
      "step": 12345
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.303983320631144,
      "learning_rate": 4.4230299994981925e-06,
      "loss": 0.5582,
      "step": 12346
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.623996830633488,
      "learning_rate": 4.422312459836778e-06,
      "loss": 0.6984,
      "step": 12347
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.654012700535251,
      "learning_rate": 4.421594932233325e-06,
      "loss": 0.5855,
      "step": 12348
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.108672551249345,
      "learning_rate": 4.420877416702811e-06,
      "loss": 0.5485,
      "step": 12349
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.988541788889543,
      "learning_rate": 4.420159913260214e-06,
      "loss": 0.6496,
      "step": 12350
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.0855714822936555,
      "learning_rate": 4.419442421920507e-06,
      "loss": 0.5646,
      "step": 12351
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.528477028728079,
      "learning_rate": 4.418724942698668e-06,
      "loss": 0.6512,
      "step": 12352
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.311732985940998,
      "learning_rate": 4.418007475609675e-06,
      "loss": 0.598,
      "step": 12353
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.129219990402013,
      "learning_rate": 4.4172900206685e-06,
      "loss": 0.6463,
      "step": 12354
    },
    {
      "epoch": 1.1,
      "grad_norm": 4.823236676426535,
      "learning_rate": 4.416572577890119e-06,
      "loss": 0.5717,
      "step": 12355
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.521081307296095,
      "learning_rate": 4.4158551472895095e-06,
      "loss": 0.5318,
      "step": 12356
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.5726882722869755,
      "learning_rate": 4.4151377288816435e-06,
      "loss": 0.5633,
      "step": 12357
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.061898299965168,
      "learning_rate": 4.4144203226814976e-06,
      "loss": 0.5421,
      "step": 12358
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.208486041286212,
      "learning_rate": 4.413702928704043e-06,
      "loss": 0.6334,
      "step": 12359
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.422023332030965,
      "learning_rate": 4.412985546964257e-06,
      "loss": 0.6087,
      "step": 12360
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.220483077826912,
      "learning_rate": 4.412268177477114e-06,
      "loss": 0.621,
      "step": 12361
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.993377165106248,
      "learning_rate": 4.411550820257584e-06,
      "loss": 0.6033,
      "step": 12362
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.37843666510677,
      "learning_rate": 4.4108334753206415e-06,
      "loss": 0.6109,
      "step": 12363
    },
    {
      "epoch": 1.1,
      "grad_norm": 3.881749816634972,
      "learning_rate": 4.410116142681261e-06,
      "loss": 0.5963,
      "step": 12364
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.188729966087592,
      "learning_rate": 4.409398822354413e-06,
      "loss": 0.599,
      "step": 12365
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.949533545731097,
      "learning_rate": 4.4086815143550714e-06,
      "loss": 0.6371,
      "step": 12366
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.230467369545347,
      "learning_rate": 4.407964218698209e-06,
      "loss": 0.4971,
      "step": 12367
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.159056186811219,
      "learning_rate": 4.407246935398797e-06,
      "loss": 0.6419,
      "step": 12368
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.746281725943701,
      "learning_rate": 4.406529664471806e-06,
      "loss": 0.6536,
      "step": 12369
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.586043087650528,
      "learning_rate": 4.405812405932208e-06,
      "loss": 0.6357,
      "step": 12370
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.943258165223436,
      "learning_rate": 4.405095159794977e-06,
      "loss": 0.5849,
      "step": 12371
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.202479419623973,
      "learning_rate": 4.4043779260750805e-06,
      "loss": 0.5889,
      "step": 12372
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.5472318225488895,
      "learning_rate": 4.40366070478749e-06,
      "loss": 0.5765,
      "step": 12373
    },
    {
      "epoch": 1.1,
      "grad_norm": 8.340520822779478,
      "learning_rate": 4.402943495947177e-06,
      "loss": 0.6624,
      "step": 12374
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.486992070680566,
      "learning_rate": 4.402226299569112e-06,
      "loss": 0.6118,
      "step": 12375
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.155550667044679,
      "learning_rate": 4.4015091156682635e-06,
      "loss": 0.651,
      "step": 12376
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.8859233471665275,
      "learning_rate": 4.4007919442596e-06,
      "loss": 0.6039,
      "step": 12377
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.083404134234437,
      "learning_rate": 4.400074785358094e-06,
      "loss": 0.5844,
      "step": 12378
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.045470689755019,
      "learning_rate": 4.399357638978712e-06,
      "loss": 0.6164,
      "step": 12379
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.434743719673973,
      "learning_rate": 4.3986405051364235e-06,
      "loss": 0.6193,
      "step": 12380
    },
    {
      "epoch": 1.1,
      "grad_norm": 4.4099929461144765,
      "learning_rate": 4.397923383846199e-06,
      "loss": 0.6019,
      "step": 12381
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.41881763434502,
      "learning_rate": 4.3972062751230044e-06,
      "loss": 0.7064,
      "step": 12382
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.412402494977789,
      "learning_rate": 4.396489178981809e-06,
      "loss": 0.5931,
      "step": 12383
    },
    {
      "epoch": 1.1,
      "grad_norm": 4.824250187414263,
      "learning_rate": 4.39577209543758e-06,
      "loss": 0.6527,
      "step": 12384
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.57780432670521,
      "learning_rate": 4.395055024505286e-06,
      "loss": 0.6619,
      "step": 12385
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.19687217878122,
      "learning_rate": 4.394337966199892e-06,
      "loss": 0.5803,
      "step": 12386
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.707706656635849,
      "learning_rate": 4.393620920536369e-06,
      "loss": 0.5975,
      "step": 12387
    },
    {
      "epoch": 1.11,
      "grad_norm": 4.818186518169836,
      "learning_rate": 4.392903887529679e-06,
      "loss": 0.649,
      "step": 12388
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.7079894210161815,
      "learning_rate": 4.392186867194791e-06,
      "loss": 0.6001,
      "step": 12389
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.727353704296412,
      "learning_rate": 4.391469859546674e-06,
      "loss": 0.6562,
      "step": 12390
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.428996348307807,
      "learning_rate": 4.3907528646002885e-06,
      "loss": 0.5763,
      "step": 12391
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.242836744676603,
      "learning_rate": 4.390035882370603e-06,
      "loss": 0.6677,
      "step": 12392
    },
    {
      "epoch": 1.11,
      "grad_norm": 8.007000675606493,
      "learning_rate": 4.3893189128725825e-06,
      "loss": 0.6115,
      "step": 12393
    },
    {
      "epoch": 1.11,
      "grad_norm": 9.614788131239115,
      "learning_rate": 4.388601956121193e-06,
      "loss": 0.5954,
      "step": 12394
    },
    {
      "epoch": 1.11,
      "grad_norm": 7.123306355801591,
      "learning_rate": 4.387885012131398e-06,
      "loss": 0.6433,
      "step": 12395
    },
    {
      "epoch": 1.11,
      "grad_norm": 7.057619506010608,
      "learning_rate": 4.387168080918163e-06,
      "loss": 0.5624,
      "step": 12396
    },
    {
      "epoch": 1.11,
      "grad_norm": 8.43527977015133,
      "learning_rate": 4.386451162496453e-06,
      "loss": 0.5732,
      "step": 12397
    },
    {
      "epoch": 1.11,
      "grad_norm": 7.206583386841495,
      "learning_rate": 4.38573425688123e-06,
      "loss": 0.6484,
      "step": 12398
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.66446825977004,
      "learning_rate": 4.38501736408746e-06,
      "loss": 0.6717,
      "step": 12399
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.37790114849956,
      "learning_rate": 4.384300484130105e-06,
      "loss": 0.6482,
      "step": 12400
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.8611866011087494,
      "learning_rate": 4.3835836170241294e-06,
      "loss": 0.6369,
      "step": 12401
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.37478310784572,
      "learning_rate": 4.382866762784496e-06,
      "loss": 0.6997,
      "step": 12402
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.757619642759219,
      "learning_rate": 4.382149921426168e-06,
      "loss": 0.6179,
      "step": 12403
    },
    {
      "epoch": 1.11,
      "grad_norm": 8.101600946205096,
      "learning_rate": 4.3814330929641084e-06,
      "loss": 0.6682,
      "step": 12404
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.777036555503142,
      "learning_rate": 4.380716277413277e-06,
      "loss": 0.6391,
      "step": 12405
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.564618207466893,
      "learning_rate": 4.379999474788636e-06,
      "loss": 0.6304,
      "step": 12406
    },
    {
      "epoch": 1.11,
      "grad_norm": 8.084343800579402,
      "learning_rate": 4.37928268510515e-06,
      "loss": 0.6068,
      "step": 12407
    },
    {
      "epoch": 1.11,
      "grad_norm": 7.34911926123663,
      "learning_rate": 4.378565908377777e-06,
      "loss": 0.5924,
      "step": 12408
    },
    {
      "epoch": 1.11,
      "grad_norm": 7.402555656962099,
      "learning_rate": 4.37784914462148e-06,
      "loss": 0.5843,
      "step": 12409
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.153507789077746,
      "learning_rate": 4.377132393851221e-06,
      "loss": 0.5969,
      "step": 12410
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.803080181587068,
      "learning_rate": 4.376415656081957e-06,
      "loss": 0.5766,
      "step": 12411
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.611108427822724,
      "learning_rate": 4.375698931328652e-06,
      "loss": 0.6481,
      "step": 12412
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.377720683296471,
      "learning_rate": 4.374982219606265e-06,
      "loss": 0.6285,
      "step": 12413
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.566584130384131,
      "learning_rate": 4.374265520929754e-06,
      "loss": 0.6397,
      "step": 12414
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.2380933210489475,
      "learning_rate": 4.373548835314081e-06,
      "loss": 0.5738,
      "step": 12415
    },
    {
      "epoch": 1.11,
      "grad_norm": 7.227957805225629,
      "learning_rate": 4.3728321627742045e-06,
      "loss": 0.6052,
      "step": 12416
    },
    {
      "epoch": 1.11,
      "grad_norm": 7.580587815743619,
      "learning_rate": 4.372115503325082e-06,
      "loss": 0.6362,
      "step": 12417
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.018137962271837,
      "learning_rate": 4.371398856981676e-06,
      "loss": 0.6477,
      "step": 12418
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.302042772310123,
      "learning_rate": 4.370682223758939e-06,
      "loss": 0.6299,
      "step": 12419
    },
    {
      "epoch": 1.11,
      "grad_norm": 4.903882409075015,
      "learning_rate": 4.369965603671835e-06,
      "loss": 0.6136,
      "step": 12420
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.026947219538958,
      "learning_rate": 4.369248996735317e-06,
      "loss": 0.5363,
      "step": 12421
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.394808748258217,
      "learning_rate": 4.368532402964345e-06,
      "loss": 0.6638,
      "step": 12422
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.026583006770837,
      "learning_rate": 4.367815822373878e-06,
      "loss": 0.624,
      "step": 12423
    },
    {
      "epoch": 1.11,
      "grad_norm": 9.682190355272507,
      "learning_rate": 4.3670992549788695e-06,
      "loss": 0.687,
      "step": 12424
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.753024838589873,
      "learning_rate": 4.366382700794278e-06,
      "loss": 0.6196,
      "step": 12425
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.606800317503926,
      "learning_rate": 4.365666159835061e-06,
      "loss": 0.6121,
      "step": 12426
    },
    {
      "epoch": 1.11,
      "grad_norm": 7.740747846959924,
      "learning_rate": 4.364949632116173e-06,
      "loss": 0.6053,
      "step": 12427
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.365163939825341,
      "learning_rate": 4.364233117652571e-06,
      "loss": 0.5781,
      "step": 12428
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.1127559354564385,
      "learning_rate": 4.36351661645921e-06,
      "loss": 0.6422,
      "step": 12429
    },
    {
      "epoch": 1.11,
      "grad_norm": 7.398025596576054,
      "learning_rate": 4.362800128551047e-06,
      "loss": 0.68,
      "step": 12430
    },
    {
      "epoch": 1.11,
      "grad_norm": 4.688255607611663,
      "learning_rate": 4.362083653943035e-06,
      "loss": 0.6423,
      "step": 12431
    },
    {
      "epoch": 1.11,
      "grad_norm": 4.618062843465149,
      "learning_rate": 4.36136719265013e-06,
      "loss": 0.5814,
      "step": 12432
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.3580771582789035,
      "learning_rate": 4.360650744687287e-06,
      "loss": 0.6268,
      "step": 12433
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.873281376717463,
      "learning_rate": 4.359934310069459e-06,
      "loss": 0.6502,
      "step": 12434
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.678365900411953,
      "learning_rate": 4.359217888811599e-06,
      "loss": 0.5564,
      "step": 12435
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.526321240589909,
      "learning_rate": 4.358501480928664e-06,
      "loss": 0.5636,
      "step": 12436
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.848404341323435,
      "learning_rate": 4.357785086435606e-06,
      "loss": 0.6099,
      "step": 12437
    },
    {
      "epoch": 1.11,
      "grad_norm": 7.6531470434488895,
      "learning_rate": 4.357068705347376e-06,
      "loss": 0.6007,
      "step": 12438
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.423178350183519,
      "learning_rate": 4.3563523376789305e-06,
      "loss": 0.5997,
      "step": 12439
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.975362659270571,
      "learning_rate": 4.355635983445219e-06,
      "loss": 0.5793,
      "step": 12440
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.992931754902949,
      "learning_rate": 4.354919642661196e-06,
      "loss": 0.6186,
      "step": 12441
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.944494568500657,
      "learning_rate": 4.3542033153418125e-06,
      "loss": 0.6264,
      "step": 12442
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.645468188325184,
      "learning_rate": 4.353487001502021e-06,
      "loss": 0.5628,
      "step": 12443
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.332459353316634,
      "learning_rate": 4.352770701156773e-06,
      "loss": 0.5967,
      "step": 12444
    },
    {
      "epoch": 1.11,
      "grad_norm": 4.857239640242378,
      "learning_rate": 4.352054414321017e-06,
      "loss": 0.5704,
      "step": 12445
    },
    {
      "epoch": 1.11,
      "grad_norm": 9.711345877767736,
      "learning_rate": 4.351338141009707e-06,
      "loss": 0.5747,
      "step": 12446
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.534222193789139,
      "learning_rate": 4.350621881237796e-06,
      "loss": 0.5915,
      "step": 12447
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.63022087233751,
      "learning_rate": 4.349905635020228e-06,
      "loss": 0.6257,
      "step": 12448
    },
    {
      "epoch": 1.11,
      "grad_norm": 7.648655366447596,
      "learning_rate": 4.349189402371957e-06,
      "loss": 0.5991,
      "step": 12449
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.947573647739892,
      "learning_rate": 4.348473183307931e-06,
      "loss": 0.6269,
      "step": 12450
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.272823555543137,
      "learning_rate": 4.3477569778431e-06,
      "loss": 0.6076,
      "step": 12451
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.5522211252028,
      "learning_rate": 4.347040785992416e-06,
      "loss": 0.5709,
      "step": 12452
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.697756422723997,
      "learning_rate": 4.346324607770824e-06,
      "loss": 0.5811,
      "step": 12453
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.7018061148913155,
      "learning_rate": 4.345608443193276e-06,
      "loss": 0.6291,
      "step": 12454
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.102227921239702,
      "learning_rate": 4.344892292274717e-06,
      "loss": 0.6709,
      "step": 12455
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.276516018848707,
      "learning_rate": 4.344176155030097e-06,
      "loss": 0.5638,
      "step": 12456
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.492320153876466,
      "learning_rate": 4.343460031474363e-06,
      "loss": 0.5303,
      "step": 12457
    },
    {
      "epoch": 1.11,
      "grad_norm": 9.0718688600014,
      "learning_rate": 4.342743921622464e-06,
      "loss": 0.6302,
      "step": 12458
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.261531803497046,
      "learning_rate": 4.342027825489346e-06,
      "loss": 0.6338,
      "step": 12459
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.205018183694389,
      "learning_rate": 4.341311743089956e-06,
      "loss": 0.6444,
      "step": 12460
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.501725863083191,
      "learning_rate": 4.3405956744392406e-06,
      "loss": 0.5892,
      "step": 12461
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.292504809848594,
      "learning_rate": 4.339879619552149e-06,
      "loss": 0.5514,
      "step": 12462
    },
    {
      "epoch": 1.11,
      "grad_norm": 10.19470357076131,
      "learning_rate": 4.339163578443621e-06,
      "loss": 0.6051,
      "step": 12463
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.284857003072968,
      "learning_rate": 4.338447551128609e-06,
      "loss": 0.6008,
      "step": 12464
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.26584153759166,
      "learning_rate": 4.337731537622054e-06,
      "loss": 0.5329,
      "step": 12465
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.015777222956886,
      "learning_rate": 4.337015537938902e-06,
      "loss": 0.6349,
      "step": 12466
    },
    {
      "epoch": 1.11,
      "grad_norm": 4.080442623891034,
      "learning_rate": 4.336299552094101e-06,
      "loss": 0.6579,
      "step": 12467
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.120084567398552,
      "learning_rate": 4.3355835801025916e-06,
      "loss": 0.5706,
      "step": 12468
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.29673921573495,
      "learning_rate": 4.334867621979319e-06,
      "loss": 0.5807,
      "step": 12469
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.904192019295189,
      "learning_rate": 4.334151677739231e-06,
      "loss": 0.642,
      "step": 12470
    },
    {
      "epoch": 1.11,
      "grad_norm": 8.64445017362723,
      "learning_rate": 4.333435747397267e-06,
      "loss": 0.6387,
      "step": 12471
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.899354577504616,
      "learning_rate": 4.3327198309683714e-06,
      "loss": 0.6397,
      "step": 12472
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.6337836643974395,
      "learning_rate": 4.332003928467489e-06,
      "loss": 0.5508,
      "step": 12473
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.620037102957254,
      "learning_rate": 4.331288039909562e-06,
      "loss": 0.5864,
      "step": 12474
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.8953244871441175,
      "learning_rate": 4.3305721653095315e-06,
      "loss": 0.6538,
      "step": 12475
    },
    {
      "epoch": 1.11,
      "grad_norm": 7.687014403385439,
      "learning_rate": 4.329856304682344e-06,
      "loss": 0.5875,
      "step": 12476
    },
    {
      "epoch": 1.11,
      "grad_norm": 10.512247297688809,
      "learning_rate": 4.329140458042937e-06,
      "loss": 0.6375,
      "step": 12477
    },
    {
      "epoch": 1.11,
      "grad_norm": 4.771293271935824,
      "learning_rate": 4.328424625406252e-06,
      "loss": 0.5864,
      "step": 12478
    },
    {
      "epoch": 1.11,
      "grad_norm": 8.652545428663629,
      "learning_rate": 4.327708806787233e-06,
      "loss": 0.6564,
      "step": 12479
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.479175633557988,
      "learning_rate": 4.326993002200821e-06,
      "loss": 0.6717,
      "step": 12480
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.666530378587797,
      "learning_rate": 4.326277211661955e-06,
      "loss": 0.5923,
      "step": 12481
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.800401294752259,
      "learning_rate": 4.325561435185578e-06,
      "loss": 0.5945,
      "step": 12482
    },
    {
      "epoch": 1.11,
      "grad_norm": 7.282074071767701,
      "learning_rate": 4.3248456727866295e-06,
      "loss": 0.5863,
      "step": 12483
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.28403227202662,
      "learning_rate": 4.324129924480048e-06,
      "loss": 0.6617,
      "step": 12484
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.014911094259924,
      "learning_rate": 4.323414190280774e-06,
      "loss": 0.6725,
      "step": 12485
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.804543694542524,
      "learning_rate": 4.322698470203748e-06,
      "loss": 0.6394,
      "step": 12486
    },
    {
      "epoch": 1.11,
      "grad_norm": 7.246969209843704,
      "learning_rate": 4.3219827642639075e-06,
      "loss": 0.6243,
      "step": 12487
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.623589438042041,
      "learning_rate": 4.321267072476191e-06,
      "loss": 0.5928,
      "step": 12488
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.839547263816092,
      "learning_rate": 4.32055139485554e-06,
      "loss": 0.6166,
      "step": 12489
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.038959113012882,
      "learning_rate": 4.319835731416892e-06,
      "loss": 0.6802,
      "step": 12490
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.087174651842763,
      "learning_rate": 4.3191200821751805e-06,
      "loss": 0.6025,
      "step": 12491
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.8796404001446225,
      "learning_rate": 4.318404447145347e-06,
      "loss": 0.6087,
      "step": 12492
    },
    {
      "epoch": 1.11,
      "grad_norm": 7.416923316230232,
      "learning_rate": 4.3176888263423285e-06,
      "loss": 0.5934,
      "step": 12493
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.88999139999343,
      "learning_rate": 4.316973219781061e-06,
      "loss": 0.582,
      "step": 12494
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.632695198496162,
      "learning_rate": 4.316257627476481e-06,
      "loss": 0.6619,
      "step": 12495
    },
    {
      "epoch": 1.11,
      "grad_norm": 4.596205656113535,
      "learning_rate": 4.315542049443527e-06,
      "loss": 0.6781,
      "step": 12496
    },
    {
      "epoch": 1.11,
      "grad_norm": 9.91531155214133,
      "learning_rate": 4.314826485697133e-06,
      "loss": 0.6264,
      "step": 12497
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.933905512762475,
      "learning_rate": 4.314110936252235e-06,
      "loss": 0.5342,
      "step": 12498
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.7134579605899365,
      "learning_rate": 4.313395401123771e-06,
      "loss": 0.6118,
      "step": 12499
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.441317437712676,
      "learning_rate": 4.312679880326672e-06,
      "loss": 0.6088,
      "step": 12500
    },
    {
      "epoch": 1.12,
      "grad_norm": 7.247213545953423,
      "learning_rate": 4.311964373875877e-06,
      "loss": 0.6521,
      "step": 12501
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.653324851007315,
      "learning_rate": 4.3112488817863186e-06,
      "loss": 0.5797,
      "step": 12502
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.0305300304034395,
      "learning_rate": 4.310533404072931e-06,
      "loss": 0.5913,
      "step": 12503
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.7100721957474185,
      "learning_rate": 4.309817940750649e-06,
      "loss": 0.5473,
      "step": 12504
    },
    {
      "epoch": 1.12,
      "grad_norm": 7.789056445119292,
      "learning_rate": 4.309102491834408e-06,
      "loss": 0.578,
      "step": 12505
    },
    {
      "epoch": 1.12,
      "grad_norm": 7.703240101293288,
      "learning_rate": 4.308387057339138e-06,
      "loss": 0.6053,
      "step": 12506
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.266163317888491,
      "learning_rate": 4.307671637279772e-06,
      "loss": 0.6054,
      "step": 12507
    },
    {
      "epoch": 1.12,
      "grad_norm": 8.808345907961963,
      "learning_rate": 4.306956231671245e-06,
      "loss": 0.6264,
      "step": 12508
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.5523556821321565,
      "learning_rate": 4.30624084052849e-06,
      "loss": 0.5925,
      "step": 12509
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.030779240530797,
      "learning_rate": 4.305525463866438e-06,
      "loss": 0.6844,
      "step": 12510
    },
    {
      "epoch": 1.12,
      "grad_norm": 7.265723569847724,
      "learning_rate": 4.30481010170002e-06,
      "loss": 0.5844,
      "step": 12511
    },
    {
      "epoch": 1.12,
      "grad_norm": 7.084113222295643,
      "learning_rate": 4.3040947540441705e-06,
      "loss": 0.6148,
      "step": 12512
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.0876815522526115,
      "learning_rate": 4.303379420913817e-06,
      "loss": 0.6332,
      "step": 12513
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.14428910290338,
      "learning_rate": 4.302664102323893e-06,
      "loss": 0.6116,
      "step": 12514
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.228048183349472,
      "learning_rate": 4.301948798289329e-06,
      "loss": 0.6365,
      "step": 12515
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.937987867433675,
      "learning_rate": 4.301233508825055e-06,
      "loss": 0.6187,
      "step": 12516
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.318244485067579,
      "learning_rate": 4.300518233946001e-06,
      "loss": 0.6663,
      "step": 12517
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.903630660028428,
      "learning_rate": 4.299802973667098e-06,
      "loss": 0.6623,
      "step": 12518
    },
    {
      "epoch": 1.12,
      "grad_norm": 7.511419558976613,
      "learning_rate": 4.299087728003276e-06,
      "loss": 0.5997,
      "step": 12519
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.553843127604773,
      "learning_rate": 4.29837249696946e-06,
      "loss": 0.6268,
      "step": 12520
    },
    {
      "epoch": 1.12,
      "grad_norm": 4.066676864247577,
      "learning_rate": 4.297657280580582e-06,
      "loss": 0.5872,
      "step": 12521
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.774241679291997,
      "learning_rate": 4.296942078851571e-06,
      "loss": 0.5809,
      "step": 12522
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.7297972980097285,
      "learning_rate": 4.2962268917973535e-06,
      "loss": 0.6514,
      "step": 12523
    },
    {
      "epoch": 1.12,
      "grad_norm": 7.240355376812263,
      "learning_rate": 4.295511719432858e-06,
      "loss": 0.7017,
      "step": 12524
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.731004362429373,
      "learning_rate": 4.294796561773014e-06,
      "loss": 0.6514,
      "step": 12525
    },
    {
      "epoch": 1.12,
      "grad_norm": 8.615624176410025,
      "learning_rate": 4.294081418832746e-06,
      "loss": 0.6822,
      "step": 12526
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.632605730172293,
      "learning_rate": 4.293366290626984e-06,
      "loss": 0.6165,
      "step": 12527
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.546942323678026,
      "learning_rate": 4.292651177170652e-06,
      "loss": 0.5991,
      "step": 12528
    },
    {
      "epoch": 1.12,
      "grad_norm": 4.837376174574886,
      "learning_rate": 4.291936078478678e-06,
      "loss": 0.5799,
      "step": 12529
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.4113561550944205,
      "learning_rate": 4.2912209945659885e-06,
      "loss": 0.5509,
      "step": 12530
    },
    {
      "epoch": 1.12,
      "grad_norm": 7.720073173635836,
      "learning_rate": 4.290505925447507e-06,
      "loss": 0.5805,
      "step": 12531
    },
    {
      "epoch": 1.12,
      "grad_norm": 7.073765502608775,
      "learning_rate": 4.289790871138161e-06,
      "loss": 0.5361,
      "step": 12532
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.698559074096666,
      "learning_rate": 4.2890758316528755e-06,
      "loss": 0.6139,
      "step": 12533
    },
    {
      "epoch": 1.12,
      "grad_norm": 7.92010389570392,
      "learning_rate": 4.2883608070065765e-06,
      "loss": 0.6643,
      "step": 12534
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.360420409190641,
      "learning_rate": 4.287645797214186e-06,
      "loss": 0.6453,
      "step": 12535
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.92851103903131,
      "learning_rate": 4.286930802290629e-06,
      "loss": 0.6266,
      "step": 12536
    },
    {
      "epoch": 1.12,
      "grad_norm": 7.446982826261007,
      "learning_rate": 4.286215822250831e-06,
      "loss": 0.6104,
      "step": 12537
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.85985106970029,
      "learning_rate": 4.285500857109713e-06,
      "loss": 0.6176,
      "step": 12538
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.737865135053264,
      "learning_rate": 4.2847859068822004e-06,
      "loss": 0.6011,
      "step": 12539
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.0605909697910585,
      "learning_rate": 4.284070971583216e-06,
      "loss": 0.5646,
      "step": 12540
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.290337526787252,
      "learning_rate": 4.283356051227682e-06,
      "loss": 0.6131,
      "step": 12541
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.870423278533271,
      "learning_rate": 4.282641145830521e-06,
      "loss": 0.6623,
      "step": 12542
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.990075902693554,
      "learning_rate": 4.281926255406655e-06,
      "loss": 0.6441,
      "step": 12543
    },
    {
      "epoch": 1.12,
      "grad_norm": 8.046875197671858,
      "learning_rate": 4.281211379971006e-06,
      "loss": 0.5887,
      "step": 12544
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.644751911767105,
      "learning_rate": 4.280496519538495e-06,
      "loss": 0.6915,
      "step": 12545
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.64651775605895,
      "learning_rate": 4.279781674124044e-06,
      "loss": 0.6326,
      "step": 12546
    },
    {
      "epoch": 1.12,
      "grad_norm": 8.458152262899448,
      "learning_rate": 4.279066843742573e-06,
      "loss": 0.6242,
      "step": 12547
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.716092399522176,
      "learning_rate": 4.2783520284090056e-06,
      "loss": 0.6278,
      "step": 12548
    },
    {
      "epoch": 1.12,
      "grad_norm": 8.495129903061246,
      "learning_rate": 4.2776372281382555e-06,
      "loss": 0.6204,
      "step": 12549
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.859250351027002,
      "learning_rate": 4.2769224429452485e-06,
      "loss": 0.6128,
      "step": 12550
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.6383455903845325,
      "learning_rate": 4.276207672844901e-06,
      "loss": 0.6076,
      "step": 12551
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.54431183311348,
      "learning_rate": 4.275492917852132e-06,
      "loss": 0.5916,
      "step": 12552
    },
    {
      "epoch": 1.12,
      "grad_norm": 7.250890108293123,
      "learning_rate": 4.274778177981863e-06,
      "loss": 0.5999,
      "step": 12553
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.805617472680836,
      "learning_rate": 4.274063453249011e-06,
      "loss": 0.6062,
      "step": 12554
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.434437259733753,
      "learning_rate": 4.273348743668494e-06,
      "loss": 0.5893,
      "step": 12555
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.916842749480253,
      "learning_rate": 4.272634049255233e-06,
      "loss": 0.5349,
      "step": 12556
    },
    {
      "epoch": 1.12,
      "grad_norm": 7.428018979257786,
      "learning_rate": 4.271919370024142e-06,
      "loss": 0.6945,
      "step": 12557
    },
    {
      "epoch": 1.12,
      "grad_norm": 8.645703099745653,
      "learning_rate": 4.271204705990138e-06,
      "loss": 0.625,
      "step": 12558
    },
    {
      "epoch": 1.12,
      "grad_norm": 7.4814715216574,
      "learning_rate": 4.270490057168142e-06,
      "loss": 0.6055,
      "step": 12559
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.940894450436005,
      "learning_rate": 4.269775423573067e-06,
      "loss": 0.5758,
      "step": 12560
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.000167840556448,
      "learning_rate": 4.269060805219831e-06,
      "loss": 0.6564,
      "step": 12561
    },
    {
      "epoch": 1.12,
      "grad_norm": 7.133457918040288,
      "learning_rate": 4.268346202123352e-06,
      "loss": 0.6399,
      "step": 12562
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.872302051587071,
      "learning_rate": 4.267631614298543e-06,
      "loss": 0.6193,
      "step": 12563
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.560978196055702,
      "learning_rate": 4.266917041760319e-06,
      "loss": 0.6135,
      "step": 12564
    },
    {
      "epoch": 1.12,
      "grad_norm": 7.385390873332481,
      "learning_rate": 4.266202484523597e-06,
      "loss": 0.5498,
      "step": 12565
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.487515154735273,
      "learning_rate": 4.265487942603291e-06,
      "loss": 0.6646,
      "step": 12566
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.663119525327174,
      "learning_rate": 4.264773416014315e-06,
      "loss": 0.6148,
      "step": 12567
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.907333322972671,
      "learning_rate": 4.264058904771584e-06,
      "loss": 0.6838,
      "step": 12568
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.62998157967311,
      "learning_rate": 4.263344408890013e-06,
      "loss": 0.5541,
      "step": 12569
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.091646347992674,
      "learning_rate": 4.262629928384512e-06,
      "loss": 0.6487,
      "step": 12570
    },
    {
      "epoch": 1.12,
      "grad_norm": 7.1232226141445265,
      "learning_rate": 4.261915463269997e-06,
      "loss": 0.66,
      "step": 12571
    },
    {
      "epoch": 1.12,
      "grad_norm": 9.716415726282506,
      "learning_rate": 4.261201013561382e-06,
      "loss": 0.6201,
      "step": 12572
    },
    {
      "epoch": 1.12,
      "grad_norm": 4.353528590735751,
      "learning_rate": 4.260486579273577e-06,
      "loss": 0.5861,
      "step": 12573
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.7599399564986875,
      "learning_rate": 4.259772160421495e-06,
      "loss": 0.6063,
      "step": 12574
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.023318942672598,
      "learning_rate": 4.259057757020048e-06,
      "loss": 0.6731,
      "step": 12575
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.20617711537705,
      "learning_rate": 4.258343369084147e-06,
      "loss": 0.5863,
      "step": 12576
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.533455501667167,
      "learning_rate": 4.257628996628707e-06,
      "loss": 0.5712,
      "step": 12577
    },
    {
      "epoch": 1.12,
      "grad_norm": 7.924636033386092,
      "learning_rate": 4.2569146396686335e-06,
      "loss": 0.5742,
      "step": 12578
    },
    {
      "epoch": 1.12,
      "grad_norm": 4.854630442719464,
      "learning_rate": 4.25620029821884e-06,
      "loss": 0.5651,
      "step": 12579
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.4648697030662605,
      "learning_rate": 4.255485972294236e-06,
      "loss": 0.5723,
      "step": 12580
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.779738673282413,
      "learning_rate": 4.254771661909732e-06,
      "loss": 0.5691,
      "step": 12581
    },
    {
      "epoch": 1.12,
      "grad_norm": 8.03151137797915,
      "learning_rate": 4.254057367080237e-06,
      "loss": 0.6242,
      "step": 12582
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.969144080385405,
      "learning_rate": 4.253343087820662e-06,
      "loss": 0.5834,
      "step": 12583
    },
    {
      "epoch": 1.12,
      "grad_norm": 4.957695694540873,
      "learning_rate": 4.252628824145913e-06,
      "loss": 0.5794,
      "step": 12584
    },
    {
      "epoch": 1.12,
      "grad_norm": 7.16738261192193,
      "learning_rate": 4.251914576070904e-06,
      "loss": 0.5688,
      "step": 12585
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.310930878727806,
      "learning_rate": 4.251200343610537e-06,
      "loss": 0.61,
      "step": 12586
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.430806467512137,
      "learning_rate": 4.2504861267797235e-06,
      "loss": 0.6435,
      "step": 12587
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.288129616329784,
      "learning_rate": 4.249771925593371e-06,
      "loss": 0.6023,
      "step": 12588
    },
    {
      "epoch": 1.12,
      "grad_norm": 7.275389170974548,
      "learning_rate": 4.249057740066387e-06,
      "loss": 0.6678,
      "step": 12589
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.59452582144278,
      "learning_rate": 4.248343570213677e-06,
      "loss": 0.5462,
      "step": 12590
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.477370382870268,
      "learning_rate": 4.247629416050152e-06,
      "loss": 0.5485,
      "step": 12591
    },
    {
      "epoch": 1.12,
      "grad_norm": 7.072146654692142,
      "learning_rate": 4.246915277590713e-06,
      "loss": 0.7186,
      "step": 12592
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.6341279317006485,
      "learning_rate": 4.246201154850268e-06,
      "loss": 0.5879,
      "step": 12593
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.031028223532872,
      "learning_rate": 4.245487047843722e-06,
      "loss": 0.6045,
      "step": 12594
    },
    {
      "epoch": 1.12,
      "grad_norm": 8.169968311930544,
      "learning_rate": 4.244772956585984e-06,
      "loss": 0.6503,
      "step": 12595
    },
    {
      "epoch": 1.12,
      "grad_norm": 4.770817414272366,
      "learning_rate": 4.2440588810919556e-06,
      "loss": 0.7011,
      "step": 12596
    },
    {
      "epoch": 1.12,
      "grad_norm": 7.420205228086168,
      "learning_rate": 4.243344821376541e-06,
      "loss": 0.5677,
      "step": 12597
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.539940078599746,
      "learning_rate": 4.242630777454649e-06,
      "loss": 0.589,
      "step": 12598
    },
    {
      "epoch": 1.12,
      "grad_norm": 4.4275348856243895,
      "learning_rate": 4.2419167493411794e-06,
      "loss": 0.5345,
      "step": 12599
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.18953761728353,
      "learning_rate": 4.2412027370510365e-06,
      "loss": 0.6245,
      "step": 12600
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.017957215767398,
      "learning_rate": 4.2404887405991264e-06,
      "loss": 0.6089,
      "step": 12601
    },
    {
      "epoch": 1.12,
      "grad_norm": 8.912039653653796,
      "learning_rate": 4.23977476000035e-06,
      "loss": 0.6128,
      "step": 12602
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.34966727286289,
      "learning_rate": 4.2390607952696106e-06,
      "loss": 0.6058,
      "step": 12603
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.8519158463911705,
      "learning_rate": 4.238346846421811e-06,
      "loss": 0.5385,
      "step": 12604
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.088551482706895,
      "learning_rate": 4.2376329134718526e-06,
      "loss": 0.5375,
      "step": 12605
    },
    {
      "epoch": 1.12,
      "grad_norm": 4.617207121207288,
      "learning_rate": 4.236918996434639e-06,
      "loss": 0.6688,
      "step": 12606
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.357094574082609,
      "learning_rate": 4.236205095325069e-06,
      "loss": 0.6014,
      "step": 12607
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.949445411706624,
      "learning_rate": 4.235491210158045e-06,
      "loss": 0.6116,
      "step": 12608
    },
    {
      "epoch": 1.12,
      "grad_norm": 4.836761575830953,
      "learning_rate": 4.234777340948468e-06,
      "loss": 0.5581,
      "step": 12609
    },
    {
      "epoch": 1.12,
      "grad_norm": 7.125425113484446,
      "learning_rate": 4.234063487711237e-06,
      "loss": 0.6425,
      "step": 12610
    },
    {
      "epoch": 1.13,
      "grad_norm": 4.9552841003218,
      "learning_rate": 4.2333496504612544e-06,
      "loss": 0.6238,
      "step": 12611
    },
    {
      "epoch": 1.13,
      "grad_norm": 7.844297311056438,
      "learning_rate": 4.232635829213419e-06,
      "loss": 0.6612,
      "step": 12612
    },
    {
      "epoch": 1.13,
      "grad_norm": 3.976919915712569,
      "learning_rate": 4.23192202398263e-06,
      "loss": 0.5286,
      "step": 12613
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.410874947924332,
      "learning_rate": 4.231208234783786e-06,
      "loss": 0.6068,
      "step": 12614
    },
    {
      "epoch": 1.13,
      "grad_norm": 7.193329231437338,
      "learning_rate": 4.230494461631786e-06,
      "loss": 0.6949,
      "step": 12615
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.46414875884019,
      "learning_rate": 4.229780704541531e-06,
      "loss": 0.6332,
      "step": 12616
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.962758084194527,
      "learning_rate": 4.229066963527915e-06,
      "loss": 0.6674,
      "step": 12617
    },
    {
      "epoch": 1.13,
      "grad_norm": 8.502094469393777,
      "learning_rate": 4.228353238605839e-06,
      "loss": 0.5961,
      "step": 12618
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.6339336469914665,
      "learning_rate": 4.227639529790199e-06,
      "loss": 0.5663,
      "step": 12619
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.819457661223192,
      "learning_rate": 4.226925837095894e-06,
      "loss": 0.561,
      "step": 12620
    },
    {
      "epoch": 1.13,
      "grad_norm": 8.076832392882768,
      "learning_rate": 4.226212160537818e-06,
      "loss": 0.623,
      "step": 12621
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.962524814103424,
      "learning_rate": 4.225498500130868e-06,
      "loss": 0.697,
      "step": 12622
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.444347984686597,
      "learning_rate": 4.2247848558899415e-06,
      "loss": 0.6445,
      "step": 12623
    },
    {
      "epoch": 1.13,
      "grad_norm": 8.875845333231622,
      "learning_rate": 4.224071227829932e-06,
      "loss": 0.5706,
      "step": 12624
    },
    {
      "epoch": 1.13,
      "grad_norm": 4.2990162082842875,
      "learning_rate": 4.223357615965737e-06,
      "loss": 0.5659,
      "step": 12625
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.1639851791217195,
      "learning_rate": 4.222644020312251e-06,
      "loss": 0.6286,
      "step": 12626
    },
    {
      "epoch": 1.13,
      "grad_norm": 7.142686118827435,
      "learning_rate": 4.221930440884369e-06,
      "loss": 0.6713,
      "step": 12627
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.606556344887824,
      "learning_rate": 4.221216877696984e-06,
      "loss": 0.6128,
      "step": 12628
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.825885758856009,
      "learning_rate": 4.220503330764993e-06,
      "loss": 0.5959,
      "step": 12629
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.919752072333492,
      "learning_rate": 4.2197898001032864e-06,
      "loss": 0.5779,
      "step": 12630
    },
    {
      "epoch": 1.13,
      "grad_norm": 8.315978496799756,
      "learning_rate": 4.21907628572676e-06,
      "loss": 0.5516,
      "step": 12631
    },
    {
      "epoch": 1.13,
      "grad_norm": 7.931908316588876,
      "learning_rate": 4.218362787650305e-06,
      "loss": 0.6139,
      "step": 12632
    },
    {
      "epoch": 1.13,
      "grad_norm": 7.807503741748099,
      "learning_rate": 4.217649305888816e-06,
      "loss": 0.558,
      "step": 12633
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.152737266495149,
      "learning_rate": 4.216935840457183e-06,
      "loss": 0.6647,
      "step": 12634
    },
    {
      "epoch": 1.13,
      "grad_norm": 4.719033698010361,
      "learning_rate": 4.2162223913703025e-06,
      "loss": 0.6129,
      "step": 12635
    },
    {
      "epoch": 1.13,
      "grad_norm": 8.21001507201182,
      "learning_rate": 4.215508958643062e-06,
      "loss": 0.625,
      "step": 12636
    },
    {
      "epoch": 1.13,
      "grad_norm": 4.799218633764392,
      "learning_rate": 4.214795542290353e-06,
      "loss": 0.5796,
      "step": 12637
    },
    {
      "epoch": 1.13,
      "grad_norm": 10.768683716493918,
      "learning_rate": 4.214082142327066e-06,
      "loss": 0.6423,
      "step": 12638
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.5463201947189305,
      "learning_rate": 4.213368758768096e-06,
      "loss": 0.5739,
      "step": 12639
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.240257711424719,
      "learning_rate": 4.212655391628329e-06,
      "loss": 0.5774,
      "step": 12640
    },
    {
      "epoch": 1.13,
      "grad_norm": 7.128317232122081,
      "learning_rate": 4.211942040922656e-06,
      "loss": 0.6164,
      "step": 12641
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.123918484759255,
      "learning_rate": 4.211228706665969e-06,
      "loss": 0.6001,
      "step": 12642
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.460897375480822,
      "learning_rate": 4.210515388873153e-06,
      "loss": 0.5799,
      "step": 12643
    },
    {
      "epoch": 1.13,
      "grad_norm": 9.797370459393301,
      "learning_rate": 4.209802087559101e-06,
      "loss": 0.6111,
      "step": 12644
    },
    {
      "epoch": 1.13,
      "grad_norm": 8.714295799633994,
      "learning_rate": 4.209088802738699e-06,
      "loss": 0.5975,
      "step": 12645
    },
    {
      "epoch": 1.13,
      "grad_norm": 4.953990790805406,
      "learning_rate": 4.208375534426837e-06,
      "loss": 0.6299,
      "step": 12646
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.711421118634388,
      "learning_rate": 4.207662282638401e-06,
      "loss": 0.5981,
      "step": 12647
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.1069776209122,
      "learning_rate": 4.20694904738828e-06,
      "loss": 0.5487,
      "step": 12648
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.166576316330994,
      "learning_rate": 4.206235828691363e-06,
      "loss": 0.5666,
      "step": 12649
    },
    {
      "epoch": 1.13,
      "grad_norm": 7.908183987889166,
      "learning_rate": 4.205522626562533e-06,
      "loss": 0.6116,
      "step": 12650
    },
    {
      "epoch": 1.13,
      "grad_norm": 7.143491140045141,
      "learning_rate": 4.204809441016677e-06,
      "loss": 0.6172,
      "step": 12651
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.271974062529674,
      "learning_rate": 4.204096272068685e-06,
      "loss": 0.5864,
      "step": 12652
    },
    {
      "epoch": 1.13,
      "grad_norm": 9.081076796014594,
      "learning_rate": 4.203383119733439e-06,
      "loss": 0.6474,
      "step": 12653
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.322175724906395,
      "learning_rate": 4.202669984025825e-06,
      "loss": 0.6181,
      "step": 12654
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.058714280042682,
      "learning_rate": 4.20195686496073e-06,
      "loss": 0.616,
      "step": 12655
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.027504728221585,
      "learning_rate": 4.201243762553038e-06,
      "loss": 0.6322,
      "step": 12656
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.575385707803813,
      "learning_rate": 4.200530676817632e-06,
      "loss": 0.6112,
      "step": 12657
    },
    {
      "epoch": 1.13,
      "grad_norm": 4.828332688309542,
      "learning_rate": 4.199817607769399e-06,
      "loss": 0.6411,
      "step": 12658
    },
    {
      "epoch": 1.13,
      "grad_norm": 4.928712119123263,
      "learning_rate": 4.199104555423219e-06,
      "loss": 0.5565,
      "step": 12659
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.726832603140673,
      "learning_rate": 4.198391519793979e-06,
      "loss": 0.6786,
      "step": 12660
    },
    {
      "epoch": 1.13,
      "grad_norm": 8.056123519710118,
      "learning_rate": 4.197678500896561e-06,
      "loss": 0.6974,
      "step": 12661
    },
    {
      "epoch": 1.13,
      "grad_norm": 7.954240246553722,
      "learning_rate": 4.196965498745845e-06,
      "loss": 0.569,
      "step": 12662
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.819523592431655,
      "learning_rate": 4.196252513356717e-06,
      "loss": 0.6152,
      "step": 12663
    },
    {
      "epoch": 1.13,
      "grad_norm": 4.7666273788002345,
      "learning_rate": 4.1955395447440585e-06,
      "loss": 0.5805,
      "step": 12664
    },
    {
      "epoch": 1.13,
      "grad_norm": 8.28439395408195,
      "learning_rate": 4.194826592922751e-06,
      "loss": 0.5406,
      "step": 12665
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.186520893297952,
      "learning_rate": 4.194113657907672e-06,
      "loss": 0.5387,
      "step": 12666
    },
    {
      "epoch": 1.13,
      "grad_norm": 4.89582197075012,
      "learning_rate": 4.193400739713707e-06,
      "loss": 0.587,
      "step": 12667
    },
    {
      "epoch": 1.13,
      "grad_norm": 8.467188976767641,
      "learning_rate": 4.192687838355736e-06,
      "loss": 0.6653,
      "step": 12668
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.037381235541991,
      "learning_rate": 4.191974953848637e-06,
      "loss": 0.5686,
      "step": 12669
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.572959212276394,
      "learning_rate": 4.191262086207291e-06,
      "loss": 0.6427,
      "step": 12670
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.210053869088005,
      "learning_rate": 4.190549235446579e-06,
      "loss": 0.5804,
      "step": 12671
    },
    {
      "epoch": 1.13,
      "grad_norm": 7.326169962240459,
      "learning_rate": 4.189836401581378e-06,
      "loss": 0.6197,
      "step": 12672
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.333692452935304,
      "learning_rate": 4.189123584626569e-06,
      "loss": 0.5907,
      "step": 12673
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.118125662132285,
      "learning_rate": 4.188410784597029e-06,
      "loss": 0.5918,
      "step": 12674
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.681988467837658,
      "learning_rate": 4.187698001507637e-06,
      "loss": 0.6622,
      "step": 12675
    },
    {
      "epoch": 1.13,
      "grad_norm": 7.465401883676141,
      "learning_rate": 4.186985235373269e-06,
      "loss": 0.6295,
      "step": 12676
    },
    {
      "epoch": 1.13,
      "grad_norm": 7.296987835785575,
      "learning_rate": 4.186272486208805e-06,
      "loss": 0.6114,
      "step": 12677
    },
    {
      "epoch": 1.13,
      "grad_norm": 7.639720527971992,
      "learning_rate": 4.185559754029122e-06,
      "loss": 0.6302,
      "step": 12678
    },
    {
      "epoch": 1.13,
      "grad_norm": 4.688682945601475,
      "learning_rate": 4.184847038849094e-06,
      "loss": 0.6027,
      "step": 12679
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.350043471093085,
      "learning_rate": 4.184134340683598e-06,
      "loss": 0.6058,
      "step": 12680
    },
    {
      "epoch": 1.13,
      "grad_norm": 7.458072245147057,
      "learning_rate": 4.183421659547512e-06,
      "loss": 0.5657,
      "step": 12681
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.5493362656049925,
      "learning_rate": 4.18270899545571e-06,
      "loss": 0.6538,
      "step": 12682
    },
    {
      "epoch": 1.13,
      "grad_norm": 7.500736731559442,
      "learning_rate": 4.181996348423068e-06,
      "loss": 0.6143,
      "step": 12683
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.2576764710640855,
      "learning_rate": 4.181283718464462e-06,
      "loss": 0.5741,
      "step": 12684
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.800755571559783,
      "learning_rate": 4.180571105594764e-06,
      "loss": 0.6002,
      "step": 12685
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.471930041886409,
      "learning_rate": 4.179858509828849e-06,
      "loss": 0.5819,
      "step": 12686
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.96610695322445,
      "learning_rate": 4.179145931181594e-06,
      "loss": 0.5855,
      "step": 12687
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.927183113306261,
      "learning_rate": 4.178433369667868e-06,
      "loss": 0.6984,
      "step": 12688
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.70317171687799,
      "learning_rate": 4.177720825302547e-06,
      "loss": 0.5914,
      "step": 12689
    },
    {
      "epoch": 1.13,
      "grad_norm": 8.236083615690184,
      "learning_rate": 4.177008298100502e-06,
      "loss": 0.6124,
      "step": 12690
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.637475939020546,
      "learning_rate": 4.176295788076608e-06,
      "loss": 0.6446,
      "step": 12691
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.770102139228452,
      "learning_rate": 4.175583295245736e-06,
      "loss": 0.6591,
      "step": 12692
    },
    {
      "epoch": 1.13,
      "grad_norm": 7.350057745896262,
      "learning_rate": 4.174870819622755e-06,
      "loss": 0.5849,
      "step": 12693
    },
    {
      "epoch": 1.13,
      "grad_norm": 4.9855297849665865,
      "learning_rate": 4.174158361222541e-06,
      "loss": 0.5413,
      "step": 12694
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.064068768018749,
      "learning_rate": 4.173445920059961e-06,
      "loss": 0.6503,
      "step": 12695
    },
    {
      "epoch": 1.13,
      "grad_norm": 8.202963995744277,
      "learning_rate": 4.172733496149887e-06,
      "loss": 0.6136,
      "step": 12696
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.306456038250808,
      "learning_rate": 4.172021089507191e-06,
      "loss": 0.5673,
      "step": 12697
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.409561043291567,
      "learning_rate": 4.171308700146741e-06,
      "loss": 0.6294,
      "step": 12698
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.697725374062372,
      "learning_rate": 4.170596328083407e-06,
      "loss": 0.6039,
      "step": 12699
    },
    {
      "epoch": 1.13,
      "grad_norm": 7.323796131847582,
      "learning_rate": 4.169883973332058e-06,
      "loss": 0.5681,
      "step": 12700
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.674420703859286,
      "learning_rate": 4.169171635907563e-06,
      "loss": 0.5468,
      "step": 12701
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.239631491299969,
      "learning_rate": 4.168459315824791e-06,
      "loss": 0.603,
      "step": 12702
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.312389100461924,
      "learning_rate": 4.16774701309861e-06,
      "loss": 0.5326,
      "step": 12703
    },
    {
      "epoch": 1.13,
      "grad_norm": 7.379880029324593,
      "learning_rate": 4.167034727743886e-06,
      "loss": 0.5908,
      "step": 12704
    },
    {
      "epoch": 1.13,
      "grad_norm": 4.996491974536102,
      "learning_rate": 4.1663224597754905e-06,
      "loss": 0.6279,
      "step": 12705
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.9410683986714545,
      "learning_rate": 4.165610209208286e-06,
      "loss": 0.5961,
      "step": 12706
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.79877507566606,
      "learning_rate": 4.1648979760571425e-06,
      "loss": 0.5601,
      "step": 12707
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.759308860300633,
      "learning_rate": 4.164185760336924e-06,
      "loss": 0.6363,
      "step": 12708
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.403143590025384,
      "learning_rate": 4.163473562062498e-06,
      "loss": 0.6306,
      "step": 12709
    },
    {
      "epoch": 1.13,
      "grad_norm": 8.088698030591823,
      "learning_rate": 4.1627613812487284e-06,
      "loss": 0.5612,
      "step": 12710
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.25837451932744,
      "learning_rate": 4.162049217910481e-06,
      "loss": 0.6393,
      "step": 12711
    },
    {
      "epoch": 1.13,
      "grad_norm": 7.355127702528974,
      "learning_rate": 4.161337072062623e-06,
      "loss": 0.6171,
      "step": 12712
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.404315763727701,
      "learning_rate": 4.160624943720015e-06,
      "loss": 0.5954,
      "step": 12713
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.757445688024219,
      "learning_rate": 4.159912832897524e-06,
      "loss": 0.599,
      "step": 12714
    },
    {
      "epoch": 1.13,
      "grad_norm": 4.5739824279960635,
      "learning_rate": 4.159200739610013e-06,
      "loss": 0.5546,
      "step": 12715
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.570251189514082,
      "learning_rate": 4.158488663872346e-06,
      "loss": 0.6777,
      "step": 12716
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.3087020230117945,
      "learning_rate": 4.157776605699383e-06,
      "loss": 0.6052,
      "step": 12717
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.516401456893279,
      "learning_rate": 4.157064565105991e-06,
      "loss": 0.587,
      "step": 12718
    },
    {
      "epoch": 1.13,
      "grad_norm": 7.967350149704901,
      "learning_rate": 4.156352542107029e-06,
      "loss": 0.6188,
      "step": 12719
    },
    {
      "epoch": 1.13,
      "grad_norm": 4.654118097714581,
      "learning_rate": 4.15564053671736e-06,
      "loss": 0.5781,
      "step": 12720
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.691401781781373,
      "learning_rate": 4.154928548951848e-06,
      "loss": 0.6757,
      "step": 12721
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.226859324028045,
      "learning_rate": 4.154216578825351e-06,
      "loss": 0.5911,
      "step": 12722
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.3251342668830315,
      "learning_rate": 4.15350462635273e-06,
      "loss": 0.5871,
      "step": 12723
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.143308841401558,
      "learning_rate": 4.1527926915488445e-06,
      "loss": 0.6205,
      "step": 12724
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.699136838272306,
      "learning_rate": 4.152080774428558e-06,
      "loss": 0.5602,
      "step": 12725
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.605763130199257,
      "learning_rate": 4.1513688750067284e-06,
      "loss": 0.6026,
      "step": 12726
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.300114079442129,
      "learning_rate": 4.150656993298214e-06,
      "loss": 0.6425,
      "step": 12727
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.842733429162357,
      "learning_rate": 4.149945129317877e-06,
      "loss": 0.5617,
      "step": 12728
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.380793756538788,
      "learning_rate": 4.149233283080571e-06,
      "loss": 0.5706,
      "step": 12729
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.028300939575589,
      "learning_rate": 4.148521454601159e-06,
      "loss": 0.6265,
      "step": 12730
    },
    {
      "epoch": 1.14,
      "grad_norm": 7.8889565086954665,
      "learning_rate": 4.147809643894497e-06,
      "loss": 0.625,
      "step": 12731
    },
    {
      "epoch": 1.14,
      "grad_norm": 4.865211723973894,
      "learning_rate": 4.1470978509754416e-06,
      "loss": 0.5891,
      "step": 12732
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.808786602730541,
      "learning_rate": 4.14638607585885e-06,
      "loss": 0.6265,
      "step": 12733
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.766070203249128,
      "learning_rate": 4.1456743185595815e-06,
      "loss": 0.5744,
      "step": 12734
    },
    {
      "epoch": 1.14,
      "grad_norm": 8.075150449958093,
      "learning_rate": 4.14496257909249e-06,
      "loss": 0.6031,
      "step": 12735
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.354861107191467,
      "learning_rate": 4.144250857472434e-06,
      "loss": 0.6466,
      "step": 12736
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.527357840624476,
      "learning_rate": 4.143539153714264e-06,
      "loss": 0.5781,
      "step": 12737
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.6616421839879285,
      "learning_rate": 4.14282746783284e-06,
      "loss": 0.598,
      "step": 12738
    },
    {
      "epoch": 1.14,
      "grad_norm": 7.333362555937387,
      "learning_rate": 4.142115799843015e-06,
      "loss": 0.6448,
      "step": 12739
    },
    {
      "epoch": 1.14,
      "grad_norm": 7.453391436371966,
      "learning_rate": 4.141404149759644e-06,
      "loss": 0.6512,
      "step": 12740
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.970562121879597,
      "learning_rate": 4.1406925175975825e-06,
      "loss": 0.6456,
      "step": 12741
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.947067527403919,
      "learning_rate": 4.139980903371681e-06,
      "loss": 0.5786,
      "step": 12742
    },
    {
      "epoch": 1.14,
      "grad_norm": 7.125958269744058,
      "learning_rate": 4.139269307096795e-06,
      "loss": 0.6059,
      "step": 12743
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.047556641680677,
      "learning_rate": 4.138557728787778e-06,
      "loss": 0.6695,
      "step": 12744
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.031019876509816,
      "learning_rate": 4.137846168459481e-06,
      "loss": 0.5728,
      "step": 12745
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.880993271217456,
      "learning_rate": 4.137134626126757e-06,
      "loss": 0.6501,
      "step": 12746
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.081469797923237,
      "learning_rate": 4.13642310180446e-06,
      "loss": 0.6063,
      "step": 12747
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.942432961520711,
      "learning_rate": 4.135711595507437e-06,
      "loss": 0.6652,
      "step": 12748
    },
    {
      "epoch": 1.14,
      "grad_norm": 8.136049596210688,
      "learning_rate": 4.135000107250543e-06,
      "loss": 0.6025,
      "step": 12749
    },
    {
      "epoch": 1.14,
      "grad_norm": 7.350976950096308,
      "learning_rate": 4.134288637048629e-06,
      "loss": 0.5891,
      "step": 12750
    },
    {
      "epoch": 1.14,
      "grad_norm": 4.758320824760471,
      "learning_rate": 4.133577184916543e-06,
      "loss": 0.6853,
      "step": 12751
    },
    {
      "epoch": 1.14,
      "grad_norm": 12.199076974291925,
      "learning_rate": 4.132865750869134e-06,
      "loss": 0.6169,
      "step": 12752
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.944342950291794,
      "learning_rate": 4.132154334921255e-06,
      "loss": 0.606,
      "step": 12753
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.931219516405805,
      "learning_rate": 4.131442937087754e-06,
      "loss": 0.63,
      "step": 12754
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.157822474561904,
      "learning_rate": 4.130731557383477e-06,
      "loss": 0.6257,
      "step": 12755
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.10263260902672,
      "learning_rate": 4.130020195823277e-06,
      "loss": 0.5934,
      "step": 12756
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.75764409861294,
      "learning_rate": 4.129308852422001e-06,
      "loss": 0.5929,
      "step": 12757
    },
    {
      "epoch": 1.14,
      "grad_norm": 11.647357216690482,
      "learning_rate": 4.128597527194495e-06,
      "loss": 0.6152,
      "step": 12758
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.543494380550673,
      "learning_rate": 4.127886220155606e-06,
      "loss": 0.591,
      "step": 12759
    },
    {
      "epoch": 1.14,
      "grad_norm": 4.539793705869565,
      "learning_rate": 4.127174931320184e-06,
      "loss": 0.6275,
      "step": 12760
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.889793883793437,
      "learning_rate": 4.126463660703073e-06,
      "loss": 0.5949,
      "step": 12761
    },
    {
      "epoch": 1.14,
      "grad_norm": 7.092178034925467,
      "learning_rate": 4.12575240831912e-06,
      "loss": 0.6176,
      "step": 12762
    },
    {
      "epoch": 1.14,
      "grad_norm": 7.670853216669609,
      "learning_rate": 4.125041174183172e-06,
      "loss": 0.577,
      "step": 12763
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.859521412804961,
      "learning_rate": 4.124329958310075e-06,
      "loss": 0.6117,
      "step": 12764
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.973185397904458,
      "learning_rate": 4.12361876071467e-06,
      "loss": 0.6102,
      "step": 12765
    },
    {
      "epoch": 1.14,
      "grad_norm": 7.288597881500008,
      "learning_rate": 4.122907581411803e-06,
      "loss": 0.5992,
      "step": 12766
    },
    {
      "epoch": 1.14,
      "grad_norm": 8.436943258155077,
      "learning_rate": 4.122196420416321e-06,
      "loss": 0.609,
      "step": 12767
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.138179883334849,
      "learning_rate": 4.121485277743066e-06,
      "loss": 0.5633,
      "step": 12768
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.128256097209497,
      "learning_rate": 4.120774153406882e-06,
      "loss": 0.5615,
      "step": 12769
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.321727757305374,
      "learning_rate": 4.120063047422612e-06,
      "loss": 0.5916,
      "step": 12770
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.592273565257728,
      "learning_rate": 4.119351959805099e-06,
      "loss": 0.5967,
      "step": 12771
    },
    {
      "epoch": 1.14,
      "grad_norm": 4.6479033584633545,
      "learning_rate": 4.118640890569185e-06,
      "loss": 0.6123,
      "step": 12772
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.095865924443309,
      "learning_rate": 4.117929839729713e-06,
      "loss": 0.5944,
      "step": 12773
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.394881126178612,
      "learning_rate": 4.117218807301522e-06,
      "loss": 0.5825,
      "step": 12774
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.832065361073542,
      "learning_rate": 4.116507793299458e-06,
      "loss": 0.5901,
      "step": 12775
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.236404395882163,
      "learning_rate": 4.115796797738357e-06,
      "loss": 0.588,
      "step": 12776
    },
    {
      "epoch": 1.14,
      "grad_norm": 4.883805925776225,
      "learning_rate": 4.115085820633061e-06,
      "loss": 0.6807,
      "step": 12777
    },
    {
      "epoch": 1.14,
      "grad_norm": 8.295645737960045,
      "learning_rate": 4.114374861998412e-06,
      "loss": 0.591,
      "step": 12778
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.339563205810006,
      "learning_rate": 4.113663921849249e-06,
      "loss": 0.6469,
      "step": 12779
    },
    {
      "epoch": 1.14,
      "grad_norm": 8.045728616981826,
      "learning_rate": 4.11295300020041e-06,
      "loss": 0.6168,
      "step": 12780
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.986106198077401,
      "learning_rate": 4.112242097066733e-06,
      "loss": 0.5533,
      "step": 12781
    },
    {
      "epoch": 1.14,
      "grad_norm": 4.616816668337794,
      "learning_rate": 4.111531212463058e-06,
      "loss": 0.6175,
      "step": 12782
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.867884790340103,
      "learning_rate": 4.110820346404226e-06,
      "loss": 0.6197,
      "step": 12783
    },
    {
      "epoch": 1.14,
      "grad_norm": 7.989078606775965,
      "learning_rate": 4.110109498905069e-06,
      "loss": 0.628,
      "step": 12784
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.447944089010971,
      "learning_rate": 4.10939866998043e-06,
      "loss": 0.648,
      "step": 12785
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.684374747059337,
      "learning_rate": 4.10868785964514e-06,
      "loss": 0.6126,
      "step": 12786
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.550044145885111,
      "learning_rate": 4.107977067914041e-06,
      "loss": 0.6205,
      "step": 12787
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.918509800947694,
      "learning_rate": 4.107266294801968e-06,
      "loss": 0.6203,
      "step": 12788
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.175138519646643,
      "learning_rate": 4.106555540323755e-06,
      "loss": 0.584,
      "step": 12789
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.244555923210068,
      "learning_rate": 4.105844804494238e-06,
      "loss": 0.6128,
      "step": 12790
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.875723140691153,
      "learning_rate": 4.1051340873282545e-06,
      "loss": 0.5932,
      "step": 12791
    },
    {
      "epoch": 1.14,
      "grad_norm": 7.6101211710140495,
      "learning_rate": 4.104423388840635e-06,
      "loss": 0.6155,
      "step": 12792
    },
    {
      "epoch": 1.14,
      "grad_norm": 7.337509126754695,
      "learning_rate": 4.103712709046219e-06,
      "loss": 0.6458,
      "step": 12793
    },
    {
      "epoch": 1.14,
      "grad_norm": 7.806206478829382,
      "learning_rate": 4.103002047959834e-06,
      "loss": 0.6489,
      "step": 12794
    },
    {
      "epoch": 1.14,
      "grad_norm": 4.123203951112772,
      "learning_rate": 4.102291405596319e-06,
      "loss": 0.6035,
      "step": 12795
    },
    {
      "epoch": 1.14,
      "grad_norm": 9.555776568653464,
      "learning_rate": 4.1015807819705036e-06,
      "loss": 0.6194,
      "step": 12796
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.700463455818202,
      "learning_rate": 4.100870177097222e-06,
      "loss": 0.6108,
      "step": 12797
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.059683023690938,
      "learning_rate": 4.1001595909913075e-06,
      "loss": 0.5479,
      "step": 12798
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.127899678297387,
      "learning_rate": 4.09944902366759e-06,
      "loss": 0.5469,
      "step": 12799
    },
    {
      "epoch": 1.14,
      "grad_norm": 7.0156331235176435,
      "learning_rate": 4.0987384751409015e-06,
      "loss": 0.6052,
      "step": 12800
    },
    {
      "epoch": 1.14,
      "grad_norm": 9.402652326663736,
      "learning_rate": 4.098027945426075e-06,
      "loss": 0.5805,
      "step": 12801
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.214074715173855,
      "learning_rate": 4.097317434537939e-06,
      "loss": 0.5724,
      "step": 12802
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.122618038910121,
      "learning_rate": 4.096606942491324e-06,
      "loss": 0.6138,
      "step": 12803
    },
    {
      "epoch": 1.14,
      "grad_norm": 7.324854168897886,
      "learning_rate": 4.095896469301062e-06,
      "loss": 0.6904,
      "step": 12804
    },
    {
      "epoch": 1.14,
      "grad_norm": 4.958021496848178,
      "learning_rate": 4.09518601498198e-06,
      "loss": 0.6138,
      "step": 12805
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.791586359127114,
      "learning_rate": 4.094475579548909e-06,
      "loss": 0.6001,
      "step": 12806
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.627803443853279,
      "learning_rate": 4.093765163016678e-06,
      "loss": 0.55,
      "step": 12807
    },
    {
      "epoch": 1.14,
      "grad_norm": 7.974814155850224,
      "learning_rate": 4.0930547654001154e-06,
      "loss": 0.5925,
      "step": 12808
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.893931761876743,
      "learning_rate": 4.092344386714047e-06,
      "loss": 0.6525,
      "step": 12809
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.581244452895232,
      "learning_rate": 4.091634026973301e-06,
      "loss": 0.6254,
      "step": 12810
    },
    {
      "epoch": 1.14,
      "grad_norm": 8.185131563953298,
      "learning_rate": 4.090923686192707e-06,
      "loss": 0.6148,
      "step": 12811
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.018644560787843,
      "learning_rate": 4.090213364387089e-06,
      "loss": 0.5823,
      "step": 12812
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.273722739574082,
      "learning_rate": 4.089503061571274e-06,
      "loss": 0.5465,
      "step": 12813
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.372631725903289,
      "learning_rate": 4.08879277776009e-06,
      "loss": 0.5956,
      "step": 12814
    },
    {
      "epoch": 1.14,
      "grad_norm": 7.7677059168467295,
      "learning_rate": 4.088082512968361e-06,
      "loss": 0.6257,
      "step": 12815
    },
    {
      "epoch": 1.14,
      "grad_norm": 7.899159156747473,
      "learning_rate": 4.0873722672109115e-06,
      "loss": 0.5434,
      "step": 12816
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.303273939884204,
      "learning_rate": 4.086662040502569e-06,
      "loss": 0.5995,
      "step": 12817
    },
    {
      "epoch": 1.14,
      "grad_norm": 9.790862909693988,
      "learning_rate": 4.085951832858155e-06,
      "loss": 0.5866,
      "step": 12818
    },
    {
      "epoch": 1.14,
      "grad_norm": 8.773710760791847,
      "learning_rate": 4.085241644292495e-06,
      "loss": 0.6231,
      "step": 12819
    },
    {
      "epoch": 1.14,
      "grad_norm": 4.816050416873687,
      "learning_rate": 4.084531474820413e-06,
      "loss": 0.539,
      "step": 12820
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.252981499276761,
      "learning_rate": 4.083821324456731e-06,
      "loss": 0.5737,
      "step": 12821
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.403551955250211,
      "learning_rate": 4.083111193216274e-06,
      "loss": 0.6352,
      "step": 12822
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.499492197440524,
      "learning_rate": 4.082401081113861e-06,
      "loss": 0.6833,
      "step": 12823
    },
    {
      "epoch": 1.14,
      "grad_norm": 7.25817667440091,
      "learning_rate": 4.081690988164317e-06,
      "loss": 0.6811,
      "step": 12824
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.292793425763199,
      "learning_rate": 4.080980914382461e-06,
      "loss": 0.5822,
      "step": 12825
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.241834975960904,
      "learning_rate": 4.080270859783117e-06,
      "loss": 0.6337,
      "step": 12826
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.36673943803984,
      "learning_rate": 4.079560824381105e-06,
      "loss": 0.5574,
      "step": 12827
    },
    {
      "epoch": 1.14,
      "grad_norm": 8.441263619417304,
      "learning_rate": 4.078850808191243e-06,
      "loss": 0.5733,
      "step": 12828
    },
    {
      "epoch": 1.14,
      "grad_norm": 7.339624316272556,
      "learning_rate": 4.078140811228355e-06,
      "loss": 0.6111,
      "step": 12829
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.916008062555011,
      "learning_rate": 4.077430833507259e-06,
      "loss": 0.6722,
      "step": 12830
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.632471359294994,
      "learning_rate": 4.0767208750427725e-06,
      "loss": 0.5919,
      "step": 12831
    },
    {
      "epoch": 1.14,
      "grad_norm": 8.372138304110585,
      "learning_rate": 4.076010935849716e-06,
      "loss": 0.6031,
      "step": 12832
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.029163707120544,
      "learning_rate": 4.075301015942909e-06,
      "loss": 0.6016,
      "step": 12833
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.824890235988416,
      "learning_rate": 4.074591115337167e-06,
      "loss": 0.5474,
      "step": 12834
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.362884615608136,
      "learning_rate": 4.073881234047309e-06,
      "loss": 0.5992,
      "step": 12835
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.333278801761608,
      "learning_rate": 4.0731713720881526e-06,
      "loss": 0.5698,
      "step": 12836
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.208397677427569,
      "learning_rate": 4.072461529474516e-06,
      "loss": 0.6053,
      "step": 12837
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.173229520615358,
      "learning_rate": 4.071751706221212e-06,
      "loss": 0.5854,
      "step": 12838
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.726421716103314,
      "learning_rate": 4.071041902343057e-06,
      "loss": 0.5852,
      "step": 12839
    },
    {
      "epoch": 1.15,
      "grad_norm": 7.311094120248424,
      "learning_rate": 4.070332117854869e-06,
      "loss": 0.586,
      "step": 12840
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.254355299643176,
      "learning_rate": 4.069622352771462e-06,
      "loss": 0.5996,
      "step": 12841
    },
    {
      "epoch": 1.15,
      "grad_norm": 7.5260880288762095,
      "learning_rate": 4.06891260710765e-06,
      "loss": 0.6103,
      "step": 12842
    },
    {
      "epoch": 1.15,
      "grad_norm": 7.6769890931178395,
      "learning_rate": 4.06820288087825e-06,
      "loss": 0.593,
      "step": 12843
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.824417505602332,
      "learning_rate": 4.0674931740980725e-06,
      "loss": 0.6334,
      "step": 12844
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.702695288803621,
      "learning_rate": 4.066783486781934e-06,
      "loss": 0.5568,
      "step": 12845
    },
    {
      "epoch": 1.15,
      "grad_norm": 8.902140913020927,
      "learning_rate": 4.0660738189446455e-06,
      "loss": 0.6335,
      "step": 12846
    },
    {
      "epoch": 1.15,
      "grad_norm": 7.750322244524445,
      "learning_rate": 4.065364170601022e-06,
      "loss": 0.5937,
      "step": 12847
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.219518044125492,
      "learning_rate": 4.064654541765873e-06,
      "loss": 0.6074,
      "step": 12848
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.7976813563133645,
      "learning_rate": 4.063944932454014e-06,
      "loss": 0.6504,
      "step": 12849
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.329784321131278,
      "learning_rate": 4.063235342680253e-06,
      "loss": 0.6322,
      "step": 12850
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.408937774643539,
      "learning_rate": 4.062525772459406e-06,
      "loss": 0.6168,
      "step": 12851
    },
    {
      "epoch": 1.15,
      "grad_norm": 7.655391881049637,
      "learning_rate": 4.061816221806276e-06,
      "loss": 0.6282,
      "step": 12852
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.555664216417251,
      "learning_rate": 4.0611066907356815e-06,
      "loss": 0.6407,
      "step": 12853
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.046652080314439,
      "learning_rate": 4.060397179262426e-06,
      "loss": 0.5624,
      "step": 12854
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.0731467601213325,
      "learning_rate": 4.059687687401323e-06,
      "loss": 0.7119,
      "step": 12855
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.092792978439218,
      "learning_rate": 4.05897821516718e-06,
      "loss": 0.6854,
      "step": 12856
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.027003200515344,
      "learning_rate": 4.058268762574805e-06,
      "loss": 0.5935,
      "step": 12857
    },
    {
      "epoch": 1.15,
      "grad_norm": 10.484743995597071,
      "learning_rate": 4.057559329639009e-06,
      "loss": 0.5196,
      "step": 12858
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.028704907707448,
      "learning_rate": 4.056849916374597e-06,
      "loss": 0.6068,
      "step": 12859
    },
    {
      "epoch": 1.15,
      "grad_norm": 7.9471582758263395,
      "learning_rate": 4.056140522796377e-06,
      "loss": 0.5888,
      "step": 12860
    },
    {
      "epoch": 1.15,
      "grad_norm": 7.323088985046348,
      "learning_rate": 4.055431148919158e-06,
      "loss": 0.6431,
      "step": 12861
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.850591541806382,
      "learning_rate": 4.054721794757744e-06,
      "loss": 0.666,
      "step": 12862
    },
    {
      "epoch": 1.15,
      "grad_norm": 8.466758536432264,
      "learning_rate": 4.054012460326942e-06,
      "loss": 0.5766,
      "step": 12863
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.232755544749327,
      "learning_rate": 4.05330314564156e-06,
      "loss": 0.5546,
      "step": 12864
    },
    {
      "epoch": 1.15,
      "grad_norm": 7.738421201072238,
      "learning_rate": 4.052593850716401e-06,
      "loss": 0.6597,
      "step": 12865
    },
    {
      "epoch": 1.15,
      "grad_norm": 7.324660711368953,
      "learning_rate": 4.051884575566271e-06,
      "loss": 0.5845,
      "step": 12866
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.785817337006844,
      "learning_rate": 4.0511753202059735e-06,
      "loss": 0.581,
      "step": 12867
    },
    {
      "epoch": 1.15,
      "grad_norm": 4.728488083068472,
      "learning_rate": 4.050466084650314e-06,
      "loss": 0.6193,
      "step": 12868
    },
    {
      "epoch": 1.15,
      "grad_norm": 12.262278881169339,
      "learning_rate": 4.049756868914094e-06,
      "loss": 0.6783,
      "step": 12869
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.754396868791832,
      "learning_rate": 4.049047673012119e-06,
      "loss": 0.6547,
      "step": 12870
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.762727559654872,
      "learning_rate": 4.048338496959192e-06,
      "loss": 0.6784,
      "step": 12871
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.124918721979807,
      "learning_rate": 4.047629340770113e-06,
      "loss": 0.5947,
      "step": 12872
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.416438891243214,
      "learning_rate": 4.046920204459687e-06,
      "loss": 0.5155,
      "step": 12873
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.715530101349614,
      "learning_rate": 4.046211088042715e-06,
      "loss": 0.5956,
      "step": 12874
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.68798714187965,
      "learning_rate": 4.0455019915339964e-06,
      "loss": 0.6065,
      "step": 12875
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.710219835867501,
      "learning_rate": 4.044792914948335e-06,
      "loss": 0.5652,
      "step": 12876
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.285799349308264,
      "learning_rate": 4.0440838583005295e-06,
      "loss": 0.6255,
      "step": 12877
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.582549425778738,
      "learning_rate": 4.043374821605379e-06,
      "loss": 0.6226,
      "step": 12878
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.13527730243023,
      "learning_rate": 4.042665804877683e-06,
      "loss": 0.5936,
      "step": 12879
    },
    {
      "epoch": 1.15,
      "grad_norm": 7.998567597244722,
      "learning_rate": 4.041956808132246e-06,
      "loss": 0.6136,
      "step": 12880
    },
    {
      "epoch": 1.15,
      "grad_norm": 4.674853484921768,
      "learning_rate": 4.041247831383861e-06,
      "loss": 0.622,
      "step": 12881
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.61713501277379,
      "learning_rate": 4.040538874647325e-06,
      "loss": 0.5933,
      "step": 12882
    },
    {
      "epoch": 1.15,
      "grad_norm": 12.064019993350454,
      "learning_rate": 4.039829937937441e-06,
      "loss": 0.6597,
      "step": 12883
    },
    {
      "epoch": 1.15,
      "grad_norm": 4.793041162295543,
      "learning_rate": 4.039121021269005e-06,
      "loss": 0.5805,
      "step": 12884
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.63716929067069,
      "learning_rate": 4.038412124656811e-06,
      "loss": 0.5674,
      "step": 12885
    },
    {
      "epoch": 1.15,
      "grad_norm": 7.267135561357943,
      "learning_rate": 4.03770324811566e-06,
      "loss": 0.5962,
      "step": 12886
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.415399790833891,
      "learning_rate": 4.036994391660346e-06,
      "loss": 0.685,
      "step": 12887
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.097844314674362,
      "learning_rate": 4.0362855553056646e-06,
      "loss": 0.6498,
      "step": 12888
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.5044093783752865,
      "learning_rate": 4.035576739066411e-06,
      "loss": 0.6295,
      "step": 12889
    },
    {
      "epoch": 1.15,
      "grad_norm": 8.341189359945046,
      "learning_rate": 4.034867942957383e-06,
      "loss": 0.5968,
      "step": 12890
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.91013106763395,
      "learning_rate": 4.034159166993371e-06,
      "loss": 0.5802,
      "step": 12891
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.17667503415963,
      "learning_rate": 4.033450411189172e-06,
      "loss": 0.5983,
      "step": 12892
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.216968826718032,
      "learning_rate": 4.03274167555958e-06,
      "loss": 0.6585,
      "step": 12893
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.164144173998226,
      "learning_rate": 4.032032960119388e-06,
      "loss": 0.6776,
      "step": 12894
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.00529538806573,
      "learning_rate": 4.0313242648833865e-06,
      "loss": 0.5401,
      "step": 12895
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.556242762324982,
      "learning_rate": 4.030615589866369e-06,
      "loss": 0.6407,
      "step": 12896
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.266595833477032,
      "learning_rate": 4.029906935083129e-06,
      "loss": 0.5925,
      "step": 12897
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.450261018060676,
      "learning_rate": 4.029198300548457e-06,
      "loss": 0.5699,
      "step": 12898
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.1503394899620885,
      "learning_rate": 4.028489686277145e-06,
      "loss": 0.587,
      "step": 12899
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.696315274523511,
      "learning_rate": 4.0277810922839824e-06,
      "loss": 0.5791,
      "step": 12900
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.478380931244562,
      "learning_rate": 4.02707251858376e-06,
      "loss": 0.601,
      "step": 12901
    },
    {
      "epoch": 1.15,
      "grad_norm": 4.938362554736422,
      "learning_rate": 4.026363965191269e-06,
      "loss": 0.6834,
      "step": 12902
    },
    {
      "epoch": 1.15,
      "grad_norm": 7.262734009340027,
      "learning_rate": 4.025655432121299e-06,
      "loss": 0.6065,
      "step": 12903
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.843808971286931,
      "learning_rate": 4.0249469193886355e-06,
      "loss": 0.5908,
      "step": 12904
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.057167342921941,
      "learning_rate": 4.0242384270080715e-06,
      "loss": 0.6446,
      "step": 12905
    },
    {
      "epoch": 1.15,
      "grad_norm": 4.899461529320566,
      "learning_rate": 4.0235299549943935e-06,
      "loss": 0.5436,
      "step": 12906
    },
    {
      "epoch": 1.15,
      "grad_norm": 7.349229949305356,
      "learning_rate": 4.022821503362388e-06,
      "loss": 0.6769,
      "step": 12907
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.624426351035255,
      "learning_rate": 4.022113072126845e-06,
      "loss": 0.5597,
      "step": 12908
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.391669634530487,
      "learning_rate": 4.021404661302551e-06,
      "loss": 0.6199,
      "step": 12909
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.831207460454864,
      "learning_rate": 4.02069627090429e-06,
      "loss": 0.6278,
      "step": 12910
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.873230444466613,
      "learning_rate": 4.01998790094685e-06,
      "loss": 0.6232,
      "step": 12911
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.266884672211902,
      "learning_rate": 4.0192795514450165e-06,
      "loss": 0.632,
      "step": 12912
    },
    {
      "epoch": 1.15,
      "grad_norm": 4.894407259696568,
      "learning_rate": 4.018571222413575e-06,
      "loss": 0.6186,
      "step": 12913
    },
    {
      "epoch": 1.15,
      "grad_norm": 7.042329040379265,
      "learning_rate": 4.017862913867309e-06,
      "loss": 0.6311,
      "step": 12914
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.469520317146092,
      "learning_rate": 4.017154625821003e-06,
      "loss": 0.5782,
      "step": 12915
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.720201659902491,
      "learning_rate": 4.016446358289444e-06,
      "loss": 0.6451,
      "step": 12916
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.834522335972365,
      "learning_rate": 4.015738111287412e-06,
      "loss": 0.5956,
      "step": 12917
    },
    {
      "epoch": 1.15,
      "grad_norm": 7.005359422646475,
      "learning_rate": 4.015029884829691e-06,
      "loss": 0.6228,
      "step": 12918
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.010213049187701,
      "learning_rate": 4.014321678931065e-06,
      "loss": 0.6034,
      "step": 12919
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.318113505934496,
      "learning_rate": 4.013613493606315e-06,
      "loss": 0.637,
      "step": 12920
    },
    {
      "epoch": 1.15,
      "grad_norm": 8.456426201722001,
      "learning_rate": 4.012905328870223e-06,
      "loss": 0.5506,
      "step": 12921
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.325230261444733,
      "learning_rate": 4.012197184737572e-06,
      "loss": 0.5617,
      "step": 12922
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.878872081746013,
      "learning_rate": 4.011489061223142e-06,
      "loss": 0.552,
      "step": 12923
    },
    {
      "epoch": 1.15,
      "grad_norm": 8.447291423814715,
      "learning_rate": 4.010780958341711e-06,
      "loss": 0.6592,
      "step": 12924
    },
    {
      "epoch": 1.15,
      "grad_norm": 7.136925288731995,
      "learning_rate": 4.010072876108061e-06,
      "loss": 0.6096,
      "step": 12925
    },
    {
      "epoch": 1.15,
      "grad_norm": 4.7735551525400854,
      "learning_rate": 4.009364814536974e-06,
      "loss": 0.572,
      "step": 12926
    },
    {
      "epoch": 1.15,
      "grad_norm": 4.610777470220228,
      "learning_rate": 4.008656773643224e-06,
      "loss": 0.6069,
      "step": 12927
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.229757944790915,
      "learning_rate": 4.007948753441594e-06,
      "loss": 0.5917,
      "step": 12928
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.240319092679932,
      "learning_rate": 4.007240753946862e-06,
      "loss": 0.591,
      "step": 12929
    },
    {
      "epoch": 1.15,
      "grad_norm": 4.857702847974442,
      "learning_rate": 4.006532775173805e-06,
      "loss": 0.5804,
      "step": 12930
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.601304181465091,
      "learning_rate": 4.005824817137199e-06,
      "loss": 0.6259,
      "step": 12931
    },
    {
      "epoch": 1.15,
      "grad_norm": 7.028212465461521,
      "learning_rate": 4.005116879851825e-06,
      "loss": 0.6131,
      "step": 12932
    },
    {
      "epoch": 1.15,
      "grad_norm": 7.009572914792403,
      "learning_rate": 4.004408963332456e-06,
      "loss": 0.6109,
      "step": 12933
    },
    {
      "epoch": 1.15,
      "grad_norm": 11.034448736563673,
      "learning_rate": 4.003701067593871e-06,
      "loss": 0.6847,
      "step": 12934
    },
    {
      "epoch": 1.15,
      "grad_norm": 7.018727209609907,
      "learning_rate": 4.002993192650843e-06,
      "loss": 0.5612,
      "step": 12935
    },
    {
      "epoch": 1.15,
      "grad_norm": 7.398749039704398,
      "learning_rate": 4.0022853385181484e-06,
      "loss": 0.6514,
      "step": 12936
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.34905754032123,
      "learning_rate": 4.001577505210564e-06,
      "loss": 0.5986,
      "step": 12937
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.332238507782955,
      "learning_rate": 4.000869692742862e-06,
      "loss": 0.6638,
      "step": 12938
    },
    {
      "epoch": 1.15,
      "grad_norm": 8.236282493853684,
      "learning_rate": 4.000161901129818e-06,
      "loss": 0.6232,
      "step": 12939
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.179038575653843,
      "learning_rate": 3.999454130386203e-06,
      "loss": 0.6635,
      "step": 12940
    },
    {
      "epoch": 1.15,
      "grad_norm": 8.425671142289678,
      "learning_rate": 3.9987463805267914e-06,
      "loss": 0.6218,
      "step": 12941
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.985831364104041,
      "learning_rate": 3.998038651566359e-06,
      "loss": 0.6259,
      "step": 12942
    },
    {
      "epoch": 1.15,
      "grad_norm": 8.799333842306265,
      "learning_rate": 3.997330943519673e-06,
      "loss": 0.6574,
      "step": 12943
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.548162156102045,
      "learning_rate": 3.996623256401509e-06,
      "loss": 0.6571,
      "step": 12944
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.488667078780645,
      "learning_rate": 3.9959155902266365e-06,
      "loss": 0.6368,
      "step": 12945
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.096531459219551,
      "learning_rate": 3.995207945009827e-06,
      "loss": 0.5616,
      "step": 12946
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.135233585678908,
      "learning_rate": 3.994500320765852e-06,
      "loss": 0.6185,
      "step": 12947
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.117003878272828,
      "learning_rate": 3.993792717509481e-06,
      "loss": 0.5966,
      "step": 12948
    },
    {
      "epoch": 1.16,
      "grad_norm": 7.205804767360758,
      "learning_rate": 3.993085135255482e-06,
      "loss": 0.5098,
      "step": 12949
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.247721642017168,
      "learning_rate": 3.992377574018627e-06,
      "loss": 0.5973,
      "step": 12950
    },
    {
      "epoch": 1.16,
      "grad_norm": 8.732337919947808,
      "learning_rate": 3.991670033813683e-06,
      "loss": 0.5776,
      "step": 12951
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.808583985680122,
      "learning_rate": 3.990962514655421e-06,
      "loss": 0.619,
      "step": 12952
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.5386362157793405,
      "learning_rate": 3.990255016558604e-06,
      "loss": 0.5711,
      "step": 12953
    },
    {
      "epoch": 1.16,
      "grad_norm": 8.382117100636847,
      "learning_rate": 3.989547539538004e-06,
      "loss": 0.6067,
      "step": 12954
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.106438004258891,
      "learning_rate": 3.9888400836083844e-06,
      "loss": 0.5881,
      "step": 12955
    },
    {
      "epoch": 1.16,
      "grad_norm": 8.225845523732604,
      "learning_rate": 3.988132648784514e-06,
      "loss": 0.6306,
      "step": 12956
    },
    {
      "epoch": 1.16,
      "grad_norm": 7.076381821596656,
      "learning_rate": 3.98742523508116e-06,
      "loss": 0.5598,
      "step": 12957
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.998279670270793,
      "learning_rate": 3.9867178425130856e-06,
      "loss": 0.5854,
      "step": 12958
    },
    {
      "epoch": 1.16,
      "grad_norm": 7.572892684706871,
      "learning_rate": 3.986010471095057e-06,
      "loss": 0.635,
      "step": 12959
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.4374491011574335,
      "learning_rate": 3.98530312084184e-06,
      "loss": 0.6253,
      "step": 12960
    },
    {
      "epoch": 1.16,
      "grad_norm": 4.559749376764543,
      "learning_rate": 3.9845957917681975e-06,
      "loss": 0.5984,
      "step": 12961
    },
    {
      "epoch": 1.16,
      "grad_norm": 4.315177411202582,
      "learning_rate": 3.983888483888893e-06,
      "loss": 0.5957,
      "step": 12962
    },
    {
      "epoch": 1.16,
      "grad_norm": 8.807879081974322,
      "learning_rate": 3.983181197218694e-06,
      "loss": 0.589,
      "step": 12963
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.821924924069717,
      "learning_rate": 3.9824739317723585e-06,
      "loss": 0.6013,
      "step": 12964
    },
    {
      "epoch": 1.16,
      "grad_norm": 7.952922065383657,
      "learning_rate": 3.981766687564652e-06,
      "loss": 0.5928,
      "step": 12965
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.073007473876,
      "learning_rate": 3.981059464610338e-06,
      "loss": 0.6084,
      "step": 12966
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.2214790207171875,
      "learning_rate": 3.980352262924175e-06,
      "loss": 0.6294,
      "step": 12967
    },
    {
      "epoch": 1.16,
      "grad_norm": 7.011536638037049,
      "learning_rate": 3.979645082520924e-06,
      "loss": 0.6033,
      "step": 12968
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.381083232034313,
      "learning_rate": 3.978937923415348e-06,
      "loss": 0.5547,
      "step": 12969
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.087215600429319,
      "learning_rate": 3.978230785622208e-06,
      "loss": 0.5747,
      "step": 12970
    },
    {
      "epoch": 1.16,
      "grad_norm": 7.719575712256528,
      "learning_rate": 3.977523669156261e-06,
      "loss": 0.6574,
      "step": 12971
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.4682768140732465,
      "learning_rate": 3.9768165740322685e-06,
      "loss": 0.6653,
      "step": 12972
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.085554925423577,
      "learning_rate": 3.976109500264991e-06,
      "loss": 0.658,
      "step": 12973
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.64776042493028,
      "learning_rate": 3.975402447869183e-06,
      "loss": 0.5746,
      "step": 12974
    },
    {
      "epoch": 1.16,
      "grad_norm": 7.133397846551657,
      "learning_rate": 3.974695416859606e-06,
      "loss": 0.6128,
      "step": 12975
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.5579077019004135,
      "learning_rate": 3.973988407251018e-06,
      "loss": 0.5986,
      "step": 12976
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.156625648590385,
      "learning_rate": 3.9732814190581736e-06,
      "loss": 0.6228,
      "step": 12977
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.665695063169347,
      "learning_rate": 3.972574452295831e-06,
      "loss": 0.5422,
      "step": 12978
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.69833412722042,
      "learning_rate": 3.971867506978748e-06,
      "loss": 0.6542,
      "step": 12979
    },
    {
      "epoch": 1.16,
      "grad_norm": 8.791863524189502,
      "learning_rate": 3.971160583121678e-06,
      "loss": 0.6429,
      "step": 12980
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.632376419562274,
      "learning_rate": 3.97045368073938e-06,
      "loss": 0.614,
      "step": 12981
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.809302172332681,
      "learning_rate": 3.969746799846605e-06,
      "loss": 0.629,
      "step": 12982
    },
    {
      "epoch": 1.16,
      "grad_norm": 4.805766213068198,
      "learning_rate": 3.969039940458109e-06,
      "loss": 0.5685,
      "step": 12983
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.943879478451886,
      "learning_rate": 3.968333102588646e-06,
      "loss": 0.6437,
      "step": 12984
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.161374955857747,
      "learning_rate": 3.967626286252971e-06,
      "loss": 0.5965,
      "step": 12985
    },
    {
      "epoch": 1.16,
      "grad_norm": 7.655589665133632,
      "learning_rate": 3.966919491465837e-06,
      "loss": 0.5884,
      "step": 12986
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.438425702386194,
      "learning_rate": 3.966212718241995e-06,
      "loss": 0.6151,
      "step": 12987
    },
    {
      "epoch": 1.16,
      "grad_norm": 4.633190352018585,
      "learning_rate": 3.965505966596199e-06,
      "loss": 0.6698,
      "step": 12988
    },
    {
      "epoch": 1.16,
      "grad_norm": 7.512873862448835,
      "learning_rate": 3.964799236543201e-06,
      "loss": 0.5931,
      "step": 12989
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.74831177559796,
      "learning_rate": 3.964092528097751e-06,
      "loss": 0.5918,
      "step": 12990
    },
    {
      "epoch": 1.16,
      "grad_norm": 8.377836647078722,
      "learning_rate": 3.9633858412746e-06,
      "loss": 0.682,
      "step": 12991
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.464391536541433,
      "learning_rate": 3.962679176088502e-06,
      "loss": 0.6789,
      "step": 12992
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.578944076367026,
      "learning_rate": 3.961972532554203e-06,
      "loss": 0.6341,
      "step": 12993
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.053424478823554,
      "learning_rate": 3.961265910686454e-06,
      "loss": 0.5545,
      "step": 12994
    },
    {
      "epoch": 1.16,
      "grad_norm": 8.79194251902041,
      "learning_rate": 3.960559310500006e-06,
      "loss": 0.7113,
      "step": 12995
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.311557473901445,
      "learning_rate": 3.959852732009606e-06,
      "loss": 0.5586,
      "step": 12996
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.998757266510518,
      "learning_rate": 3.95914617523e-06,
      "loss": 0.5903,
      "step": 12997
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.491434918618766,
      "learning_rate": 3.958439640175939e-06,
      "loss": 0.6245,
      "step": 12998
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.66220241939905,
      "learning_rate": 3.957733126862171e-06,
      "loss": 0.6211,
      "step": 12999
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.814238222288148,
      "learning_rate": 3.957026635303439e-06,
      "loss": 0.5416,
      "step": 13000
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.306995974195617,
      "learning_rate": 3.956320165514493e-06,
      "loss": 0.6487,
      "step": 13001
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.035796972102513,
      "learning_rate": 3.9556137175100786e-06,
      "loss": 0.5434,
      "step": 13002
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.657102822186134,
      "learning_rate": 3.95490729130494e-06,
      "loss": 0.6036,
      "step": 13003
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.838548432768052,
      "learning_rate": 3.9542008869138226e-06,
      "loss": 0.5719,
      "step": 13004
    },
    {
      "epoch": 1.16,
      "grad_norm": 7.717549281554352,
      "learning_rate": 3.953494504351473e-06,
      "loss": 0.5642,
      "step": 13005
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.116543351412588,
      "learning_rate": 3.952788143632632e-06,
      "loss": 0.6085,
      "step": 13006
    },
    {
      "epoch": 1.16,
      "grad_norm": 7.11017652768376,
      "learning_rate": 3.9520818047720454e-06,
      "loss": 0.6382,
      "step": 13007
    },
    {
      "epoch": 1.16,
      "grad_norm": 7.292829767253248,
      "learning_rate": 3.951375487784457e-06,
      "loss": 0.6158,
      "step": 13008
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.424063490813165,
      "learning_rate": 3.950669192684609e-06,
      "loss": 0.5952,
      "step": 13009
    },
    {
      "epoch": 1.16,
      "grad_norm": 7.898971423169459,
      "learning_rate": 3.949962919487245e-06,
      "loss": 0.6337,
      "step": 13010
    },
    {
      "epoch": 1.16,
      "grad_norm": 8.813872882492872,
      "learning_rate": 3.9492566682071025e-06,
      "loss": 0.6639,
      "step": 13011
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.2334693327831,
      "learning_rate": 3.948550438858928e-06,
      "loss": 0.6211,
      "step": 13012
    },
    {
      "epoch": 1.16,
      "grad_norm": 8.002893682889669,
      "learning_rate": 3.947844231457459e-06,
      "loss": 0.6904,
      "step": 13013
    },
    {
      "epoch": 1.16,
      "grad_norm": 8.29305328517966,
      "learning_rate": 3.947138046017438e-06,
      "loss": 0.601,
      "step": 13014
    },
    {
      "epoch": 1.16,
      "grad_norm": 7.963350086670499,
      "learning_rate": 3.946431882553605e-06,
      "loss": 0.6072,
      "step": 13015
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.7294365711590665,
      "learning_rate": 3.945725741080699e-06,
      "loss": 0.577,
      "step": 13016
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.8003413777655854,
      "learning_rate": 3.945019621613459e-06,
      "loss": 0.6008,
      "step": 13017
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.120669689166336,
      "learning_rate": 3.944313524166624e-06,
      "loss": 0.5705,
      "step": 13018
    },
    {
      "epoch": 1.16,
      "grad_norm": 7.846715806069802,
      "learning_rate": 3.943607448754932e-06,
      "loss": 0.551,
      "step": 13019
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.697288165173782,
      "learning_rate": 3.9429013953931225e-06,
      "loss": 0.6392,
      "step": 13020
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.2101551995713375,
      "learning_rate": 3.94219536409593e-06,
      "loss": 0.6159,
      "step": 13021
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.557074014057159,
      "learning_rate": 3.941489354878092e-06,
      "loss": 0.5902,
      "step": 13022
    },
    {
      "epoch": 1.16,
      "grad_norm": 7.324671095183377,
      "learning_rate": 3.940783367754348e-06,
      "loss": 0.6556,
      "step": 13023
    },
    {
      "epoch": 1.16,
      "grad_norm": 8.084648410221194,
      "learning_rate": 3.940077402739431e-06,
      "loss": 0.5513,
      "step": 13024
    },
    {
      "epoch": 1.16,
      "grad_norm": 8.617682893983577,
      "learning_rate": 3.939371459848077e-06,
      "loss": 0.6166,
      "step": 13025
    },
    {
      "epoch": 1.16,
      "grad_norm": 4.515876656907204,
      "learning_rate": 3.93866553909502e-06,
      "loss": 0.5537,
      "step": 13026
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.1047014362838485,
      "learning_rate": 3.9379596404949946e-06,
      "loss": 0.5834,
      "step": 13027
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.650562809176478,
      "learning_rate": 3.937253764062738e-06,
      "loss": 0.5607,
      "step": 13028
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.536342784491075,
      "learning_rate": 3.936547909812981e-06,
      "loss": 0.568,
      "step": 13029
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.342780264509894,
      "learning_rate": 3.9358420777604574e-06,
      "loss": 0.6148,
      "step": 13030
    },
    {
      "epoch": 1.16,
      "grad_norm": 9.428270898556068,
      "learning_rate": 3.9351362679199e-06,
      "loss": 0.5843,
      "step": 13031
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.47963437244726,
      "learning_rate": 3.934430480306041e-06,
      "loss": 0.5978,
      "step": 13032
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.960527303633001,
      "learning_rate": 3.933724714933612e-06,
      "loss": 0.6527,
      "step": 13033
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.04560279964691,
      "learning_rate": 3.933018971817344e-06,
      "loss": 0.5302,
      "step": 13034
    },
    {
      "epoch": 1.16,
      "grad_norm": 8.475576444622297,
      "learning_rate": 3.932313250971969e-06,
      "loss": 0.5723,
      "step": 13035
    },
    {
      "epoch": 1.16,
      "grad_norm": 7.528136085185549,
      "learning_rate": 3.931607552412217e-06,
      "loss": 0.615,
      "step": 13036
    },
    {
      "epoch": 1.16,
      "grad_norm": 7.0575569920201575,
      "learning_rate": 3.930901876152817e-06,
      "loss": 0.5799,
      "step": 13037
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.055816558079321,
      "learning_rate": 3.930196222208501e-06,
      "loss": 0.6236,
      "step": 13038
    },
    {
      "epoch": 1.16,
      "grad_norm": 7.23980310206584,
      "learning_rate": 3.929490590593994e-06,
      "loss": 0.6325,
      "step": 13039
    },
    {
      "epoch": 1.16,
      "grad_norm": 9.26627310604706,
      "learning_rate": 3.9287849813240265e-06,
      "loss": 0.5973,
      "step": 13040
    },
    {
      "epoch": 1.16,
      "grad_norm": 8.77397695484451,
      "learning_rate": 3.928079394413327e-06,
      "loss": 0.6377,
      "step": 13041
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.765782726298847,
      "learning_rate": 3.927373829876622e-06,
      "loss": 0.6016,
      "step": 13042
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.663810923043396,
      "learning_rate": 3.9266682877286405e-06,
      "loss": 0.6199,
      "step": 13043
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.647399375397755,
      "learning_rate": 3.925962767984107e-06,
      "loss": 0.5285,
      "step": 13044
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.595437831850956,
      "learning_rate": 3.925257270657749e-06,
      "loss": 0.5957,
      "step": 13045
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.673434830295425,
      "learning_rate": 3.924551795764292e-06,
      "loss": 0.6482,
      "step": 13046
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.295389970660264,
      "learning_rate": 3.92384634331846e-06,
      "loss": 0.556,
      "step": 13047
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.803718525959643,
      "learning_rate": 3.92314091333498e-06,
      "loss": 0.6506,
      "step": 13048
    },
    {
      "epoch": 1.16,
      "grad_norm": 7.907367962224505,
      "learning_rate": 3.9224355058285755e-06,
      "loss": 0.5515,
      "step": 13049
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.859039437303726,
      "learning_rate": 3.921730120813968e-06,
      "loss": 0.5886,
      "step": 13050
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.419256837303687,
      "learning_rate": 3.921024758305884e-06,
      "loss": 0.5744,
      "step": 13051
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.0350692170564,
      "learning_rate": 3.920319418319046e-06,
      "loss": 0.6123,
      "step": 13052
    },
    {
      "epoch": 1.16,
      "grad_norm": 8.591031276284657,
      "learning_rate": 3.919614100868177e-06,
      "loss": 0.6212,
      "step": 13053
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.4696260864712185,
      "learning_rate": 3.9189088059679955e-06,
      "loss": 0.6299,
      "step": 13054
    },
    {
      "epoch": 1.16,
      "grad_norm": 4.382606749937304,
      "learning_rate": 3.918203533633225e-06,
      "loss": 0.5746,
      "step": 13055
    },
    {
      "epoch": 1.16,
      "grad_norm": 9.360086269628882,
      "learning_rate": 3.9174982838785875e-06,
      "loss": 0.6519,
      "step": 13056
    },
    {
      "epoch": 1.16,
      "grad_norm": 4.825289869499037,
      "learning_rate": 3.916793056718802e-06,
      "loss": 0.5773,
      "step": 13057
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.303396750281862,
      "learning_rate": 3.916087852168588e-06,
      "loss": 0.6891,
      "step": 13058
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.4232250096541605,
      "learning_rate": 3.915382670242669e-06,
      "loss": 0.5975,
      "step": 13059
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.739321471134495,
      "learning_rate": 3.91467751095576e-06,
      "loss": 0.556,
      "step": 13060
    },
    {
      "epoch": 1.17,
      "grad_norm": 4.5234425282383475,
      "learning_rate": 3.913972374322581e-06,
      "loss": 0.5785,
      "step": 13061
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.295995856776078,
      "learning_rate": 3.913267260357852e-06,
      "loss": 0.5756,
      "step": 13062
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.657182058263523,
      "learning_rate": 3.912562169076287e-06,
      "loss": 0.6339,
      "step": 13063
    },
    {
      "epoch": 1.17,
      "grad_norm": 4.5569762811890495,
      "learning_rate": 3.911857100492605e-06,
      "loss": 0.5532,
      "step": 13064
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.769926487014623,
      "learning_rate": 3.9111520546215255e-06,
      "loss": 0.5575,
      "step": 13065
    },
    {
      "epoch": 1.17,
      "grad_norm": 6.35077762964298,
      "learning_rate": 3.91044703147776e-06,
      "loss": 0.662,
      "step": 13066
    },
    {
      "epoch": 1.17,
      "grad_norm": 6.891584086086863,
      "learning_rate": 3.909742031076029e-06,
      "loss": 0.5653,
      "step": 13067
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.9575296102349755,
      "learning_rate": 3.909037053431044e-06,
      "loss": 0.5632,
      "step": 13068
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.588207350387445,
      "learning_rate": 3.908332098557523e-06,
      "loss": 0.571,
      "step": 13069
    },
    {
      "epoch": 1.17,
      "grad_norm": 4.906235502922406,
      "learning_rate": 3.907627166470176e-06,
      "loss": 0.6163,
      "step": 13070
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.770251806959943,
      "learning_rate": 3.90692225718372e-06,
      "loss": 0.5602,
      "step": 13071
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.204008855596645,
      "learning_rate": 3.90621737071287e-06,
      "loss": 0.6856,
      "step": 13072
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.23175817118081,
      "learning_rate": 3.905512507072336e-06,
      "loss": 0.6206,
      "step": 13073
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.738104857263245,
      "learning_rate": 3.904807666276831e-06,
      "loss": 0.5794,
      "step": 13074
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.228177568655165,
      "learning_rate": 3.904102848341068e-06,
      "loss": 0.5972,
      "step": 13075
    },
    {
      "epoch": 1.17,
      "grad_norm": 6.0774810778863575,
      "learning_rate": 3.903398053279757e-06,
      "loss": 0.6308,
      "step": 13076
    },
    {
      "epoch": 1.17,
      "grad_norm": 8.02918837488522,
      "learning_rate": 3.902693281107611e-06,
      "loss": 0.6125,
      "step": 13077
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.284985421628524,
      "learning_rate": 3.901988531839341e-06,
      "loss": 0.6074,
      "step": 13078
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.403344839933403,
      "learning_rate": 3.901283805489654e-06,
      "loss": 0.624,
      "step": 13079
    },
    {
      "epoch": 1.17,
      "grad_norm": 8.689008689432383,
      "learning_rate": 3.900579102073262e-06,
      "loss": 0.6386,
      "step": 13080
    },
    {
      "epoch": 1.17,
      "grad_norm": 6.22590554860552,
      "learning_rate": 3.899874421604873e-06,
      "loss": 0.5736,
      "step": 13081
    },
    {
      "epoch": 1.17,
      "grad_norm": 6.53600547668058,
      "learning_rate": 3.899169764099199e-06,
      "loss": 0.633,
      "step": 13082
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.5288998685311785,
      "learning_rate": 3.898465129570943e-06,
      "loss": 0.6216,
      "step": 13083
    },
    {
      "epoch": 1.17,
      "grad_norm": 8.481900718823557,
      "learning_rate": 3.897760518034815e-06,
      "loss": 0.582,
      "step": 13084
    },
    {
      "epoch": 1.17,
      "grad_norm": 6.05133806209224,
      "learning_rate": 3.897055929505523e-06,
      "loss": 0.6241,
      "step": 13085
    },
    {
      "epoch": 1.17,
      "grad_norm": 9.436327479474544,
      "learning_rate": 3.896351363997772e-06,
      "loss": 0.622,
      "step": 13086
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.435300030243588,
      "learning_rate": 3.89564682152627e-06,
      "loss": 0.5921,
      "step": 13087
    },
    {
      "epoch": 1.17,
      "grad_norm": 6.363143953399953,
      "learning_rate": 3.894942302105723e-06,
      "loss": 0.6041,
      "step": 13088
    },
    {
      "epoch": 1.17,
      "grad_norm": 4.951531391521561,
      "learning_rate": 3.894237805750834e-06,
      "loss": 0.6322,
      "step": 13089
    },
    {
      "epoch": 1.17,
      "grad_norm": 6.554669069377846,
      "learning_rate": 3.893533332476309e-06,
      "loss": 0.6279,
      "step": 13090
    },
    {
      "epoch": 1.17,
      "grad_norm": 6.814255114066726,
      "learning_rate": 3.892828882296854e-06,
      "loss": 0.6017,
      "step": 13091
    },
    {
      "epoch": 1.17,
      "grad_norm": 8.922956837119923,
      "learning_rate": 3.892124455227171e-06,
      "loss": 0.6165,
      "step": 13092
    },
    {
      "epoch": 1.17,
      "grad_norm": 8.617459965961334,
      "learning_rate": 3.8914200512819626e-06,
      "loss": 0.6223,
      "step": 13093
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.203476049972661,
      "learning_rate": 3.8907156704759344e-06,
      "loss": 0.66,
      "step": 13094
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.83257745916291,
      "learning_rate": 3.8900113128237865e-06,
      "loss": 0.5459,
      "step": 13095
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.961540856629339,
      "learning_rate": 3.889306978340223e-06,
      "loss": 0.5427,
      "step": 13096
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.300691447045469,
      "learning_rate": 3.888602667039943e-06,
      "loss": 0.6251,
      "step": 13097
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.727507634393755,
      "learning_rate": 3.887898378937649e-06,
      "loss": 0.6074,
      "step": 13098
    },
    {
      "epoch": 1.17,
      "grad_norm": 4.990901270673559,
      "learning_rate": 3.88719411404804e-06,
      "loss": 0.5976,
      "step": 13099
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.106082400185761,
      "learning_rate": 3.8864898723858155e-06,
      "loss": 0.5682,
      "step": 13100
    },
    {
      "epoch": 1.17,
      "grad_norm": 4.270537214662647,
      "learning_rate": 3.885785653965678e-06,
      "loss": 0.6049,
      "step": 13101
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.003440328559636,
      "learning_rate": 3.885081458802325e-06,
      "loss": 0.621,
      "step": 13102
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.879410440772941,
      "learning_rate": 3.884377286910454e-06,
      "loss": 0.5902,
      "step": 13103
    },
    {
      "epoch": 1.17,
      "grad_norm": 6.366635983973927,
      "learning_rate": 3.883673138304765e-06,
      "loss": 0.5961,
      "step": 13104
    },
    {
      "epoch": 1.17,
      "grad_norm": 6.620583603027561,
      "learning_rate": 3.882969012999954e-06,
      "loss": 0.6036,
      "step": 13105
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.603844786210148,
      "learning_rate": 3.882264911010719e-06,
      "loss": 0.5664,
      "step": 13106
    },
    {
      "epoch": 1.17,
      "grad_norm": 6.845710719703136,
      "learning_rate": 3.8815608323517555e-06,
      "loss": 0.564,
      "step": 13107
    },
    {
      "epoch": 1.17,
      "grad_norm": 6.531370222691904,
      "learning_rate": 3.8808567770377605e-06,
      "loss": 0.5582,
      "step": 13108
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.145407624095916,
      "learning_rate": 3.88015274508343e-06,
      "loss": 0.6039,
      "step": 13109
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.148181286168265,
      "learning_rate": 3.879448736503458e-06,
      "loss": 0.6172,
      "step": 13110
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.755471130672694,
      "learning_rate": 3.878744751312542e-06,
      "loss": 0.5624,
      "step": 13111
    },
    {
      "epoch": 1.17,
      "grad_norm": 6.460231436370971,
      "learning_rate": 3.87804078952537e-06,
      "loss": 0.5787,
      "step": 13112
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.4518605952844705,
      "learning_rate": 3.877336851156642e-06,
      "loss": 0.6139,
      "step": 13113
    },
    {
      "epoch": 1.17,
      "grad_norm": 6.261750079520075,
      "learning_rate": 3.876632936221047e-06,
      "loss": 0.5859,
      "step": 13114
    },
    {
      "epoch": 1.17,
      "grad_norm": 9.406566212529814,
      "learning_rate": 3.87592904473328e-06,
      "loss": 0.7079,
      "step": 13115
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.751445276641179,
      "learning_rate": 3.875225176708033e-06,
      "loss": 0.59,
      "step": 13116
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.7898044581478505,
      "learning_rate": 3.874521332159996e-06,
      "loss": 0.5909,
      "step": 13117
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.799519186078594,
      "learning_rate": 3.873817511103862e-06,
      "loss": 0.621,
      "step": 13118
    },
    {
      "epoch": 1.17,
      "grad_norm": 8.053185341928462,
      "learning_rate": 3.8731137135543216e-06,
      "loss": 0.5938,
      "step": 13119
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.749671556629064,
      "learning_rate": 3.8724099395260635e-06,
      "loss": 0.611,
      "step": 13120
    },
    {
      "epoch": 1.17,
      "grad_norm": 6.109910660447366,
      "learning_rate": 3.871706189033779e-06,
      "loss": 0.5638,
      "step": 13121
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.720315338068922,
      "learning_rate": 3.8710024620921586e-06,
      "loss": 0.5806,
      "step": 13122
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.689276258460644,
      "learning_rate": 3.870298758715888e-06,
      "loss": 0.5627,
      "step": 13123
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.492206940913018,
      "learning_rate": 3.869595078919655e-06,
      "loss": 0.6511,
      "step": 13124
    },
    {
      "epoch": 1.17,
      "grad_norm": 6.316523620767012,
      "learning_rate": 3.8688914227181534e-06,
      "loss": 0.6109,
      "step": 13125
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.502326252774817,
      "learning_rate": 3.868187790126064e-06,
      "loss": 0.6025,
      "step": 13126
    },
    {
      "epoch": 1.17,
      "grad_norm": 8.12602342859233,
      "learning_rate": 3.867484181158076e-06,
      "loss": 0.5605,
      "step": 13127
    },
    {
      "epoch": 1.17,
      "grad_norm": 6.367728858075991,
      "learning_rate": 3.866780595828874e-06,
      "loss": 0.6237,
      "step": 13128
    },
    {
      "epoch": 1.17,
      "grad_norm": 8.67963829277927,
      "learning_rate": 3.866077034153147e-06,
      "loss": 0.6456,
      "step": 13129
    },
    {
      "epoch": 1.17,
      "grad_norm": 8.431035422706968,
      "learning_rate": 3.8653734961455786e-06,
      "loss": 0.6159,
      "step": 13130
    },
    {
      "epoch": 1.17,
      "grad_norm": 8.049136954745777,
      "learning_rate": 3.864669981820853e-06,
      "loss": 0.6428,
      "step": 13131
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.05467948109974,
      "learning_rate": 3.863966491193656e-06,
      "loss": 0.576,
      "step": 13132
    },
    {
      "epoch": 1.17,
      "grad_norm": 8.597968310594995,
      "learning_rate": 3.863263024278669e-06,
      "loss": 0.6204,
      "step": 13133
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.499246525751215,
      "learning_rate": 3.862559581090578e-06,
      "loss": 0.6086,
      "step": 13134
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.325690905350083,
      "learning_rate": 3.861856161644066e-06,
      "loss": 0.6034,
      "step": 13135
    },
    {
      "epoch": 1.17,
      "grad_norm": 8.523044288625295,
      "learning_rate": 3.8611527659538125e-06,
      "loss": 0.681,
      "step": 13136
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.853815249076578,
      "learning_rate": 3.860449394034501e-06,
      "loss": 0.5793,
      "step": 13137
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.609702290793866,
      "learning_rate": 3.859746045900814e-06,
      "loss": 0.5884,
      "step": 13138
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.236393574553493,
      "learning_rate": 3.859042721567432e-06,
      "loss": 0.6753,
      "step": 13139
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.782241058988325,
      "learning_rate": 3.858339421049033e-06,
      "loss": 0.6076,
      "step": 13140
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.019723673610625,
      "learning_rate": 3.857636144360298e-06,
      "loss": 0.6538,
      "step": 13141
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.3155479605752305,
      "learning_rate": 3.856932891515908e-06,
      "loss": 0.6119,
      "step": 13142
    },
    {
      "epoch": 1.17,
      "grad_norm": 6.172754714235043,
      "learning_rate": 3.856229662530539e-06,
      "loss": 0.604,
      "step": 13143
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.3300155384264265,
      "learning_rate": 3.855526457418871e-06,
      "loss": 0.6136,
      "step": 13144
    },
    {
      "epoch": 1.17,
      "grad_norm": 9.474951542008082,
      "learning_rate": 3.854823276195584e-06,
      "loss": 0.6301,
      "step": 13145
    },
    {
      "epoch": 1.17,
      "grad_norm": 8.618798306147506,
      "learning_rate": 3.8541201188753505e-06,
      "loss": 0.5933,
      "step": 13146
    },
    {
      "epoch": 1.17,
      "grad_norm": 6.301752541749486,
      "learning_rate": 3.853416985472851e-06,
      "loss": 0.57,
      "step": 13147
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.194995784976399,
      "learning_rate": 3.8527138760027625e-06,
      "loss": 0.566,
      "step": 13148
    },
    {
      "epoch": 1.17,
      "grad_norm": 4.782089503732792,
      "learning_rate": 3.852010790479758e-06,
      "loss": 0.5894,
      "step": 13149
    },
    {
      "epoch": 1.17,
      "grad_norm": 6.550366673635147,
      "learning_rate": 3.851307728918513e-06,
      "loss": 0.5854,
      "step": 13150
    },
    {
      "epoch": 1.17,
      "grad_norm": 8.009008709232404,
      "learning_rate": 3.850604691333707e-06,
      "loss": 0.5922,
      "step": 13151
    },
    {
      "epoch": 1.17,
      "grad_norm": 6.777689459605355,
      "learning_rate": 3.849901677740008e-06,
      "loss": 0.6091,
      "step": 13152
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.639235064820066,
      "learning_rate": 3.849198688152093e-06,
      "loss": 0.6037,
      "step": 13153
    },
    {
      "epoch": 1.17,
      "grad_norm": 3.7338570093759094,
      "learning_rate": 3.848495722584637e-06,
      "loss": 0.5773,
      "step": 13154
    },
    {
      "epoch": 1.17,
      "grad_norm": 4.932110868013976,
      "learning_rate": 3.847792781052312e-06,
      "loss": 0.6116,
      "step": 13155
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.483975580918394,
      "learning_rate": 3.8470898635697865e-06,
      "loss": 0.6359,
      "step": 13156
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.135931707829017,
      "learning_rate": 3.846386970151735e-06,
      "loss": 0.5768,
      "step": 13157
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.171791283464882,
      "learning_rate": 3.845684100812831e-06,
      "loss": 0.6024,
      "step": 13158
    },
    {
      "epoch": 1.17,
      "grad_norm": 4.77972339164141,
      "learning_rate": 3.844981255567742e-06,
      "loss": 0.5674,
      "step": 13159
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.223079184455958,
      "learning_rate": 3.844278434431141e-06,
      "loss": 0.576,
      "step": 13160
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.041031677132561,
      "learning_rate": 3.843575637417696e-06,
      "loss": 0.6083,
      "step": 13161
    },
    {
      "epoch": 1.17,
      "grad_norm": 8.280997444682933,
      "learning_rate": 3.842872864542077e-06,
      "loss": 0.7025,
      "step": 13162
    },
    {
      "epoch": 1.17,
      "grad_norm": 4.944981007285902,
      "learning_rate": 3.8421701158189525e-06,
      "loss": 0.6304,
      "step": 13163
    },
    {
      "epoch": 1.17,
      "grad_norm": 6.996941525027087,
      "learning_rate": 3.841467391262993e-06,
      "loss": 0.6453,
      "step": 13164
    },
    {
      "epoch": 1.17,
      "grad_norm": 8.057070001192715,
      "learning_rate": 3.840764690888863e-06,
      "loss": 0.5732,
      "step": 13165
    },
    {
      "epoch": 1.17,
      "grad_norm": 6.302275058954713,
      "learning_rate": 3.840062014711232e-06,
      "loss": 0.6284,
      "step": 13166
    },
    {
      "epoch": 1.17,
      "grad_norm": 6.203503656619742,
      "learning_rate": 3.839359362744766e-06,
      "loss": 0.5412,
      "step": 13167
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.094757247561121,
      "learning_rate": 3.838656735004134e-06,
      "loss": 0.6401,
      "step": 13168
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.650541856092161,
      "learning_rate": 3.837954131503996e-06,
      "loss": 0.6458,
      "step": 13169
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.754594358883974,
      "learning_rate": 3.837251552259021e-06,
      "loss": 0.5263,
      "step": 13170
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.265323803836049,
      "learning_rate": 3.836548997283876e-06,
      "loss": 0.6305,
      "step": 13171
    },
    {
      "epoch": 1.18,
      "grad_norm": 4.948198077878267,
      "learning_rate": 3.835846466593219e-06,
      "loss": 0.6037,
      "step": 13172
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.945206844423957,
      "learning_rate": 3.835143960201719e-06,
      "loss": 0.6175,
      "step": 13173
    },
    {
      "epoch": 1.18,
      "grad_norm": 7.245099390056637,
      "learning_rate": 3.8344414781240395e-06,
      "loss": 0.582,
      "step": 13174
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.28294939504522,
      "learning_rate": 3.83373902037484e-06,
      "loss": 0.5542,
      "step": 13175
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.417907208507796,
      "learning_rate": 3.8330365869687845e-06,
      "loss": 0.6689,
      "step": 13176
    },
    {
      "epoch": 1.18,
      "grad_norm": 8.546569193457564,
      "learning_rate": 3.832334177920535e-06,
      "loss": 0.684,
      "step": 13177
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.7596159689019615,
      "learning_rate": 3.831631793244752e-06,
      "loss": 0.5488,
      "step": 13178
    },
    {
      "epoch": 1.18,
      "grad_norm": 9.301007571183177,
      "learning_rate": 3.830929432956097e-06,
      "loss": 0.5768,
      "step": 13179
    },
    {
      "epoch": 1.18,
      "grad_norm": 7.396348941705788,
      "learning_rate": 3.83022709706923e-06,
      "loss": 0.578,
      "step": 13180
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.018489918179628,
      "learning_rate": 3.829524785598811e-06,
      "loss": 0.5622,
      "step": 13181
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.586007370308069,
      "learning_rate": 3.828822498559498e-06,
      "loss": 0.6688,
      "step": 13182
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.608359072932163,
      "learning_rate": 3.828120235965952e-06,
      "loss": 0.6159,
      "step": 13183
    },
    {
      "epoch": 1.18,
      "grad_norm": 7.426969731812003,
      "learning_rate": 3.827417997832828e-06,
      "loss": 0.6029,
      "step": 13184
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.799087998865868,
      "learning_rate": 3.826715784174786e-06,
      "loss": 0.6191,
      "step": 13185
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.530763546105602,
      "learning_rate": 3.826013595006482e-06,
      "loss": 0.602,
      "step": 13186
    },
    {
      "epoch": 1.18,
      "grad_norm": 7.612883333465717,
      "learning_rate": 3.825311430342575e-06,
      "loss": 0.6488,
      "step": 13187
    },
    {
      "epoch": 1.18,
      "grad_norm": 7.42220551112495,
      "learning_rate": 3.824609290197716e-06,
      "loss": 0.6287,
      "step": 13188
    },
    {
      "epoch": 1.18,
      "grad_norm": 8.127347148017519,
      "learning_rate": 3.823907174586568e-06,
      "loss": 0.6506,
      "step": 13189
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.954993209563043,
      "learning_rate": 3.823205083523779e-06,
      "loss": 0.5899,
      "step": 13190
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.179945254715884,
      "learning_rate": 3.822503017024007e-06,
      "loss": 0.5687,
      "step": 13191
    },
    {
      "epoch": 1.18,
      "grad_norm": 7.351674509100959,
      "learning_rate": 3.821800975101908e-06,
      "loss": 0.6416,
      "step": 13192
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.393741220404368,
      "learning_rate": 3.821098957772132e-06,
      "loss": 0.611,
      "step": 13193
    },
    {
      "epoch": 1.18,
      "grad_norm": 8.647179031731183,
      "learning_rate": 3.820396965049332e-06,
      "loss": 0.5695,
      "step": 13194
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.281014566765714,
      "learning_rate": 3.8196949969481645e-06,
      "loss": 0.6073,
      "step": 13195
    },
    {
      "epoch": 1.18,
      "grad_norm": 4.537891772787161,
      "learning_rate": 3.818993053483278e-06,
      "loss": 0.5313,
      "step": 13196
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.590657468638105,
      "learning_rate": 3.818291134669328e-06,
      "loss": 0.6193,
      "step": 13197
    },
    {
      "epoch": 1.18,
      "grad_norm": 10.146222918283696,
      "learning_rate": 3.817589240520959e-06,
      "loss": 0.5697,
      "step": 13198
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.663575967152685,
      "learning_rate": 3.816887371052827e-06,
      "loss": 0.5568,
      "step": 13199
    },
    {
      "epoch": 1.18,
      "grad_norm": 7.361293799322229,
      "learning_rate": 3.816185526279579e-06,
      "loss": 0.5968,
      "step": 13200
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.719444284799625,
      "learning_rate": 3.815483706215866e-06,
      "loss": 0.6294,
      "step": 13201
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.347761300437313,
      "learning_rate": 3.814781910876337e-06,
      "loss": 0.627,
      "step": 13202
    },
    {
      "epoch": 1.18,
      "grad_norm": 11.462790234371068,
      "learning_rate": 3.814080140275639e-06,
      "loss": 0.657,
      "step": 13203
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.0866315872042325,
      "learning_rate": 3.8133783944284208e-06,
      "loss": 0.6257,
      "step": 13204
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.43256800342738,
      "learning_rate": 3.8126766733493315e-06,
      "loss": 0.6821,
      "step": 13205
    },
    {
      "epoch": 1.18,
      "grad_norm": 7.112844067881695,
      "learning_rate": 3.8119749770530156e-06,
      "loss": 0.6339,
      "step": 13206
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.310850292709379,
      "learning_rate": 3.81127330555412e-06,
      "loss": 0.5606,
      "step": 13207
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.125483434917712,
      "learning_rate": 3.810571658867292e-06,
      "loss": 0.6158,
      "step": 13208
    },
    {
      "epoch": 1.18,
      "grad_norm": 8.040806908805983,
      "learning_rate": 3.8098700370071746e-06,
      "loss": 0.6371,
      "step": 13209
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.0604166184261254,
      "learning_rate": 3.8091684399884145e-06,
      "loss": 0.5755,
      "step": 13210
    },
    {
      "epoch": 1.18,
      "grad_norm": 7.413286094195066,
      "learning_rate": 3.8084668678256566e-06,
      "loss": 0.6647,
      "step": 13211
    },
    {
      "epoch": 1.18,
      "grad_norm": 7.391750070465308,
      "learning_rate": 3.8077653205335424e-06,
      "loss": 0.6458,
      "step": 13212
    },
    {
      "epoch": 1.18,
      "grad_norm": 9.11981539978656,
      "learning_rate": 3.8070637981267154e-06,
      "loss": 0.6247,
      "step": 13213
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.462681142292151,
      "learning_rate": 3.806362300619819e-06,
      "loss": 0.6049,
      "step": 13214
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.399418519794362,
      "learning_rate": 3.805660828027497e-06,
      "loss": 0.5698,
      "step": 13215
    },
    {
      "epoch": 1.18,
      "grad_norm": 7.748295216649823,
      "learning_rate": 3.804959380364388e-06,
      "loss": 0.5842,
      "step": 13216
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.884095179822025,
      "learning_rate": 3.8042579576451354e-06,
      "loss": 0.6035,
      "step": 13217
    },
    {
      "epoch": 1.18,
      "grad_norm": 7.505648163349529,
      "learning_rate": 3.8035565598843803e-06,
      "loss": 0.602,
      "step": 13218
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.319434379300883,
      "learning_rate": 3.8028551870967606e-06,
      "loss": 0.6509,
      "step": 13219
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.375517202565116,
      "learning_rate": 3.802153839296917e-06,
      "loss": 0.5999,
      "step": 13220
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.949610461389127,
      "learning_rate": 3.8014525164994903e-06,
      "loss": 0.5484,
      "step": 13221
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.048975103510584,
      "learning_rate": 3.8007512187191165e-06,
      "loss": 0.6512,
      "step": 13222
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.508492638069277,
      "learning_rate": 3.8000499459704345e-06,
      "loss": 0.5734,
      "step": 13223
    },
    {
      "epoch": 1.18,
      "grad_norm": 7.060789261054142,
      "learning_rate": 3.7993486982680826e-06,
      "loss": 0.5722,
      "step": 13224
    },
    {
      "epoch": 1.18,
      "grad_norm": 7.648121968998392,
      "learning_rate": 3.7986474756266967e-06,
      "loss": 0.6652,
      "step": 13225
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.653296374801162,
      "learning_rate": 3.7979462780609157e-06,
      "loss": 0.6165,
      "step": 13226
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.594883576088282,
      "learning_rate": 3.797245105585372e-06,
      "loss": 0.5927,
      "step": 13227
    },
    {
      "epoch": 1.18,
      "grad_norm": 8.01665206685435,
      "learning_rate": 3.7965439582147034e-06,
      "loss": 0.56,
      "step": 13228
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.391254319164415,
      "learning_rate": 3.795842835963543e-06,
      "loss": 0.6248,
      "step": 13229
    },
    {
      "epoch": 1.18,
      "grad_norm": 8.88211341795923,
      "learning_rate": 3.7951417388465267e-06,
      "loss": 0.5783,
      "step": 13230
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.566379749638653,
      "learning_rate": 3.7944406668782884e-06,
      "loss": 0.6597,
      "step": 13231
    },
    {
      "epoch": 1.18,
      "grad_norm": 8.40596420084098,
      "learning_rate": 3.793739620073461e-06,
      "loss": 0.646,
      "step": 13232
    },
    {
      "epoch": 1.18,
      "grad_norm": 4.501862450967066,
      "learning_rate": 3.7930385984466765e-06,
      "loss": 0.5636,
      "step": 13233
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.100769400178658,
      "learning_rate": 3.7923376020125695e-06,
      "loss": 0.578,
      "step": 13234
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.910944398082639,
      "learning_rate": 3.7916366307857695e-06,
      "loss": 0.5889,
      "step": 13235
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.385611246147002,
      "learning_rate": 3.7909356847809083e-06,
      "loss": 0.5772,
      "step": 13236
    },
    {
      "epoch": 1.18,
      "grad_norm": 8.317599667652292,
      "learning_rate": 3.790234764012618e-06,
      "loss": 0.5768,
      "step": 13237
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.338641447885498,
      "learning_rate": 3.7895338684955265e-06,
      "loss": 0.6223,
      "step": 13238
    },
    {
      "epoch": 1.18,
      "grad_norm": 8.14470365668281,
      "learning_rate": 3.7888329982442645e-06,
      "loss": 0.5897,
      "step": 13239
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.535561989130127,
      "learning_rate": 3.788132153273464e-06,
      "loss": 0.6137,
      "step": 13240
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.440169201279031,
      "learning_rate": 3.7874313335977486e-06,
      "loss": 0.6441,
      "step": 13241
    },
    {
      "epoch": 1.18,
      "grad_norm": 7.841297096789114,
      "learning_rate": 3.7867305392317487e-06,
      "loss": 0.6267,
      "step": 13242
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.940754531040744,
      "learning_rate": 3.786029770190091e-06,
      "loss": 0.5876,
      "step": 13243
    },
    {
      "epoch": 1.18,
      "grad_norm": 8.156419852464348,
      "learning_rate": 3.785329026487405e-06,
      "loss": 0.6463,
      "step": 13244
    },
    {
      "epoch": 1.18,
      "grad_norm": 9.980607311821721,
      "learning_rate": 3.784628308138314e-06,
      "loss": 0.6243,
      "step": 13245
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.602694513251037,
      "learning_rate": 3.7839276151574454e-06,
      "loss": 0.6361,
      "step": 13246
    },
    {
      "epoch": 1.18,
      "grad_norm": 8.576985364428046,
      "learning_rate": 3.7832269475594264e-06,
      "loss": 0.5945,
      "step": 13247
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.664315858664254,
      "learning_rate": 3.7825263053588784e-06,
      "loss": 0.5689,
      "step": 13248
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.5470681706782825,
      "learning_rate": 3.781825688570428e-06,
      "loss": 0.6106,
      "step": 13249
    },
    {
      "epoch": 1.18,
      "grad_norm": 4.896469984249298,
      "learning_rate": 3.781125097208699e-06,
      "loss": 0.5564,
      "step": 13250
    },
    {
      "epoch": 1.18,
      "grad_norm": 8.240674168278007,
      "learning_rate": 3.7804245312883143e-06,
      "loss": 0.6658,
      "step": 13251
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.7718166489550695,
      "learning_rate": 3.779723990823896e-06,
      "loss": 0.5789,
      "step": 13252
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.989728306379843,
      "learning_rate": 3.779023475830068e-06,
      "loss": 0.7092,
      "step": 13253
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.500942951599489,
      "learning_rate": 3.7783229863214503e-06,
      "loss": 0.605,
      "step": 13254
    },
    {
      "epoch": 1.18,
      "grad_norm": 7.666505927375248,
      "learning_rate": 3.7776225223126668e-06,
      "loss": 0.5962,
      "step": 13255
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.40636839716343,
      "learning_rate": 3.776922083818334e-06,
      "loss": 0.532,
      "step": 13256
    },
    {
      "epoch": 1.18,
      "grad_norm": 7.630788685669185,
      "learning_rate": 3.7762216708530752e-06,
      "loss": 0.6458,
      "step": 13257
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.035297666169042,
      "learning_rate": 3.7755212834315086e-06,
      "loss": 0.5701,
      "step": 13258
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.378867132589014,
      "learning_rate": 3.774820921568253e-06,
      "loss": 0.6499,
      "step": 13259
    },
    {
      "epoch": 1.18,
      "grad_norm": 7.958589691222815,
      "learning_rate": 3.774120585277929e-06,
      "loss": 0.6083,
      "step": 13260
    },
    {
      "epoch": 1.18,
      "grad_norm": 8.392173707471922,
      "learning_rate": 3.7734202745751513e-06,
      "loss": 0.6344,
      "step": 13261
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.92480870133496,
      "learning_rate": 3.7727199894745403e-06,
      "loss": 0.6006,
      "step": 13262
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.267763790569879,
      "learning_rate": 3.7720197299907125e-06,
      "loss": 0.5968,
      "step": 13263
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.060929570616743,
      "learning_rate": 3.771319496138282e-06,
      "loss": 0.6187,
      "step": 13264
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.104468331594591,
      "learning_rate": 3.7706192879318677e-06,
      "loss": 0.5615,
      "step": 13265
    },
    {
      "epoch": 1.18,
      "grad_norm": 7.191561124696481,
      "learning_rate": 3.769919105386082e-06,
      "loss": 0.6241,
      "step": 13266
    },
    {
      "epoch": 1.18,
      "grad_norm": 8.657674235112134,
      "learning_rate": 3.769218948515542e-06,
      "loss": 0.6335,
      "step": 13267
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.6828150271855735,
      "learning_rate": 3.7685188173348617e-06,
      "loss": 0.5148,
      "step": 13268
    },
    {
      "epoch": 1.18,
      "grad_norm": 4.815823582785311,
      "learning_rate": 3.767818711858654e-06,
      "loss": 0.5578,
      "step": 13269
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.931385042715725,
      "learning_rate": 3.767118632101533e-06,
      "loss": 0.5911,
      "step": 13270
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.36604679466121,
      "learning_rate": 3.7664185780781086e-06,
      "loss": 0.5914,
      "step": 13271
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.526798259526903,
      "learning_rate": 3.7657185498029953e-06,
      "loss": 0.5897,
      "step": 13272
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.087691965343447,
      "learning_rate": 3.765018547290805e-06,
      "loss": 0.5558,
      "step": 13273
    },
    {
      "epoch": 1.18,
      "grad_norm": 7.596228389798004,
      "learning_rate": 3.7643185705561474e-06,
      "loss": 0.6214,
      "step": 13274
    },
    {
      "epoch": 1.18,
      "grad_norm": 4.946643460316142,
      "learning_rate": 3.7636186196136344e-06,
      "loss": 0.6083,
      "step": 13275
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.347269396300193,
      "learning_rate": 3.762918694477874e-06,
      "loss": 0.5773,
      "step": 13276
    },
    {
      "epoch": 1.18,
      "grad_norm": 7.4929287401877485,
      "learning_rate": 3.762218795163477e-06,
      "loss": 0.6728,
      "step": 13277
    },
    {
      "epoch": 1.18,
      "grad_norm": 4.625121906964087,
      "learning_rate": 3.761518921685052e-06,
      "loss": 0.5948,
      "step": 13278
    },
    {
      "epoch": 1.18,
      "grad_norm": 8.412082163470508,
      "learning_rate": 3.760819074057207e-06,
      "loss": 0.6219,
      "step": 13279
    },
    {
      "epoch": 1.18,
      "grad_norm": 7.985843124683444,
      "learning_rate": 3.76011925229455e-06,
      "loss": 0.566,
      "step": 13280
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.548101312657256,
      "learning_rate": 3.75941945641169e-06,
      "loss": 0.6511,
      "step": 13281
    },
    {
      "epoch": 1.18,
      "grad_norm": 7.240070132273118,
      "learning_rate": 3.75871968642323e-06,
      "loss": 0.5319,
      "step": 13282
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.667293647052264,
      "learning_rate": 3.7580199423437785e-06,
      "loss": 0.6617,
      "step": 13283
    },
    {
      "epoch": 1.19,
      "grad_norm": 8.837019475422412,
      "learning_rate": 3.757320224187943e-06,
      "loss": 0.6347,
      "step": 13284
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.0935397922269985,
      "learning_rate": 3.756620531970325e-06,
      "loss": 0.6361,
      "step": 13285
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.115331221320091,
      "learning_rate": 3.7559208657055303e-06,
      "loss": 0.6281,
      "step": 13286
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.837845988772841,
      "learning_rate": 3.755221225408162e-06,
      "loss": 0.6325,
      "step": 13287
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.316747432985996,
      "learning_rate": 3.7545216110928263e-06,
      "loss": 0.6276,
      "step": 13288
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.199342894312467,
      "learning_rate": 3.753822022774123e-06,
      "loss": 0.5602,
      "step": 13289
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.9296696470862615,
      "learning_rate": 3.753122460466656e-06,
      "loss": 0.5909,
      "step": 13290
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.3088975727473064,
      "learning_rate": 3.7524229241850284e-06,
      "loss": 0.5641,
      "step": 13291
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.414238619615227,
      "learning_rate": 3.7517234139438384e-06,
      "loss": 0.6554,
      "step": 13292
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.841049319880134,
      "learning_rate": 3.751023929757689e-06,
      "loss": 0.5834,
      "step": 13293
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.010294441135963,
      "learning_rate": 3.750324471641181e-06,
      "loss": 0.639,
      "step": 13294
    },
    {
      "epoch": 1.19,
      "grad_norm": 9.229695815841394,
      "learning_rate": 3.749625039608911e-06,
      "loss": 0.635,
      "step": 13295
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.540344754532188,
      "learning_rate": 3.7489256336754808e-06,
      "loss": 0.5727,
      "step": 13296
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.151932074001832,
      "learning_rate": 3.748226253855489e-06,
      "loss": 0.6235,
      "step": 13297
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.168453831367233,
      "learning_rate": 3.747526900163534e-06,
      "loss": 0.5883,
      "step": 13298
    },
    {
      "epoch": 1.19,
      "grad_norm": 4.935356159178155,
      "learning_rate": 3.74682757261421e-06,
      "loss": 0.5572,
      "step": 13299
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.239923943658096,
      "learning_rate": 3.746128271222116e-06,
      "loss": 0.6167,
      "step": 13300
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.423374920120001,
      "learning_rate": 3.74542899600185e-06,
      "loss": 0.6183,
      "step": 13301
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.1237377099317225,
      "learning_rate": 3.7447297469680037e-06,
      "loss": 0.6318,
      "step": 13302
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.387916506163475,
      "learning_rate": 3.744030524135176e-06,
      "loss": 0.5456,
      "step": 13303
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.773770048278127,
      "learning_rate": 3.7433313275179615e-06,
      "loss": 0.5879,
      "step": 13304
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.642016470015299,
      "learning_rate": 3.742632157130952e-06,
      "loss": 0.5694,
      "step": 13305
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.599297983853513,
      "learning_rate": 3.7419330129887433e-06,
      "loss": 0.6193,
      "step": 13306
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.6446573905629425,
      "learning_rate": 3.741233895105929e-06,
      "loss": 0.65,
      "step": 13307
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.210638884693305,
      "learning_rate": 3.740534803497099e-06,
      "loss": 0.6439,
      "step": 13308
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.381276249951244,
      "learning_rate": 3.739835738176847e-06,
      "loss": 0.5418,
      "step": 13309
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.767867866686351,
      "learning_rate": 3.7391366991597645e-06,
      "loss": 0.5676,
      "step": 13310
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.296950697007662,
      "learning_rate": 3.7384376864604422e-06,
      "loss": 0.5709,
      "step": 13311
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.285868834222443,
      "learning_rate": 3.7377387000934718e-06,
      "loss": 0.5684,
      "step": 13312
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.763670166663195,
      "learning_rate": 3.73703974007344e-06,
      "loss": 0.6617,
      "step": 13313
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.537799752909669,
      "learning_rate": 3.7363408064149387e-06,
      "loss": 0.6376,
      "step": 13314
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.849846124156815,
      "learning_rate": 3.735641899132555e-06,
      "loss": 0.6191,
      "step": 13315
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.808325279229659,
      "learning_rate": 3.734943018240877e-06,
      "loss": 0.5652,
      "step": 13316
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.392338458294858,
      "learning_rate": 3.7342441637544946e-06,
      "loss": 0.6199,
      "step": 13317
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.812279635082237,
      "learning_rate": 3.7335453356879925e-06,
      "loss": 0.6859,
      "step": 13318
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.402024635774427,
      "learning_rate": 3.7328465340559572e-06,
      "loss": 0.5737,
      "step": 13319
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.44177630983827,
      "learning_rate": 3.732147758872977e-06,
      "loss": 0.6287,
      "step": 13320
    },
    {
      "epoch": 1.19,
      "grad_norm": 8.34228154208312,
      "learning_rate": 3.731449010153635e-06,
      "loss": 0.6008,
      "step": 13321
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.577441400818133,
      "learning_rate": 3.7307502879125168e-06,
      "loss": 0.6183,
      "step": 13322
    },
    {
      "epoch": 1.19,
      "grad_norm": 9.274844485097661,
      "learning_rate": 3.7300515921642077e-06,
      "loss": 0.5931,
      "step": 13323
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.333971070286751,
      "learning_rate": 3.7293529229232896e-06,
      "loss": 0.5541,
      "step": 13324
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.617372528403955,
      "learning_rate": 3.7286542802043456e-06,
      "loss": 0.6152,
      "step": 13325
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.898132804161141,
      "learning_rate": 3.7279556640219617e-06,
      "loss": 0.6085,
      "step": 13326
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.557944291987163,
      "learning_rate": 3.727257074390718e-06,
      "loss": 0.6431,
      "step": 13327
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.096262078492384,
      "learning_rate": 3.7265585113251933e-06,
      "loss": 0.58,
      "step": 13328
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.0923032264784815,
      "learning_rate": 3.7258599748399717e-06,
      "loss": 0.596,
      "step": 13329
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.204784179701877,
      "learning_rate": 3.7251614649496327e-06,
      "loss": 0.5687,
      "step": 13330
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.163593417308658,
      "learning_rate": 3.724462981668756e-06,
      "loss": 0.5634,
      "step": 13331
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.216488183742146,
      "learning_rate": 3.723764525011921e-06,
      "loss": 0.6073,
      "step": 13332
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.908128317353288,
      "learning_rate": 3.7230660949937077e-06,
      "loss": 0.5712,
      "step": 13333
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.395384841171362,
      "learning_rate": 3.7223676916286924e-06,
      "loss": 0.5783,
      "step": 13334
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.206625552311317,
      "learning_rate": 3.721669314931453e-06,
      "loss": 0.6437,
      "step": 13335
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.138890824520557,
      "learning_rate": 3.720970964916568e-06,
      "loss": 0.5508,
      "step": 13336
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.907259011999707,
      "learning_rate": 3.720272641598613e-06,
      "loss": 0.5539,
      "step": 13337
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.77308647027385,
      "learning_rate": 3.7195743449921636e-06,
      "loss": 0.5643,
      "step": 13338
    },
    {
      "epoch": 1.19,
      "grad_norm": 8.050993946611047,
      "learning_rate": 3.718876075111797e-06,
      "loss": 0.5897,
      "step": 13339
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.668025934411838,
      "learning_rate": 3.718177831972085e-06,
      "loss": 0.6356,
      "step": 13340
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.101800675397439,
      "learning_rate": 3.7174796155876056e-06,
      "loss": 0.5768,
      "step": 13341
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.0751932016519925,
      "learning_rate": 3.7167814259729296e-06,
      "loss": 0.5779,
      "step": 13342
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.998333013343968,
      "learning_rate": 3.7160832631426313e-06,
      "loss": 0.6689,
      "step": 13343
    },
    {
      "epoch": 1.19,
      "grad_norm": 8.708210994188647,
      "learning_rate": 3.7153851271112823e-06,
      "loss": 0.6379,
      "step": 13344
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.2749219821951066,
      "learning_rate": 3.714687017893456e-06,
      "loss": 0.5954,
      "step": 13345
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.741420807799858,
      "learning_rate": 3.713988935503724e-06,
      "loss": 0.6066,
      "step": 13346
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.792812730450356,
      "learning_rate": 3.7132908799566552e-06,
      "loss": 0.6012,
      "step": 13347
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.50524200313823,
      "learning_rate": 3.712592851266822e-06,
      "loss": 0.64,
      "step": 13348
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.481621204089206,
      "learning_rate": 3.7118948494487943e-06,
      "loss": 0.5935,
      "step": 13349
    },
    {
      "epoch": 1.19,
      "grad_norm": 8.153970854546364,
      "learning_rate": 3.71119687451714e-06,
      "loss": 0.5718,
      "step": 13350
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.368904877323651,
      "learning_rate": 3.7104989264864294e-06,
      "loss": 0.635,
      "step": 13351
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.820112626328267,
      "learning_rate": 3.7098010053712285e-06,
      "loss": 0.6577,
      "step": 13352
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.3518502113369255,
      "learning_rate": 3.7091031111861064e-06,
      "loss": 0.5468,
      "step": 13353
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.760270766985963,
      "learning_rate": 3.7084052439456314e-06,
      "loss": 0.6516,
      "step": 13354
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.118158925536969,
      "learning_rate": 3.707707403664367e-06,
      "loss": 0.5769,
      "step": 13355
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.894813199152525,
      "learning_rate": 3.707009590356882e-06,
      "loss": 0.6032,
      "step": 13356
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.058749299610843,
      "learning_rate": 3.7063118040377387e-06,
      "loss": 0.5395,
      "step": 13357
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.636230658910119,
      "learning_rate": 3.7056140447215038e-06,
      "loss": 0.6086,
      "step": 13358
    },
    {
      "epoch": 1.19,
      "grad_norm": 8.20518466879551,
      "learning_rate": 3.7049163124227404e-06,
      "loss": 0.5478,
      "step": 13359
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.0024708210386075,
      "learning_rate": 3.704218607156013e-06,
      "loss": 0.6487,
      "step": 13360
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.111633604856421,
      "learning_rate": 3.7035209289358853e-06,
      "loss": 0.617,
      "step": 13361
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.001762036576767,
      "learning_rate": 3.702823277776918e-06,
      "loss": 0.6183,
      "step": 13362
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.728208792333243,
      "learning_rate": 3.7021256536936745e-06,
      "loss": 0.5905,
      "step": 13363
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.367934634851974,
      "learning_rate": 3.7014280567007156e-06,
      "loss": 0.6489,
      "step": 13364
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.041847191060068,
      "learning_rate": 3.700730486812602e-06,
      "loss": 0.5555,
      "step": 13365
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.365171179129494,
      "learning_rate": 3.7000329440438943e-06,
      "loss": 0.5594,
      "step": 13366
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.465580709639025,
      "learning_rate": 3.699335428409152e-06,
      "loss": 0.6252,
      "step": 13367
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.212286361538738,
      "learning_rate": 3.698637939922934e-06,
      "loss": 0.6195,
      "step": 13368
    },
    {
      "epoch": 1.19,
      "grad_norm": 4.7696498958472455,
      "learning_rate": 3.6979404785997987e-06,
      "loss": 0.5948,
      "step": 13369
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.689260460015791,
      "learning_rate": 3.697243044454307e-06,
      "loss": 0.5936,
      "step": 13370
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.390039822540484,
      "learning_rate": 3.6965456375010123e-06,
      "loss": 0.6192,
      "step": 13371
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.815751775174619,
      "learning_rate": 3.695848257754472e-06,
      "loss": 0.5975,
      "step": 13372
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.276173778013968,
      "learning_rate": 3.6951509052292434e-06,
      "loss": 0.6367,
      "step": 13373
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.0723634437266245,
      "learning_rate": 3.694453579939883e-06,
      "loss": 0.6596,
      "step": 13374
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.92943420646439,
      "learning_rate": 3.6937562819009443e-06,
      "loss": 0.6453,
      "step": 13375
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.234335758242693,
      "learning_rate": 3.6930590111269816e-06,
      "loss": 0.6025,
      "step": 13376
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.531777454037257,
      "learning_rate": 3.692361767632552e-06,
      "loss": 0.6328,
      "step": 13377
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.712010850284064,
      "learning_rate": 3.6916645514322053e-06,
      "loss": 0.5869,
      "step": 13378
    },
    {
      "epoch": 1.19,
      "grad_norm": 8.359634302580409,
      "learning_rate": 3.690967362540496e-06,
      "loss": 0.6161,
      "step": 13379
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.481288670367188,
      "learning_rate": 3.690270200971977e-06,
      "loss": 0.6286,
      "step": 13380
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.832034131502601,
      "learning_rate": 3.689573066741199e-06,
      "loss": 0.6539,
      "step": 13381
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.684148107185263,
      "learning_rate": 3.688875959862712e-06,
      "loss": 0.6819,
      "step": 13382
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.166003390579808,
      "learning_rate": 3.688178880351071e-06,
      "loss": 0.5682,
      "step": 13383
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.279794586850889,
      "learning_rate": 3.6874818282208202e-06,
      "loss": 0.5893,
      "step": 13384
    },
    {
      "epoch": 1.19,
      "grad_norm": 10.200082078579573,
      "learning_rate": 3.686784803486515e-06,
      "loss": 0.6844,
      "step": 13385
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.4499319426330315,
      "learning_rate": 3.686087806162699e-06,
      "loss": 0.5994,
      "step": 13386
    },
    {
      "epoch": 1.19,
      "grad_norm": 8.381656677468134,
      "learning_rate": 3.6853908362639233e-06,
      "loss": 0.6476,
      "step": 13387
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.363877402522761,
      "learning_rate": 3.6846938938047345e-06,
      "loss": 0.6404,
      "step": 13388
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.012799199197222,
      "learning_rate": 3.68399697879968e-06,
      "loss": 0.5979,
      "step": 13389
    },
    {
      "epoch": 1.19,
      "grad_norm": 9.688515158515365,
      "learning_rate": 3.683300091263307e-06,
      "loss": 0.65,
      "step": 13390
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.340990871851641,
      "learning_rate": 3.682603231210161e-06,
      "loss": 0.5997,
      "step": 13391
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.180246543453739,
      "learning_rate": 3.6819063986547865e-06,
      "loss": 0.5668,
      "step": 13392
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.525451847076877,
      "learning_rate": 3.681209593611731e-06,
      "loss": 0.6236,
      "step": 13393
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.278110805525595,
      "learning_rate": 3.680512816095536e-06,
      "loss": 0.5717,
      "step": 13394
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.0761307915444,
      "learning_rate": 3.679816066120746e-06,
      "loss": 0.6066,
      "step": 13395
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.4382547642093515,
      "learning_rate": 3.6791193437019057e-06,
      "loss": 0.6302,
      "step": 13396
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.026694764826297,
      "learning_rate": 3.6784226488535557e-06,
      "loss": 0.6258,
      "step": 13397
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.583972786627504,
      "learning_rate": 3.6777259815902385e-06,
      "loss": 0.6069,
      "step": 13398
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.410957446317676,
      "learning_rate": 3.677029341926498e-06,
      "loss": 0.6454,
      "step": 13399
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.642764136223984,
      "learning_rate": 3.6763327298768713e-06,
      "loss": 0.5532,
      "step": 13400
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.806724778840881,
      "learning_rate": 3.6756361454558987e-06,
      "loss": 0.5714,
      "step": 13401
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.109153636810562,
      "learning_rate": 3.674939588678122e-06,
      "loss": 0.5934,
      "step": 13402
    },
    {
      "epoch": 1.2,
      "grad_norm": 7.5558213319209235,
      "learning_rate": 3.6742430595580803e-06,
      "loss": 0.5528,
      "step": 13403
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.396656410559189,
      "learning_rate": 3.67354655811031e-06,
      "loss": 0.6158,
      "step": 13404
    },
    {
      "epoch": 1.2,
      "grad_norm": 7.025110289598913,
      "learning_rate": 3.672850084349351e-06,
      "loss": 0.6211,
      "step": 13405
    },
    {
      "epoch": 1.2,
      "grad_norm": 7.631370099735983,
      "learning_rate": 3.672153638289741e-06,
      "loss": 0.5557,
      "step": 13406
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.436376643303969,
      "learning_rate": 3.6714572199460145e-06,
      "loss": 0.6381,
      "step": 13407
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.360879261903947,
      "learning_rate": 3.6707608293327092e-06,
      "loss": 0.6128,
      "step": 13408
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.688349962289531,
      "learning_rate": 3.670064466464362e-06,
      "loss": 0.5485,
      "step": 13409
    },
    {
      "epoch": 1.2,
      "grad_norm": 7.225358542943175,
      "learning_rate": 3.6693681313555053e-06,
      "loss": 0.5611,
      "step": 13410
    },
    {
      "epoch": 1.2,
      "grad_norm": 7.4853367045870005,
      "learning_rate": 3.6686718240206744e-06,
      "loss": 0.6258,
      "step": 13411
    },
    {
      "epoch": 1.2,
      "grad_norm": 7.117383553738129,
      "learning_rate": 3.6679755444744055e-06,
      "loss": 0.5555,
      "step": 13412
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.55056395502896,
      "learning_rate": 3.6672792927312307e-06,
      "loss": 0.562,
      "step": 13413
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.9510342837356225,
      "learning_rate": 3.6665830688056797e-06,
      "loss": 0.5794,
      "step": 13414
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.7470150233605555,
      "learning_rate": 3.665886872712287e-06,
      "loss": 0.5904,
      "step": 13415
    },
    {
      "epoch": 1.2,
      "grad_norm": 7.416530551095121,
      "learning_rate": 3.6651907044655854e-06,
      "loss": 0.6729,
      "step": 13416
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.785884103178055,
      "learning_rate": 3.6644945640801035e-06,
      "loss": 0.6192,
      "step": 13417
    },
    {
      "epoch": 1.2,
      "grad_norm": 8.47907456455891,
      "learning_rate": 3.663798451570373e-06,
      "loss": 0.5795,
      "step": 13418
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.413815094207633,
      "learning_rate": 3.6631023669509245e-06,
      "loss": 0.5701,
      "step": 13419
    },
    {
      "epoch": 1.2,
      "grad_norm": 7.5730509754455015,
      "learning_rate": 3.6624063102362854e-06,
      "loss": 0.63,
      "step": 13420
    },
    {
      "epoch": 1.2,
      "grad_norm": 4.95195520875387,
      "learning_rate": 3.6617102814409856e-06,
      "loss": 0.5744,
      "step": 13421
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.28903149340452,
      "learning_rate": 3.6610142805795534e-06,
      "loss": 0.5378,
      "step": 13422
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.170294841867409,
      "learning_rate": 3.660318307666515e-06,
      "loss": 0.5193,
      "step": 13423
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.1120929242656095,
      "learning_rate": 3.659622362716398e-06,
      "loss": 0.6029,
      "step": 13424
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.2150607636132404,
      "learning_rate": 3.65892644574373e-06,
      "loss": 0.5764,
      "step": 13425
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.839085036099872,
      "learning_rate": 3.658230556763034e-06,
      "loss": 0.6325,
      "step": 13426
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.790711140060222,
      "learning_rate": 3.657534695788838e-06,
      "loss": 0.6223,
      "step": 13427
    },
    {
      "epoch": 1.2,
      "grad_norm": 7.268448662204009,
      "learning_rate": 3.6568388628356654e-06,
      "loss": 0.6811,
      "step": 13428
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.802607309271588,
      "learning_rate": 3.65614305791804e-06,
      "loss": 0.6674,
      "step": 13429
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.553434981697832,
      "learning_rate": 3.655447281050485e-06,
      "loss": 0.5806,
      "step": 13430
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.75191898522017,
      "learning_rate": 3.6547515322475226e-06,
      "loss": 0.5723,
      "step": 13431
    },
    {
      "epoch": 1.2,
      "grad_norm": 8.420756003216937,
      "learning_rate": 3.6540558115236766e-06,
      "loss": 0.6518,
      "step": 13432
    },
    {
      "epoch": 1.2,
      "grad_norm": 4.8719181280495025,
      "learning_rate": 3.6533601188934677e-06,
      "loss": 0.5099,
      "step": 13433
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.219725796661219,
      "learning_rate": 3.652664454371418e-06,
      "loss": 0.6194,
      "step": 13434
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.1649765486890455,
      "learning_rate": 3.651968817972046e-06,
      "loss": 0.6152,
      "step": 13435
    },
    {
      "epoch": 1.2,
      "grad_norm": 7.401132509805776,
      "learning_rate": 3.6512732097098735e-06,
      "loss": 0.6425,
      "step": 13436
    },
    {
      "epoch": 1.2,
      "grad_norm": 7.475634055010808,
      "learning_rate": 3.650577629599419e-06,
      "loss": 0.5959,
      "step": 13437
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.866006466905289,
      "learning_rate": 3.6498820776552012e-06,
      "loss": 0.5129,
      "step": 13438
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.365268849156915,
      "learning_rate": 3.649186553891738e-06,
      "loss": 0.5199,
      "step": 13439
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.852090716109708,
      "learning_rate": 3.648491058323548e-06,
      "loss": 0.6229,
      "step": 13440
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.545140374887758,
      "learning_rate": 3.6477955909651465e-06,
      "loss": 0.5909,
      "step": 13441
    },
    {
      "epoch": 1.2,
      "grad_norm": 7.007464495513034,
      "learning_rate": 3.6471001518310533e-06,
      "loss": 0.68,
      "step": 13442
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.649516078457325,
      "learning_rate": 3.6464047409357793e-06,
      "loss": 0.6019,
      "step": 13443
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.0638617064300915,
      "learning_rate": 3.645709358293843e-06,
      "loss": 0.5948,
      "step": 13444
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.2767983954504505,
      "learning_rate": 3.645014003919757e-06,
      "loss": 0.6186,
      "step": 13445
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.903209155005053,
      "learning_rate": 3.644318677828036e-06,
      "loss": 0.5957,
      "step": 13446
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.0433584869065236,
      "learning_rate": 3.643623380033195e-06,
      "loss": 0.601,
      "step": 13447
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.279335490355251,
      "learning_rate": 3.6429281105497443e-06,
      "loss": 0.6755,
      "step": 13448
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.745026484132621,
      "learning_rate": 3.642232869392197e-06,
      "loss": 0.5892,
      "step": 13449
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.570097558881129,
      "learning_rate": 3.641537656575067e-06,
      "loss": 0.6562,
      "step": 13450
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.316970160286934,
      "learning_rate": 3.6408424721128623e-06,
      "loss": 0.6534,
      "step": 13451
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.953060113527509,
      "learning_rate": 3.640147316020094e-06,
      "loss": 0.6128,
      "step": 13452
    },
    {
      "epoch": 1.2,
      "grad_norm": 9.334902464848572,
      "learning_rate": 3.639452188311274e-06,
      "loss": 0.6058,
      "step": 13453
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.260760494810699,
      "learning_rate": 3.6387570890009083e-06,
      "loss": 0.5569,
      "step": 13454
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.419175287614351,
      "learning_rate": 3.6380620181035074e-06,
      "loss": 0.5966,
      "step": 13455
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.254001382729129,
      "learning_rate": 3.637366975633581e-06,
      "loss": 0.6313,
      "step": 13456
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.887830111327473,
      "learning_rate": 3.6366719616056354e-06,
      "loss": 0.6336,
      "step": 13457
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.246545503275529,
      "learning_rate": 3.6359769760341756e-06,
      "loss": 0.5795,
      "step": 13458
    },
    {
      "epoch": 1.2,
      "grad_norm": 7.609166419548819,
      "learning_rate": 3.6352820189337085e-06,
      "loss": 0.6346,
      "step": 13459
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.886720766770024,
      "learning_rate": 3.6345870903187417e-06,
      "loss": 0.6283,
      "step": 13460
    },
    {
      "epoch": 1.2,
      "grad_norm": 4.312832307825357,
      "learning_rate": 3.6338921902037782e-06,
      "loss": 0.6527,
      "step": 13461
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.186255809462418,
      "learning_rate": 3.633197318603323e-06,
      "loss": 0.5829,
      "step": 13462
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.46061794843479,
      "learning_rate": 3.6325024755318826e-06,
      "loss": 0.5466,
      "step": 13463
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.696618396058396,
      "learning_rate": 3.6318076610039563e-06,
      "loss": 0.6098,
      "step": 13464
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.39099216724389,
      "learning_rate": 3.6311128750340495e-06,
      "loss": 0.5559,
      "step": 13465
    },
    {
      "epoch": 1.2,
      "grad_norm": 7.715127282462917,
      "learning_rate": 3.6304181176366647e-06,
      "loss": 0.6186,
      "step": 13466
    },
    {
      "epoch": 1.2,
      "grad_norm": 7.430671364930992,
      "learning_rate": 3.6297233888263004e-06,
      "loss": 0.5894,
      "step": 13467
    },
    {
      "epoch": 1.2,
      "grad_norm": 11.32433630042538,
      "learning_rate": 3.629028688617461e-06,
      "loss": 0.5864,
      "step": 13468
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.76485883462344,
      "learning_rate": 3.628334017024645e-06,
      "loss": 0.6051,
      "step": 13469
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.862312266734907,
      "learning_rate": 3.6276393740623515e-06,
      "loss": 0.5758,
      "step": 13470
    },
    {
      "epoch": 1.2,
      "grad_norm": 7.084893719507433,
      "learning_rate": 3.6269447597450834e-06,
      "loss": 0.6346,
      "step": 13471
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.740075927238261,
      "learning_rate": 3.6262501740873344e-06,
      "loss": 0.6236,
      "step": 13472
    },
    {
      "epoch": 1.2,
      "grad_norm": 9.290273262851208,
      "learning_rate": 3.6255556171036056e-06,
      "loss": 0.6173,
      "step": 13473
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.171461344662817,
      "learning_rate": 3.624861088808392e-06,
      "loss": 0.5847,
      "step": 13474
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.9268736223860925,
      "learning_rate": 3.6241665892161914e-06,
      "loss": 0.6013,
      "step": 13475
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.4546088228048335,
      "learning_rate": 3.6234721183415016e-06,
      "loss": 0.6107,
      "step": 13476
    },
    {
      "epoch": 1.2,
      "grad_norm": 7.590112483658299,
      "learning_rate": 3.6227776761988158e-06,
      "loss": 0.5984,
      "step": 13477
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.642075570392588,
      "learning_rate": 3.6220832628026303e-06,
      "loss": 0.5921,
      "step": 13478
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.227625016150862,
      "learning_rate": 3.6213888781674396e-06,
      "loss": 0.5815,
      "step": 13479
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.8834600629175275,
      "learning_rate": 3.620694522307736e-06,
      "loss": 0.5707,
      "step": 13480
    },
    {
      "epoch": 1.2,
      "grad_norm": 7.144045281716197,
      "learning_rate": 3.6200001952380138e-06,
      "loss": 0.6746,
      "step": 13481
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.199366796315545,
      "learning_rate": 3.6193058969727657e-06,
      "loss": 0.6035,
      "step": 13482
    },
    {
      "epoch": 1.2,
      "grad_norm": 7.3313862945462995,
      "learning_rate": 3.6186116275264834e-06,
      "loss": 0.643,
      "step": 13483
    },
    {
      "epoch": 1.2,
      "grad_norm": 8.115140724911324,
      "learning_rate": 3.6179173869136575e-06,
      "loss": 0.6173,
      "step": 13484
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.215081643300225,
      "learning_rate": 3.617223175148782e-06,
      "loss": 0.5843,
      "step": 13485
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.394690674045719,
      "learning_rate": 3.616528992246343e-06,
      "loss": 0.5799,
      "step": 13486
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.8027945380135835,
      "learning_rate": 3.61583483822083e-06,
      "loss": 0.5733,
      "step": 13487
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.743858826347586,
      "learning_rate": 3.6151407130867346e-06,
      "loss": 0.5827,
      "step": 13488
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.614220286873898,
      "learning_rate": 3.6144466168585447e-06,
      "loss": 0.578,
      "step": 13489
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.7753686731642615,
      "learning_rate": 3.6137525495507456e-06,
      "loss": 0.5954,
      "step": 13490
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.685945364385989,
      "learning_rate": 3.613058511177827e-06,
      "loss": 0.6542,
      "step": 13491
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.82382789205328,
      "learning_rate": 3.6123645017542754e-06,
      "loss": 0.586,
      "step": 13492
    },
    {
      "epoch": 1.2,
      "grad_norm": 8.757346902782816,
      "learning_rate": 3.6116705212945754e-06,
      "loss": 0.6658,
      "step": 13493
    },
    {
      "epoch": 1.2,
      "grad_norm": 4.6031511035343415,
      "learning_rate": 3.610976569813212e-06,
      "loss": 0.6565,
      "step": 13494
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.679110651127734,
      "learning_rate": 3.610282647324672e-06,
      "loss": 0.5674,
      "step": 13495
    },
    {
      "epoch": 1.2,
      "grad_norm": 7.303850456199928,
      "learning_rate": 3.609588753843438e-06,
      "loss": 0.6317,
      "step": 13496
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.305525666307039,
      "learning_rate": 3.6088948893839938e-06,
      "loss": 0.6372,
      "step": 13497
    },
    {
      "epoch": 1.2,
      "grad_norm": 9.007270545148677,
      "learning_rate": 3.6082010539608224e-06,
      "loss": 0.5913,
      "step": 13498
    },
    {
      "epoch": 1.2,
      "grad_norm": 9.599923266591107,
      "learning_rate": 3.6075072475884053e-06,
      "loss": 0.5965,
      "step": 13499
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.82696728154465,
      "learning_rate": 3.6068134702812275e-06,
      "loss": 0.5871,
      "step": 13500
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.942248415524232,
      "learning_rate": 3.606119722053765e-06,
      "loss": 0.6002,
      "step": 13501
    },
    {
      "epoch": 1.2,
      "grad_norm": 7.554624312448088,
      "learning_rate": 3.605426002920502e-06,
      "loss": 0.5731,
      "step": 13502
    },
    {
      "epoch": 1.2,
      "grad_norm": 7.018830101239312,
      "learning_rate": 3.6047323128959155e-06,
      "loss": 0.6769,
      "step": 13503
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.817969522441525,
      "learning_rate": 3.604038651994487e-06,
      "loss": 0.5928,
      "step": 13504
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.8294227249977855,
      "learning_rate": 3.603345020230696e-06,
      "loss": 0.6481,
      "step": 13505
    },
    {
      "epoch": 1.2,
      "grad_norm": 7.918528355117688,
      "learning_rate": 3.6026514176190173e-06,
      "loss": 0.5975,
      "step": 13506
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.209877336487939,
      "learning_rate": 3.60195784417393e-06,
      "loss": 0.5648,
      "step": 13507
    },
    {
      "epoch": 1.21,
      "grad_norm": 7.821566509572182,
      "learning_rate": 3.6012642999099124e-06,
      "loss": 0.5838,
      "step": 13508
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.34115095906706,
      "learning_rate": 3.6005707848414385e-06,
      "loss": 0.612,
      "step": 13509
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.034771982717397,
      "learning_rate": 3.5998772989829856e-06,
      "loss": 0.5443,
      "step": 13510
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.747182229983236,
      "learning_rate": 3.5991838423490265e-06,
      "loss": 0.6105,
      "step": 13511
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.691849233078796,
      "learning_rate": 3.5984904149540375e-06,
      "loss": 0.5771,
      "step": 13512
    },
    {
      "epoch": 1.21,
      "grad_norm": 4.85816319229172,
      "learning_rate": 3.5977970168124925e-06,
      "loss": 0.565,
      "step": 13513
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.183793716407959,
      "learning_rate": 3.5971036479388645e-06,
      "loss": 0.5822,
      "step": 13514
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.396954914237455,
      "learning_rate": 3.596410308347625e-06,
      "loss": 0.5746,
      "step": 13515
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.446578996356294,
      "learning_rate": 3.595716998053246e-06,
      "loss": 0.5874,
      "step": 13516
    },
    {
      "epoch": 1.21,
      "grad_norm": 4.913065810572012,
      "learning_rate": 3.5950237170701985e-06,
      "loss": 0.6196,
      "step": 13517
    },
    {
      "epoch": 1.21,
      "grad_norm": 7.238654536215153,
      "learning_rate": 3.5943304654129556e-06,
      "loss": 0.5446,
      "step": 13518
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.608655369086084,
      "learning_rate": 3.593637243095985e-06,
      "loss": 0.5951,
      "step": 13519
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.021294730942938,
      "learning_rate": 3.5929440501337574e-06,
      "loss": 0.6996,
      "step": 13520
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.188487757806333,
      "learning_rate": 3.5922508865407414e-06,
      "loss": 0.5199,
      "step": 13521
    },
    {
      "epoch": 1.21,
      "grad_norm": 7.578262845405902,
      "learning_rate": 3.5915577523314047e-06,
      "loss": 0.6462,
      "step": 13522
    },
    {
      "epoch": 1.21,
      "grad_norm": 8.308587444249335,
      "learning_rate": 3.5908646475202163e-06,
      "loss": 0.6338,
      "step": 13523
    },
    {
      "epoch": 1.21,
      "grad_norm": 7.161080599981999,
      "learning_rate": 3.5901715721216417e-06,
      "loss": 0.5935,
      "step": 13524
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.665958829984157,
      "learning_rate": 3.589478526150148e-06,
      "loss": 0.5468,
      "step": 13525
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.247252408884917,
      "learning_rate": 3.588785509620202e-06,
      "loss": 0.5954,
      "step": 13526
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.2605726063989815,
      "learning_rate": 3.588092522546267e-06,
      "loss": 0.6094,
      "step": 13527
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.125745486744999,
      "learning_rate": 3.5873995649428085e-06,
      "loss": 0.6178,
      "step": 13528
    },
    {
      "epoch": 1.21,
      "grad_norm": 7.212053369477381,
      "learning_rate": 3.586706636824293e-06,
      "loss": 0.6391,
      "step": 13529
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.697577093246554,
      "learning_rate": 3.5860137382051795e-06,
      "loss": 0.5617,
      "step": 13530
    },
    {
      "epoch": 1.21,
      "grad_norm": 7.608779831226991,
      "learning_rate": 3.5853208690999325e-06,
      "loss": 0.702,
      "step": 13531
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.111523340904321,
      "learning_rate": 3.584628029523014e-06,
      "loss": 0.6916,
      "step": 13532
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.048511635879818,
      "learning_rate": 3.5839352194888867e-06,
      "loss": 0.5896,
      "step": 13533
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.316130757934675,
      "learning_rate": 3.58324243901201e-06,
      "loss": 0.5785,
      "step": 13534
    },
    {
      "epoch": 1.21,
      "grad_norm": 4.9789709282241486,
      "learning_rate": 3.5825496881068437e-06,
      "loss": 0.6514,
      "step": 13535
    },
    {
      "epoch": 1.21,
      "grad_norm": 4.772108612526293,
      "learning_rate": 3.58185696678785e-06,
      "loss": 0.6201,
      "step": 13536
    },
    {
      "epoch": 1.21,
      "grad_norm": 8.216509841134839,
      "learning_rate": 3.5811642750694857e-06,
      "loss": 0.6363,
      "step": 13537
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.175431395922183,
      "learning_rate": 3.58047161296621e-06,
      "loss": 0.5856,
      "step": 13538
    },
    {
      "epoch": 1.21,
      "grad_norm": 4.845018754886855,
      "learning_rate": 3.5797789804924815e-06,
      "loss": 0.6225,
      "step": 13539
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.245404901505736,
      "learning_rate": 3.5790863776627556e-06,
      "loss": 0.6366,
      "step": 13540
    },
    {
      "epoch": 1.21,
      "grad_norm": 8.480958676570017,
      "learning_rate": 3.5783938044914895e-06,
      "loss": 0.6037,
      "step": 13541
    },
    {
      "epoch": 1.21,
      "grad_norm": 7.58019185631316,
      "learning_rate": 3.5777012609931406e-06,
      "loss": 0.5737,
      "step": 13542
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.847005554219167,
      "learning_rate": 3.5770087471821645e-06,
      "loss": 0.6153,
      "step": 13543
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.090265085342041,
      "learning_rate": 3.576316263073012e-06,
      "loss": 0.6116,
      "step": 13544
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.673642059023037,
      "learning_rate": 3.57562380868014e-06,
      "loss": 0.6067,
      "step": 13545
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.557212777336027,
      "learning_rate": 3.5749313840180026e-06,
      "loss": 0.5878,
      "step": 13546
    },
    {
      "epoch": 1.21,
      "grad_norm": 8.537432240662312,
      "learning_rate": 3.574238989101051e-06,
      "loss": 0.649,
      "step": 13547
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.958083884256966,
      "learning_rate": 3.573546623943738e-06,
      "loss": 0.5846,
      "step": 13548
    },
    {
      "epoch": 1.21,
      "grad_norm": 7.009278459758131,
      "learning_rate": 3.5728542885605155e-06,
      "loss": 0.6578,
      "step": 13549
    },
    {
      "epoch": 1.21,
      "grad_norm": 7.8317902998830204,
      "learning_rate": 3.5721619829658338e-06,
      "loss": 0.6283,
      "step": 13550
    },
    {
      "epoch": 1.21,
      "grad_norm": 7.19746750980491,
      "learning_rate": 3.5714697071741444e-06,
      "loss": 0.5641,
      "step": 13551
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.064444794587696,
      "learning_rate": 3.570777461199897e-06,
      "loss": 0.572,
      "step": 13552
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.699465718327881,
      "learning_rate": 3.570085245057539e-06,
      "loss": 0.5462,
      "step": 13553
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.358162214064198,
      "learning_rate": 3.56939305876152e-06,
      "loss": 0.512,
      "step": 13554
    },
    {
      "epoch": 1.21,
      "grad_norm": 7.965856438700478,
      "learning_rate": 3.5687009023262897e-06,
      "loss": 0.6306,
      "step": 13555
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.0811818247818525,
      "learning_rate": 3.568008775766292e-06,
      "loss": 0.6627,
      "step": 13556
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.949226354563478,
      "learning_rate": 3.5673166790959745e-06,
      "loss": 0.5543,
      "step": 13557
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.7686978814773,
      "learning_rate": 3.566624612329787e-06,
      "loss": 0.6154,
      "step": 13558
    },
    {
      "epoch": 1.21,
      "grad_norm": 10.008112930725066,
      "learning_rate": 3.56593257548217e-06,
      "loss": 0.5387,
      "step": 13559
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.262690014972771,
      "learning_rate": 3.5652405685675696e-06,
      "loss": 0.5828,
      "step": 13560
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.724215502508707,
      "learning_rate": 3.564548591600429e-06,
      "loss": 0.5595,
      "step": 13561
    },
    {
      "epoch": 1.21,
      "grad_norm": 4.935514420405982,
      "learning_rate": 3.563856644595195e-06,
      "loss": 0.5629,
      "step": 13562
    },
    {
      "epoch": 1.21,
      "grad_norm": 7.136352426902465,
      "learning_rate": 3.5631647275663073e-06,
      "loss": 0.5795,
      "step": 13563
    },
    {
      "epoch": 1.21,
      "grad_norm": 7.691878849404457,
      "learning_rate": 3.562472840528209e-06,
      "loss": 0.5861,
      "step": 13564
    },
    {
      "epoch": 1.21,
      "grad_norm": 7.185483630908752,
      "learning_rate": 3.5617809834953433e-06,
      "loss": 0.6019,
      "step": 13565
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.22843177093469,
      "learning_rate": 3.5610891564821482e-06,
      "loss": 0.5446,
      "step": 13566
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.3476173357644,
      "learning_rate": 3.560397359503066e-06,
      "loss": 0.5779,
      "step": 13567
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.771327529535014,
      "learning_rate": 3.5597055925725375e-06,
      "loss": 0.6011,
      "step": 13568
    },
    {
      "epoch": 1.21,
      "grad_norm": 7.814000158522739,
      "learning_rate": 3.559013855705e-06,
      "loss": 0.5922,
      "step": 13569
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.434577369544952,
      "learning_rate": 3.5583221489148923e-06,
      "loss": 0.5568,
      "step": 13570
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.037034089091607,
      "learning_rate": 3.5576304722166533e-06,
      "loss": 0.598,
      "step": 13571
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.679545792571334,
      "learning_rate": 3.5569388256247206e-06,
      "loss": 0.6661,
      "step": 13572
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.287390696154459,
      "learning_rate": 3.556247209153528e-06,
      "loss": 0.5799,
      "step": 13573
    },
    {
      "epoch": 1.21,
      "grad_norm": 4.731024940803107,
      "learning_rate": 3.5555556228175126e-06,
      "loss": 0.6259,
      "step": 13574
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.298304327728533,
      "learning_rate": 3.554864066631112e-06,
      "loss": 0.5847,
      "step": 13575
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.8888470622305675,
      "learning_rate": 3.554172540608757e-06,
      "loss": 0.5767,
      "step": 13576
    },
    {
      "epoch": 1.21,
      "grad_norm": 7.770712682237114,
      "learning_rate": 3.553481044764885e-06,
      "loss": 0.5803,
      "step": 13577
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.399882620080166,
      "learning_rate": 3.552789579113929e-06,
      "loss": 0.7094,
      "step": 13578
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.27193692060216,
      "learning_rate": 3.55209814367032e-06,
      "loss": 0.6033,
      "step": 13579
    },
    {
      "epoch": 1.21,
      "grad_norm": 7.804444640116811,
      "learning_rate": 3.5514067384484918e-06,
      "loss": 0.6298,
      "step": 13580
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.643739799188149,
      "learning_rate": 3.5507153634628765e-06,
      "loss": 0.608,
      "step": 13581
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.392851998981026,
      "learning_rate": 3.5500240187279024e-06,
      "loss": 0.5508,
      "step": 13582
    },
    {
      "epoch": 1.21,
      "grad_norm": 9.19786163085266,
      "learning_rate": 3.549332704258002e-06,
      "loss": 0.634,
      "step": 13583
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.658542767268432,
      "learning_rate": 3.5486414200676056e-06,
      "loss": 0.6057,
      "step": 13584
    },
    {
      "epoch": 1.21,
      "grad_norm": 7.488448371941349,
      "learning_rate": 3.54795016617114e-06,
      "loss": 0.5837,
      "step": 13585
    },
    {
      "epoch": 1.21,
      "grad_norm": 4.751706048737387,
      "learning_rate": 3.547258942583037e-06,
      "loss": 0.6008,
      "step": 13586
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.161856639633811,
      "learning_rate": 3.5465677493177202e-06,
      "loss": 0.6627,
      "step": 13587
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.083215262675206,
      "learning_rate": 3.5458765863896193e-06,
      "loss": 0.5857,
      "step": 13588
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.252792084557848,
      "learning_rate": 3.5451854538131593e-06,
      "loss": 0.6216,
      "step": 13589
    },
    {
      "epoch": 1.21,
      "grad_norm": 7.4753792648109325,
      "learning_rate": 3.5444943516027673e-06,
      "loss": 0.6323,
      "step": 13590
    },
    {
      "epoch": 1.21,
      "grad_norm": 8.299125672989351,
      "learning_rate": 3.543803279772869e-06,
      "loss": 0.6339,
      "step": 13591
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.870168991739116,
      "learning_rate": 3.5431122383378867e-06,
      "loss": 0.6892,
      "step": 13592
    },
    {
      "epoch": 1.21,
      "grad_norm": 4.749067434263115,
      "learning_rate": 3.5424212273122466e-06,
      "loss": 0.6186,
      "step": 13593
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.46922010491614,
      "learning_rate": 3.541730246710372e-06,
      "loss": 0.5789,
      "step": 13594
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.167847006382334,
      "learning_rate": 3.541039296546684e-06,
      "loss": 0.6145,
      "step": 13595
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.475966029601449,
      "learning_rate": 3.540348376835606e-06,
      "loss": 0.5981,
      "step": 13596
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.505954464377029,
      "learning_rate": 3.539657487591559e-06,
      "loss": 0.6835,
      "step": 13597
    },
    {
      "epoch": 1.21,
      "grad_norm": 7.055002830628653,
      "learning_rate": 3.5389666288289636e-06,
      "loss": 0.6123,
      "step": 13598
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.048872772717313,
      "learning_rate": 3.538275800562241e-06,
      "loss": 0.5986,
      "step": 13599
    },
    {
      "epoch": 1.21,
      "grad_norm": 7.297833155321413,
      "learning_rate": 3.537585002805809e-06,
      "loss": 0.6389,
      "step": 13600
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.108885373183691,
      "learning_rate": 3.53689423557409e-06,
      "loss": 0.5926,
      "step": 13601
    },
    {
      "epoch": 1.21,
      "grad_norm": 8.432027351649202,
      "learning_rate": 3.536203498881497e-06,
      "loss": 0.6126,
      "step": 13602
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.934804474551162,
      "learning_rate": 3.5355127927424503e-06,
      "loss": 0.6002,
      "step": 13603
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.573655443672592,
      "learning_rate": 3.534822117171368e-06,
      "loss": 0.6502,
      "step": 13604
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.3426984798113155,
      "learning_rate": 3.5341314721826648e-06,
      "loss": 0.6042,
      "step": 13605
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.826241625278029,
      "learning_rate": 3.5334408577907576e-06,
      "loss": 0.6366,
      "step": 13606
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.293728297765808,
      "learning_rate": 3.5327502740100595e-06,
      "loss": 0.5609,
      "step": 13607
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.591031327740003,
      "learning_rate": 3.532059720854986e-06,
      "loss": 0.6025,
      "step": 13608
    },
    {
      "epoch": 1.21,
      "grad_norm": 7.109288836358396,
      "learning_rate": 3.531369198339953e-06,
      "loss": 0.6073,
      "step": 13609
    },
    {
      "epoch": 1.21,
      "grad_norm": 8.129298005338526,
      "learning_rate": 3.5306787064793702e-06,
      "loss": 0.5673,
      "step": 13610
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.337513331728844,
      "learning_rate": 3.5299882452876512e-06,
      "loss": 0.581,
      "step": 13611
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.688341191027541,
      "learning_rate": 3.52929781477921e-06,
      "loss": 0.5538,
      "step": 13612
    },
    {
      "epoch": 1.21,
      "grad_norm": 7.032925484353675,
      "learning_rate": 3.5286074149684547e-06,
      "loss": 0.6012,
      "step": 13613
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.621584534829211,
      "learning_rate": 3.5279170458697974e-06,
      "loss": 0.6712,
      "step": 13614
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.6504114617296635,
      "learning_rate": 3.527226707497651e-06,
      "loss": 0.6059,
      "step": 13615
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.382215044998568,
      "learning_rate": 3.5265363998664195e-06,
      "loss": 0.5328,
      "step": 13616
    },
    {
      "epoch": 1.21,
      "grad_norm": 7.055524677340241,
      "learning_rate": 3.5258461229905127e-06,
      "loss": 0.5189,
      "step": 13617
    },
    {
      "epoch": 1.21,
      "grad_norm": 8.677874487686541,
      "learning_rate": 3.5251558768843407e-06,
      "loss": 0.6176,
      "step": 13618
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.709031712703059,
      "learning_rate": 3.52446566156231e-06,
      "loss": 0.6207,
      "step": 13619
    },
    {
      "epoch": 1.22,
      "grad_norm": 7.7522118303487515,
      "learning_rate": 3.5237754770388266e-06,
      "loss": 0.6134,
      "step": 13620
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.842348690106948,
      "learning_rate": 3.5230853233282973e-06,
      "loss": 0.5741,
      "step": 13621
    },
    {
      "epoch": 1.22,
      "grad_norm": 4.712162914931422,
      "learning_rate": 3.522395200445129e-06,
      "loss": 0.5906,
      "step": 13622
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.931746814081168,
      "learning_rate": 3.5217051084037234e-06,
      "loss": 0.6257,
      "step": 13623
    },
    {
      "epoch": 1.22,
      "grad_norm": 8.90105844462885,
      "learning_rate": 3.521015047218486e-06,
      "loss": 0.6049,
      "step": 13624
    },
    {
      "epoch": 1.22,
      "grad_norm": 7.897725953464819,
      "learning_rate": 3.5203250169038224e-06,
      "loss": 0.68,
      "step": 13625
    },
    {
      "epoch": 1.22,
      "grad_norm": 10.05183484412896,
      "learning_rate": 3.519635017474132e-06,
      "loss": 0.6454,
      "step": 13626
    },
    {
      "epoch": 1.22,
      "grad_norm": 10.911588825782172,
      "learning_rate": 3.5189450489438186e-06,
      "loss": 0.584,
      "step": 13627
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.384114681509384,
      "learning_rate": 3.5182551113272856e-06,
      "loss": 0.5876,
      "step": 13628
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.965074899257299,
      "learning_rate": 3.51756520463893e-06,
      "loss": 0.5926,
      "step": 13629
    },
    {
      "epoch": 1.22,
      "grad_norm": 7.104897477094417,
      "learning_rate": 3.5168753288931573e-06,
      "loss": 0.595,
      "step": 13630
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.156096393268672,
      "learning_rate": 3.516185484104362e-06,
      "loss": 0.6082,
      "step": 13631
    },
    {
      "epoch": 1.22,
      "grad_norm": 7.934671736175371,
      "learning_rate": 3.5154956702869458e-06,
      "loss": 0.6235,
      "step": 13632
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.572038244748409,
      "learning_rate": 3.5148058874553053e-06,
      "loss": 0.6237,
      "step": 13633
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.161589079322032,
      "learning_rate": 3.5141161356238395e-06,
      "loss": 0.6385,
      "step": 13634
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.750867208485521,
      "learning_rate": 3.513426414806947e-06,
      "loss": 0.6465,
      "step": 13635
    },
    {
      "epoch": 1.22,
      "grad_norm": 7.432961772962671,
      "learning_rate": 3.5127367250190214e-06,
      "loss": 0.5916,
      "step": 13636
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.531570560101088,
      "learning_rate": 3.512047066274459e-06,
      "loss": 0.5797,
      "step": 13637
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.858078392919785,
      "learning_rate": 3.511357438587657e-06,
      "loss": 0.6343,
      "step": 13638
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.403997021452365,
      "learning_rate": 3.5106678419730073e-06,
      "loss": 0.6436,
      "step": 13639
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.41242625899477,
      "learning_rate": 3.5099782764449043e-06,
      "loss": 0.6087,
      "step": 13640
    },
    {
      "epoch": 1.22,
      "grad_norm": 11.474015051655877,
      "learning_rate": 3.5092887420177437e-06,
      "loss": 0.5978,
      "step": 13641
    },
    {
      "epoch": 1.22,
      "grad_norm": 7.0520918870069105,
      "learning_rate": 3.508599238705914e-06,
      "loss": 0.5919,
      "step": 13642
    },
    {
      "epoch": 1.22,
      "grad_norm": 8.15430365636395,
      "learning_rate": 3.5079097665238094e-06,
      "loss": 0.6149,
      "step": 13643
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.544066321223188,
      "learning_rate": 3.5072203254858234e-06,
      "loss": 0.6559,
      "step": 13644
    },
    {
      "epoch": 1.22,
      "grad_norm": 8.374510285451224,
      "learning_rate": 3.5065309156063424e-06,
      "loss": 0.5872,
      "step": 13645
    },
    {
      "epoch": 1.22,
      "grad_norm": 7.8094833668540895,
      "learning_rate": 3.5058415368997572e-06,
      "loss": 0.6453,
      "step": 13646
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.8337935450445455,
      "learning_rate": 3.5051521893804576e-06,
      "loss": 0.6319,
      "step": 13647
    },
    {
      "epoch": 1.22,
      "grad_norm": 7.462943200175098,
      "learning_rate": 3.5044628730628333e-06,
      "loss": 0.6311,
      "step": 13648
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.042105511355303,
      "learning_rate": 3.5037735879612704e-06,
      "loss": 0.6884,
      "step": 13649
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.643086314276672,
      "learning_rate": 3.503084334090157e-06,
      "loss": 0.5331,
      "step": 13650
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.924219723632321,
      "learning_rate": 3.502395111463881e-06,
      "loss": 0.6517,
      "step": 13651
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.20928582277627,
      "learning_rate": 3.5017059200968266e-06,
      "loss": 0.651,
      "step": 13652
    },
    {
      "epoch": 1.22,
      "grad_norm": 7.469244963902982,
      "learning_rate": 3.50101676000338e-06,
      "loss": 0.5803,
      "step": 13653
    },
    {
      "epoch": 1.22,
      "grad_norm": 8.312863024084436,
      "learning_rate": 3.5003276311979268e-06,
      "loss": 0.5989,
      "step": 13654
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.144202534261614,
      "learning_rate": 3.499638533694849e-06,
      "loss": 0.5665,
      "step": 13655
    },
    {
      "epoch": 1.22,
      "grad_norm": 9.587468724546763,
      "learning_rate": 3.4989494675085307e-06,
      "loss": 0.5859,
      "step": 13656
    },
    {
      "epoch": 1.22,
      "grad_norm": 7.223086322069463,
      "learning_rate": 3.4982604326533564e-06,
      "loss": 0.686,
      "step": 13657
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.117887076690238,
      "learning_rate": 3.497571429143708e-06,
      "loss": 0.6427,
      "step": 13658
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.165763739887072,
      "learning_rate": 3.496882456993963e-06,
      "loss": 0.5731,
      "step": 13659
    },
    {
      "epoch": 1.22,
      "grad_norm": 8.09498622097635,
      "learning_rate": 3.4961935162185056e-06,
      "loss": 0.5587,
      "step": 13660
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.72486533936278,
      "learning_rate": 3.4955046068317154e-06,
      "loss": 0.6503,
      "step": 13661
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.023933415007256,
      "learning_rate": 3.494815728847972e-06,
      "loss": 0.5835,
      "step": 13662
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.011070614499679,
      "learning_rate": 3.4941268822816533e-06,
      "loss": 0.5541,
      "step": 13663
    },
    {
      "epoch": 1.22,
      "grad_norm": 10.591474001354547,
      "learning_rate": 3.493438067147139e-06,
      "loss": 0.6553,
      "step": 13664
    },
    {
      "epoch": 1.22,
      "grad_norm": 7.144185767060184,
      "learning_rate": 3.492749283458805e-06,
      "loss": 0.5777,
      "step": 13665
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.504227521965311,
      "learning_rate": 3.4920605312310283e-06,
      "loss": 0.6418,
      "step": 13666
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.287541750859735,
      "learning_rate": 3.491371810478188e-06,
      "loss": 0.6427,
      "step": 13667
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.608583146730501,
      "learning_rate": 3.490683121214655e-06,
      "loss": 0.6035,
      "step": 13668
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.299171854694549,
      "learning_rate": 3.489994463454807e-06,
      "loss": 0.5516,
      "step": 13669
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.127466173594637,
      "learning_rate": 3.489305837213019e-06,
      "loss": 0.581,
      "step": 13670
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.348778427654451,
      "learning_rate": 3.488617242503662e-06,
      "loss": 0.603,
      "step": 13671
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.398051689099296,
      "learning_rate": 3.487928679341111e-06,
      "loss": 0.6764,
      "step": 13672
    },
    {
      "epoch": 1.22,
      "grad_norm": 8.50113187552315,
      "learning_rate": 3.487240147739738e-06,
      "loss": 0.5823,
      "step": 13673
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.010880745358703,
      "learning_rate": 3.486551647713914e-06,
      "loss": 0.6221,
      "step": 13674
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.224583280749061,
      "learning_rate": 3.4858631792780084e-06,
      "loss": 0.5835,
      "step": 13675
    },
    {
      "epoch": 1.22,
      "grad_norm": 7.538297862491756,
      "learning_rate": 3.4851747424463933e-06,
      "loss": 0.6184,
      "step": 13676
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.626137788212427,
      "learning_rate": 3.4844863372334386e-06,
      "loss": 0.5878,
      "step": 13677
    },
    {
      "epoch": 1.22,
      "grad_norm": 4.758321681791451,
      "learning_rate": 3.4837979636535125e-06,
      "loss": 0.5776,
      "step": 13678
    },
    {
      "epoch": 1.22,
      "grad_norm": 9.067182589008237,
      "learning_rate": 3.4831096217209824e-06,
      "loss": 0.5972,
      "step": 13679
    },
    {
      "epoch": 1.22,
      "grad_norm": 7.910604666708848,
      "learning_rate": 3.4824213114502194e-06,
      "loss": 0.5989,
      "step": 13680
    },
    {
      "epoch": 1.22,
      "grad_norm": 8.002398584513303,
      "learning_rate": 3.481733032855586e-06,
      "loss": 0.6592,
      "step": 13681
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.135466437030478,
      "learning_rate": 3.4810447859514517e-06,
      "loss": 0.6162,
      "step": 13682
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.121029674490818,
      "learning_rate": 3.48035657075218e-06,
      "loss": 0.5445,
      "step": 13683
    },
    {
      "epoch": 1.22,
      "grad_norm": 9.657794510852804,
      "learning_rate": 3.479668387272138e-06,
      "loss": 0.5755,
      "step": 13684
    },
    {
      "epoch": 1.22,
      "grad_norm": 8.002570491172465,
      "learning_rate": 3.4789802355256886e-06,
      "loss": 0.6348,
      "step": 13685
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.863625871417886,
      "learning_rate": 3.4782921155271955e-06,
      "loss": 0.6016,
      "step": 13686
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.748697924626328,
      "learning_rate": 3.477604027291024e-06,
      "loss": 0.6247,
      "step": 13687
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.194435251414398,
      "learning_rate": 3.4769159708315325e-06,
      "loss": 0.5957,
      "step": 13688
    },
    {
      "epoch": 1.22,
      "grad_norm": 7.318623970362775,
      "learning_rate": 3.4762279461630854e-06,
      "loss": 0.5877,
      "step": 13689
    },
    {
      "epoch": 1.22,
      "grad_norm": 4.492506882571419,
      "learning_rate": 3.4755399533000415e-06,
      "loss": 0.5818,
      "step": 13690
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.660255247698129,
      "learning_rate": 3.4748519922567635e-06,
      "loss": 0.6284,
      "step": 13691
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.165666952655602,
      "learning_rate": 3.4741640630476103e-06,
      "loss": 0.6211,
      "step": 13692
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.302844061391524,
      "learning_rate": 3.4734761656869404e-06,
      "loss": 0.6277,
      "step": 13693
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.47509193220591,
      "learning_rate": 3.4727883001891123e-06,
      "loss": 0.6217,
      "step": 13694
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.97898364845394,
      "learning_rate": 3.4721004665684844e-06,
      "loss": 0.6012,
      "step": 13695
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.899759510333936,
      "learning_rate": 3.471412664839413e-06,
      "loss": 0.6356,
      "step": 13696
    },
    {
      "epoch": 1.22,
      "grad_norm": 7.1652801621887585,
      "learning_rate": 3.470724895016254e-06,
      "loss": 0.64,
      "step": 13697
    },
    {
      "epoch": 1.22,
      "grad_norm": 8.554186527615782,
      "learning_rate": 3.470037157113365e-06,
      "loss": 0.6346,
      "step": 13698
    },
    {
      "epoch": 1.22,
      "grad_norm": 3.945789277809698,
      "learning_rate": 3.4693494511450987e-06,
      "loss": 0.6249,
      "step": 13699
    },
    {
      "epoch": 1.22,
      "grad_norm": 7.93985263334273,
      "learning_rate": 3.468661777125811e-06,
      "loss": 0.5837,
      "step": 13700
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.0448620376442825,
      "learning_rate": 3.467974135069856e-06,
      "loss": 0.633,
      "step": 13701
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.021546296705873,
      "learning_rate": 3.4672865249915864e-06,
      "loss": 0.6323,
      "step": 13702
    },
    {
      "epoch": 1.22,
      "grad_norm": 7.319041769079637,
      "learning_rate": 3.466598946905353e-06,
      "loss": 0.5491,
      "step": 13703
    },
    {
      "epoch": 1.22,
      "grad_norm": 4.41857156286032,
      "learning_rate": 3.4659114008255073e-06,
      "loss": 0.583,
      "step": 13704
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.423507891545533,
      "learning_rate": 3.465223886766403e-06,
      "loss": 0.6125,
      "step": 13705
    },
    {
      "epoch": 1.22,
      "grad_norm": 7.897584245984497,
      "learning_rate": 3.464536404742388e-06,
      "loss": 0.5999,
      "step": 13706
    },
    {
      "epoch": 1.22,
      "grad_norm": 7.462102382313217,
      "learning_rate": 3.4638489547678126e-06,
      "loss": 0.5913,
      "step": 13707
    },
    {
      "epoch": 1.22,
      "grad_norm": 8.588243638263608,
      "learning_rate": 3.4631615368570275e-06,
      "loss": 0.5783,
      "step": 13708
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.629166074858269,
      "learning_rate": 3.4624741510243787e-06,
      "loss": 0.6265,
      "step": 13709
    },
    {
      "epoch": 1.22,
      "grad_norm": 8.426405063558105,
      "learning_rate": 3.4617867972842146e-06,
      "loss": 0.6383,
      "step": 13710
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.471963568692138,
      "learning_rate": 3.4610994756508844e-06,
      "loss": 0.5508,
      "step": 13711
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.934229493168135,
      "learning_rate": 3.4604121861387308e-06,
      "loss": 0.6287,
      "step": 13712
    },
    {
      "epoch": 1.22,
      "grad_norm": 7.65844871668796,
      "learning_rate": 3.4597249287621004e-06,
      "loss": 0.6279,
      "step": 13713
    },
    {
      "epoch": 1.22,
      "grad_norm": 8.068638360336003,
      "learning_rate": 3.4590377035353412e-06,
      "loss": 0.5806,
      "step": 13714
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.739169334062597,
      "learning_rate": 3.458350510472794e-06,
      "loss": 0.6107,
      "step": 13715
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.731840569681148,
      "learning_rate": 3.4576633495888055e-06,
      "loss": 0.6111,
      "step": 13716
    },
    {
      "epoch": 1.22,
      "grad_norm": 8.029395218125194,
      "learning_rate": 3.456976220897715e-06,
      "loss": 0.6139,
      "step": 13717
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.560664634787917,
      "learning_rate": 3.4562891244138673e-06,
      "loss": 0.611,
      "step": 13718
    },
    {
      "epoch": 1.22,
      "grad_norm": 8.231727208473416,
      "learning_rate": 3.455602060151603e-06,
      "loss": 0.6081,
      "step": 13719
    },
    {
      "epoch": 1.22,
      "grad_norm": 4.162461830374463,
      "learning_rate": 3.4549150281252635e-06,
      "loss": 0.6066,
      "step": 13720
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.072580154935637,
      "learning_rate": 3.45422802834919e-06,
      "loss": 0.5566,
      "step": 13721
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.852888006938338,
      "learning_rate": 3.4535410608377206e-06,
      "loss": 0.6079,
      "step": 13722
    },
    {
      "epoch": 1.22,
      "grad_norm": 7.238629582356523,
      "learning_rate": 3.452854125605194e-06,
      "loss": 0.586,
      "step": 13723
    },
    {
      "epoch": 1.22,
      "grad_norm": 8.44460878540945,
      "learning_rate": 3.452167222665951e-06,
      "loss": 0.6166,
      "step": 13724
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.071057968557668,
      "learning_rate": 3.4514803520343266e-06,
      "loss": 0.6116,
      "step": 13725
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.440454964130839,
      "learning_rate": 3.4507935137246584e-06,
      "loss": 0.6096,
      "step": 13726
    },
    {
      "epoch": 1.22,
      "grad_norm": 7.352271576079093,
      "learning_rate": 3.450106707751284e-06,
      "loss": 0.5797,
      "step": 13727
    },
    {
      "epoch": 1.22,
      "grad_norm": 4.7382830608654185,
      "learning_rate": 3.449419934128536e-06,
      "loss": 0.5818,
      "step": 13728
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.806600843786048,
      "learning_rate": 3.4487331928707522e-06,
      "loss": 0.6335,
      "step": 13729
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.091271184703984,
      "learning_rate": 3.4480464839922665e-06,
      "loss": 0.5975,
      "step": 13730
    },
    {
      "epoch": 1.22,
      "grad_norm": 4.737315887974828,
      "learning_rate": 3.447359807507413e-06,
      "loss": 0.578,
      "step": 13731
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.621983206801497,
      "learning_rate": 3.446673163430521e-06,
      "loss": 0.6085,
      "step": 13732
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.350291413618931,
      "learning_rate": 3.445986551775925e-06,
      "loss": 0.5824,
      "step": 13733
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.546765446452017,
      "learning_rate": 3.4452999725579577e-06,
      "loss": 0.5421,
      "step": 13734
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.268518222855655,
      "learning_rate": 3.4446134257909474e-06,
      "loss": 0.6305,
      "step": 13735
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.888291633438685,
      "learning_rate": 3.443926911489226e-06,
      "loss": 0.599,
      "step": 13736
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.83516913225478,
      "learning_rate": 3.443240429667123e-06,
      "loss": 0.5442,
      "step": 13737
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.133745252992368,
      "learning_rate": 3.4425539803389663e-06,
      "loss": 0.6044,
      "step": 13738
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.650552730660366,
      "learning_rate": 3.4418675635190844e-06,
      "loss": 0.5784,
      "step": 13739
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.180726383210313,
      "learning_rate": 3.441181179221806e-06,
      "loss": 0.571,
      "step": 13740
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.260247412255486,
      "learning_rate": 3.4404948274614554e-06,
      "loss": 0.5982,
      "step": 13741
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.208784260931408,
      "learning_rate": 3.4398085082523604e-06,
      "loss": 0.6217,
      "step": 13742
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.874968863367118,
      "learning_rate": 3.4391222216088467e-06,
      "loss": 0.6362,
      "step": 13743
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.163676296293012,
      "learning_rate": 3.4384359675452373e-06,
      "loss": 0.6549,
      "step": 13744
    },
    {
      "epoch": 1.23,
      "grad_norm": 4.942995236625064,
      "learning_rate": 3.437749746075859e-06,
      "loss": 0.6791,
      "step": 13745
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.385802563310941,
      "learning_rate": 3.4370635572150324e-06,
      "loss": 0.5785,
      "step": 13746
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.628853323665255,
      "learning_rate": 3.4363774009770817e-06,
      "loss": 0.5475,
      "step": 13747
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.107363138269978,
      "learning_rate": 3.4356912773763274e-06,
      "loss": 0.5745,
      "step": 13748
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.239732325654228,
      "learning_rate": 3.4350051864270923e-06,
      "loss": 0.6126,
      "step": 13749
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.376257752567354,
      "learning_rate": 3.4343191281436973e-06,
      "loss": 0.6064,
      "step": 13750
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.527535559973688,
      "learning_rate": 3.4336331025404613e-06,
      "loss": 0.5809,
      "step": 13751
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.133310473979701,
      "learning_rate": 3.4329471096317035e-06,
      "loss": 0.659,
      "step": 13752
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.316843532550631,
      "learning_rate": 3.432261149431744e-06,
      "loss": 0.6286,
      "step": 13753
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.980576901338581,
      "learning_rate": 3.4315752219548985e-06,
      "loss": 0.6268,
      "step": 13754
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.732802479643587,
      "learning_rate": 3.430889327215486e-06,
      "loss": 0.5918,
      "step": 13755
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.9217898434268434,
      "learning_rate": 3.4302034652278226e-06,
      "loss": 0.5873,
      "step": 13756
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.243299592992757,
      "learning_rate": 3.4295176360062244e-06,
      "loss": 0.5632,
      "step": 13757
    },
    {
      "epoch": 1.23,
      "grad_norm": 8.265570259431275,
      "learning_rate": 3.4288318395650066e-06,
      "loss": 0.6179,
      "step": 13758
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.044740321188384,
      "learning_rate": 3.4281460759184825e-06,
      "loss": 0.5506,
      "step": 13759
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.684671330764707,
      "learning_rate": 3.4274603450809686e-06,
      "loss": 0.6246,
      "step": 13760
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.275206070781751,
      "learning_rate": 3.426774647066774e-06,
      "loss": 0.6113,
      "step": 13761
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.536551062964168,
      "learning_rate": 3.4260889818902133e-06,
      "loss": 0.5916,
      "step": 13762
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.087734492598801,
      "learning_rate": 3.425403349565599e-06,
      "loss": 0.5939,
      "step": 13763
    },
    {
      "epoch": 1.23,
      "grad_norm": 8.554672226656763,
      "learning_rate": 3.4247177501072414e-06,
      "loss": 0.5723,
      "step": 13764
    },
    {
      "epoch": 1.23,
      "grad_norm": 8.814128736950552,
      "learning_rate": 3.424032183529451e-06,
      "loss": 0.599,
      "step": 13765
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.96271757032685,
      "learning_rate": 3.4233466498465375e-06,
      "loss": 0.5832,
      "step": 13766
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.47901840707784,
      "learning_rate": 3.422661149072809e-06,
      "loss": 0.5625,
      "step": 13767
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.073526204867449,
      "learning_rate": 3.4219756812225763e-06,
      "loss": 0.6381,
      "step": 13768
    },
    {
      "epoch": 1.23,
      "grad_norm": 8.283941346070618,
      "learning_rate": 3.421290246310144e-06,
      "loss": 0.6649,
      "step": 13769
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.874826238269134,
      "learning_rate": 3.420604844349821e-06,
      "loss": 0.5874,
      "step": 13770
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.485772770941453,
      "learning_rate": 3.4199194753559136e-06,
      "loss": 0.6396,
      "step": 13771
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.505555517534428,
      "learning_rate": 3.4192341393427257e-06,
      "loss": 0.642,
      "step": 13772
    },
    {
      "epoch": 1.23,
      "grad_norm": 8.079182563210349,
      "learning_rate": 3.418548836324563e-06,
      "loss": 0.6432,
      "step": 13773
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.366760566509017,
      "learning_rate": 3.4178635663157333e-06,
      "loss": 0.586,
      "step": 13774
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.99856690471874,
      "learning_rate": 3.417178329330535e-06,
      "loss": 0.6047,
      "step": 13775
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.09555258764799,
      "learning_rate": 3.4164931253832724e-06,
      "loss": 0.6109,
      "step": 13776
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.201360711269401,
      "learning_rate": 3.4158079544882474e-06,
      "loss": 0.5848,
      "step": 13777
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.516912977370182,
      "learning_rate": 3.4151228166597627e-06,
      "loss": 0.6389,
      "step": 13778
    },
    {
      "epoch": 1.23,
      "grad_norm": 8.46292951364681,
      "learning_rate": 3.414437711912118e-06,
      "loss": 0.6119,
      "step": 13779
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.9028876320684525,
      "learning_rate": 3.4137526402596135e-06,
      "loss": 0.6213,
      "step": 13780
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.272898241021619,
      "learning_rate": 3.4130676017165497e-06,
      "loss": 0.5905,
      "step": 13781
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.056132214435461,
      "learning_rate": 3.4123825962972236e-06,
      "loss": 0.5966,
      "step": 13782
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.614495605729986,
      "learning_rate": 3.4116976240159337e-06,
      "loss": 0.5658,
      "step": 13783
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.370092959658471,
      "learning_rate": 3.4110126848869798e-06,
      "loss": 0.5586,
      "step": 13784
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.658755901739205,
      "learning_rate": 3.4103277789246537e-06,
      "loss": 0.6569,
      "step": 13785
    },
    {
      "epoch": 1.23,
      "grad_norm": 8.359420472165821,
      "learning_rate": 3.4096429061432544e-06,
      "loss": 0.6212,
      "step": 13786
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.398281183844312,
      "learning_rate": 3.4089580665570775e-06,
      "loss": 0.5778,
      "step": 13787
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.049647571079885,
      "learning_rate": 3.408273260180418e-06,
      "loss": 0.6448,
      "step": 13788
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.492346171167204,
      "learning_rate": 3.4075884870275667e-06,
      "loss": 0.5659,
      "step": 13789
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.8041512664030455,
      "learning_rate": 3.4069037471128173e-06,
      "loss": 0.6039,
      "step": 13790
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.394143938803646,
      "learning_rate": 3.4062190404504654e-06,
      "loss": 0.5668,
      "step": 13791
    },
    {
      "epoch": 1.23,
      "grad_norm": 8.711275653626348,
      "learning_rate": 3.4055343670547993e-06,
      "loss": 0.5659,
      "step": 13792
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.03300205176709,
      "learning_rate": 3.4048497269401114e-06,
      "loss": 0.6199,
      "step": 13793
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.264464209572731,
      "learning_rate": 3.4041651201206925e-06,
      "loss": 0.575,
      "step": 13794
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.104780004814482,
      "learning_rate": 3.4034805466108324e-06,
      "loss": 0.6275,
      "step": 13795
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.342228665765283,
      "learning_rate": 3.4027960064248183e-06,
      "loss": 0.5858,
      "step": 13796
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.008383516219498,
      "learning_rate": 3.402111499576941e-06,
      "loss": 0.5789,
      "step": 13797
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.652815789509286,
      "learning_rate": 3.4014270260814864e-06,
      "loss": 0.5333,
      "step": 13798
    },
    {
      "epoch": 1.23,
      "grad_norm": 8.439345461799373,
      "learning_rate": 3.400742585952741e-06,
      "loss": 0.6463,
      "step": 13799
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.091178086429984,
      "learning_rate": 3.4000581792049937e-06,
      "loss": 0.5408,
      "step": 13800
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.886366345203147,
      "learning_rate": 3.399373805852526e-06,
      "loss": 0.5939,
      "step": 13801
    },
    {
      "epoch": 1.23,
      "grad_norm": 8.027574769572873,
      "learning_rate": 3.3986894659096257e-06,
      "loss": 0.5597,
      "step": 13802
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.066205263703978,
      "learning_rate": 3.3980051593905784e-06,
      "loss": 0.5938,
      "step": 13803
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.223139143452273,
      "learning_rate": 3.3973208863096628e-06,
      "loss": 0.5391,
      "step": 13804
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.302200513792785,
      "learning_rate": 3.396636646681164e-06,
      "loss": 0.6796,
      "step": 13805
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.8603332403046915,
      "learning_rate": 3.3959524405193634e-06,
      "loss": 0.6222,
      "step": 13806
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.064098174291945,
      "learning_rate": 3.3952682678385437e-06,
      "loss": 0.5643,
      "step": 13807
    },
    {
      "epoch": 1.23,
      "grad_norm": 9.321930834602918,
      "learning_rate": 3.3945841286529833e-06,
      "loss": 0.5889,
      "step": 13808
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.445381032130245,
      "learning_rate": 3.3939000229769637e-06,
      "loss": 0.6274,
      "step": 13809
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.225767120114856,
      "learning_rate": 3.3932159508247653e-06,
      "loss": 0.556,
      "step": 13810
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.240831379631339,
      "learning_rate": 3.3925319122106637e-06,
      "loss": 0.5938,
      "step": 13811
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.356868316352738,
      "learning_rate": 3.391847907148938e-06,
      "loss": 0.6394,
      "step": 13812
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.591946369639385,
      "learning_rate": 3.3911639356538666e-06,
      "loss": 0.5651,
      "step": 13813
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.730617109324473,
      "learning_rate": 3.390479997739724e-06,
      "loss": 0.4826,
      "step": 13814
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.423344655866369,
      "learning_rate": 3.3897960934207863e-06,
      "loss": 0.5878,
      "step": 13815
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.957952268398784,
      "learning_rate": 3.3891122227113298e-06,
      "loss": 0.6509,
      "step": 13816
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.868810673812395,
      "learning_rate": 3.3884283856256295e-06,
      "loss": 0.5637,
      "step": 13817
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.117163318406728,
      "learning_rate": 3.3877445821779548e-06,
      "loss": 0.5561,
      "step": 13818
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.960162593175425,
      "learning_rate": 3.387060812382581e-06,
      "loss": 0.5727,
      "step": 13819
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.6040776156262435,
      "learning_rate": 3.386377076253782e-06,
      "loss": 0.6372,
      "step": 13820
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.663369105859036,
      "learning_rate": 3.385693373805827e-06,
      "loss": 0.5975,
      "step": 13821
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.065868476541639,
      "learning_rate": 3.3850097050529867e-06,
      "loss": 0.6123,
      "step": 13822
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.558087024508988,
      "learning_rate": 3.3843260700095336e-06,
      "loss": 0.6088,
      "step": 13823
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.83926611704635,
      "learning_rate": 3.383642468689734e-06,
      "loss": 0.5856,
      "step": 13824
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.536724242089029,
      "learning_rate": 3.382958901107859e-06,
      "loss": 0.5655,
      "step": 13825
    },
    {
      "epoch": 1.23,
      "grad_norm": 8.955462900309438,
      "learning_rate": 3.382275367278176e-06,
      "loss": 0.5679,
      "step": 13826
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.423887103565648,
      "learning_rate": 3.3815918672149505e-06,
      "loss": 0.5305,
      "step": 13827
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.803187189764014,
      "learning_rate": 3.3809084009324512e-06,
      "loss": 0.6026,
      "step": 13828
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.972950532994315,
      "learning_rate": 3.3802249684449445e-06,
      "loss": 0.5457,
      "step": 13829
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.139203312561252,
      "learning_rate": 3.3795415697666935e-06,
      "loss": 0.6229,
      "step": 13830
    },
    {
      "epoch": 1.23,
      "grad_norm": 8.429347743674175,
      "learning_rate": 3.3788582049119622e-06,
      "loss": 0.5935,
      "step": 13831
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.716283593459564,
      "learning_rate": 3.3781748738950186e-06,
      "loss": 0.6444,
      "step": 13832
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.851309906996595,
      "learning_rate": 3.377491576730122e-06,
      "loss": 0.5633,
      "step": 13833
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.347437490178266,
      "learning_rate": 3.3768083134315336e-06,
      "loss": 0.7057,
      "step": 13834
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.007478878279024,
      "learning_rate": 3.376125084013518e-06,
      "loss": 0.5915,
      "step": 13835
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.1383571268190105,
      "learning_rate": 3.375441888490335e-06,
      "loss": 0.5933,
      "step": 13836
    },
    {
      "epoch": 1.23,
      "grad_norm": 8.49990888404691,
      "learning_rate": 3.374758726876245e-06,
      "loss": 0.6307,
      "step": 13837
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.1092539465595435,
      "learning_rate": 3.3740755991855063e-06,
      "loss": 0.603,
      "step": 13838
    },
    {
      "epoch": 1.23,
      "grad_norm": 12.438065399448549,
      "learning_rate": 3.3733925054323802e-06,
      "loss": 0.6006,
      "step": 13839
    },
    {
      "epoch": 1.23,
      "grad_norm": 8.386851136374954,
      "learning_rate": 3.3727094456311227e-06,
      "loss": 0.582,
      "step": 13840
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.714682309771767,
      "learning_rate": 3.372026419795992e-06,
      "loss": 0.5546,
      "step": 13841
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.287703828468095,
      "learning_rate": 3.3713434279412437e-06,
      "loss": 0.6194,
      "step": 13842
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.436042608619036,
      "learning_rate": 3.3706604700811353e-06,
      "loss": 0.6053,
      "step": 13843
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.6547446074045675,
      "learning_rate": 3.3699775462299224e-06,
      "loss": 0.5812,
      "step": 13844
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.998792279460274,
      "learning_rate": 3.3692946564018567e-06,
      "loss": 0.569,
      "step": 13845
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.89599515321578,
      "learning_rate": 3.368611800611197e-06,
      "loss": 0.6312,
      "step": 13846
    },
    {
      "epoch": 1.24,
      "grad_norm": 4.720132327775423,
      "learning_rate": 3.36792897887219e-06,
      "loss": 0.5919,
      "step": 13847
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.314852840694335,
      "learning_rate": 3.367246191199092e-06,
      "loss": 0.5844,
      "step": 13848
    },
    {
      "epoch": 1.24,
      "grad_norm": 7.4835013929533165,
      "learning_rate": 3.3665634376061553e-06,
      "loss": 0.6061,
      "step": 13849
    },
    {
      "epoch": 1.24,
      "grad_norm": 4.866757715312992,
      "learning_rate": 3.3658807181076282e-06,
      "loss": 0.5646,
      "step": 13850
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.459109643119882,
      "learning_rate": 3.3651980327177635e-06,
      "loss": 0.5896,
      "step": 13851
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.748618680906519,
      "learning_rate": 3.364515381450809e-06,
      "loss": 0.5752,
      "step": 13852
    },
    {
      "epoch": 1.24,
      "grad_norm": 8.500540691977445,
      "learning_rate": 3.3638327643210146e-06,
      "loss": 0.5663,
      "step": 13853
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.935904536868631,
      "learning_rate": 3.363150181342629e-06,
      "loss": 0.5755,
      "step": 13854
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.79234544378916,
      "learning_rate": 3.3624676325298974e-06,
      "loss": 0.6437,
      "step": 13855
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.910852941472072,
      "learning_rate": 3.3617851178970675e-06,
      "loss": 0.5789,
      "step": 13856
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.748201795773867,
      "learning_rate": 3.3611026374583867e-06,
      "loss": 0.5369,
      "step": 13857
    },
    {
      "epoch": 1.24,
      "grad_norm": 7.355561238021794,
      "learning_rate": 3.360420191228099e-06,
      "loss": 0.6014,
      "step": 13858
    },
    {
      "epoch": 1.24,
      "grad_norm": 7.816483051637275,
      "learning_rate": 3.359737779220449e-06,
      "loss": 0.5995,
      "step": 13859
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.8680668432124,
      "learning_rate": 3.3590554014496822e-06,
      "loss": 0.612,
      "step": 13860
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.7736035368089755,
      "learning_rate": 3.3583730579300395e-06,
      "loss": 0.603,
      "step": 13861
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.021797996647345,
      "learning_rate": 3.3576907486757624e-06,
      "loss": 0.5434,
      "step": 13862
    },
    {
      "epoch": 1.24,
      "grad_norm": 7.023400251767776,
      "learning_rate": 3.3570084737010955e-06,
      "loss": 0.5884,
      "step": 13863
    },
    {
      "epoch": 1.24,
      "grad_norm": 7.706593096474654,
      "learning_rate": 3.3563262330202794e-06,
      "loss": 0.6153,
      "step": 13864
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.911287300481809,
      "learning_rate": 3.355644026647552e-06,
      "loss": 0.6135,
      "step": 13865
    },
    {
      "epoch": 1.24,
      "grad_norm": 4.370784763008989,
      "learning_rate": 3.3549618545971552e-06,
      "loss": 0.6348,
      "step": 13866
    },
    {
      "epoch": 1.24,
      "grad_norm": 7.582992648057642,
      "learning_rate": 3.3542797168833274e-06,
      "loss": 0.6024,
      "step": 13867
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.098431327471657,
      "learning_rate": 3.353597613520306e-06,
      "loss": 0.6541,
      "step": 13868
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.731793667342119,
      "learning_rate": 3.352915544522329e-06,
      "loss": 0.5625,
      "step": 13869
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.176387696693082,
      "learning_rate": 3.3522335099036336e-06,
      "loss": 0.6352,
      "step": 13870
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.167264657918009,
      "learning_rate": 3.351551509678454e-06,
      "loss": 0.6581,
      "step": 13871
    },
    {
      "epoch": 1.24,
      "grad_norm": 7.54501724410477,
      "learning_rate": 3.3508695438610273e-06,
      "loss": 0.6405,
      "step": 13872
    },
    {
      "epoch": 1.24,
      "grad_norm": 7.153118358778522,
      "learning_rate": 3.3501876124655874e-06,
      "loss": 0.5866,
      "step": 13873
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.291358473005948,
      "learning_rate": 3.3495057155063673e-06,
      "loss": 0.6125,
      "step": 13874
    },
    {
      "epoch": 1.24,
      "grad_norm": 8.214560423136342,
      "learning_rate": 3.348823852997604e-06,
      "loss": 0.5992,
      "step": 13875
    },
    {
      "epoch": 1.24,
      "grad_norm": 4.1187240087199255,
      "learning_rate": 3.348142024953523e-06,
      "loss": 0.5912,
      "step": 13876
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.386724419656915,
      "learning_rate": 3.347460231388361e-06,
      "loss": 0.6342,
      "step": 13877
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.835162587074977,
      "learning_rate": 3.346778472316348e-06,
      "loss": 0.544,
      "step": 13878
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.719989010017606,
      "learning_rate": 3.3460967477517127e-06,
      "loss": 0.563,
      "step": 13879
    },
    {
      "epoch": 1.24,
      "grad_norm": 8.573657042383761,
      "learning_rate": 3.3454150577086876e-06,
      "loss": 0.5899,
      "step": 13880
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.649239763244389,
      "learning_rate": 3.344733402201498e-06,
      "loss": 0.5953,
      "step": 13881
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.471414105137186,
      "learning_rate": 3.344051781244374e-06,
      "loss": 0.6382,
      "step": 13882
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.674510955425737,
      "learning_rate": 3.3433701948515435e-06,
      "loss": 0.5887,
      "step": 13883
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.767751906944727,
      "learning_rate": 3.3426886430372307e-06,
      "loss": 0.6019,
      "step": 13884
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.49385661001288,
      "learning_rate": 3.3420071258156638e-06,
      "loss": 0.5548,
      "step": 13885
    },
    {
      "epoch": 1.24,
      "grad_norm": 7.108482465337122,
      "learning_rate": 3.3413256432010683e-06,
      "loss": 0.5754,
      "step": 13886
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.077595090303393,
      "learning_rate": 3.3406441952076664e-06,
      "loss": 0.5733,
      "step": 13887
    },
    {
      "epoch": 1.24,
      "grad_norm": 9.84980552106509,
      "learning_rate": 3.3399627818496827e-06,
      "loss": 0.5762,
      "step": 13888
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.207316334677537,
      "learning_rate": 3.339281403141342e-06,
      "loss": 0.576,
      "step": 13889
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.347552823153011,
      "learning_rate": 3.3386000590968654e-06,
      "loss": 0.5775,
      "step": 13890
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.5630194323878,
      "learning_rate": 3.3379187497304723e-06,
      "loss": 0.6647,
      "step": 13891
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.367205384942764,
      "learning_rate": 3.337237475056386e-06,
      "loss": 0.5469,
      "step": 13892
    },
    {
      "epoch": 1.24,
      "grad_norm": 4.879093139408942,
      "learning_rate": 3.3365562350888263e-06,
      "loss": 0.5868,
      "step": 13893
    },
    {
      "epoch": 1.24,
      "grad_norm": 8.702342431011823,
      "learning_rate": 3.335875029842012e-06,
      "loss": 0.6109,
      "step": 13894
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.516611780673539,
      "learning_rate": 3.3351938593301613e-06,
      "loss": 0.5633,
      "step": 13895
    },
    {
      "epoch": 1.24,
      "grad_norm": 7.024793299945083,
      "learning_rate": 3.3345127235674944e-06,
      "loss": 0.5977,
      "step": 13896
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.48420557814582,
      "learning_rate": 3.333831622568226e-06,
      "loss": 0.6032,
      "step": 13897
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.034627900049218,
      "learning_rate": 3.3331505563465736e-06,
      "loss": 0.6097,
      "step": 13898
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.651525007556082,
      "learning_rate": 3.3324695249167538e-06,
      "loss": 0.5998,
      "step": 13899
    },
    {
      "epoch": 1.24,
      "grad_norm": 7.592706365074424,
      "learning_rate": 3.3317885282929806e-06,
      "loss": 0.632,
      "step": 13900
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.760941586381532,
      "learning_rate": 3.3311075664894676e-06,
      "loss": 0.625,
      "step": 13901
    },
    {
      "epoch": 1.24,
      "grad_norm": 8.367649246904879,
      "learning_rate": 3.330426639520431e-06,
      "loss": 0.6341,
      "step": 13902
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.723750576591715,
      "learning_rate": 3.3297457474000804e-06,
      "loss": 0.5599,
      "step": 13903
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.779393195185581,
      "learning_rate": 3.3290648901426327e-06,
      "loss": 0.6007,
      "step": 13904
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.79752474451026,
      "learning_rate": 3.3283840677622925e-06,
      "loss": 0.6244,
      "step": 13905
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.940958944924962,
      "learning_rate": 3.3277032802732765e-06,
      "loss": 0.5736,
      "step": 13906
    },
    {
      "epoch": 1.24,
      "grad_norm": 7.040517495967791,
      "learning_rate": 3.3270225276897903e-06,
      "loss": 0.6132,
      "step": 13907
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.740321827810527,
      "learning_rate": 3.326341810026046e-06,
      "loss": 0.5642,
      "step": 13908
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.758758085012527,
      "learning_rate": 3.3256611272962514e-06,
      "loss": 0.5724,
      "step": 13909
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.960155427388263,
      "learning_rate": 3.3249804795146135e-06,
      "loss": 0.6378,
      "step": 13910
    },
    {
      "epoch": 1.24,
      "grad_norm": 8.627263009727804,
      "learning_rate": 3.324299866695339e-06,
      "loss": 0.6433,
      "step": 13911
    },
    {
      "epoch": 1.24,
      "grad_norm": 8.965456731788608,
      "learning_rate": 3.323619288852637e-06,
      "loss": 0.6392,
      "step": 13912
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.998165146975378,
      "learning_rate": 3.3229387460007095e-06,
      "loss": 0.6126,
      "step": 13913
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.6404390822475365,
      "learning_rate": 3.322258238153764e-06,
      "loss": 0.5596,
      "step": 13914
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.56688601281048,
      "learning_rate": 3.3215777653260033e-06,
      "loss": 0.6059,
      "step": 13915
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.089792394264766,
      "learning_rate": 3.3208973275316305e-06,
      "loss": 0.552,
      "step": 13916
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.390239928363491,
      "learning_rate": 3.3202169247848505e-06,
      "loss": 0.6008,
      "step": 13917
    },
    {
      "epoch": 1.24,
      "grad_norm": 7.560480473525067,
      "learning_rate": 3.3195365570998627e-06,
      "loss": 0.5421,
      "step": 13918
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.875405081928904,
      "learning_rate": 3.3188562244908697e-06,
      "loss": 0.6854,
      "step": 13919
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.851865931709931,
      "learning_rate": 3.31817592697207e-06,
      "loss": 0.6307,
      "step": 13920
    },
    {
      "epoch": 1.24,
      "grad_norm": 4.597350905407898,
      "learning_rate": 3.3174956645576644e-06,
      "loss": 0.6356,
      "step": 13921
    },
    {
      "epoch": 1.24,
      "grad_norm": 4.689732060899616,
      "learning_rate": 3.316815437261853e-06,
      "loss": 0.5917,
      "step": 13922
    },
    {
      "epoch": 1.24,
      "grad_norm": 8.696663644990037,
      "learning_rate": 3.3161352450988337e-06,
      "loss": 0.6373,
      "step": 13923
    },
    {
      "epoch": 1.24,
      "grad_norm": 4.525005208182841,
      "learning_rate": 3.3154550880828025e-06,
      "loss": 0.5755,
      "step": 13924
    },
    {
      "epoch": 1.24,
      "grad_norm": 7.714016680325924,
      "learning_rate": 3.314774966227958e-06,
      "loss": 0.587,
      "step": 13925
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.581185838256802,
      "learning_rate": 3.3140948795484952e-06,
      "loss": 0.6407,
      "step": 13926
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.9696007597262435,
      "learning_rate": 3.3134148280586104e-06,
      "loss": 0.565,
      "step": 13927
    },
    {
      "epoch": 1.24,
      "grad_norm": 8.226026397727015,
      "learning_rate": 3.312734811772497e-06,
      "loss": 0.612,
      "step": 13928
    },
    {
      "epoch": 1.24,
      "grad_norm": 7.337696780749672,
      "learning_rate": 3.312054830704348e-06,
      "loss": 0.6475,
      "step": 13929
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.901229962925826,
      "learning_rate": 3.31137488486836e-06,
      "loss": 0.6021,
      "step": 13930
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.307436812868264,
      "learning_rate": 3.310694974278722e-06,
      "loss": 0.5735,
      "step": 13931
    },
    {
      "epoch": 1.24,
      "grad_norm": 4.19860041895276,
      "learning_rate": 3.310015098949629e-06,
      "loss": 0.62,
      "step": 13932
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.701268129243264,
      "learning_rate": 3.3093352588952677e-06,
      "loss": 0.5944,
      "step": 13933
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.785709252767324,
      "learning_rate": 3.308655454129831e-06,
      "loss": 0.5876,
      "step": 13934
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.525936507433927,
      "learning_rate": 3.307975684667507e-06,
      "loss": 0.5913,
      "step": 13935
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.0085616853388775,
      "learning_rate": 3.3072959505224855e-06,
      "loss": 0.5911,
      "step": 13936
    },
    {
      "epoch": 1.24,
      "grad_norm": 9.018474483307616,
      "learning_rate": 3.306616251708955e-06,
      "loss": 0.6248,
      "step": 13937
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.494178558542679,
      "learning_rate": 3.3059365882411004e-06,
      "loss": 0.5909,
      "step": 13938
    },
    {
      "epoch": 1.24,
      "grad_norm": 8.937744527601746,
      "learning_rate": 3.3052569601331096e-06,
      "loss": 0.6206,
      "step": 13939
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.665924616097426,
      "learning_rate": 3.30457736739917e-06,
      "loss": 0.5967,
      "step": 13940
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.170349005519378,
      "learning_rate": 3.303897810053463e-06,
      "loss": 0.6451,
      "step": 13941
    },
    {
      "epoch": 1.24,
      "grad_norm": 7.132660501036077,
      "learning_rate": 3.3032182881101755e-06,
      "loss": 0.5742,
      "step": 13942
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.142602706370649,
      "learning_rate": 3.3025388015834907e-06,
      "loss": 0.6059,
      "step": 13943
    },
    {
      "epoch": 1.24,
      "grad_norm": 4.893511869243785,
      "learning_rate": 3.3018593504875905e-06,
      "loss": 0.6091,
      "step": 13944
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.794473494485555,
      "learning_rate": 3.3011799348366573e-06,
      "loss": 0.6165,
      "step": 13945
    },
    {
      "epoch": 1.24,
      "grad_norm": 7.596216671876299,
      "learning_rate": 3.3005005546448742e-06,
      "loss": 0.6361,
      "step": 13946
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.1601907055207645,
      "learning_rate": 3.2998212099264206e-06,
      "loss": 0.5645,
      "step": 13947
    },
    {
      "epoch": 1.24,
      "grad_norm": 7.225906915174299,
      "learning_rate": 3.299141900695474e-06,
      "loss": 0.6181,
      "step": 13948
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.813350845986244,
      "learning_rate": 3.2984626269662156e-06,
      "loss": 0.5525,
      "step": 13949
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.340626314796184,
      "learning_rate": 3.297783388752824e-06,
      "loss": 0.5871,
      "step": 13950
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.8822046604436515,
      "learning_rate": 3.2971041860694765e-06,
      "loss": 0.5647,
      "step": 13951
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.685727765431623,
      "learning_rate": 3.2964250189303486e-06,
      "loss": 0.6093,
      "step": 13952
    },
    {
      "epoch": 1.24,
      "grad_norm": 7.117931502683133,
      "learning_rate": 3.2957458873496194e-06,
      "loss": 0.5839,
      "step": 13953
    },
    {
      "epoch": 1.24,
      "grad_norm": 8.837130059155726,
      "learning_rate": 3.2950667913414614e-06,
      "loss": 0.6292,
      "step": 13954
    },
    {
      "epoch": 1.24,
      "grad_norm": 8.430477627141649,
      "learning_rate": 3.29438773092005e-06,
      "loss": 0.6248,
      "step": 13955
    },
    {
      "epoch": 1.25,
      "grad_norm": 7.248600283537663,
      "learning_rate": 3.293708706099561e-06,
      "loss": 0.6677,
      "step": 13956
    },
    {
      "epoch": 1.25,
      "grad_norm": 10.242730343810818,
      "learning_rate": 3.293029716894165e-06,
      "loss": 0.6643,
      "step": 13957
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.076699233082732,
      "learning_rate": 3.2923507633180353e-06,
      "loss": 0.636,
      "step": 13958
    },
    {
      "epoch": 1.25,
      "grad_norm": 9.333338623987936,
      "learning_rate": 3.2916718453853445e-06,
      "loss": 0.5913,
      "step": 13959
    },
    {
      "epoch": 1.25,
      "grad_norm": 9.467258616390868,
      "learning_rate": 3.290992963110262e-06,
      "loss": 0.5665,
      "step": 13960
    },
    {
      "epoch": 1.25,
      "grad_norm": 7.29823500933735,
      "learning_rate": 3.2903141165069607e-06,
      "loss": 0.6299,
      "step": 13961
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.2532297487837365,
      "learning_rate": 3.289635305589606e-06,
      "loss": 0.6508,
      "step": 13962
    },
    {
      "epoch": 1.25,
      "grad_norm": 5.60076781389035,
      "learning_rate": 3.28895653037237e-06,
      "loss": 0.6628,
      "step": 13963
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.401855859392039,
      "learning_rate": 3.2882777908694175e-06,
      "loss": 0.6101,
      "step": 13964
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.602857243994107,
      "learning_rate": 3.2875990870949184e-06,
      "loss": 0.5767,
      "step": 13965
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.675746954187519,
      "learning_rate": 3.286920419063039e-06,
      "loss": 0.5749,
      "step": 13966
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.651322710025811,
      "learning_rate": 3.2862417867879432e-06,
      "loss": 0.5619,
      "step": 13967
    },
    {
      "epoch": 1.25,
      "grad_norm": 8.106140528568869,
      "learning_rate": 3.285563190283797e-06,
      "loss": 0.6639,
      "step": 13968
    },
    {
      "epoch": 1.25,
      "grad_norm": 5.006326743248452,
      "learning_rate": 3.2848846295647653e-06,
      "loss": 0.5506,
      "step": 13969
    },
    {
      "epoch": 1.25,
      "grad_norm": 9.031778398428461,
      "learning_rate": 3.28420610464501e-06,
      "loss": 0.5631,
      "step": 13970
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.539257698421576,
      "learning_rate": 3.283527615538695e-06,
      "loss": 0.5742,
      "step": 13971
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.33487060869525,
      "learning_rate": 3.2828491622599836e-06,
      "loss": 0.5836,
      "step": 13972
    },
    {
      "epoch": 1.25,
      "grad_norm": 9.950658585131695,
      "learning_rate": 3.282170744823034e-06,
      "loss": 0.6077,
      "step": 13973
    },
    {
      "epoch": 1.25,
      "grad_norm": 8.115665151983794,
      "learning_rate": 3.281492363242008e-06,
      "loss": 0.5845,
      "step": 13974
    },
    {
      "epoch": 1.25,
      "grad_norm": 8.47193439688196,
      "learning_rate": 3.2808140175310667e-06,
      "loss": 0.6087,
      "step": 13975
    },
    {
      "epoch": 1.25,
      "grad_norm": 4.699346551043925,
      "learning_rate": 3.2801357077043682e-06,
      "loss": 0.5472,
      "step": 13976
    },
    {
      "epoch": 1.25,
      "grad_norm": 4.772496905651359,
      "learning_rate": 3.2794574337760687e-06,
      "loss": 0.5852,
      "step": 13977
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.579831838089926,
      "learning_rate": 3.2787791957603276e-06,
      "loss": 0.653,
      "step": 13978
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.406047262141717,
      "learning_rate": 3.2781009936713025e-06,
      "loss": 0.5939,
      "step": 13979
    },
    {
      "epoch": 1.25,
      "grad_norm": 7.740205873365283,
      "learning_rate": 3.2774228275231467e-06,
      "loss": 0.6005,
      "step": 13980
    },
    {
      "epoch": 1.25,
      "grad_norm": 8.050623770906041,
      "learning_rate": 3.276744697330017e-06,
      "loss": 0.5578,
      "step": 13981
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.465004719658058,
      "learning_rate": 3.2760666031060696e-06,
      "loss": 0.6001,
      "step": 13982
    },
    {
      "epoch": 1.25,
      "grad_norm": 10.2661484780485,
      "learning_rate": 3.275388544865455e-06,
      "loss": 0.61,
      "step": 13983
    },
    {
      "epoch": 1.25,
      "grad_norm": 5.9270552099218685,
      "learning_rate": 3.2747105226223274e-06,
      "loss": 0.5817,
      "step": 13984
    },
    {
      "epoch": 1.25,
      "grad_norm": 8.289370457237832,
      "learning_rate": 3.274032536390841e-06,
      "loss": 0.6086,
      "step": 13985
    },
    {
      "epoch": 1.25,
      "grad_norm": 8.133253410738387,
      "learning_rate": 3.273354586185144e-06,
      "loss": 0.6088,
      "step": 13986
    },
    {
      "epoch": 1.25,
      "grad_norm": 4.254965304405587,
      "learning_rate": 3.272676672019389e-06,
      "loss": 0.6175,
      "step": 13987
    },
    {
      "epoch": 1.25,
      "grad_norm": 7.047914348978449,
      "learning_rate": 3.271998793907727e-06,
      "loss": 0.5885,
      "step": 13988
    },
    {
      "epoch": 1.25,
      "grad_norm": 7.129035482451844,
      "learning_rate": 3.271320951864304e-06,
      "loss": 0.6039,
      "step": 13989
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.625092939230338,
      "learning_rate": 3.270643145903273e-06,
      "loss": 0.5668,
      "step": 13990
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.4817455218576,
      "learning_rate": 3.2699653760387774e-06,
      "loss": 0.6306,
      "step": 13991
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.865737669961547,
      "learning_rate": 3.269287642284966e-06,
      "loss": 0.589,
      "step": 13992
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.700508681165714,
      "learning_rate": 3.268609944655985e-06,
      "loss": 0.6489,
      "step": 13993
    },
    {
      "epoch": 1.25,
      "grad_norm": 7.240287628185725,
      "learning_rate": 3.2679322831659786e-06,
      "loss": 0.6193,
      "step": 13994
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.578733561261825,
      "learning_rate": 3.2672546578290943e-06,
      "loss": 0.5626,
      "step": 13995
    },
    {
      "epoch": 1.25,
      "grad_norm": 7.499349563288458,
      "learning_rate": 3.266577068659473e-06,
      "loss": 0.678,
      "step": 13996
    },
    {
      "epoch": 1.25,
      "grad_norm": 7.514611635801977,
      "learning_rate": 3.2658995156712605e-06,
      "loss": 0.6471,
      "step": 13997
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.82055186940561,
      "learning_rate": 3.2652219988785973e-06,
      "loss": 0.524,
      "step": 13998
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.4895029189305085,
      "learning_rate": 3.264544518295626e-06,
      "loss": 0.592,
      "step": 13999
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.223296733041059,
      "learning_rate": 3.2638670739364865e-06,
      "loss": 0.6026,
      "step": 14000
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.7762669518419925,
      "learning_rate": 3.2631896658153205e-06,
      "loss": 0.6264,
      "step": 14001
    },
    {
      "epoch": 1.25,
      "grad_norm": 8.187894812259207,
      "learning_rate": 3.262512293946267e-06,
      "loss": 0.6013,
      "step": 14002
    },
    {
      "epoch": 1.25,
      "grad_norm": 5.285959247229007,
      "learning_rate": 3.261834958343464e-06,
      "loss": 0.6019,
      "step": 14003
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.763980672833974,
      "learning_rate": 3.2611576590210492e-06,
      "loss": 0.5354,
      "step": 14004
    },
    {
      "epoch": 1.25,
      "grad_norm": 4.717877662923218,
      "learning_rate": 3.2604803959931623e-06,
      "loss": 0.6162,
      "step": 14005
    },
    {
      "epoch": 1.25,
      "grad_norm": 7.452128334960911,
      "learning_rate": 3.2598031692739365e-06,
      "loss": 0.6122,
      "step": 14006
    },
    {
      "epoch": 1.25,
      "grad_norm": 7.141084431190575,
      "learning_rate": 3.2591259788775078e-06,
      "loss": 0.6436,
      "step": 14007
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.85355605203259,
      "learning_rate": 3.2584488248180124e-06,
      "loss": 0.5693,
      "step": 14008
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.437154818801018,
      "learning_rate": 3.2577717071095827e-06,
      "loss": 0.6302,
      "step": 14009
    },
    {
      "epoch": 1.25,
      "grad_norm": 10.103423573237823,
      "learning_rate": 3.2570946257663548e-06,
      "loss": 0.6152,
      "step": 14010
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.494533312422914,
      "learning_rate": 3.2564175808024588e-06,
      "loss": 0.5978,
      "step": 14011
    },
    {
      "epoch": 1.25,
      "grad_norm": 7.397635371123472,
      "learning_rate": 3.255740572232027e-06,
      "loss": 0.5769,
      "step": 14012
    },
    {
      "epoch": 1.25,
      "grad_norm": 7.127067363830593,
      "learning_rate": 3.2550636000691925e-06,
      "loss": 0.5979,
      "step": 14013
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.504548071172773,
      "learning_rate": 3.254386664328083e-06,
      "loss": 0.6572,
      "step": 14014
    },
    {
      "epoch": 1.25,
      "grad_norm": 7.3382298123127745,
      "learning_rate": 3.2537097650228284e-06,
      "loss": 0.6494,
      "step": 14015
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.368460014365391,
      "learning_rate": 3.2530329021675594e-06,
      "loss": 0.6349,
      "step": 14016
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.354730723266199,
      "learning_rate": 3.252356075776402e-06,
      "loss": 0.6334,
      "step": 14017
    },
    {
      "epoch": 1.25,
      "grad_norm": 7.6006898669149905,
      "learning_rate": 3.251679285863484e-06,
      "loss": 0.608,
      "step": 14018
    },
    {
      "epoch": 1.25,
      "grad_norm": 5.716010870220275,
      "learning_rate": 3.251002532442934e-06,
      "loss": 0.5879,
      "step": 14019
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.191421136181028,
      "learning_rate": 3.2503258155288753e-06,
      "loss": 0.6326,
      "step": 14020
    },
    {
      "epoch": 1.25,
      "grad_norm": 5.664309025663215,
      "learning_rate": 3.249649135135432e-06,
      "loss": 0.6345,
      "step": 14021
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.252364430084975,
      "learning_rate": 3.2489724912767307e-06,
      "loss": 0.653,
      "step": 14022
    },
    {
      "epoch": 1.25,
      "grad_norm": 7.952643112715204,
      "learning_rate": 3.2482958839668943e-06,
      "loss": 0.6471,
      "step": 14023
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.352516092276112,
      "learning_rate": 3.247619313220044e-06,
      "loss": 0.6243,
      "step": 14024
    },
    {
      "epoch": 1.25,
      "grad_norm": 5.751697129902406,
      "learning_rate": 3.246942779050303e-06,
      "loss": 0.5707,
      "step": 14025
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.154606974028246,
      "learning_rate": 3.2462662814717937e-06,
      "loss": 0.5536,
      "step": 14026
    },
    {
      "epoch": 1.25,
      "grad_norm": 5.775743085263744,
      "learning_rate": 3.2455898204986337e-06,
      "loss": 0.6454,
      "step": 14027
    },
    {
      "epoch": 1.25,
      "grad_norm": 7.162330855616766,
      "learning_rate": 3.2449133961449442e-06,
      "loss": 0.5301,
      "step": 14028
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.88266655719037,
      "learning_rate": 3.2442370084248452e-06,
      "loss": 0.5358,
      "step": 14029
    },
    {
      "epoch": 1.25,
      "grad_norm": 5.922234468299968,
      "learning_rate": 3.2435606573524526e-06,
      "loss": 0.5711,
      "step": 14030
    },
    {
      "epoch": 1.25,
      "grad_norm": 5.370064059684683,
      "learning_rate": 3.242884342941884e-06,
      "loss": 0.6056,
      "step": 14031
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.043947766989689,
      "learning_rate": 3.2422080652072584e-06,
      "loss": 0.579,
      "step": 14032
    },
    {
      "epoch": 1.25,
      "grad_norm": 5.530701200418585,
      "learning_rate": 3.2415318241626914e-06,
      "loss": 0.5681,
      "step": 14033
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.977076706723662,
      "learning_rate": 3.240855619822294e-06,
      "loss": 0.618,
      "step": 14034
    },
    {
      "epoch": 1.25,
      "grad_norm": 4.911226015940273,
      "learning_rate": 3.2401794522001834e-06,
      "loss": 0.6203,
      "step": 14035
    },
    {
      "epoch": 1.25,
      "grad_norm": 7.429399018636236,
      "learning_rate": 3.2395033213104733e-06,
      "loss": 0.5781,
      "step": 14036
    },
    {
      "epoch": 1.25,
      "grad_norm": 5.227652894084293,
      "learning_rate": 3.2388272271672755e-06,
      "loss": 0.6354,
      "step": 14037
    },
    {
      "epoch": 1.25,
      "grad_norm": 9.794424396011177,
      "learning_rate": 3.2381511697847023e-06,
      "loss": 0.5291,
      "step": 14038
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.137166664493985,
      "learning_rate": 3.2374751491768654e-06,
      "loss": 0.6176,
      "step": 14039
    },
    {
      "epoch": 1.25,
      "grad_norm": 5.872228377459595,
      "learning_rate": 3.2367991653578747e-06,
      "loss": 0.6223,
      "step": 14040
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.327793655332198,
      "learning_rate": 3.2361232183418396e-06,
      "loss": 0.5805,
      "step": 14041
    },
    {
      "epoch": 1.25,
      "grad_norm": 5.057526886048322,
      "learning_rate": 3.2354473081428704e-06,
      "loss": 0.5553,
      "step": 14042
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.877476898441335,
      "learning_rate": 3.2347714347750735e-06,
      "loss": 0.6284,
      "step": 14043
    },
    {
      "epoch": 1.25,
      "grad_norm": 8.098282895244523,
      "learning_rate": 3.234095598252558e-06,
      "loss": 0.6255,
      "step": 14044
    },
    {
      "epoch": 1.25,
      "grad_norm": 4.858865527794053,
      "learning_rate": 3.2334197985894293e-06,
      "loss": 0.612,
      "step": 14045
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.838535982802868,
      "learning_rate": 3.232744035799793e-06,
      "loss": 0.6503,
      "step": 14046
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.916111469378982,
      "learning_rate": 3.232068309897755e-06,
      "loss": 0.6635,
      "step": 14047
    },
    {
      "epoch": 1.25,
      "grad_norm": 5.086104763482101,
      "learning_rate": 3.2313926208974213e-06,
      "loss": 0.589,
      "step": 14048
    },
    {
      "epoch": 1.25,
      "grad_norm": 8.212442369289526,
      "learning_rate": 3.2307169688128924e-06,
      "loss": 0.5985,
      "step": 14049
    },
    {
      "epoch": 1.25,
      "grad_norm": 8.0933866702693,
      "learning_rate": 3.2300413536582708e-06,
      "loss": 0.6096,
      "step": 14050
    },
    {
      "epoch": 1.25,
      "grad_norm": 5.9955381500736715,
      "learning_rate": 3.2293657754476603e-06,
      "loss": 0.6113,
      "step": 14051
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.912352665417203,
      "learning_rate": 3.228690234195162e-06,
      "loss": 0.6096,
      "step": 14052
    },
    {
      "epoch": 1.25,
      "grad_norm": 7.017728273888986,
      "learning_rate": 3.2280147299148754e-06,
      "loss": 0.6133,
      "step": 14053
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.118768034781843,
      "learning_rate": 3.2273392626209005e-06,
      "loss": 0.639,
      "step": 14054
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.064692534375806,
      "learning_rate": 3.2266638323273376e-06,
      "loss": 0.5613,
      "step": 14055
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.524256294823813,
      "learning_rate": 3.2259884390482828e-06,
      "loss": 0.604,
      "step": 14056
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.91168380553558,
      "learning_rate": 3.225313082797834e-06,
      "loss": 0.5876,
      "step": 14057
    },
    {
      "epoch": 1.25,
      "grad_norm": 5.871226559791408,
      "learning_rate": 3.224637763590089e-06,
      "loss": 0.5646,
      "step": 14058
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.461172584248612,
      "learning_rate": 3.2239624814391422e-06,
      "loss": 0.6178,
      "step": 14059
    },
    {
      "epoch": 1.25,
      "grad_norm": 5.5957970224309435,
      "learning_rate": 3.223287236359088e-06,
      "loss": 0.597,
      "step": 14060
    },
    {
      "epoch": 1.25,
      "grad_norm": 7.5037045824599815,
      "learning_rate": 3.2226120283640246e-06,
      "loss": 0.625,
      "step": 14061
    },
    {
      "epoch": 1.25,
      "grad_norm": 5.770002719089063,
      "learning_rate": 3.2219368574680427e-06,
      "loss": 0.6022,
      "step": 14062
    },
    {
      "epoch": 1.25,
      "grad_norm": 7.105514596629109,
      "learning_rate": 3.221261723685233e-06,
      "loss": 0.5898,
      "step": 14063
    },
    {
      "epoch": 1.25,
      "grad_norm": 7.349512656303862,
      "learning_rate": 3.2205866270296894e-06,
      "loss": 0.5582,
      "step": 14064
    },
    {
      "epoch": 1.25,
      "grad_norm": 7.228682225740329,
      "learning_rate": 3.2199115675155047e-06,
      "loss": 0.6351,
      "step": 14065
    },
    {
      "epoch": 1.25,
      "grad_norm": 4.717447970681148,
      "learning_rate": 3.219236545156766e-06,
      "loss": 0.5666,
      "step": 14066
    },
    {
      "epoch": 1.25,
      "grad_norm": 7.34503655980662,
      "learning_rate": 3.2185615599675644e-06,
      "loss": 0.6204,
      "step": 14067
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.375518686439419,
      "learning_rate": 3.217886611961991e-06,
      "loss": 0.6295,
      "step": 14068
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.582980458269471,
      "learning_rate": 3.21721170115413e-06,
      "loss": 0.6181,
      "step": 14069
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.4940951793199115,
      "learning_rate": 3.216536827558071e-06,
      "loss": 0.6215,
      "step": 14070
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.352982394230798,
      "learning_rate": 3.2158619911879006e-06,
      "loss": 0.6249,
      "step": 14071
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.557162244642264,
      "learning_rate": 3.2151871920577027e-06,
      "loss": 0.5511,
      "step": 14072
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.585553960495882,
      "learning_rate": 3.214512430181564e-06,
      "loss": 0.6552,
      "step": 14073
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.980666124117477,
      "learning_rate": 3.213837705573569e-06,
      "loss": 0.6051,
      "step": 14074
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.88610994155848,
      "learning_rate": 3.213163018247799e-06,
      "loss": 0.6219,
      "step": 14075
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.165266763434706,
      "learning_rate": 3.2124883682183383e-06,
      "loss": 0.5902,
      "step": 14076
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.008425408333757,
      "learning_rate": 3.21181375549927e-06,
      "loss": 0.5713,
      "step": 14077
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.65602781235058,
      "learning_rate": 3.2111391801046732e-06,
      "loss": 0.6194,
      "step": 14078
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.516092309930541,
      "learning_rate": 3.2104646420486268e-06,
      "loss": 0.5737,
      "step": 14079
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.398987715829651,
      "learning_rate": 3.209790141345213e-06,
      "loss": 0.6025,
      "step": 14080
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.456942850662524,
      "learning_rate": 3.2091156780085087e-06,
      "loss": 0.5516,
      "step": 14081
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.9105840715343785,
      "learning_rate": 3.2084412520525933e-06,
      "loss": 0.6267,
      "step": 14082
    },
    {
      "epoch": 1.26,
      "grad_norm": 4.777697733915076,
      "learning_rate": 3.2077668634915427e-06,
      "loss": 0.6089,
      "step": 14083
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.3263791192739145,
      "learning_rate": 3.2070925123394346e-06,
      "loss": 0.5705,
      "step": 14084
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.640599910622072,
      "learning_rate": 3.2064181986103428e-06,
      "loss": 0.6203,
      "step": 14085
    },
    {
      "epoch": 1.26,
      "grad_norm": 8.381911654619529,
      "learning_rate": 3.205743922318345e-06,
      "loss": 0.5969,
      "step": 14086
    },
    {
      "epoch": 1.26,
      "grad_norm": 9.23345393161523,
      "learning_rate": 3.2050696834775117e-06,
      "loss": 0.6113,
      "step": 14087
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.470943311337223,
      "learning_rate": 3.2043954821019185e-06,
      "loss": 0.6052,
      "step": 14088
    },
    {
      "epoch": 1.26,
      "grad_norm": 8.531897006932326,
      "learning_rate": 3.203721318205638e-06,
      "loss": 0.5821,
      "step": 14089
    },
    {
      "epoch": 1.26,
      "grad_norm": 4.628243889667987,
      "learning_rate": 3.20304719180274e-06,
      "loss": 0.6404,
      "step": 14090
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.020132308061576,
      "learning_rate": 3.202373102907299e-06,
      "loss": 0.5998,
      "step": 14091
    },
    {
      "epoch": 1.26,
      "grad_norm": 4.846589469385847,
      "learning_rate": 3.2016990515333807e-06,
      "loss": 0.5683,
      "step": 14092
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.0800028546283515,
      "learning_rate": 3.201025037695056e-06,
      "loss": 0.6133,
      "step": 14093
    },
    {
      "epoch": 1.26,
      "grad_norm": 8.144582123391693,
      "learning_rate": 3.200351061406396e-06,
      "loss": 0.6223,
      "step": 14094
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.990833408354744,
      "learning_rate": 3.199677122681465e-06,
      "loss": 0.5656,
      "step": 14095
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.110028118299155,
      "learning_rate": 3.1990032215343316e-06,
      "loss": 0.6087,
      "step": 14096
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.326134812119718,
      "learning_rate": 3.198329357979062e-06,
      "loss": 0.5586,
      "step": 14097
    },
    {
      "epoch": 1.26,
      "grad_norm": 8.913633186874835,
      "learning_rate": 3.197655532029721e-06,
      "loss": 0.6473,
      "step": 14098
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.979730222666931,
      "learning_rate": 3.1969817437003743e-06,
      "loss": 0.6096,
      "step": 14099
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.58602061451955,
      "learning_rate": 3.1963079930050846e-06,
      "loss": 0.6076,
      "step": 14100
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.11562058986664,
      "learning_rate": 3.1956342799579154e-06,
      "loss": 0.606,
      "step": 14101
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.6349703063701355,
      "learning_rate": 3.19496060457293e-06,
      "loss": 0.6299,
      "step": 14102
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.244702639160508,
      "learning_rate": 3.194286966864188e-06,
      "loss": 0.6417,
      "step": 14103
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.795554538535375,
      "learning_rate": 3.1936133668457516e-06,
      "loss": 0.5842,
      "step": 14104
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.423737124255248,
      "learning_rate": 3.192939804531681e-06,
      "loss": 0.5723,
      "step": 14105
    },
    {
      "epoch": 1.26,
      "grad_norm": 9.131434711476878,
      "learning_rate": 3.192266279936035e-06,
      "loss": 0.5891,
      "step": 14106
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.115364691362743,
      "learning_rate": 3.1915927930728695e-06,
      "loss": 0.5603,
      "step": 14107
    },
    {
      "epoch": 1.26,
      "grad_norm": 9.13187605486909,
      "learning_rate": 3.1909193439562445e-06,
      "loss": 0.6661,
      "step": 14108
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.672533579132917,
      "learning_rate": 3.190245932600218e-06,
      "loss": 0.5161,
      "step": 14109
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.588612999070721,
      "learning_rate": 3.189572559018843e-06,
      "loss": 0.5627,
      "step": 14110
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.9783156312184165,
      "learning_rate": 3.1888992232261763e-06,
      "loss": 0.6071,
      "step": 14111
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.139964631202621,
      "learning_rate": 3.1882259252362724e-06,
      "loss": 0.5318,
      "step": 14112
    },
    {
      "epoch": 1.26,
      "grad_norm": 8.222890383282024,
      "learning_rate": 3.187552665063184e-06,
      "loss": 0.5837,
      "step": 14113
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.274210314824888,
      "learning_rate": 3.1868794427209636e-06,
      "loss": 0.5464,
      "step": 14114
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.179620590108454,
      "learning_rate": 3.1862062582236664e-06,
      "loss": 0.5937,
      "step": 14115
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.483164541210393,
      "learning_rate": 3.1855331115853398e-06,
      "loss": 0.5583,
      "step": 14116
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.209258549408169,
      "learning_rate": 3.184860002820036e-06,
      "loss": 0.593,
      "step": 14117
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.163140042605447,
      "learning_rate": 3.1841869319418063e-06,
      "loss": 0.5967,
      "step": 14118
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.558458251520648,
      "learning_rate": 3.183513898964696e-06,
      "loss": 0.6175,
      "step": 14119
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.121149138112046,
      "learning_rate": 3.1828409039027568e-06,
      "loss": 0.5603,
      "step": 14120
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.213865146458191,
      "learning_rate": 3.1821679467700328e-06,
      "loss": 0.6213,
      "step": 14121
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.0368464368196495,
      "learning_rate": 3.1814950275805727e-06,
      "loss": 0.6484,
      "step": 14122
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.554890613382404,
      "learning_rate": 3.18082214634842e-06,
      "loss": 0.5154,
      "step": 14123
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.962880022061275,
      "learning_rate": 3.180149303087622e-06,
      "loss": 0.5608,
      "step": 14124
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.644969443239955,
      "learning_rate": 3.179476497812223e-06,
      "loss": 0.5356,
      "step": 14125
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.095671495879797,
      "learning_rate": 3.178803730536263e-06,
      "loss": 0.5807,
      "step": 14126
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.589783169221604,
      "learning_rate": 3.1781310012737875e-06,
      "loss": 0.6465,
      "step": 14127
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.56471163591941,
      "learning_rate": 3.1774583100388384e-06,
      "loss": 0.5792,
      "step": 14128
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.600816897701473,
      "learning_rate": 3.176785656845455e-06,
      "loss": 0.555,
      "step": 14129
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.192400339800826,
      "learning_rate": 3.176113041707678e-06,
      "loss": 0.5905,
      "step": 14130
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.198299329290172,
      "learning_rate": 3.175440464639548e-06,
      "loss": 0.6211,
      "step": 14131
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.834340001191554,
      "learning_rate": 3.1747679256551017e-06,
      "loss": 0.6648,
      "step": 14132
    },
    {
      "epoch": 1.26,
      "grad_norm": 8.267287598992556,
      "learning_rate": 3.1740954247683776e-06,
      "loss": 0.6002,
      "step": 14133
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.302201742955359,
      "learning_rate": 3.1734229619934152e-06,
      "loss": 0.5412,
      "step": 14134
    },
    {
      "epoch": 1.26,
      "grad_norm": 11.219836003502474,
      "learning_rate": 3.172750537344247e-06,
      "loss": 0.5841,
      "step": 14135
    },
    {
      "epoch": 1.26,
      "grad_norm": 4.8893354557429385,
      "learning_rate": 3.1720781508349088e-06,
      "loss": 0.5507,
      "step": 14136
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.008734839533168,
      "learning_rate": 3.171405802479436e-06,
      "loss": 0.6141,
      "step": 14137
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.562817912046149,
      "learning_rate": 3.170733492291864e-06,
      "loss": 0.6277,
      "step": 14138
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.773892380061328,
      "learning_rate": 3.1700612202862225e-06,
      "loss": 0.6391,
      "step": 14139
    },
    {
      "epoch": 1.26,
      "grad_norm": 9.434446282513486,
      "learning_rate": 3.1693889864765464e-06,
      "loss": 0.5655,
      "step": 14140
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.386640838658993,
      "learning_rate": 3.1687167908768667e-06,
      "loss": 0.5878,
      "step": 14141
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.248533301097855,
      "learning_rate": 3.1680446335012127e-06,
      "loss": 0.5839,
      "step": 14142
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.786753471239798,
      "learning_rate": 3.167372514363615e-06,
      "loss": 0.6191,
      "step": 14143
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.007300929580455,
      "learning_rate": 3.166700433478104e-06,
      "loss": 0.659,
      "step": 14144
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.875141880568826,
      "learning_rate": 3.166028390858705e-06,
      "loss": 0.5883,
      "step": 14145
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.199897981137047,
      "learning_rate": 3.1653563865194474e-06,
      "loss": 0.6088,
      "step": 14146
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.762887981197458,
      "learning_rate": 3.1646844204743585e-06,
      "loss": 0.5439,
      "step": 14147
    },
    {
      "epoch": 1.26,
      "grad_norm": 8.308055234903481,
      "learning_rate": 3.164012492737462e-06,
      "loss": 0.6333,
      "step": 14148
    },
    {
      "epoch": 1.26,
      "grad_norm": 8.567579373020022,
      "learning_rate": 3.163340603322786e-06,
      "loss": 0.591,
      "step": 14149
    },
    {
      "epoch": 1.26,
      "grad_norm": 4.454007350599428,
      "learning_rate": 3.1626687522443508e-06,
      "loss": 0.5587,
      "step": 14150
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.24843520550573,
      "learning_rate": 3.161996939516182e-06,
      "loss": 0.6222,
      "step": 14151
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.698131516576222,
      "learning_rate": 3.1613251651523014e-06,
      "loss": 0.5988,
      "step": 14152
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.018300918304478,
      "learning_rate": 3.1606534291667317e-06,
      "loss": 0.6014,
      "step": 14153
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.329832645434805,
      "learning_rate": 3.1599817315734944e-06,
      "loss": 0.624,
      "step": 14154
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.470105756553675,
      "learning_rate": 3.1593100723866075e-06,
      "loss": 0.5744,
      "step": 14155
    },
    {
      "epoch": 1.26,
      "grad_norm": 10.475704580143065,
      "learning_rate": 3.158638451620092e-06,
      "loss": 0.6055,
      "step": 14156
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.275631950307534,
      "learning_rate": 3.157966869287967e-06,
      "loss": 0.5892,
      "step": 14157
    },
    {
      "epoch": 1.26,
      "grad_norm": 8.976031971233075,
      "learning_rate": 3.157295325404249e-06,
      "loss": 0.6645,
      "step": 14158
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.830912851027432,
      "learning_rate": 3.1566238199829556e-06,
      "loss": 0.5381,
      "step": 14159
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.172642167010969,
      "learning_rate": 3.1559523530381044e-06,
      "loss": 0.6004,
      "step": 14160
    },
    {
      "epoch": 1.26,
      "grad_norm": 4.651443111597223,
      "learning_rate": 3.155280924583708e-06,
      "loss": 0.6057,
      "step": 14161
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.809900001974165,
      "learning_rate": 3.1546095346337834e-06,
      "loss": 0.578,
      "step": 14162
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.194748920194539,
      "learning_rate": 3.1539381832023436e-06,
      "loss": 0.5593,
      "step": 14163
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.301999716506646,
      "learning_rate": 3.1532668703034015e-06,
      "loss": 0.575,
      "step": 14164
    },
    {
      "epoch": 1.26,
      "grad_norm": 11.422045344549227,
      "learning_rate": 3.152595595950968e-06,
      "loss": 0.6911,
      "step": 14165
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.754563147395034,
      "learning_rate": 3.1519243601590555e-06,
      "loss": 0.5758,
      "step": 14166
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.04495346712206,
      "learning_rate": 3.1512531629416765e-06,
      "loss": 0.6328,
      "step": 14167
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.268071540827965,
      "learning_rate": 3.150582004312837e-06,
      "loss": 0.5732,
      "step": 14168
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.248758159492901,
      "learning_rate": 3.149910884286549e-06,
      "loss": 0.5898,
      "step": 14169
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.524830157928618,
      "learning_rate": 3.14923980287682e-06,
      "loss": 0.5744,
      "step": 14170
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.833774137232275,
      "learning_rate": 3.148568760097656e-06,
      "loss": 0.6075,
      "step": 14171
    },
    {
      "epoch": 1.26,
      "grad_norm": 8.83540598615519,
      "learning_rate": 3.147897755963066e-06,
      "loss": 0.6947,
      "step": 14172
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.412120982449695,
      "learning_rate": 3.147226790487053e-06,
      "loss": 0.6104,
      "step": 14173
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.441377585468815,
      "learning_rate": 3.1465558636836235e-06,
      "loss": 0.5161,
      "step": 14174
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.557013295369439,
      "learning_rate": 3.145884975566782e-06,
      "loss": 0.5673,
      "step": 14175
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.5969036503025595,
      "learning_rate": 3.1452141261505302e-06,
      "loss": 0.5509,
      "step": 14176
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.387687635941349,
      "learning_rate": 3.144543315448872e-06,
      "loss": 0.6066,
      "step": 14177
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.89059027690157,
      "learning_rate": 3.1438725434758103e-06,
      "loss": 0.6495,
      "step": 14178
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.205531193282285,
      "learning_rate": 3.1432018102453425e-06,
      "loss": 0.5976,
      "step": 14179
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.788578657967702,
      "learning_rate": 3.142531115771472e-06,
      "loss": 0.6095,
      "step": 14180
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.498455488771687,
      "learning_rate": 3.1418604600681955e-06,
      "loss": 0.6182,
      "step": 14181
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.536943268243391,
      "learning_rate": 3.1411898431495134e-06,
      "loss": 0.5981,
      "step": 14182
    },
    {
      "epoch": 1.27,
      "grad_norm": 8.010657010345819,
      "learning_rate": 3.1405192650294214e-06,
      "loss": 0.6932,
      "step": 14183
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.919520878895749,
      "learning_rate": 3.1398487257219177e-06,
      "loss": 0.6559,
      "step": 14184
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.646922695154604,
      "learning_rate": 3.1391782252409996e-06,
      "loss": 0.5938,
      "step": 14185
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.154626532136132,
      "learning_rate": 3.13850776360066e-06,
      "loss": 0.584,
      "step": 14186
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.5671544344804405,
      "learning_rate": 3.1378373408148937e-06,
      "loss": 0.5897,
      "step": 14187
    },
    {
      "epoch": 1.27,
      "grad_norm": 4.990441106710454,
      "learning_rate": 3.137166956897696e-06,
      "loss": 0.5819,
      "step": 14188
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.893635449788463,
      "learning_rate": 3.136496611863058e-06,
      "loss": 0.5886,
      "step": 14189
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.164289525197329,
      "learning_rate": 3.1358263057249714e-06,
      "loss": 0.6184,
      "step": 14190
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.50713273732411,
      "learning_rate": 3.13515603849743e-06,
      "loss": 0.6383,
      "step": 14191
    },
    {
      "epoch": 1.27,
      "grad_norm": 8.381388738024818,
      "learning_rate": 3.134485810194423e-06,
      "loss": 0.6764,
      "step": 14192
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.638633468682515,
      "learning_rate": 3.1338156208299376e-06,
      "loss": 0.5935,
      "step": 14193
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.3553750918849605,
      "learning_rate": 3.133145470417963e-06,
      "loss": 0.585,
      "step": 14194
    },
    {
      "epoch": 1.27,
      "grad_norm": 8.909058332302738,
      "learning_rate": 3.132475358972491e-06,
      "loss": 0.5939,
      "step": 14195
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.556654218771383,
      "learning_rate": 3.131805286507504e-06,
      "loss": 0.5823,
      "step": 14196
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.2262978451907705,
      "learning_rate": 3.131135253036991e-06,
      "loss": 0.58,
      "step": 14197
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.768669259223704,
      "learning_rate": 3.1304652585749375e-06,
      "loss": 0.6495,
      "step": 14198
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.277392208883144,
      "learning_rate": 3.1297953031353267e-06,
      "loss": 0.5741,
      "step": 14199
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.503469595682173,
      "learning_rate": 3.1291253867321425e-06,
      "loss": 0.5864,
      "step": 14200
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.745831337259602,
      "learning_rate": 3.1284555093793713e-06,
      "loss": 0.5299,
      "step": 14201
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.666942037502495,
      "learning_rate": 3.1277856710909906e-06,
      "loss": 0.5353,
      "step": 14202
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.194199978769812,
      "learning_rate": 3.1271158718809835e-06,
      "loss": 0.5569,
      "step": 14203
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.4002925355548905,
      "learning_rate": 3.126446111763333e-06,
      "loss": 0.5895,
      "step": 14204
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.669597814249999,
      "learning_rate": 3.1257763907520157e-06,
      "loss": 0.547,
      "step": 14205
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.13875042033398,
      "learning_rate": 3.1251067088610134e-06,
      "loss": 0.5721,
      "step": 14206
    },
    {
      "epoch": 1.27,
      "grad_norm": 9.160312276270883,
      "learning_rate": 3.124437066104301e-06,
      "loss": 0.6371,
      "step": 14207
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.8240299852373525,
      "learning_rate": 3.123767462495858e-06,
      "loss": 0.6185,
      "step": 14208
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.191085451679857,
      "learning_rate": 3.1230978980496594e-06,
      "loss": 0.5798,
      "step": 14209
    },
    {
      "epoch": 1.27,
      "grad_norm": 8.247850563114849,
      "learning_rate": 3.1224283727796824e-06,
      "loss": 0.5724,
      "step": 14210
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.046461606035232,
      "learning_rate": 3.1217588866999017e-06,
      "loss": 0.559,
      "step": 14211
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.813814197497029,
      "learning_rate": 3.121089439824291e-06,
      "loss": 0.6003,
      "step": 14212
    },
    {
      "epoch": 1.27,
      "grad_norm": 4.984119235874705,
      "learning_rate": 3.120420032166822e-06,
      "loss": 0.6263,
      "step": 14213
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.335981322248142,
      "learning_rate": 3.1197506637414702e-06,
      "loss": 0.5611,
      "step": 14214
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.783742820369118,
      "learning_rate": 3.119081334562205e-06,
      "loss": 0.6224,
      "step": 14215
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.1060029094150226,
      "learning_rate": 3.1184120446429977e-06,
      "loss": 0.5845,
      "step": 14216
    },
    {
      "epoch": 1.27,
      "grad_norm": 4.628802882489749,
      "learning_rate": 3.1177427939978187e-06,
      "loss": 0.5838,
      "step": 14217
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.729326600447655,
      "learning_rate": 3.1170735826406367e-06,
      "loss": 0.649,
      "step": 14218
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.239336196574895,
      "learning_rate": 3.1164044105854197e-06,
      "loss": 0.6201,
      "step": 14219
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.102852234236515,
      "learning_rate": 3.115735277846137e-06,
      "loss": 0.5593,
      "step": 14220
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.995396505325374,
      "learning_rate": 3.115066184436754e-06,
      "loss": 0.5682,
      "step": 14221
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.079799004528221,
      "learning_rate": 3.1143971303712352e-06,
      "loss": 0.6187,
      "step": 14222
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.320089070263943,
      "learning_rate": 3.113728115663547e-06,
      "loss": 0.6077,
      "step": 14223
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.749677453447969,
      "learning_rate": 3.1130591403276543e-06,
      "loss": 0.6679,
      "step": 14224
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.660356567377498,
      "learning_rate": 3.1123902043775184e-06,
      "loss": 0.6059,
      "step": 14225
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.083363663416029,
      "learning_rate": 3.1117213078271036e-06,
      "loss": 0.5508,
      "step": 14226
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.998723469062759,
      "learning_rate": 3.1110524506903715e-06,
      "loss": 0.5632,
      "step": 14227
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.8944294365451695,
      "learning_rate": 3.1103836329812833e-06,
      "loss": 0.5906,
      "step": 14228
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.856703683825819,
      "learning_rate": 3.1097148547137967e-06,
      "loss": 0.569,
      "step": 14229
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.912603394898779,
      "learning_rate": 3.1090461159018753e-06,
      "loss": 0.5499,
      "step": 14230
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.691173160590024,
      "learning_rate": 3.1083774165594737e-06,
      "loss": 0.6673,
      "step": 14231
    },
    {
      "epoch": 1.27,
      "grad_norm": 8.269041304521982,
      "learning_rate": 3.1077087567005504e-06,
      "loss": 0.659,
      "step": 14232
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.8013448668031895,
      "learning_rate": 3.1070401363390647e-06,
      "loss": 0.643,
      "step": 14233
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.419292694077384,
      "learning_rate": 3.106371555488969e-06,
      "loss": 0.6305,
      "step": 14234
    },
    {
      "epoch": 1.27,
      "grad_norm": 8.943833539430145,
      "learning_rate": 3.105703014164222e-06,
      "loss": 0.5998,
      "step": 14235
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.903888829930047,
      "learning_rate": 3.1050345123787747e-06,
      "loss": 0.6119,
      "step": 14236
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.074583214424002,
      "learning_rate": 3.104366050146582e-06,
      "loss": 0.5222,
      "step": 14237
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.9277232880543655,
      "learning_rate": 3.1036976274815967e-06,
      "loss": 0.5942,
      "step": 14238
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.554341856569497,
      "learning_rate": 3.1030292443977705e-06,
      "loss": 0.6437,
      "step": 14239
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.168673845974766,
      "learning_rate": 3.102360900909055e-06,
      "loss": 0.6388,
      "step": 14240
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.025031508591829,
      "learning_rate": 3.1016925970293997e-06,
      "loss": 0.6086,
      "step": 14241
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.045135326347537,
      "learning_rate": 3.1010243327727535e-06,
      "loss": 0.5403,
      "step": 14242
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.986581791264296,
      "learning_rate": 3.100356108153067e-06,
      "loss": 0.6597,
      "step": 14243
    },
    {
      "epoch": 1.27,
      "grad_norm": 8.891436296707225,
      "learning_rate": 3.0996879231842856e-06,
      "loss": 0.6157,
      "step": 14244
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.832163532580829,
      "learning_rate": 3.0990197778803573e-06,
      "loss": 0.6091,
      "step": 14245
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.583140208531847,
      "learning_rate": 3.0983516722552287e-06,
      "loss": 0.5786,
      "step": 14246
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.792662164063253,
      "learning_rate": 3.0976836063228433e-06,
      "loss": 0.6062,
      "step": 14247
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.790451857620161,
      "learning_rate": 3.097015580097148e-06,
      "loss": 0.6118,
      "step": 14248
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.2213016928643565,
      "learning_rate": 3.0963475935920837e-06,
      "loss": 0.6149,
      "step": 14249
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.234415628691221,
      "learning_rate": 3.0956796468215965e-06,
      "loss": 0.6413,
      "step": 14250
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.813345218559186,
      "learning_rate": 3.095011739799625e-06,
      "loss": 0.5486,
      "step": 14251
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.066415365896051,
      "learning_rate": 3.094343872540111e-06,
      "loss": 0.5765,
      "step": 14252
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.434258299681204,
      "learning_rate": 3.093676045056996e-06,
      "loss": 0.5298,
      "step": 14253
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.473804072843131,
      "learning_rate": 3.0930082573642183e-06,
      "loss": 0.6463,
      "step": 14254
    },
    {
      "epoch": 1.27,
      "grad_norm": 10.698363818516853,
      "learning_rate": 3.0923405094757165e-06,
      "loss": 0.5869,
      "step": 14255
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.818556474619349,
      "learning_rate": 3.0916728014054305e-06,
      "loss": 0.6414,
      "step": 14256
    },
    {
      "epoch": 1.27,
      "grad_norm": 8.861939405602447,
      "learning_rate": 3.091005133167294e-06,
      "loss": 0.5301,
      "step": 14257
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.156114565594416,
      "learning_rate": 3.090337504775246e-06,
      "loss": 0.6125,
      "step": 14258
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.917903221441524,
      "learning_rate": 3.08966991624322e-06,
      "loss": 0.6235,
      "step": 14259
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.1613935770679475,
      "learning_rate": 3.08900236758515e-06,
      "loss": 0.6246,
      "step": 14260
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.644894811836973,
      "learning_rate": 3.088334858814972e-06,
      "loss": 0.5428,
      "step": 14261
    },
    {
      "epoch": 1.27,
      "grad_norm": 4.810045082312236,
      "learning_rate": 3.0876673899466164e-06,
      "loss": 0.5939,
      "step": 14262
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.262029851398582,
      "learning_rate": 3.086999960994016e-06,
      "loss": 0.5773,
      "step": 14263
    },
    {
      "epoch": 1.27,
      "grad_norm": 10.195939801569406,
      "learning_rate": 3.0863325719711047e-06,
      "loss": 0.5769,
      "step": 14264
    },
    {
      "epoch": 1.27,
      "grad_norm": 8.168682597372745,
      "learning_rate": 3.085665222891808e-06,
      "loss": 0.5828,
      "step": 14265
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.920976493917427,
      "learning_rate": 3.0849979137700568e-06,
      "loss": 0.5562,
      "step": 14266
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.791298442647365,
      "learning_rate": 3.084330644619781e-06,
      "loss": 0.5954,
      "step": 14267
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.9297839110308175,
      "learning_rate": 3.0836634154549077e-06,
      "loss": 0.5653,
      "step": 14268
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.070154395491449,
      "learning_rate": 3.082996226289363e-06,
      "loss": 0.5905,
      "step": 14269
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.144580455257197,
      "learning_rate": 3.082329077137075e-06,
      "loss": 0.5983,
      "step": 14270
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.232464275014895,
      "learning_rate": 3.081661968011968e-06,
      "loss": 0.6038,
      "step": 14271
    },
    {
      "epoch": 1.27,
      "grad_norm": 8.293164832321724,
      "learning_rate": 3.080994898927966e-06,
      "loss": 0.6305,
      "step": 14272
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.944609032632751,
      "learning_rate": 3.0803278698989926e-06,
      "loss": 0.5789,
      "step": 14273
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.8831910617440295,
      "learning_rate": 3.0796608809389715e-06,
      "loss": 0.5642,
      "step": 14274
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.7962676264187225,
      "learning_rate": 3.0789939320618233e-06,
      "loss": 0.5564,
      "step": 14275
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.467791626918249,
      "learning_rate": 3.07832702328147e-06,
      "loss": 0.6439,
      "step": 14276
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.6361439924557555,
      "learning_rate": 3.0776601546118325e-06,
      "loss": 0.5547,
      "step": 14277
    },
    {
      "epoch": 1.27,
      "grad_norm": 8.808965677239383,
      "learning_rate": 3.0769933260668284e-06,
      "loss": 0.6319,
      "step": 14278
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.667533811502846,
      "learning_rate": 3.076326537660379e-06,
      "loss": 0.5855,
      "step": 14279
    },
    {
      "epoch": 1.27,
      "grad_norm": 8.085992125082479,
      "learning_rate": 3.075659789406399e-06,
      "loss": 0.5724,
      "step": 14280
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.883805384217571,
      "learning_rate": 3.074993081318807e-06,
      "loss": 0.653,
      "step": 14281
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.25627421531941,
      "learning_rate": 3.0743264134115186e-06,
      "loss": 0.6408,
      "step": 14282
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.493316759165138,
      "learning_rate": 3.0736597856984486e-06,
      "loss": 0.6284,
      "step": 14283
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.1024514090059485,
      "learning_rate": 3.072993198193513e-06,
      "loss": 0.5951,
      "step": 14284
    },
    {
      "epoch": 1.27,
      "grad_norm": 4.793914059615104,
      "learning_rate": 3.0723266509106232e-06,
      "loss": 0.6171,
      "step": 14285
    },
    {
      "epoch": 1.27,
      "grad_norm": 8.65980062502108,
      "learning_rate": 3.0716601438636934e-06,
      "loss": 0.6547,
      "step": 14286
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.42235348248907,
      "learning_rate": 3.070993677066636e-06,
      "loss": 0.6163,
      "step": 14287
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.228579719428768,
      "learning_rate": 3.07032725053336e-06,
      "loss": 0.593,
      "step": 14288
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.870585553822756,
      "learning_rate": 3.069660864277777e-06,
      "loss": 0.5687,
      "step": 14289
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.895604003551127,
      "learning_rate": 3.068994518313797e-06,
      "loss": 0.5403,
      "step": 14290
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.819614149707555,
      "learning_rate": 3.068328212655326e-06,
      "loss": 0.6028,
      "step": 14291
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.449255400515749,
      "learning_rate": 3.067661947316274e-06,
      "loss": 0.5628,
      "step": 14292
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.8167478707992535,
      "learning_rate": 3.0669957223105485e-06,
      "loss": 0.576,
      "step": 14293
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.6098646413118765,
      "learning_rate": 3.066329537652054e-06,
      "loss": 0.5584,
      "step": 14294
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.235940136366945,
      "learning_rate": 3.065663393354694e-06,
      "loss": 0.6056,
      "step": 14295
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.845955800152011,
      "learning_rate": 3.0649972894323744e-06,
      "loss": 0.5746,
      "step": 14296
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.659744099532797,
      "learning_rate": 3.0643312258990007e-06,
      "loss": 0.5565,
      "step": 14297
    },
    {
      "epoch": 1.28,
      "grad_norm": 7.092797715473327,
      "learning_rate": 3.0636652027684717e-06,
      "loss": 0.5302,
      "step": 14298
    },
    {
      "epoch": 1.28,
      "grad_norm": 8.528806482700434,
      "learning_rate": 3.062999220054691e-06,
      "loss": 0.5916,
      "step": 14299
    },
    {
      "epoch": 1.28,
      "grad_norm": 7.402670631395741,
      "learning_rate": 3.0623332777715613e-06,
      "loss": 0.6222,
      "step": 14300
    },
    {
      "epoch": 1.28,
      "grad_norm": 8.583312177906585,
      "learning_rate": 3.0616673759329794e-06,
      "loss": 0.5671,
      "step": 14301
    },
    {
      "epoch": 1.28,
      "grad_norm": 8.46379892501733,
      "learning_rate": 3.061001514552847e-06,
      "loss": 0.6064,
      "step": 14302
    },
    {
      "epoch": 1.28,
      "grad_norm": 7.22413564101537,
      "learning_rate": 3.0603356936450625e-06,
      "loss": 0.6405,
      "step": 14303
    },
    {
      "epoch": 1.28,
      "grad_norm": 4.837745172169967,
      "learning_rate": 3.0596699132235207e-06,
      "loss": 0.5624,
      "step": 14304
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.593361176182623,
      "learning_rate": 3.059004173302121e-06,
      "loss": 0.5621,
      "step": 14305
    },
    {
      "epoch": 1.28,
      "grad_norm": 8.614762976905338,
      "learning_rate": 3.058338473894759e-06,
      "loss": 0.6393,
      "step": 14306
    },
    {
      "epoch": 1.28,
      "grad_norm": 9.044602794266224,
      "learning_rate": 3.0576728150153306e-06,
      "loss": 0.6063,
      "step": 14307
    },
    {
      "epoch": 1.28,
      "grad_norm": 7.693069153526909,
      "learning_rate": 3.0570071966777267e-06,
      "loss": 0.5407,
      "step": 14308
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.555061104533352,
      "learning_rate": 3.056341618895842e-06,
      "loss": 0.5213,
      "step": 14309
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.851284808719114,
      "learning_rate": 3.0556760816835706e-06,
      "loss": 0.5975,
      "step": 14310
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.468626144861592,
      "learning_rate": 3.055010585054802e-06,
      "loss": 0.5807,
      "step": 14311
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.923367024490816,
      "learning_rate": 3.0543451290234283e-06,
      "loss": 0.5736,
      "step": 14312
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.4097998292197715,
      "learning_rate": 3.0536797136033395e-06,
      "loss": 0.5828,
      "step": 14313
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.370782621579315,
      "learning_rate": 3.053014338808424e-06,
      "loss": 0.6132,
      "step": 14314
    },
    {
      "epoch": 1.28,
      "grad_norm": 7.721757983250321,
      "learning_rate": 3.0523490046525693e-06,
      "loss": 0.6115,
      "step": 14315
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.493781272618809,
      "learning_rate": 3.0516837111496663e-06,
      "loss": 0.5358,
      "step": 14316
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.6402423824366945,
      "learning_rate": 3.051018458313597e-06,
      "loss": 0.5552,
      "step": 14317
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.622464268004977,
      "learning_rate": 3.0503532461582495e-06,
      "loss": 0.61,
      "step": 14318
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.984473716136694,
      "learning_rate": 3.0496880746975095e-06,
      "loss": 0.5739,
      "step": 14319
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.175806184352043,
      "learning_rate": 3.049022943945259e-06,
      "loss": 0.6168,
      "step": 14320
    },
    {
      "epoch": 1.28,
      "grad_norm": 4.949973474645171,
      "learning_rate": 3.048357853915381e-06,
      "loss": 0.5835,
      "step": 14321
    },
    {
      "epoch": 1.28,
      "grad_norm": 8.757594011401247,
      "learning_rate": 3.0476928046217618e-06,
      "loss": 0.5906,
      "step": 14322
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.4085677896361775,
      "learning_rate": 3.0470277960782784e-06,
      "loss": 0.5679,
      "step": 14323
    },
    {
      "epoch": 1.28,
      "grad_norm": 7.057962868567656,
      "learning_rate": 3.046362828298812e-06,
      "loss": 0.591,
      "step": 14324
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.6696734487736995,
      "learning_rate": 3.045697901297243e-06,
      "loss": 0.5843,
      "step": 14325
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.824674157993956,
      "learning_rate": 3.045033015087452e-06,
      "loss": 0.5389,
      "step": 14326
    },
    {
      "epoch": 1.28,
      "grad_norm": 7.348036588829832,
      "learning_rate": 3.044368169683314e-06,
      "loss": 0.6422,
      "step": 14327
    },
    {
      "epoch": 1.28,
      "grad_norm": 7.157059886998844,
      "learning_rate": 3.043703365098707e-06,
      "loss": 0.5722,
      "step": 14328
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.617081775996938,
      "learning_rate": 3.0430386013475097e-06,
      "loss": 0.5986,
      "step": 14329
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.727959659727873,
      "learning_rate": 3.042373878443595e-06,
      "loss": 0.5821,
      "step": 14330
    },
    {
      "epoch": 1.28,
      "grad_norm": 8.7947060066047,
      "learning_rate": 3.041709196400839e-06,
      "loss": 0.5981,
      "step": 14331
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.9803431739196515,
      "learning_rate": 3.0410445552331155e-06,
      "loss": 0.6101,
      "step": 14332
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.213762409379191,
      "learning_rate": 3.0403799549542958e-06,
      "loss": 0.5317,
      "step": 14333
    },
    {
      "epoch": 1.28,
      "grad_norm": 7.355672485492928,
      "learning_rate": 3.0397153955782532e-06,
      "loss": 0.5986,
      "step": 14334
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.760193223718291,
      "learning_rate": 3.039050877118859e-06,
      "loss": 0.6066,
      "step": 14335
    },
    {
      "epoch": 1.28,
      "grad_norm": 8.789958809959167,
      "learning_rate": 3.038386399589985e-06,
      "loss": 0.6223,
      "step": 14336
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.312145189369302,
      "learning_rate": 3.0377219630054967e-06,
      "loss": 0.5362,
      "step": 14337
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.069933981319586,
      "learning_rate": 3.037057567379266e-06,
      "loss": 0.5434,
      "step": 14338
    },
    {
      "epoch": 1.28,
      "grad_norm": 8.017657864510372,
      "learning_rate": 3.0363932127251595e-06,
      "loss": 0.6481,
      "step": 14339
    },
    {
      "epoch": 1.28,
      "grad_norm": 8.912913069664853,
      "learning_rate": 3.035728899057044e-06,
      "loss": 0.6206,
      "step": 14340
    },
    {
      "epoch": 1.28,
      "grad_norm": 7.545767784314333,
      "learning_rate": 3.0350646263887875e-06,
      "loss": 0.6048,
      "step": 14341
    },
    {
      "epoch": 1.28,
      "grad_norm": 8.602418686082135,
      "learning_rate": 3.0344003947342527e-06,
      "loss": 0.5883,
      "step": 14342
    },
    {
      "epoch": 1.28,
      "grad_norm": 8.064167558461232,
      "learning_rate": 3.033736204107305e-06,
      "loss": 0.5877,
      "step": 14343
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.009610810023753,
      "learning_rate": 3.0330720545218086e-06,
      "loss": 0.6243,
      "step": 14344
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.8157846897559216,
      "learning_rate": 3.0324079459916243e-06,
      "loss": 0.5912,
      "step": 14345
    },
    {
      "epoch": 1.28,
      "grad_norm": 7.8942669618033,
      "learning_rate": 3.0317438785306164e-06,
      "loss": 0.5509,
      "step": 14346
    },
    {
      "epoch": 1.28,
      "grad_norm": 7.046989192574817,
      "learning_rate": 3.031079852152644e-06,
      "loss": 0.6456,
      "step": 14347
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.794560843753916,
      "learning_rate": 3.030415866871568e-06,
      "loss": 0.5684,
      "step": 14348
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.828012648488844,
      "learning_rate": 3.0297519227012472e-06,
      "loss": 0.5838,
      "step": 14349
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.691471404379447,
      "learning_rate": 3.029088019655542e-06,
      "loss": 0.6336,
      "step": 14350
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.036297068939976,
      "learning_rate": 3.0284241577483074e-06,
      "loss": 0.6367,
      "step": 14351
    },
    {
      "epoch": 1.28,
      "grad_norm": 4.8253795336043614,
      "learning_rate": 3.0277603369934004e-06,
      "loss": 0.5978,
      "step": 14352
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.746917094150679,
      "learning_rate": 3.0270965574046762e-06,
      "loss": 0.6017,
      "step": 14353
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.4607631820179,
      "learning_rate": 3.0264328189959924e-06,
      "loss": 0.6215,
      "step": 14354
    },
    {
      "epoch": 1.28,
      "grad_norm": 7.463362677406458,
      "learning_rate": 3.0257691217812003e-06,
      "loss": 0.6159,
      "step": 14355
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.759879548734567,
      "learning_rate": 3.025105465774154e-06,
      "loss": 0.5901,
      "step": 14356
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.697949753698949,
      "learning_rate": 3.024441850988708e-06,
      "loss": 0.6006,
      "step": 14357
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.11725984203276,
      "learning_rate": 3.02377827743871e-06,
      "loss": 0.5641,
      "step": 14358
    },
    {
      "epoch": 1.28,
      "grad_norm": 7.640768898172056,
      "learning_rate": 3.0231147451380134e-06,
      "loss": 0.6524,
      "step": 14359
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.269534441951477,
      "learning_rate": 3.022451254100468e-06,
      "loss": 0.5804,
      "step": 14360
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.296494171047499,
      "learning_rate": 3.0217878043399205e-06,
      "loss": 0.6182,
      "step": 14361
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.384066763854973,
      "learning_rate": 3.0211243958702207e-06,
      "loss": 0.5797,
      "step": 14362
    },
    {
      "epoch": 1.28,
      "grad_norm": 4.869319366820157,
      "learning_rate": 3.020461028705216e-06,
      "loss": 0.6399,
      "step": 14363
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.9306178710307815,
      "learning_rate": 3.0197977028587523e-06,
      "loss": 0.5688,
      "step": 14364
    },
    {
      "epoch": 1.28,
      "grad_norm": 9.102073271754799,
      "learning_rate": 3.0191344183446756e-06,
      "loss": 0.6137,
      "step": 14365
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.755903058489476,
      "learning_rate": 3.0184711751768284e-06,
      "loss": 0.6039,
      "step": 14366
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.99306986163534,
      "learning_rate": 3.0178079733690575e-06,
      "loss": 0.5615,
      "step": 14367
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.821766197420608,
      "learning_rate": 3.0171448129352025e-06,
      "loss": 0.5061,
      "step": 14368
    },
    {
      "epoch": 1.28,
      "grad_norm": 8.011335284526892,
      "learning_rate": 3.016481693889108e-06,
      "loss": 0.5738,
      "step": 14369
    },
    {
      "epoch": 1.28,
      "grad_norm": 8.581667925303389,
      "learning_rate": 3.0158186162446144e-06,
      "loss": 0.5803,
      "step": 14370
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.804071797266645,
      "learning_rate": 3.015155580015562e-06,
      "loss": 0.6375,
      "step": 14371
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.543247062622981,
      "learning_rate": 3.014492585215789e-06,
      "loss": 0.5714,
      "step": 14372
    },
    {
      "epoch": 1.28,
      "grad_norm": 7.497285761895214,
      "learning_rate": 3.0138296318591365e-06,
      "loss": 0.5792,
      "step": 14373
    },
    {
      "epoch": 1.28,
      "grad_norm": 9.791748847589739,
      "learning_rate": 3.0131667199594396e-06,
      "loss": 0.5473,
      "step": 14374
    },
    {
      "epoch": 1.28,
      "grad_norm": 7.962116724997619,
      "learning_rate": 3.012503849530537e-06,
      "loss": 0.6473,
      "step": 14375
    },
    {
      "epoch": 1.28,
      "grad_norm": 8.805485935512662,
      "learning_rate": 3.011841020586264e-06,
      "loss": 0.6466,
      "step": 14376
    },
    {
      "epoch": 1.28,
      "grad_norm": 7.141315008362686,
      "learning_rate": 3.0111782331404556e-06,
      "loss": 0.6109,
      "step": 14377
    },
    {
      "epoch": 1.28,
      "grad_norm": 7.385817080194886,
      "learning_rate": 3.010515487206945e-06,
      "loss": 0.5511,
      "step": 14378
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.531865460087033,
      "learning_rate": 3.0098527827995705e-06,
      "loss": 0.5865,
      "step": 14379
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.254039209004664,
      "learning_rate": 3.0091901199321582e-06,
      "loss": 0.5662,
      "step": 14380
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.193749963469741,
      "learning_rate": 3.008527498618542e-06,
      "loss": 0.6018,
      "step": 14381
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.649871901771433,
      "learning_rate": 3.007864918872553e-06,
      "loss": 0.5874,
      "step": 14382
    },
    {
      "epoch": 1.28,
      "grad_norm": 4.9688691911898735,
      "learning_rate": 3.007202380708022e-06,
      "loss": 0.5827,
      "step": 14383
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.275159180190917,
      "learning_rate": 3.006539884138777e-06,
      "loss": 0.6815,
      "step": 14384
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.673871408210668,
      "learning_rate": 3.0058774291786452e-06,
      "loss": 0.5902,
      "step": 14385
    },
    {
      "epoch": 1.28,
      "grad_norm": 7.117984593874003,
      "learning_rate": 3.005215015841456e-06,
      "loss": 0.5869,
      "step": 14386
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.810975711845501,
      "learning_rate": 3.0045526441410345e-06,
      "loss": 0.6097,
      "step": 14387
    },
    {
      "epoch": 1.28,
      "grad_norm": 8.551732812221912,
      "learning_rate": 3.0038903140912064e-06,
      "loss": 0.577,
      "step": 14388
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.77656003481528,
      "learning_rate": 3.003228025705797e-06,
      "loss": 0.5484,
      "step": 14389
    },
    {
      "epoch": 1.28,
      "grad_norm": 8.072311964229831,
      "learning_rate": 3.0025657789986294e-06,
      "loss": 0.6813,
      "step": 14390
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.091237050436139,
      "learning_rate": 3.0019035739835265e-06,
      "loss": 0.5719,
      "step": 14391
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.02138209346955,
      "learning_rate": 3.001241410674312e-06,
      "loss": 0.5339,
      "step": 14392
    },
    {
      "epoch": 1.28,
      "grad_norm": 7.2785118233909,
      "learning_rate": 3.000579289084805e-06,
      "loss": 0.6815,
      "step": 14393
    },
    {
      "epoch": 1.28,
      "grad_norm": 7.465857273112296,
      "learning_rate": 2.999917209228829e-06,
      "loss": 0.5916,
      "step": 14394
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.394551575051437,
      "learning_rate": 2.9992551711201977e-06,
      "loss": 0.629,
      "step": 14395
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.717746201154778,
      "learning_rate": 2.998593174772735e-06,
      "loss": 0.6452,
      "step": 14396
    },
    {
      "epoch": 1.28,
      "grad_norm": 9.20931204771821,
      "learning_rate": 2.9979312202002563e-06,
      "loss": 0.5965,
      "step": 14397
    },
    {
      "epoch": 1.28,
      "grad_norm": 7.171815444819069,
      "learning_rate": 2.9972693074165782e-06,
      "loss": 0.5878,
      "step": 14398
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.322540400602786,
      "learning_rate": 2.9966074364355186e-06,
      "loss": 0.5789,
      "step": 14399
    },
    {
      "epoch": 1.28,
      "grad_norm": 4.750910988809198,
      "learning_rate": 2.9959456072708904e-06,
      "loss": 0.5846,
      "step": 14400
    },
    {
      "epoch": 1.28,
      "grad_norm": 7.07144267188725,
      "learning_rate": 2.9952838199365086e-06,
      "loss": 0.5455,
      "step": 14401
    },
    {
      "epoch": 1.28,
      "grad_norm": 7.288642631383981,
      "learning_rate": 2.9946220744461885e-06,
      "loss": 0.5809,
      "step": 14402
    },
    {
      "epoch": 1.28,
      "grad_norm": 7.778273456316964,
      "learning_rate": 2.9939603708137395e-06,
      "loss": 0.576,
      "step": 14403
    },
    {
      "epoch": 1.28,
      "grad_norm": 7.020343951142864,
      "learning_rate": 2.993298709052974e-06,
      "loss": 0.6374,
      "step": 14404
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.601210127060374,
      "learning_rate": 2.992637089177705e-06,
      "loss": 0.6025,
      "step": 14405
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.490660037863082,
      "learning_rate": 2.99197551120174e-06,
      "loss": 0.6195,
      "step": 14406
    },
    {
      "epoch": 1.29,
      "grad_norm": 4.544748043002633,
      "learning_rate": 2.991313975138888e-06,
      "loss": 0.5627,
      "step": 14407
    },
    {
      "epoch": 1.29,
      "grad_norm": 7.122196329547084,
      "learning_rate": 2.9906524810029603e-06,
      "loss": 0.5759,
      "step": 14408
    },
    {
      "epoch": 1.29,
      "grad_norm": 7.359433099485161,
      "learning_rate": 2.9899910288077605e-06,
      "loss": 0.6045,
      "step": 14409
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.591937463117704,
      "learning_rate": 2.989329618567095e-06,
      "loss": 0.6199,
      "step": 14410
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.352429490880989,
      "learning_rate": 2.9886682502947705e-06,
      "loss": 0.6135,
      "step": 14411
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.050420548421991,
      "learning_rate": 2.9880069240045928e-06,
      "loss": 0.6756,
      "step": 14412
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.719338475646623,
      "learning_rate": 2.9873456397103636e-06,
      "loss": 0.5225,
      "step": 14413
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.267706208212715,
      "learning_rate": 2.986684397425886e-06,
      "loss": 0.6683,
      "step": 14414
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.277878442931311,
      "learning_rate": 2.9860231971649635e-06,
      "loss": 0.5783,
      "step": 14415
    },
    {
      "epoch": 1.29,
      "grad_norm": 7.180729546338164,
      "learning_rate": 2.985362038941395e-06,
      "loss": 0.5667,
      "step": 14416
    },
    {
      "epoch": 1.29,
      "grad_norm": 8.175315697180206,
      "learning_rate": 2.9847009227689837e-06,
      "loss": 0.6434,
      "step": 14417
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.241542736381951,
      "learning_rate": 2.984039848661526e-06,
      "loss": 0.6247,
      "step": 14418
    },
    {
      "epoch": 1.29,
      "grad_norm": 7.619062741973584,
      "learning_rate": 2.9833788166328216e-06,
      "loss": 0.6208,
      "step": 14419
    },
    {
      "epoch": 1.29,
      "grad_norm": 8.326711253432372,
      "learning_rate": 2.9827178266966684e-06,
      "loss": 0.6166,
      "step": 14420
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.175539236247336,
      "learning_rate": 2.9820568788668623e-06,
      "loss": 0.5958,
      "step": 14421
    },
    {
      "epoch": 1.29,
      "grad_norm": 8.041452148426877,
      "learning_rate": 2.9813959731571996e-06,
      "loss": 0.632,
      "step": 14422
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.8813924779902145,
      "learning_rate": 2.9807351095814774e-06,
      "loss": 0.5999,
      "step": 14423
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.628485086893321,
      "learning_rate": 2.9800742881534857e-06,
      "loss": 0.6365,
      "step": 14424
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.9116879558409465,
      "learning_rate": 2.97941350888702e-06,
      "loss": 0.5929,
      "step": 14425
    },
    {
      "epoch": 1.29,
      "grad_norm": 8.694945382717716,
      "learning_rate": 2.978752771795872e-06,
      "loss": 0.6167,
      "step": 14426
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.5976980693651575,
      "learning_rate": 2.9780920768938345e-06,
      "loss": 0.61,
      "step": 14427
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.075185954413045,
      "learning_rate": 2.9774314241946957e-06,
      "loss": 0.5728,
      "step": 14428
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.049814540447643,
      "learning_rate": 2.9767708137122462e-06,
      "loss": 0.5773,
      "step": 14429
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.247392618575722,
      "learning_rate": 2.976110245460277e-06,
      "loss": 0.5983,
      "step": 14430
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.676823111955589,
      "learning_rate": 2.9754497194525726e-06,
      "loss": 0.6403,
      "step": 14431
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.3580829363464435,
      "learning_rate": 2.9747892357029216e-06,
      "loss": 0.5533,
      "step": 14432
    },
    {
      "epoch": 1.29,
      "grad_norm": 7.563709019204993,
      "learning_rate": 2.974128794225112e-06,
      "loss": 0.6295,
      "step": 14433
    },
    {
      "epoch": 1.29,
      "grad_norm": 7.704241637271939,
      "learning_rate": 2.9734683950329252e-06,
      "loss": 0.5879,
      "step": 14434
    },
    {
      "epoch": 1.29,
      "grad_norm": 7.384434006269605,
      "learning_rate": 2.9728080381401483e-06,
      "loss": 0.5758,
      "step": 14435
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.7199491651219825,
      "learning_rate": 2.9721477235605646e-06,
      "loss": 0.6355,
      "step": 14436
    },
    {
      "epoch": 1.29,
      "grad_norm": 7.177447766636385,
      "learning_rate": 2.971487451307957e-06,
      "loss": 0.6201,
      "step": 14437
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.012642094416646,
      "learning_rate": 2.970827221396105e-06,
      "loss": 0.609,
      "step": 14438
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.2710244661418315,
      "learning_rate": 2.970167033838791e-06,
      "loss": 0.602,
      "step": 14439
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.066599519877523,
      "learning_rate": 2.9695068886497964e-06,
      "loss": 0.6141,
      "step": 14440
    },
    {
      "epoch": 1.29,
      "grad_norm": 7.390458290205408,
      "learning_rate": 2.968846785842897e-06,
      "loss": 0.6111,
      "step": 14441
    },
    {
      "epoch": 1.29,
      "grad_norm": 4.2182718630909495,
      "learning_rate": 2.968186725431873e-06,
      "loss": 0.6417,
      "step": 14442
    },
    {
      "epoch": 1.29,
      "grad_norm": 9.09551022615898,
      "learning_rate": 2.9675267074305023e-06,
      "loss": 0.6229,
      "step": 14443
    },
    {
      "epoch": 1.29,
      "grad_norm": 7.733975075562566,
      "learning_rate": 2.9668667318525603e-06,
      "loss": 0.6691,
      "step": 14444
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.230984917763174,
      "learning_rate": 2.9662067987118227e-06,
      "loss": 0.6441,
      "step": 14445
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.789848139733644,
      "learning_rate": 2.9655469080220645e-06,
      "loss": 0.5713,
      "step": 14446
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.419212430122438,
      "learning_rate": 2.964887059797058e-06,
      "loss": 0.5805,
      "step": 14447
    },
    {
      "epoch": 1.29,
      "grad_norm": 7.4520182895240135,
      "learning_rate": 2.964227254050578e-06,
      "loss": 0.5798,
      "step": 14448
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.226393183192043,
      "learning_rate": 2.963567490796397e-06,
      "loss": 0.6302,
      "step": 14449
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.9274499968145795,
      "learning_rate": 2.9629077700482835e-06,
      "loss": 0.6177,
      "step": 14450
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.949739847549607,
      "learning_rate": 2.962248091820009e-06,
      "loss": 0.5949,
      "step": 14451
    },
    {
      "epoch": 1.29,
      "grad_norm": 7.704920399908984,
      "learning_rate": 2.9615884561253454e-06,
      "loss": 0.6517,
      "step": 14452
    },
    {
      "epoch": 1.29,
      "grad_norm": 7.065451322091057,
      "learning_rate": 2.9609288629780573e-06,
      "loss": 0.5872,
      "step": 14453
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.586936660918267,
      "learning_rate": 2.9602693123919136e-06,
      "loss": 0.5914,
      "step": 14454
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.565660920165601,
      "learning_rate": 2.95960980438068e-06,
      "loss": 0.5607,
      "step": 14455
    },
    {
      "epoch": 1.29,
      "grad_norm": 7.263796195417792,
      "learning_rate": 2.9589503389581247e-06,
      "loss": 0.5683,
      "step": 14456
    },
    {
      "epoch": 1.29,
      "grad_norm": 8.023559193371083,
      "learning_rate": 2.9582909161380102e-06,
      "loss": 0.6174,
      "step": 14457
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.550958159408224,
      "learning_rate": 2.9576315359341016e-06,
      "loss": 0.5343,
      "step": 14458
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.433858968770386,
      "learning_rate": 2.956972198360163e-06,
      "loss": 0.5611,
      "step": 14459
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.3446663607984055,
      "learning_rate": 2.9563129034299546e-06,
      "loss": 0.612,
      "step": 14460
    },
    {
      "epoch": 1.29,
      "grad_norm": 9.53026769134544,
      "learning_rate": 2.9556536511572386e-06,
      "loss": 0.5569,
      "step": 14461
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.2412641910058415,
      "learning_rate": 2.9549944415557774e-06,
      "loss": 0.6269,
      "step": 14462
    },
    {
      "epoch": 1.29,
      "grad_norm": 7.587887939921509,
      "learning_rate": 2.9543352746393274e-06,
      "loss": 0.6304,
      "step": 14463
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.974147737994717,
      "learning_rate": 2.9536761504216478e-06,
      "loss": 0.6062,
      "step": 14464
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.169903989105843,
      "learning_rate": 2.953017068916499e-06,
      "loss": 0.6347,
      "step": 14465
    },
    {
      "epoch": 1.29,
      "grad_norm": 7.32194438683674,
      "learning_rate": 2.9523580301376364e-06,
      "loss": 0.5664,
      "step": 14466
    },
    {
      "epoch": 1.29,
      "grad_norm": 4.4698692992806945,
      "learning_rate": 2.9516990340988137e-06,
      "loss": 0.5712,
      "step": 14467
    },
    {
      "epoch": 1.29,
      "grad_norm": 7.264261181345793,
      "learning_rate": 2.9510400808137885e-06,
      "loss": 0.6136,
      "step": 14468
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.1062475869052495,
      "learning_rate": 2.9503811702963157e-06,
      "loss": 0.6122,
      "step": 14469
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.606051457279306,
      "learning_rate": 2.9497223025601464e-06,
      "loss": 0.6225,
      "step": 14470
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.5665490648723,
      "learning_rate": 2.949063477619033e-06,
      "loss": 0.5786,
      "step": 14471
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.947497018375507,
      "learning_rate": 2.9484046954867295e-06,
      "loss": 0.5627,
      "step": 14472
    },
    {
      "epoch": 1.29,
      "grad_norm": 7.01405433561849,
      "learning_rate": 2.9477459561769837e-06,
      "loss": 0.5653,
      "step": 14473
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.857553890874094,
      "learning_rate": 2.947087259703548e-06,
      "loss": 0.6598,
      "step": 14474
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.299606075557728,
      "learning_rate": 2.9464286060801696e-06,
      "loss": 0.6011,
      "step": 14475
    },
    {
      "epoch": 1.29,
      "grad_norm": 8.264069576661537,
      "learning_rate": 2.9457699953205964e-06,
      "loss": 0.5911,
      "step": 14476
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.814710529979519,
      "learning_rate": 2.9451114274385763e-06,
      "loss": 0.5905,
      "step": 14477
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.62531385999506,
      "learning_rate": 2.944452902447855e-06,
      "loss": 0.6191,
      "step": 14478
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.283886613167699,
      "learning_rate": 2.943794420362177e-06,
      "loss": 0.5712,
      "step": 14479
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.32928636396176,
      "learning_rate": 2.9431359811952904e-06,
      "loss": 0.6337,
      "step": 14480
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.97385638156927,
      "learning_rate": 2.9424775849609332e-06,
      "loss": 0.6148,
      "step": 14481
    },
    {
      "epoch": 1.29,
      "grad_norm": 11.780910325628934,
      "learning_rate": 2.941819231672851e-06,
      "loss": 0.6006,
      "step": 14482
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.213129265179419,
      "learning_rate": 2.9411609213447847e-06,
      "loss": 0.538,
      "step": 14483
    },
    {
      "epoch": 1.29,
      "grad_norm": 4.221343122944482,
      "learning_rate": 2.940502653990475e-06,
      "loss": 0.5998,
      "step": 14484
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.7199667320361,
      "learning_rate": 2.9398444296236626e-06,
      "loss": 0.6114,
      "step": 14485
    },
    {
      "epoch": 1.29,
      "grad_norm": 7.904275901203021,
      "learning_rate": 2.939186248258086e-06,
      "loss": 0.6609,
      "step": 14486
    },
    {
      "epoch": 1.29,
      "grad_norm": 7.375450456953021,
      "learning_rate": 2.9385281099074837e-06,
      "loss": 0.6051,
      "step": 14487
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.340211414965274,
      "learning_rate": 2.9378700145855926e-06,
      "loss": 0.5936,
      "step": 14488
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.714977566095878,
      "learning_rate": 2.9372119623061486e-06,
      "loss": 0.6359,
      "step": 14489
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.775478708262551,
      "learning_rate": 2.936553953082887e-06,
      "loss": 0.5088,
      "step": 14490
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.63295122456318,
      "learning_rate": 2.935895986929544e-06,
      "loss": 0.6776,
      "step": 14491
    },
    {
      "epoch": 1.29,
      "grad_norm": 8.07595971174323,
      "learning_rate": 2.935238063859851e-06,
      "loss": 0.6131,
      "step": 14492
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.497652195115357,
      "learning_rate": 2.934580183887542e-06,
      "loss": 0.5429,
      "step": 14493
    },
    {
      "epoch": 1.29,
      "grad_norm": 4.539707395102691,
      "learning_rate": 2.9339223470263485e-06,
      "loss": 0.6445,
      "step": 14494
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.615063927375149,
      "learning_rate": 2.933264553290003e-06,
      "loss": 0.5858,
      "step": 14495
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.889654117658814,
      "learning_rate": 2.9326068026922317e-06,
      "loss": 0.6119,
      "step": 14496
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.666303950749915,
      "learning_rate": 2.9319490952467654e-06,
      "loss": 0.6204,
      "step": 14497
    },
    {
      "epoch": 1.29,
      "grad_norm": 7.092001272181314,
      "learning_rate": 2.9312914309673347e-06,
      "loss": 0.5945,
      "step": 14498
    },
    {
      "epoch": 1.29,
      "grad_norm": 4.906228194516457,
      "learning_rate": 2.9306338098676634e-06,
      "loss": 0.6053,
      "step": 14499
    },
    {
      "epoch": 1.29,
      "grad_norm": 7.1484423471018115,
      "learning_rate": 2.92997623196148e-06,
      "loss": 0.5658,
      "step": 14500
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.765846149156163,
      "learning_rate": 2.92931869726251e-06,
      "loss": 0.5801,
      "step": 14501
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.932822698194865,
      "learning_rate": 2.9286612057844764e-06,
      "loss": 0.5767,
      "step": 14502
    },
    {
      "epoch": 1.29,
      "grad_norm": 7.995533589782565,
      "learning_rate": 2.928003757541106e-06,
      "loss": 0.6221,
      "step": 14503
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.33556852739582,
      "learning_rate": 2.9273463525461176e-06,
      "loss": 0.505,
      "step": 14504
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.392449928716301,
      "learning_rate": 2.926688990813235e-06,
      "loss": 0.6256,
      "step": 14505
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.172297128117079,
      "learning_rate": 2.926031672356181e-06,
      "loss": 0.5515,
      "step": 14506
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.705695210112105,
      "learning_rate": 2.925374397188672e-06,
      "loss": 0.4884,
      "step": 14507
    },
    {
      "epoch": 1.29,
      "grad_norm": 7.672195736116548,
      "learning_rate": 2.9247171653244298e-06,
      "loss": 0.5507,
      "step": 14508
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.906221206106583,
      "learning_rate": 2.9240599767771737e-06,
      "loss": 0.6661,
      "step": 14509
    },
    {
      "epoch": 1.29,
      "grad_norm": 7.351080215925008,
      "learning_rate": 2.9234028315606185e-06,
      "loss": 0.6204,
      "step": 14510
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.717152182518857,
      "learning_rate": 2.9227457296884804e-06,
      "loss": 0.65,
      "step": 14511
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.265891814117842,
      "learning_rate": 2.9220886711744757e-06,
      "loss": 0.5701,
      "step": 14512
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.4579015549943355,
      "learning_rate": 2.921431656032321e-06,
      "loss": 0.5922,
      "step": 14513
    },
    {
      "epoch": 1.29,
      "grad_norm": 7.850922306700705,
      "learning_rate": 2.920774684275727e-06,
      "loss": 0.5385,
      "step": 14514
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.861687917777732,
      "learning_rate": 2.920117755918409e-06,
      "loss": 0.6257,
      "step": 14515
    },
    {
      "epoch": 1.29,
      "grad_norm": 4.764866341894267,
      "learning_rate": 2.919460870974078e-06,
      "loss": 0.5766,
      "step": 14516
    },
    {
      "epoch": 1.3,
      "grad_norm": 4.905707825057792,
      "learning_rate": 2.9188040294564436e-06,
      "loss": 0.5755,
      "step": 14517
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.171337916830878,
      "learning_rate": 2.9181472313792204e-06,
      "loss": 0.5726,
      "step": 14518
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.833513915800901,
      "learning_rate": 2.9174904767561118e-06,
      "loss": 0.5723,
      "step": 14519
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.5619989722368475,
      "learning_rate": 2.9168337656008294e-06,
      "loss": 0.5531,
      "step": 14520
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.695213895699714,
      "learning_rate": 2.9161770979270796e-06,
      "loss": 0.6885,
      "step": 14521
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.89302230585796,
      "learning_rate": 2.91552047374857e-06,
      "loss": 0.5308,
      "step": 14522
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.161419963348462,
      "learning_rate": 2.9148638930790053e-06,
      "loss": 0.6121,
      "step": 14523
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.163412961631086,
      "learning_rate": 2.9142073559320925e-06,
      "loss": 0.6622,
      "step": 14524
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.349941000318637,
      "learning_rate": 2.9135508623215335e-06,
      "loss": 0.5622,
      "step": 14525
    },
    {
      "epoch": 1.3,
      "grad_norm": 7.152082360602169,
      "learning_rate": 2.912894412261029e-06,
      "loss": 0.6355,
      "step": 14526
    },
    {
      "epoch": 1.3,
      "grad_norm": 7.7727656001951795,
      "learning_rate": 2.912238005764283e-06,
      "loss": 0.5992,
      "step": 14527
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.519862717145605,
      "learning_rate": 2.911581642844996e-06,
      "loss": 0.5763,
      "step": 14528
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.689388518891596,
      "learning_rate": 2.9109253235168693e-06,
      "loss": 0.622,
      "step": 14529
    },
    {
      "epoch": 1.3,
      "grad_norm": 7.020884728616298,
      "learning_rate": 2.9102690477936012e-06,
      "loss": 0.5354,
      "step": 14530
    },
    {
      "epoch": 1.3,
      "grad_norm": 9.929154770888006,
      "learning_rate": 2.9096128156888924e-06,
      "loss": 0.6512,
      "step": 14531
    },
    {
      "epoch": 1.3,
      "grad_norm": 8.276423943087273,
      "learning_rate": 2.908956627216436e-06,
      "loss": 0.5882,
      "step": 14532
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.772407992138237,
      "learning_rate": 2.9083004823899308e-06,
      "loss": 0.5735,
      "step": 14533
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.914403228246642,
      "learning_rate": 2.9076443812230728e-06,
      "loss": 0.6578,
      "step": 14534
    },
    {
      "epoch": 1.3,
      "grad_norm": 7.449237370704516,
      "learning_rate": 2.906988323729555e-06,
      "loss": 0.6068,
      "step": 14535
    },
    {
      "epoch": 1.3,
      "grad_norm": 7.829577667435318,
      "learning_rate": 2.906332309923075e-06,
      "loss": 0.6586,
      "step": 14536
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.910420622835594,
      "learning_rate": 2.905676339817321e-06,
      "loss": 0.6323,
      "step": 14537
    },
    {
      "epoch": 1.3,
      "grad_norm": 4.754830071241391,
      "learning_rate": 2.905020413425989e-06,
      "loss": 0.5454,
      "step": 14538
    },
    {
      "epoch": 1.3,
      "grad_norm": 7.02232816008928,
      "learning_rate": 2.9043645307627654e-06,
      "loss": 0.6239,
      "step": 14539
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.92581087067422,
      "learning_rate": 2.9037086918413435e-06,
      "loss": 0.5927,
      "step": 14540
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.854929770521914,
      "learning_rate": 2.903052896675412e-06,
      "loss": 0.6047,
      "step": 14541
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.69186715056417,
      "learning_rate": 2.9023971452786582e-06,
      "loss": 0.5911,
      "step": 14542
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.341556601943801,
      "learning_rate": 2.9017414376647713e-06,
      "loss": 0.5444,
      "step": 14543
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.171946325891566,
      "learning_rate": 2.901085773847438e-06,
      "loss": 0.6274,
      "step": 14544
    },
    {
      "epoch": 1.3,
      "grad_norm": 7.62275719532823,
      "learning_rate": 2.9004301538403408e-06,
      "loss": 0.5848,
      "step": 14545
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.030926754451648,
      "learning_rate": 2.899774577657166e-06,
      "loss": 0.6074,
      "step": 14546
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.754153915251487,
      "learning_rate": 2.899119045311598e-06,
      "loss": 0.6083,
      "step": 14547
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.372916982167611,
      "learning_rate": 2.8984635568173184e-06,
      "loss": 0.6248,
      "step": 14548
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.442547645089385,
      "learning_rate": 2.8978081121880123e-06,
      "loss": 0.6166,
      "step": 14549
    },
    {
      "epoch": 1.3,
      "grad_norm": 9.39966921393144,
      "learning_rate": 2.8971527114373566e-06,
      "loss": 0.6154,
      "step": 14550
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.000751638331322,
      "learning_rate": 2.8964973545790322e-06,
      "loss": 0.6111,
      "step": 14551
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.33573654289518,
      "learning_rate": 2.8958420416267186e-06,
      "loss": 0.599,
      "step": 14552
    },
    {
      "epoch": 1.3,
      "grad_norm": 7.9136006998574215,
      "learning_rate": 2.8951867725940967e-06,
      "loss": 0.586,
      "step": 14553
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.346644992356539,
      "learning_rate": 2.89453154749484e-06,
      "loss": 0.5904,
      "step": 14554
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.296452160201325,
      "learning_rate": 2.8938763663426256e-06,
      "loss": 0.5603,
      "step": 14555
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.565424717771355,
      "learning_rate": 2.8932212291511318e-06,
      "loss": 0.5479,
      "step": 14556
    },
    {
      "epoch": 1.3,
      "grad_norm": 4.108016342723745,
      "learning_rate": 2.8925661359340297e-06,
      "loss": 0.6041,
      "step": 14557
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.546824388534938,
      "learning_rate": 2.891911086704994e-06,
      "loss": 0.5669,
      "step": 14558
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.596318851477017,
      "learning_rate": 2.8912560814776973e-06,
      "loss": 0.7147,
      "step": 14559
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.8108105698924,
      "learning_rate": 2.8906011202658122e-06,
      "loss": 0.6083,
      "step": 14560
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.401324960002299,
      "learning_rate": 2.8899462030830093e-06,
      "loss": 0.6212,
      "step": 14561
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.82481361165915,
      "learning_rate": 2.8892913299429605e-06,
      "loss": 0.5757,
      "step": 14562
    },
    {
      "epoch": 1.3,
      "grad_norm": 4.8821333942944385,
      "learning_rate": 2.88863650085933e-06,
      "loss": 0.5555,
      "step": 14563
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.918394121809769,
      "learning_rate": 2.8879817158457895e-06,
      "loss": 0.6192,
      "step": 14564
    },
    {
      "epoch": 1.3,
      "grad_norm": 7.538116701325618,
      "learning_rate": 2.8873269749160052e-06,
      "loss": 0.5579,
      "step": 14565
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.78462738697619,
      "learning_rate": 2.8866722780836443e-06,
      "loss": 0.6139,
      "step": 14566
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.938717803554829,
      "learning_rate": 2.886017625362373e-06,
      "loss": 0.6685,
      "step": 14567
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.665252907280941,
      "learning_rate": 2.885363016765852e-06,
      "loss": 0.5613,
      "step": 14568
    },
    {
      "epoch": 1.3,
      "grad_norm": 8.699119144330691,
      "learning_rate": 2.884708452307749e-06,
      "loss": 0.6269,
      "step": 14569
    },
    {
      "epoch": 1.3,
      "grad_norm": 7.355860222905223,
      "learning_rate": 2.884053932001723e-06,
      "loss": 0.6519,
      "step": 14570
    },
    {
      "epoch": 1.3,
      "grad_norm": 8.4993881285262,
      "learning_rate": 2.883399455861437e-06,
      "loss": 0.6093,
      "step": 14571
    },
    {
      "epoch": 1.3,
      "grad_norm": 7.921579768504905,
      "learning_rate": 2.882745023900552e-06,
      "loss": 0.5943,
      "step": 14572
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.592143959612229,
      "learning_rate": 2.882090636132728e-06,
      "loss": 0.5384,
      "step": 14573
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.0174634783975955,
      "learning_rate": 2.8814362925716244e-06,
      "loss": 0.5663,
      "step": 14574
    },
    {
      "epoch": 1.3,
      "grad_norm": 7.292763073411334,
      "learning_rate": 2.8807819932309e-06,
      "loss": 0.5984,
      "step": 14575
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.408687781902184,
      "learning_rate": 2.8801277381242087e-06,
      "loss": 0.588,
      "step": 14576
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.147232410919068,
      "learning_rate": 2.8794735272652085e-06,
      "loss": 0.5854,
      "step": 14577
    },
    {
      "epoch": 1.3,
      "grad_norm": 7.4422059108977985,
      "learning_rate": 2.8788193606675553e-06,
      "loss": 0.6145,
      "step": 14578
    },
    {
      "epoch": 1.3,
      "grad_norm": 7.7427389236091955,
      "learning_rate": 2.878165238344901e-06,
      "loss": 0.5833,
      "step": 14579
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.519504109709133,
      "learning_rate": 2.877511160310902e-06,
      "loss": 0.5885,
      "step": 14580
    },
    {
      "epoch": 1.3,
      "grad_norm": 4.519805323100826,
      "learning_rate": 2.8768571265792105e-06,
      "loss": 0.574,
      "step": 14581
    },
    {
      "epoch": 1.3,
      "grad_norm": 4.1194463678806414,
      "learning_rate": 2.876203137163477e-06,
      "loss": 0.4983,
      "step": 14582
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.975662751876128,
      "learning_rate": 2.8755491920773506e-06,
      "loss": 0.5853,
      "step": 14583
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.286827143671914,
      "learning_rate": 2.874895291334482e-06,
      "loss": 0.6089,
      "step": 14584
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.699779990514072,
      "learning_rate": 2.87424143494852e-06,
      "loss": 0.5936,
      "step": 14585
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.903985672313625,
      "learning_rate": 2.8735876229331128e-06,
      "loss": 0.6102,
      "step": 14586
    },
    {
      "epoch": 1.3,
      "grad_norm": 7.317618274327215,
      "learning_rate": 2.8729338553019073e-06,
      "loss": 0.5817,
      "step": 14587
    },
    {
      "epoch": 1.3,
      "grad_norm": 4.962095243915575,
      "learning_rate": 2.872280132068551e-06,
      "loss": 0.554,
      "step": 14588
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.427462810216966,
      "learning_rate": 2.8716264532466864e-06,
      "loss": 0.5741,
      "step": 14589
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.349386448844346,
      "learning_rate": 2.8709728188499573e-06,
      "loss": 0.5996,
      "step": 14590
    },
    {
      "epoch": 1.3,
      "grad_norm": 8.006896347491942,
      "learning_rate": 2.870319228892009e-06,
      "loss": 0.5742,
      "step": 14591
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.802868827062682,
      "learning_rate": 2.869665683386482e-06,
      "loss": 0.5912,
      "step": 14592
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.258613401389508,
      "learning_rate": 2.8690121823470196e-06,
      "loss": 0.5745,
      "step": 14593
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.714941825342107,
      "learning_rate": 2.8683587257872625e-06,
      "loss": 0.6498,
      "step": 14594
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.828825279431822,
      "learning_rate": 2.8677053137208464e-06,
      "loss": 0.6005,
      "step": 14595
    },
    {
      "epoch": 1.3,
      "grad_norm": 8.641867266009958,
      "learning_rate": 2.867051946161415e-06,
      "loss": 0.621,
      "step": 14596
    },
    {
      "epoch": 1.3,
      "grad_norm": 4.876371214965587,
      "learning_rate": 2.8663986231226004e-06,
      "loss": 0.5645,
      "step": 14597
    },
    {
      "epoch": 1.3,
      "grad_norm": 7.480565581803046,
      "learning_rate": 2.8657453446180428e-06,
      "loss": 0.5694,
      "step": 14598
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.515827463282158,
      "learning_rate": 2.8650921106613767e-06,
      "loss": 0.5569,
      "step": 14599
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.024646566978927,
      "learning_rate": 2.864438921266238e-06,
      "loss": 0.6116,
      "step": 14600
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.728279032478568,
      "learning_rate": 2.863785776446261e-06,
      "loss": 0.544,
      "step": 14601
    },
    {
      "epoch": 1.3,
      "grad_norm": 10.263198572648042,
      "learning_rate": 2.863132676215076e-06,
      "loss": 0.6159,
      "step": 14602
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.271357950055915,
      "learning_rate": 2.862479620586317e-06,
      "loss": 0.5733,
      "step": 14603
    },
    {
      "epoch": 1.3,
      "grad_norm": 7.2246920739570895,
      "learning_rate": 2.861826609573615e-06,
      "loss": 0.572,
      "step": 14604
    },
    {
      "epoch": 1.3,
      "grad_norm": 8.624085305367228,
      "learning_rate": 2.8611736431906002e-06,
      "loss": 0.5622,
      "step": 14605
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.160839399764682,
      "learning_rate": 2.860520721450901e-06,
      "loss": 0.5935,
      "step": 14606
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.257261900799848,
      "learning_rate": 2.8598678443681486e-06,
      "loss": 0.5451,
      "step": 14607
    },
    {
      "epoch": 1.3,
      "grad_norm": 10.222950006718785,
      "learning_rate": 2.859215011955966e-06,
      "loss": 0.6286,
      "step": 14608
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.455460762380976,
      "learning_rate": 2.858562224227982e-06,
      "loss": 0.5946,
      "step": 14609
    },
    {
      "epoch": 1.3,
      "grad_norm": 8.974315092214637,
      "learning_rate": 2.8579094811978235e-06,
      "loss": 0.5896,
      "step": 14610
    },
    {
      "epoch": 1.3,
      "grad_norm": 7.277357614601458,
      "learning_rate": 2.857256782879112e-06,
      "loss": 0.5722,
      "step": 14611
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.145542808188252,
      "learning_rate": 2.856604129285473e-06,
      "loss": 0.5576,
      "step": 14612
    },
    {
      "epoch": 1.3,
      "grad_norm": 7.616496264160208,
      "learning_rate": 2.8559515204305275e-06,
      "loss": 0.6199,
      "step": 14613
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.22966779126883,
      "learning_rate": 2.855298956327902e-06,
      "loss": 0.6522,
      "step": 14614
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.509542506308877,
      "learning_rate": 2.8546464369912116e-06,
      "loss": 0.5721,
      "step": 14615
    },
    {
      "epoch": 1.3,
      "grad_norm": 4.8755631682132226,
      "learning_rate": 2.8539939624340786e-06,
      "loss": 0.5684,
      "step": 14616
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.746198994894159,
      "learning_rate": 2.8533415326701214e-06,
      "loss": 0.5874,
      "step": 14617
    },
    {
      "epoch": 1.3,
      "grad_norm": 7.672804759488389,
      "learning_rate": 2.8526891477129594e-06,
      "loss": 0.6105,
      "step": 14618
    },
    {
      "epoch": 1.3,
      "grad_norm": 7.047693129054449,
      "learning_rate": 2.8520368075762083e-06,
      "loss": 0.5997,
      "step": 14619
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.3873878912263775,
      "learning_rate": 2.8513845122734867e-06,
      "loss": 0.5867,
      "step": 14620
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.323409139562968,
      "learning_rate": 2.8507322618184063e-06,
      "loss": 0.5675,
      "step": 14621
    },
    {
      "epoch": 1.3,
      "grad_norm": 7.041912851952509,
      "learning_rate": 2.8500800562245833e-06,
      "loss": 0.5886,
      "step": 14622
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.924370562154884,
      "learning_rate": 2.849427895505631e-06,
      "loss": 0.5911,
      "step": 14623
    },
    {
      "epoch": 1.3,
      "grad_norm": 7.974260267775128,
      "learning_rate": 2.8487757796751616e-06,
      "loss": 0.6882,
      "step": 14624
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.991992146352666,
      "learning_rate": 2.848123708746788e-06,
      "loss": 0.5749,
      "step": 14625
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.376901198105059,
      "learning_rate": 2.847471682734118e-06,
      "loss": 0.624,
      "step": 14626
    },
    {
      "epoch": 1.3,
      "grad_norm": 8.388046037533638,
      "learning_rate": 2.8468197016507646e-06,
      "loss": 0.6,
      "step": 14627
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.281893888198189,
      "learning_rate": 2.8461677655103325e-06,
      "loss": 0.555,
      "step": 14628
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.49058759571483,
      "learning_rate": 2.845515874326432e-06,
      "loss": 0.578,
      "step": 14629
    },
    {
      "epoch": 1.31,
      "grad_norm": 8.24310675651983,
      "learning_rate": 2.8448640281126694e-06,
      "loss": 0.5426,
      "step": 14630
    },
    {
      "epoch": 1.31,
      "grad_norm": 7.125756863629804,
      "learning_rate": 2.84421222688265e-06,
      "loss": 0.5541,
      "step": 14631
    },
    {
      "epoch": 1.31,
      "grad_norm": 8.50503056284289,
      "learning_rate": 2.843560470649981e-06,
      "loss": 0.5256,
      "step": 14632
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.411022549641236,
      "learning_rate": 2.8429087594282633e-06,
      "loss": 0.6339,
      "step": 14633
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.810738433771384,
      "learning_rate": 2.842257093231101e-06,
      "loss": 0.5884,
      "step": 14634
    },
    {
      "epoch": 1.31,
      "grad_norm": 7.254298457105643,
      "learning_rate": 2.8416054720720965e-06,
      "loss": 0.5447,
      "step": 14635
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.555677488923051,
      "learning_rate": 2.8409538959648507e-06,
      "loss": 0.5321,
      "step": 14636
    },
    {
      "epoch": 1.31,
      "grad_norm": 7.926985056074106,
      "learning_rate": 2.840302364922964e-06,
      "loss": 0.5895,
      "step": 14637
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.8515248514586515,
      "learning_rate": 2.839650878960039e-06,
      "loss": 0.6222,
      "step": 14638
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.1382569063424945,
      "learning_rate": 2.83899943808967e-06,
      "loss": 0.4932,
      "step": 14639
    },
    {
      "epoch": 1.31,
      "grad_norm": 7.213444639645714,
      "learning_rate": 2.8383480423254533e-06,
      "loss": 0.5691,
      "step": 14640
    },
    {
      "epoch": 1.31,
      "grad_norm": 7.516735519860875,
      "learning_rate": 2.8376966916809873e-06,
      "loss": 0.5912,
      "step": 14641
    },
    {
      "epoch": 1.31,
      "grad_norm": 7.22457883484646,
      "learning_rate": 2.837045386169867e-06,
      "loss": 0.5543,
      "step": 14642
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.5706122546570365,
      "learning_rate": 2.8363941258056882e-06,
      "loss": 0.5895,
      "step": 14643
    },
    {
      "epoch": 1.31,
      "grad_norm": 7.1809636592361015,
      "learning_rate": 2.8357429106020433e-06,
      "loss": 0.5902,
      "step": 14644
    },
    {
      "epoch": 1.31,
      "grad_norm": 7.259697021088739,
      "learning_rate": 2.8350917405725276e-06,
      "loss": 0.5852,
      "step": 14645
    },
    {
      "epoch": 1.31,
      "grad_norm": 8.26865289136976,
      "learning_rate": 2.834440615730729e-06,
      "loss": 0.6448,
      "step": 14646
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.3767112702828745,
      "learning_rate": 2.83378953609024e-06,
      "loss": 0.645,
      "step": 14647
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.3077570308320094,
      "learning_rate": 2.8331385016646505e-06,
      "loss": 0.6084,
      "step": 14648
    },
    {
      "epoch": 1.31,
      "grad_norm": 4.611053219388775,
      "learning_rate": 2.8324875124675486e-06,
      "loss": 0.5629,
      "step": 14649
    },
    {
      "epoch": 1.31,
      "grad_norm": 8.945938310722376,
      "learning_rate": 2.831836568512525e-06,
      "loss": 0.6268,
      "step": 14650
    },
    {
      "epoch": 1.31,
      "grad_norm": 4.922159822298804,
      "learning_rate": 2.8311856698131653e-06,
      "loss": 0.5808,
      "step": 14651
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.406849992512114,
      "learning_rate": 2.8305348163830536e-06,
      "loss": 0.5604,
      "step": 14652
    },
    {
      "epoch": 1.31,
      "grad_norm": 7.883642567149509,
      "learning_rate": 2.829884008235778e-06,
      "loss": 0.5781,
      "step": 14653
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.76383644991604,
      "learning_rate": 2.8292332453849203e-06,
      "loss": 0.6056,
      "step": 14654
    },
    {
      "epoch": 1.31,
      "grad_norm": 7.997609083422815,
      "learning_rate": 2.828582527844064e-06,
      "loss": 0.6069,
      "step": 14655
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.443818463123299,
      "learning_rate": 2.827931855626792e-06,
      "loss": 0.6437,
      "step": 14656
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.325701412945984,
      "learning_rate": 2.8272812287466855e-06,
      "loss": 0.635,
      "step": 14657
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.166774160814114,
      "learning_rate": 2.8266306472173275e-06,
      "loss": 0.626,
      "step": 14658
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.898914164405058,
      "learning_rate": 2.8259801110522938e-06,
      "loss": 0.6125,
      "step": 14659
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.054574362914878,
      "learning_rate": 2.825329620265164e-06,
      "loss": 0.6048,
      "step": 14660
    },
    {
      "epoch": 1.31,
      "grad_norm": 4.784374453714737,
      "learning_rate": 2.8246791748695156e-06,
      "loss": 0.587,
      "step": 14661
    },
    {
      "epoch": 1.31,
      "grad_norm": 7.829627942492753,
      "learning_rate": 2.8240287748789254e-06,
      "loss": 0.6135,
      "step": 14662
    },
    {
      "epoch": 1.31,
      "grad_norm": 8.107675920992314,
      "learning_rate": 2.82337842030697e-06,
      "loss": 0.6037,
      "step": 14663
    },
    {
      "epoch": 1.31,
      "grad_norm": 7.045663102114819,
      "learning_rate": 2.8227281111672243e-06,
      "loss": 0.5633,
      "step": 14664
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.5845017407080935,
      "learning_rate": 2.82207784747326e-06,
      "loss": 0.5555,
      "step": 14665
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.884136104029214,
      "learning_rate": 2.8214276292386506e-06,
      "loss": 0.557,
      "step": 14666
    },
    {
      "epoch": 1.31,
      "grad_norm": 8.174991713891545,
      "learning_rate": 2.8207774564769686e-06,
      "loss": 0.617,
      "step": 14667
    },
    {
      "epoch": 1.31,
      "grad_norm": 7.863439447404439,
      "learning_rate": 2.8201273292017872e-06,
      "loss": 0.5968,
      "step": 14668
    },
    {
      "epoch": 1.31,
      "grad_norm": 7.7306874801848195,
      "learning_rate": 2.8194772474266713e-06,
      "loss": 0.5876,
      "step": 14669
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.143109173098023,
      "learning_rate": 2.8188272111651927e-06,
      "loss": 0.6464,
      "step": 14670
    },
    {
      "epoch": 1.31,
      "grad_norm": 7.919870746999614,
      "learning_rate": 2.8181772204309216e-06,
      "loss": 0.6045,
      "step": 14671
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.929705207313971,
      "learning_rate": 2.8175272752374204e-06,
      "loss": 0.65,
      "step": 14672
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.140757151274942,
      "learning_rate": 2.816877375598258e-06,
      "loss": 0.5786,
      "step": 14673
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.791864241591136,
      "learning_rate": 2.816227521527e-06,
      "loss": 0.6113,
      "step": 14674
    },
    {
      "epoch": 1.31,
      "grad_norm": 8.355617659014257,
      "learning_rate": 2.81557771303721e-06,
      "loss": 0.5406,
      "step": 14675
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.917026591837284,
      "learning_rate": 2.814927950142451e-06,
      "loss": 0.632,
      "step": 14676
    },
    {
      "epoch": 1.31,
      "grad_norm": 8.673949072292034,
      "learning_rate": 2.8142782328562878e-06,
      "loss": 0.5551,
      "step": 14677
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.843218696148027,
      "learning_rate": 2.813628561192279e-06,
      "loss": 0.582,
      "step": 14678
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.608672206312245,
      "learning_rate": 2.8129789351639847e-06,
      "loss": 0.5622,
      "step": 14679
    },
    {
      "epoch": 1.31,
      "grad_norm": 7.147133932887482,
      "learning_rate": 2.8123293547849655e-06,
      "loss": 0.625,
      "step": 14680
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.0141314189998125,
      "learning_rate": 2.8116798200687805e-06,
      "loss": 0.5535,
      "step": 14681
    },
    {
      "epoch": 1.31,
      "grad_norm": 7.738965870934562,
      "learning_rate": 2.8110303310289887e-06,
      "loss": 0.6859,
      "step": 14682
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.010945492812928,
      "learning_rate": 2.8103808876791434e-06,
      "loss": 0.6316,
      "step": 14683
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.9653910516612685,
      "learning_rate": 2.809731490032804e-06,
      "loss": 0.6269,
      "step": 14684
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.2763726930059605,
      "learning_rate": 2.8090821381035207e-06,
      "loss": 0.5564,
      "step": 14685
    },
    {
      "epoch": 1.31,
      "grad_norm": 7.798525645534601,
      "learning_rate": 2.808432831904849e-06,
      "loss": 0.5869,
      "step": 14686
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.066442208784562,
      "learning_rate": 2.8077835714503444e-06,
      "loss": 0.5589,
      "step": 14687
    },
    {
      "epoch": 1.31,
      "grad_norm": 8.605745984702406,
      "learning_rate": 2.807134356753555e-06,
      "loss": 0.6489,
      "step": 14688
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.976100456950208,
      "learning_rate": 2.8064851878280347e-06,
      "loss": 0.5416,
      "step": 14689
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.1588069408382,
      "learning_rate": 2.805836064687334e-06,
      "loss": 0.5891,
      "step": 14690
    },
    {
      "epoch": 1.31,
      "grad_norm": 4.546737460867647,
      "learning_rate": 2.805186987344999e-06,
      "loss": 0.5644,
      "step": 14691
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.855988800974197,
      "learning_rate": 2.804537955814578e-06,
      "loss": 0.5649,
      "step": 14692
    },
    {
      "epoch": 1.31,
      "grad_norm": 7.707335917945903,
      "learning_rate": 2.8038889701096196e-06,
      "loss": 0.5817,
      "step": 14693
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.82908368757806,
      "learning_rate": 2.80324003024367e-06,
      "loss": 0.629,
      "step": 14694
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.814775642346032,
      "learning_rate": 2.8025911362302753e-06,
      "loss": 0.5701,
      "step": 14695
    },
    {
      "epoch": 1.31,
      "grad_norm": 8.25854304466767,
      "learning_rate": 2.8019422880829765e-06,
      "loss": 0.6092,
      "step": 14696
    },
    {
      "epoch": 1.31,
      "grad_norm": 8.051217189193633,
      "learning_rate": 2.801293485815321e-06,
      "loss": 0.602,
      "step": 14697
    },
    {
      "epoch": 1.31,
      "grad_norm": 8.323697431113397,
      "learning_rate": 2.8006447294408466e-06,
      "loss": 0.5928,
      "step": 14698
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.203633784829223,
      "learning_rate": 2.7999960189730977e-06,
      "loss": 0.6218,
      "step": 14699
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.724469400235122,
      "learning_rate": 2.7993473544256134e-06,
      "loss": 0.5369,
      "step": 14700
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.834829329359145,
      "learning_rate": 2.7986987358119338e-06,
      "loss": 0.614,
      "step": 14701
    },
    {
      "epoch": 1.31,
      "grad_norm": 8.236452130208386,
      "learning_rate": 2.7980501631455974e-06,
      "loss": 0.5982,
      "step": 14702
    },
    {
      "epoch": 1.31,
      "grad_norm": 8.184207686513725,
      "learning_rate": 2.7974016364401434e-06,
      "loss": 0.5663,
      "step": 14703
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.944426248083473,
      "learning_rate": 2.7967531557091056e-06,
      "loss": 0.5291,
      "step": 14704
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.49335656970426,
      "learning_rate": 2.79610472096602e-06,
      "loss": 0.6356,
      "step": 14705
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.611212994060446,
      "learning_rate": 2.795456332224421e-06,
      "loss": 0.5754,
      "step": 14706
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.815370796531603,
      "learning_rate": 2.794807989497844e-06,
      "loss": 0.628,
      "step": 14707
    },
    {
      "epoch": 1.31,
      "grad_norm": 8.546173578266586,
      "learning_rate": 2.7941596927998225e-06,
      "loss": 0.6107,
      "step": 14708
    },
    {
      "epoch": 1.31,
      "grad_norm": 7.562720550244697,
      "learning_rate": 2.7935114421438857e-06,
      "loss": 0.5623,
      "step": 14709
    },
    {
      "epoch": 1.31,
      "grad_norm": 7.035297560727162,
      "learning_rate": 2.792863237543565e-06,
      "loss": 0.5928,
      "step": 14710
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.110485605084381,
      "learning_rate": 2.7922150790123924e-06,
      "loss": 0.5738,
      "step": 14711
    },
    {
      "epoch": 1.31,
      "grad_norm": 12.16711773786022,
      "learning_rate": 2.7915669665638926e-06,
      "loss": 0.6094,
      "step": 14712
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.995457314588372,
      "learning_rate": 2.7909189002115967e-06,
      "loss": 0.6154,
      "step": 14713
    },
    {
      "epoch": 1.31,
      "grad_norm": 7.474035074897762,
      "learning_rate": 2.790270879969031e-06,
      "loss": 0.5786,
      "step": 14714
    },
    {
      "epoch": 1.31,
      "grad_norm": 7.436356114953808,
      "learning_rate": 2.789622905849723e-06,
      "loss": 0.5802,
      "step": 14715
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.746459084331684,
      "learning_rate": 2.788974977867195e-06,
      "loss": 0.5965,
      "step": 14716
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.2934500579697685,
      "learning_rate": 2.788327096034972e-06,
      "loss": 0.6338,
      "step": 14717
    },
    {
      "epoch": 1.31,
      "grad_norm": 8.886502632857635,
      "learning_rate": 2.7876792603665765e-06,
      "loss": 0.6319,
      "step": 14718
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.670049907811608,
      "learning_rate": 2.787031470875532e-06,
      "loss": 0.6219,
      "step": 14719
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.525783044032521,
      "learning_rate": 2.786383727575359e-06,
      "loss": 0.6078,
      "step": 14720
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.604540455430241,
      "learning_rate": 2.785736030479581e-06,
      "loss": 0.5451,
      "step": 14721
    },
    {
      "epoch": 1.31,
      "grad_norm": 4.982999071691602,
      "learning_rate": 2.7850883796017116e-06,
      "loss": 0.5607,
      "step": 14722
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.303709232434837,
      "learning_rate": 2.7844407749552713e-06,
      "loss": 0.6218,
      "step": 14723
    },
    {
      "epoch": 1.31,
      "grad_norm": 7.273834913219414,
      "learning_rate": 2.7837932165537786e-06,
      "loss": 0.5528,
      "step": 14724
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.976318897762929,
      "learning_rate": 2.7831457044107477e-06,
      "loss": 0.585,
      "step": 14725
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.513638998201363,
      "learning_rate": 2.782498238539699e-06,
      "loss": 0.5947,
      "step": 14726
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.8018197907209546,
      "learning_rate": 2.7818508189541404e-06,
      "loss": 0.5404,
      "step": 14727
    },
    {
      "epoch": 1.31,
      "grad_norm": 8.656951695415131,
      "learning_rate": 2.7812034456675906e-06,
      "loss": 0.5563,
      "step": 14728
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.998817415757714,
      "learning_rate": 2.780556118693558e-06,
      "loss": 0.6725,
      "step": 14729
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.092548108990833,
      "learning_rate": 2.779908838045556e-06,
      "loss": 0.658,
      "step": 14730
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.369574802493496,
      "learning_rate": 2.7792616037370944e-06,
      "loss": 0.6019,
      "step": 14731
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.759855208646145,
      "learning_rate": 2.778614415781684e-06,
      "loss": 0.6037,
      "step": 14732
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.601370856623269,
      "learning_rate": 2.7779672741928326e-06,
      "loss": 0.6087,
      "step": 14733
    },
    {
      "epoch": 1.31,
      "grad_norm": 8.396933851959563,
      "learning_rate": 2.7773201789840505e-06,
      "loss": 0.5961,
      "step": 14734
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.8989775760133325,
      "learning_rate": 2.7766731301688397e-06,
      "loss": 0.6035,
      "step": 14735
    },
    {
      "epoch": 1.31,
      "grad_norm": 4.795462326738627,
      "learning_rate": 2.7760261277607086e-06,
      "loss": 0.5768,
      "step": 14736
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.276172619241128,
      "learning_rate": 2.7753791717731616e-06,
      "loss": 0.5711,
      "step": 14737
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.444923962889785,
      "learning_rate": 2.774732262219703e-06,
      "loss": 0.6165,
      "step": 14738
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.022579838976158,
      "learning_rate": 2.7740853991138344e-06,
      "loss": 0.5893,
      "step": 14739
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.529840957324658,
      "learning_rate": 2.7734385824690606e-06,
      "loss": 0.5735,
      "step": 14740
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.602731305997631,
      "learning_rate": 2.7727918122988805e-06,
      "loss": 0.6073,
      "step": 14741
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.428052904001444,
      "learning_rate": 2.7721450886167922e-06,
      "loss": 0.6118,
      "step": 14742
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.561973206802653,
      "learning_rate": 2.771498411436295e-06,
      "loss": 0.6516,
      "step": 14743
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.499416496957474,
      "learning_rate": 2.7708517807708894e-06,
      "loss": 0.5607,
      "step": 14744
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.5241073251317605,
      "learning_rate": 2.77020519663407e-06,
      "loss": 0.5933,
      "step": 14745
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.646712953813759,
      "learning_rate": 2.769558659039335e-06,
      "loss": 0.5981,
      "step": 14746
    },
    {
      "epoch": 1.32,
      "grad_norm": 7.394998880480625,
      "learning_rate": 2.768912168000181e-06,
      "loss": 0.6073,
      "step": 14747
    },
    {
      "epoch": 1.32,
      "grad_norm": 8.48662922392802,
      "learning_rate": 2.7682657235300958e-06,
      "loss": 0.5632,
      "step": 14748
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.272496587316058,
      "learning_rate": 2.767619325642577e-06,
      "loss": 0.6223,
      "step": 14749
    },
    {
      "epoch": 1.32,
      "grad_norm": 7.751315594580467,
      "learning_rate": 2.766972974351116e-06,
      "loss": 0.6282,
      "step": 14750
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.362730041266211,
      "learning_rate": 2.7663266696692036e-06,
      "loss": 0.6068,
      "step": 14751
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.699258261492003,
      "learning_rate": 2.7656804116103307e-06,
      "loss": 0.5838,
      "step": 14752
    },
    {
      "epoch": 1.32,
      "grad_norm": 8.667695521708877,
      "learning_rate": 2.765034200187988e-06,
      "loss": 0.6331,
      "step": 14753
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.625405394545013,
      "learning_rate": 2.7643880354156616e-06,
      "loss": 0.5741,
      "step": 14754
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.298152923203906,
      "learning_rate": 2.7637419173068375e-06,
      "loss": 0.5615,
      "step": 14755
    },
    {
      "epoch": 1.32,
      "grad_norm": 8.296218566368875,
      "learning_rate": 2.7630958458750036e-06,
      "loss": 0.5876,
      "step": 14756
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.6146593257599955,
      "learning_rate": 2.7624498211336446e-06,
      "loss": 0.5932,
      "step": 14757
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.577184504335389,
      "learning_rate": 2.761803843096246e-06,
      "loss": 0.5995,
      "step": 14758
    },
    {
      "epoch": 1.32,
      "grad_norm": 7.243610449161306,
      "learning_rate": 2.76115791177629e-06,
      "loss": 0.6269,
      "step": 14759
    },
    {
      "epoch": 1.32,
      "grad_norm": 7.871397616866562,
      "learning_rate": 2.7605120271872622e-06,
      "loss": 0.6025,
      "step": 14760
    },
    {
      "epoch": 1.32,
      "grad_norm": 7.849892342980423,
      "learning_rate": 2.7598661893426393e-06,
      "loss": 0.6122,
      "step": 14761
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.038411366504692,
      "learning_rate": 2.7592203982559033e-06,
      "loss": 0.6008,
      "step": 14762
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.145141728314818,
      "learning_rate": 2.7585746539405352e-06,
      "loss": 0.6266,
      "step": 14763
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.871851163533148,
      "learning_rate": 2.757928956410012e-06,
      "loss": 0.6055,
      "step": 14764
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.137863036054402,
      "learning_rate": 2.757283305677812e-06,
      "loss": 0.5787,
      "step": 14765
    },
    {
      "epoch": 1.32,
      "grad_norm": 8.239728993783089,
      "learning_rate": 2.756637701757413e-06,
      "loss": 0.6431,
      "step": 14766
    },
    {
      "epoch": 1.32,
      "grad_norm": 9.774382453594818,
      "learning_rate": 2.7559921446622877e-06,
      "loss": 0.553,
      "step": 14767
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.182062357767533,
      "learning_rate": 2.7553466344059122e-06,
      "loss": 0.6345,
      "step": 14768
    },
    {
      "epoch": 1.32,
      "grad_norm": 7.15307730314865,
      "learning_rate": 2.754701171001761e-06,
      "loss": 0.5607,
      "step": 14769
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.798204522230267,
      "learning_rate": 2.7540557544633052e-06,
      "loss": 0.5655,
      "step": 14770
    },
    {
      "epoch": 1.32,
      "grad_norm": 7.286090903635536,
      "learning_rate": 2.753410384804016e-06,
      "loss": 0.5442,
      "step": 14771
    },
    {
      "epoch": 1.32,
      "grad_norm": 4.604793727675368,
      "learning_rate": 2.7527650620373654e-06,
      "loss": 0.6301,
      "step": 14772
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.5414526826250015,
      "learning_rate": 2.7521197861768243e-06,
      "loss": 0.5535,
      "step": 14773
    },
    {
      "epoch": 1.32,
      "grad_norm": 7.554883922511573,
      "learning_rate": 2.7514745572358586e-06,
      "loss": 0.5477,
      "step": 14774
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.22578239787286,
      "learning_rate": 2.750829375227937e-06,
      "loss": 0.594,
      "step": 14775
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.351891913371103,
      "learning_rate": 2.750184240166527e-06,
      "loss": 0.6109,
      "step": 14776
    },
    {
      "epoch": 1.32,
      "grad_norm": 7.16354643340043,
      "learning_rate": 2.7495391520650935e-06,
      "loss": 0.6846,
      "step": 14777
    },
    {
      "epoch": 1.32,
      "grad_norm": 7.073654236985589,
      "learning_rate": 2.7488941109371014e-06,
      "loss": 0.653,
      "step": 14778
    },
    {
      "epoch": 1.32,
      "grad_norm": 7.007139337164889,
      "learning_rate": 2.7482491167960175e-06,
      "loss": 0.6077,
      "step": 14779
    },
    {
      "epoch": 1.32,
      "grad_norm": 4.93099566361456,
      "learning_rate": 2.7476041696553e-06,
      "loss": 0.6047,
      "step": 14780
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.595393859394191,
      "learning_rate": 2.7469592695284127e-06,
      "loss": 0.5795,
      "step": 14781
    },
    {
      "epoch": 1.32,
      "grad_norm": 7.742193299293692,
      "learning_rate": 2.7463144164288164e-06,
      "loss": 0.5943,
      "step": 14782
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.175508599838531,
      "learning_rate": 2.7456696103699733e-06,
      "loss": 0.6143,
      "step": 14783
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.498281517334388,
      "learning_rate": 2.7450248513653376e-06,
      "loss": 0.5935,
      "step": 14784
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.958046630140257,
      "learning_rate": 2.744380139428371e-06,
      "loss": 0.6097,
      "step": 14785
    },
    {
      "epoch": 1.32,
      "grad_norm": 8.082719549028747,
      "learning_rate": 2.743735474572531e-06,
      "loss": 0.5822,
      "step": 14786
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.5395063146141315,
      "learning_rate": 2.743090856811269e-06,
      "loss": 0.5928,
      "step": 14787
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.874196927153603,
      "learning_rate": 2.742446286158044e-06,
      "loss": 0.6611,
      "step": 14788
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.377205580359138,
      "learning_rate": 2.7418017626263076e-06,
      "loss": 0.5349,
      "step": 14789
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.99683657774061,
      "learning_rate": 2.7411572862295146e-06,
      "loss": 0.5622,
      "step": 14790
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.229712372929733,
      "learning_rate": 2.7405128569811186e-06,
      "loss": 0.6094,
      "step": 14791
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.484233175308531,
      "learning_rate": 2.739868474894566e-06,
      "loss": 0.5874,
      "step": 14792
    },
    {
      "epoch": 1.32,
      "grad_norm": 8.328353405207565,
      "learning_rate": 2.73922413998331e-06,
      "loss": 0.603,
      "step": 14793
    },
    {
      "epoch": 1.32,
      "grad_norm": 9.414343958017348,
      "learning_rate": 2.7385798522607986e-06,
      "loss": 0.6205,
      "step": 14794
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.470764842769586,
      "learning_rate": 2.737935611740481e-06,
      "loss": 0.5432,
      "step": 14795
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.441225151392155,
      "learning_rate": 2.7372914184358027e-06,
      "loss": 0.6129,
      "step": 14796
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.245398172182102,
      "learning_rate": 2.736647272360213e-06,
      "loss": 0.6201,
      "step": 14797
    },
    {
      "epoch": 1.32,
      "grad_norm": 4.807978534495059,
      "learning_rate": 2.7360031735271557e-06,
      "loss": 0.5572,
      "step": 14798
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.097436859760323,
      "learning_rate": 2.735359121950072e-06,
      "loss": 0.6186,
      "step": 14799
    },
    {
      "epoch": 1.32,
      "grad_norm": 7.180228672757757,
      "learning_rate": 2.7347151176424062e-06,
      "loss": 0.5514,
      "step": 14800
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.339108443108028,
      "learning_rate": 2.734071160617602e-06,
      "loss": 0.5566,
      "step": 14801
    },
    {
      "epoch": 1.32,
      "grad_norm": 7.404977027942776,
      "learning_rate": 2.7334272508891003e-06,
      "loss": 0.6,
      "step": 14802
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.597780414270351,
      "learning_rate": 2.7327833884703402e-06,
      "loss": 0.5934,
      "step": 14803
    },
    {
      "epoch": 1.32,
      "grad_norm": 7.213774227256598,
      "learning_rate": 2.7321395733747646e-06,
      "loss": 0.5736,
      "step": 14804
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.440584455531,
      "learning_rate": 2.7314958056158064e-06,
      "loss": 0.6044,
      "step": 14805
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.018778669473016,
      "learning_rate": 2.7308520852069053e-06,
      "loss": 0.5847,
      "step": 14806
    },
    {
      "epoch": 1.32,
      "grad_norm": 4.3466335436718495,
      "learning_rate": 2.730208412161498e-06,
      "loss": 0.561,
      "step": 14807
    },
    {
      "epoch": 1.32,
      "grad_norm": 7.304543026458657,
      "learning_rate": 2.7295647864930185e-06,
      "loss": 0.5473,
      "step": 14808
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.086225833690758,
      "learning_rate": 2.728921208214903e-06,
      "loss": 0.578,
      "step": 14809
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.479213717710926,
      "learning_rate": 2.728277677340585e-06,
      "loss": 0.6647,
      "step": 14810
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.328063822968609,
      "learning_rate": 2.7276341938834937e-06,
      "loss": 0.6192,
      "step": 14811
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.668889812279947,
      "learning_rate": 2.726990757857064e-06,
      "loss": 0.5999,
      "step": 14812
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.7724247836840785,
      "learning_rate": 2.7263473692747223e-06,
      "loss": 0.5419,
      "step": 14813
    },
    {
      "epoch": 1.32,
      "grad_norm": 4.734111494523718,
      "learning_rate": 2.7257040281499005e-06,
      "loss": 0.5801,
      "step": 14814
    },
    {
      "epoch": 1.32,
      "grad_norm": 7.286505761028956,
      "learning_rate": 2.7250607344960268e-06,
      "loss": 0.6432,
      "step": 14815
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.3693239938932305,
      "learning_rate": 2.7244174883265274e-06,
      "loss": 0.6629,
      "step": 14816
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.967319821354633,
      "learning_rate": 2.7237742896548313e-06,
      "loss": 0.5392,
      "step": 14817
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.733791659103921,
      "learning_rate": 2.723131138494361e-06,
      "loss": 0.6194,
      "step": 14818
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.78626179849597,
      "learning_rate": 2.7224880348585414e-06,
      "loss": 0.5897,
      "step": 14819
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.539612011862975,
      "learning_rate": 2.721844978760797e-06,
      "loss": 0.5807,
      "step": 14820
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.263541383388645,
      "learning_rate": 2.721201970214549e-06,
      "loss": 0.5478,
      "step": 14821
    },
    {
      "epoch": 1.32,
      "grad_norm": 7.389528512808675,
      "learning_rate": 2.720559009233219e-06,
      "loss": 0.5592,
      "step": 14822
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.614801336390614,
      "learning_rate": 2.719916095830231e-06,
      "loss": 0.5918,
      "step": 14823
    },
    {
      "epoch": 1.32,
      "grad_norm": 4.759551114349417,
      "learning_rate": 2.719273230018998e-06,
      "loss": 0.5447,
      "step": 14824
    },
    {
      "epoch": 1.32,
      "grad_norm": 7.245692482858888,
      "learning_rate": 2.7186304118129424e-06,
      "loss": 0.6108,
      "step": 14825
    },
    {
      "epoch": 1.32,
      "grad_norm": 8.345345972003651,
      "learning_rate": 2.7179876412254823e-06,
      "loss": 0.5697,
      "step": 14826
    },
    {
      "epoch": 1.32,
      "grad_norm": 8.119038162198306,
      "learning_rate": 2.7173449182700306e-06,
      "loss": 0.5401,
      "step": 14827
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.727097025577731,
      "learning_rate": 2.7167022429600056e-06,
      "loss": 0.5886,
      "step": 14828
    },
    {
      "epoch": 1.32,
      "grad_norm": 7.959095661410646,
      "learning_rate": 2.7160596153088204e-06,
      "loss": 0.6122,
      "step": 14829
    },
    {
      "epoch": 1.32,
      "grad_norm": 8.177043982027541,
      "learning_rate": 2.7154170353298913e-06,
      "loss": 0.5792,
      "step": 14830
    },
    {
      "epoch": 1.32,
      "grad_norm": 8.845368185880517,
      "learning_rate": 2.714774503036626e-06,
      "loss": 0.5566,
      "step": 14831
    },
    {
      "epoch": 1.32,
      "grad_norm": 9.435598406498471,
      "learning_rate": 2.7141320184424386e-06,
      "loss": 0.5624,
      "step": 14832
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.0524040972869235,
      "learning_rate": 2.7134895815607394e-06,
      "loss": 0.5856,
      "step": 14833
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.295827333264731,
      "learning_rate": 2.712847192404938e-06,
      "loss": 0.5829,
      "step": 14834
    },
    {
      "epoch": 1.32,
      "grad_norm": 7.548080884019631,
      "learning_rate": 2.7122048509884426e-06,
      "loss": 0.6418,
      "step": 14835
    },
    {
      "epoch": 1.32,
      "grad_norm": 4.504170876438622,
      "learning_rate": 2.711562557324662e-06,
      "loss": 0.6224,
      "step": 14836
    },
    {
      "epoch": 1.32,
      "grad_norm": 7.824815677358688,
      "learning_rate": 2.710920311427e-06,
      "loss": 0.5907,
      "step": 14837
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.074124970998369,
      "learning_rate": 2.7102781133088636e-06,
      "loss": 0.6111,
      "step": 14838
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.385269707240346,
      "learning_rate": 2.7096359629836565e-06,
      "loss": 0.571,
      "step": 14839
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.907987756311079,
      "learning_rate": 2.7089938604647835e-06,
      "loss": 0.5637,
      "step": 14840
    },
    {
      "epoch": 1.32,
      "grad_norm": 8.917235500073248,
      "learning_rate": 2.708351805765649e-06,
      "loss": 0.6052,
      "step": 14841
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.5405607959482595,
      "learning_rate": 2.707709798899649e-06,
      "loss": 0.6531,
      "step": 14842
    },
    {
      "epoch": 1.32,
      "grad_norm": 4.561913950110396,
      "learning_rate": 2.70706783988019e-06,
      "loss": 0.5374,
      "step": 14843
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.036629029132004,
      "learning_rate": 2.706425928720665e-06,
      "loss": 0.5725,
      "step": 14844
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.5148579941965075,
      "learning_rate": 2.7057840654344763e-06,
      "loss": 0.5953,
      "step": 14845
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.592301552226379,
      "learning_rate": 2.705142250035021e-06,
      "loss": 0.5996,
      "step": 14846
    },
    {
      "epoch": 1.32,
      "grad_norm": 8.875543086061304,
      "learning_rate": 2.704500482535696e-06,
      "loss": 0.6649,
      "step": 14847
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.428378873240603,
      "learning_rate": 2.7038587629498964e-06,
      "loss": 0.5955,
      "step": 14848
    },
    {
      "epoch": 1.32,
      "grad_norm": 7.871809016662366,
      "learning_rate": 2.703217091291018e-06,
      "loss": 0.5819,
      "step": 14849
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.993799559247861,
      "learning_rate": 2.7025754675724514e-06,
      "loss": 0.6228,
      "step": 14850
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.201038021835727,
      "learning_rate": 2.701933891807591e-06,
      "loss": 0.6393,
      "step": 14851
    },
    {
      "epoch": 1.32,
      "grad_norm": 8.472365600709844,
      "learning_rate": 2.701292364009828e-06,
      "loss": 0.6347,
      "step": 14852
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.113169775173034,
      "learning_rate": 2.7006508841925517e-06,
      "loss": 0.5422,
      "step": 14853
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.788040029677572,
      "learning_rate": 2.7000094523691534e-06,
      "loss": 0.5632,
      "step": 14854
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.339962484642328,
      "learning_rate": 2.6993680685530233e-06,
      "loss": 0.5927,
      "step": 14855
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.663984521898269,
      "learning_rate": 2.698726732757546e-06,
      "loss": 0.6136,
      "step": 14856
    },
    {
      "epoch": 1.33,
      "grad_norm": 7.285786812647119,
      "learning_rate": 2.6980854449961063e-06,
      "loss": 0.5935,
      "step": 14857
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.0219423725873344,
      "learning_rate": 2.697444205282092e-06,
      "loss": 0.6625,
      "step": 14858
    },
    {
      "epoch": 1.33,
      "grad_norm": 4.982314647730492,
      "learning_rate": 2.696803013628887e-06,
      "loss": 0.551,
      "step": 14859
    },
    {
      "epoch": 1.33,
      "grad_norm": 8.198696032457127,
      "learning_rate": 2.696161870049876e-06,
      "loss": 0.6012,
      "step": 14860
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.458813993794885,
      "learning_rate": 2.69552077455844e-06,
      "loss": 0.5727,
      "step": 14861
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.0955312843561495,
      "learning_rate": 2.6948797271679637e-06,
      "loss": 0.5698,
      "step": 14862
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.906845967034571,
      "learning_rate": 2.6942387278918223e-06,
      "loss": 0.6243,
      "step": 14863
    },
    {
      "epoch": 1.33,
      "grad_norm": 7.262808666489778,
      "learning_rate": 2.6935977767433984e-06,
      "loss": 0.5862,
      "step": 14864
    },
    {
      "epoch": 1.33,
      "grad_norm": 8.425785697187173,
      "learning_rate": 2.69295687373607e-06,
      "loss": 0.644,
      "step": 14865
    },
    {
      "epoch": 1.33,
      "grad_norm": 8.417202152937195,
      "learning_rate": 2.6923160188832143e-06,
      "loss": 0.618,
      "step": 14866
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.30468578417814,
      "learning_rate": 2.69167521219821e-06,
      "loss": 0.5685,
      "step": 14867
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.836773671429905,
      "learning_rate": 2.691034453694428e-06,
      "loss": 0.5823,
      "step": 14868
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.85690374115841,
      "learning_rate": 2.690393743385246e-06,
      "loss": 0.5764,
      "step": 14869
    },
    {
      "epoch": 1.33,
      "grad_norm": 8.326214894425881,
      "learning_rate": 2.6897530812840385e-06,
      "loss": 0.6259,
      "step": 14870
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.179927007251622,
      "learning_rate": 2.6891124674041734e-06,
      "loss": 0.6404,
      "step": 14871
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.997689948718386,
      "learning_rate": 2.6884719017590256e-06,
      "loss": 0.5844,
      "step": 14872
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.67820598109221,
      "learning_rate": 2.6878313843619645e-06,
      "loss": 0.5324,
      "step": 14873
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.87489964894729,
      "learning_rate": 2.687190915226362e-06,
      "loss": 0.5969,
      "step": 14874
    },
    {
      "epoch": 1.33,
      "grad_norm": 8.635011775030492,
      "learning_rate": 2.686550494365582e-06,
      "loss": 0.5798,
      "step": 14875
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.699149192701553,
      "learning_rate": 2.6859101217929933e-06,
      "loss": 0.5864,
      "step": 14876
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.9868578235616425,
      "learning_rate": 2.6852697975219633e-06,
      "loss": 0.5709,
      "step": 14877
    },
    {
      "epoch": 1.33,
      "grad_norm": 7.85873524701914,
      "learning_rate": 2.684629521565858e-06,
      "loss": 0.6283,
      "step": 14878
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.518228196288253,
      "learning_rate": 2.6839892939380395e-06,
      "loss": 0.6298,
      "step": 14879
    },
    {
      "epoch": 1.33,
      "grad_norm": 7.154881710436993,
      "learning_rate": 2.683349114651876e-06,
      "loss": 0.532,
      "step": 14880
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.154491964828763,
      "learning_rate": 2.6827089837207232e-06,
      "loss": 0.5733,
      "step": 14881
    },
    {
      "epoch": 1.33,
      "grad_norm": 8.032123197322356,
      "learning_rate": 2.682068901157946e-06,
      "loss": 0.578,
      "step": 14882
    },
    {
      "epoch": 1.33,
      "grad_norm": 7.471968901214238,
      "learning_rate": 2.6814288669769044e-06,
      "loss": 0.65,
      "step": 14883
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.659905221004216,
      "learning_rate": 2.68078888119096e-06,
      "loss": 0.6641,
      "step": 14884
    },
    {
      "epoch": 1.33,
      "grad_norm": 4.737069024089865,
      "learning_rate": 2.680148943813466e-06,
      "loss": 0.6009,
      "step": 14885
    },
    {
      "epoch": 1.33,
      "grad_norm": 7.464504386149451,
      "learning_rate": 2.679509054857783e-06,
      "loss": 0.6118,
      "step": 14886
    },
    {
      "epoch": 1.33,
      "grad_norm": 7.903214291785372,
      "learning_rate": 2.678869214337268e-06,
      "loss": 0.6874,
      "step": 14887
    },
    {
      "epoch": 1.33,
      "grad_norm": 7.5478334159138765,
      "learning_rate": 2.678229422265273e-06,
      "loss": 0.5732,
      "step": 14888
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.22720678580852,
      "learning_rate": 2.6775896786551546e-06,
      "loss": 0.615,
      "step": 14889
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.182824572484987,
      "learning_rate": 2.676949983520266e-06,
      "loss": 0.5197,
      "step": 14890
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.83247557256651,
      "learning_rate": 2.6763103368739583e-06,
      "loss": 0.5857,
      "step": 14891
    },
    {
      "epoch": 1.33,
      "grad_norm": 7.135792417422544,
      "learning_rate": 2.6756707387295842e-06,
      "loss": 0.5782,
      "step": 14892
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.384657584487287,
      "learning_rate": 2.6750311891004947e-06,
      "loss": 0.5711,
      "step": 14893
    },
    {
      "epoch": 1.33,
      "grad_norm": 4.601097217274062,
      "learning_rate": 2.674391688000036e-06,
      "loss": 0.6077,
      "step": 14894
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.752391221667987,
      "learning_rate": 2.673752235441558e-06,
      "loss": 0.5985,
      "step": 14895
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.206963686023177,
      "learning_rate": 2.673112831438407e-06,
      "loss": 0.5618,
      "step": 14896
    },
    {
      "epoch": 1.33,
      "grad_norm": 7.066421829978814,
      "learning_rate": 2.67247347600393e-06,
      "loss": 0.5947,
      "step": 14897
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.392067817202814,
      "learning_rate": 2.671834169151472e-06,
      "loss": 0.6044,
      "step": 14898
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.872820903587027,
      "learning_rate": 2.671194910894379e-06,
      "loss": 0.5685,
      "step": 14899
    },
    {
      "epoch": 1.33,
      "grad_norm": 7.024416601915947,
      "learning_rate": 2.670555701245993e-06,
      "loss": 0.5723,
      "step": 14900
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.319269840485124,
      "learning_rate": 2.6699165402196525e-06,
      "loss": 0.5984,
      "step": 14901
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.078437629483275,
      "learning_rate": 2.669277427828702e-06,
      "loss": 0.6219,
      "step": 14902
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.49654835823428,
      "learning_rate": 2.668638364086481e-06,
      "loss": 0.5445,
      "step": 14903
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.21907059086118,
      "learning_rate": 2.6679993490063284e-06,
      "loss": 0.5771,
      "step": 14904
    },
    {
      "epoch": 1.33,
      "grad_norm": 7.5128751133030125,
      "learning_rate": 2.667360382601582e-06,
      "loss": 0.5775,
      "step": 14905
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.782019505766232,
      "learning_rate": 2.6667214648855816e-06,
      "loss": 0.5566,
      "step": 14906
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.599662685924294,
      "learning_rate": 2.6660825958716586e-06,
      "loss": 0.607,
      "step": 14907
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.616690368006348,
      "learning_rate": 2.6654437755731504e-06,
      "loss": 0.5393,
      "step": 14908
    },
    {
      "epoch": 1.33,
      "grad_norm": 9.021852621819734,
      "learning_rate": 2.6648050040033913e-06,
      "loss": 0.5972,
      "step": 14909
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.327744582156377,
      "learning_rate": 2.664166281175713e-06,
      "loss": 0.6173,
      "step": 14910
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.250683455723718,
      "learning_rate": 2.6635276071034484e-06,
      "loss": 0.5832,
      "step": 14911
    },
    {
      "epoch": 1.33,
      "grad_norm": 7.502477834425655,
      "learning_rate": 2.6628889817999304e-06,
      "loss": 0.5535,
      "step": 14912
    },
    {
      "epoch": 1.33,
      "grad_norm": 9.989704569156656,
      "learning_rate": 2.6622504052784857e-06,
      "loss": 0.6,
      "step": 14913
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.819990273621143,
      "learning_rate": 2.6616118775524436e-06,
      "loss": 0.5888,
      "step": 14914
    },
    {
      "epoch": 1.33,
      "grad_norm": 7.5425294998105405,
      "learning_rate": 2.6609733986351314e-06,
      "loss": 0.7046,
      "step": 14915
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.195607551872185,
      "learning_rate": 2.6603349685398784e-06,
      "loss": 0.6657,
      "step": 14916
    },
    {
      "epoch": 1.33,
      "grad_norm": 7.358906523565907,
      "learning_rate": 2.6596965872800085e-06,
      "loss": 0.573,
      "step": 14917
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.379465725423374,
      "learning_rate": 2.659058254868847e-06,
      "loss": 0.5539,
      "step": 14918
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.663860723902731,
      "learning_rate": 2.65841997131972e-06,
      "loss": 0.5899,
      "step": 14919
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.743890561375794,
      "learning_rate": 2.6577817366459465e-06,
      "loss": 0.5694,
      "step": 14920
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.870811013082939,
      "learning_rate": 2.657143550860849e-06,
      "loss": 0.6272,
      "step": 14921
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.210366156964024,
      "learning_rate": 2.6565054139777504e-06,
      "loss": 0.5241,
      "step": 14922
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.701529483097314,
      "learning_rate": 2.6558673260099676e-06,
      "loss": 0.5233,
      "step": 14923
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.161788112765237,
      "learning_rate": 2.655229286970822e-06,
      "loss": 0.5792,
      "step": 14924
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.408423940751369,
      "learning_rate": 2.6545912968736316e-06,
      "loss": 0.5879,
      "step": 14925
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.200491138869476,
      "learning_rate": 2.65395335573171e-06,
      "loss": 0.5851,
      "step": 14926
    },
    {
      "epoch": 1.33,
      "grad_norm": 4.657347574117157,
      "learning_rate": 2.653315463558375e-06,
      "loss": 0.6177,
      "step": 14927
    },
    {
      "epoch": 1.33,
      "grad_norm": 7.708975294143208,
      "learning_rate": 2.6526776203669403e-06,
      "loss": 0.6223,
      "step": 14928
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.588889212974012,
      "learning_rate": 2.6520398261707182e-06,
      "loss": 0.6044,
      "step": 14929
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.259485635524976,
      "learning_rate": 2.6514020809830237e-06,
      "loss": 0.5516,
      "step": 14930
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.461406552469308,
      "learning_rate": 2.650764384817167e-06,
      "loss": 0.6083,
      "step": 14931
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.786909588439307,
      "learning_rate": 2.6501267376864615e-06,
      "loss": 0.6741,
      "step": 14932
    },
    {
      "epoch": 1.33,
      "grad_norm": 7.906024919806796,
      "learning_rate": 2.6494891396042118e-06,
      "loss": 0.6542,
      "step": 14933
    },
    {
      "epoch": 1.33,
      "grad_norm": 7.694823758770364,
      "learning_rate": 2.6488515905837286e-06,
      "loss": 0.61,
      "step": 14934
    },
    {
      "epoch": 1.33,
      "grad_norm": 7.302534323676685,
      "learning_rate": 2.64821409063832e-06,
      "loss": 0.6435,
      "step": 14935
    },
    {
      "epoch": 1.33,
      "grad_norm": 7.240375789465482,
      "learning_rate": 2.6475766397812917e-06,
      "loss": 0.5932,
      "step": 14936
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.171303075468167,
      "learning_rate": 2.6469392380259495e-06,
      "loss": 0.5341,
      "step": 14937
    },
    {
      "epoch": 1.33,
      "grad_norm": 9.519609021624706,
      "learning_rate": 2.6463018853855994e-06,
      "loss": 0.5879,
      "step": 14938
    },
    {
      "epoch": 1.33,
      "grad_norm": 4.678550872665379,
      "learning_rate": 2.645664581873541e-06,
      "loss": 0.5613,
      "step": 14939
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.543382001499024,
      "learning_rate": 2.6450273275030785e-06,
      "loss": 0.5942,
      "step": 14940
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.947793753052208,
      "learning_rate": 2.6443901222875135e-06,
      "loss": 0.5502,
      "step": 14941
    },
    {
      "epoch": 1.33,
      "grad_norm": 7.033229048818087,
      "learning_rate": 2.6437529662401472e-06,
      "loss": 0.6015,
      "step": 14942
    },
    {
      "epoch": 1.33,
      "grad_norm": 8.660840386490642,
      "learning_rate": 2.6431158593742765e-06,
      "loss": 0.6714,
      "step": 14943
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.268532053010118,
      "learning_rate": 2.642478801703201e-06,
      "loss": 0.5486,
      "step": 14944
    },
    {
      "epoch": 1.33,
      "grad_norm": 7.443469512268695,
      "learning_rate": 2.6418417932402187e-06,
      "loss": 0.6488,
      "step": 14945
    },
    {
      "epoch": 1.33,
      "grad_norm": 7.021423210088533,
      "learning_rate": 2.641204833998623e-06,
      "loss": 0.5372,
      "step": 14946
    },
    {
      "epoch": 1.33,
      "grad_norm": 4.894680177919018,
      "learning_rate": 2.640567923991711e-06,
      "loss": 0.5992,
      "step": 14947
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.9262423377777615,
      "learning_rate": 2.639931063232776e-06,
      "loss": 0.621,
      "step": 14948
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.795859631924286,
      "learning_rate": 2.6392942517351122e-06,
      "loss": 0.561,
      "step": 14949
    },
    {
      "epoch": 1.33,
      "grad_norm": 7.030887026689776,
      "learning_rate": 2.6386574895120125e-06,
      "loss": 0.629,
      "step": 14950
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.631985703200043,
      "learning_rate": 2.638020776576765e-06,
      "loss": 0.6301,
      "step": 14951
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.084298465346028,
      "learning_rate": 2.6373841129426608e-06,
      "loss": 0.5258,
      "step": 14952
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.811468073730915,
      "learning_rate": 2.636747498622989e-06,
      "loss": 0.5507,
      "step": 14953
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.738487668097852,
      "learning_rate": 2.636110933631038e-06,
      "loss": 0.5966,
      "step": 14954
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.453849509251942,
      "learning_rate": 2.6354744179800945e-06,
      "loss": 0.5577,
      "step": 14955
    },
    {
      "epoch": 1.33,
      "grad_norm": 8.652524213228832,
      "learning_rate": 2.6348379516834466e-06,
      "loss": 0.6559,
      "step": 14956
    },
    {
      "epoch": 1.33,
      "grad_norm": 7.5460351671449235,
      "learning_rate": 2.6342015347543743e-06,
      "loss": 0.6332,
      "step": 14957
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.22873556405967,
      "learning_rate": 2.6335651672061657e-06,
      "loss": 0.596,
      "step": 14958
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.527449077272407,
      "learning_rate": 2.6329288490521003e-06,
      "loss": 0.5649,
      "step": 14959
    },
    {
      "epoch": 1.33,
      "grad_norm": 8.487027856636644,
      "learning_rate": 2.6322925803054616e-06,
      "loss": 0.6292,
      "step": 14960
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.953362339302169,
      "learning_rate": 2.63165636097953e-06,
      "loss": 0.629,
      "step": 14961
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.163992345226609,
      "learning_rate": 2.631020191087585e-06,
      "loss": 0.5731,
      "step": 14962
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.313934697373516,
      "learning_rate": 2.630384070642907e-06,
      "loss": 0.5947,
      "step": 14963
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.638426384540426,
      "learning_rate": 2.6297479996587716e-06,
      "loss": 0.5695,
      "step": 14964
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.8576512355372525,
      "learning_rate": 2.629111978148455e-06,
      "loss": 0.5567,
      "step": 14965
    },
    {
      "epoch": 1.34,
      "grad_norm": 7.475858904515392,
      "learning_rate": 2.6284760061252333e-06,
      "loss": 0.5664,
      "step": 14966
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.293796164823158,
      "learning_rate": 2.627840083602382e-06,
      "loss": 0.6067,
      "step": 14967
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.9333877721911135,
      "learning_rate": 2.627204210593174e-06,
      "loss": 0.6416,
      "step": 14968
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.565001332473048,
      "learning_rate": 2.626568387110884e-06,
      "loss": 0.6011,
      "step": 14969
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.2289431214038915,
      "learning_rate": 2.625932613168779e-06,
      "loss": 0.564,
      "step": 14970
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.187534975946211,
      "learning_rate": 2.6252968887801332e-06,
      "loss": 0.6817,
      "step": 14971
    },
    {
      "epoch": 1.34,
      "grad_norm": 4.688525302259582,
      "learning_rate": 2.6246612139582126e-06,
      "loss": 0.587,
      "step": 14972
    },
    {
      "epoch": 1.34,
      "grad_norm": 7.303293878161056,
      "learning_rate": 2.6240255887162873e-06,
      "loss": 0.6176,
      "step": 14973
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.122272070299025,
      "learning_rate": 2.623390013067625e-06,
      "loss": 0.5874,
      "step": 14974
    },
    {
      "epoch": 1.34,
      "grad_norm": 8.118440325037517,
      "learning_rate": 2.6227544870254905e-06,
      "loss": 0.617,
      "step": 14975
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.71014516516398,
      "learning_rate": 2.6221190106031522e-06,
      "loss": 0.6148,
      "step": 14976
    },
    {
      "epoch": 1.34,
      "grad_norm": 8.773967283903481,
      "learning_rate": 2.621483583813871e-06,
      "loss": 0.5929,
      "step": 14977
    },
    {
      "epoch": 1.34,
      "grad_norm": 8.589442386950346,
      "learning_rate": 2.6208482066709108e-06,
      "loss": 0.5782,
      "step": 14978
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.926821170266975,
      "learning_rate": 2.6202128791875337e-06,
      "loss": 0.6018,
      "step": 14979
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.577581290948911,
      "learning_rate": 2.6195776013770015e-06,
      "loss": 0.5294,
      "step": 14980
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.066314371728381,
      "learning_rate": 2.6189423732525733e-06,
      "loss": 0.5748,
      "step": 14981
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.115904680420456,
      "learning_rate": 2.6183071948275106e-06,
      "loss": 0.5945,
      "step": 14982
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.453892143219966,
      "learning_rate": 2.6176720661150678e-06,
      "loss": 0.5546,
      "step": 14983
    },
    {
      "epoch": 1.34,
      "grad_norm": 7.010810145640086,
      "learning_rate": 2.6170369871285026e-06,
      "loss": 0.5244,
      "step": 14984
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.6020803644538315,
      "learning_rate": 2.616401957881075e-06,
      "loss": 0.5507,
      "step": 14985
    },
    {
      "epoch": 1.34,
      "grad_norm": 32.35315049942263,
      "learning_rate": 2.615766978386034e-06,
      "loss": 0.5614,
      "step": 14986
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.637888267645923,
      "learning_rate": 2.6151320486566356e-06,
      "loss": 0.5861,
      "step": 14987
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.846547198703827,
      "learning_rate": 2.614497168706133e-06,
      "loss": 0.5963,
      "step": 14988
    },
    {
      "epoch": 1.34,
      "grad_norm": 7.5464066484805805,
      "learning_rate": 2.61386233854778e-06,
      "loss": 0.6291,
      "step": 14989
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.851151587534548,
      "learning_rate": 2.613227558194823e-06,
      "loss": 0.6256,
      "step": 14990
    },
    {
      "epoch": 1.34,
      "grad_norm": 9.090168307076514,
      "learning_rate": 2.612592827660514e-06,
      "loss": 0.5906,
      "step": 14991
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.86564363067613,
      "learning_rate": 2.611958146958101e-06,
      "loss": 0.5852,
      "step": 14992
    },
    {
      "epoch": 1.34,
      "grad_norm": 7.584974338718077,
      "learning_rate": 2.611323516100832e-06,
      "loss": 0.5846,
      "step": 14993
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.3510483812407506,
      "learning_rate": 2.6106889351019536e-06,
      "loss": 0.5755,
      "step": 14994
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.880070226661775,
      "learning_rate": 2.610054403974714e-06,
      "loss": 0.5985,
      "step": 14995
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.58917499912922,
      "learning_rate": 2.609419922732352e-06,
      "loss": 0.6258,
      "step": 14996
    },
    {
      "epoch": 1.34,
      "grad_norm": 7.733671538307503,
      "learning_rate": 2.6087854913881138e-06,
      "loss": 0.6137,
      "step": 14997
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.503375419638901,
      "learning_rate": 2.608151109955242e-06,
      "loss": 0.5951,
      "step": 14998
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.43617296770431,
      "learning_rate": 2.607516778446978e-06,
      "loss": 0.5956,
      "step": 14999
    },
    {
      "epoch": 1.34,
      "grad_norm": 10.933280407214431,
      "learning_rate": 2.606882496876563e-06,
      "loss": 0.6426,
      "step": 15000
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.678015805410801,
      "learning_rate": 2.606248265257234e-06,
      "loss": 0.5677,
      "step": 15001
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.8201231030846685,
      "learning_rate": 2.6056140836022316e-06,
      "loss": 0.6013,
      "step": 15002
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.199073837076986,
      "learning_rate": 2.60497995192479e-06,
      "loss": 0.5819,
      "step": 15003
    },
    {
      "epoch": 1.34,
      "grad_norm": 8.599937861921019,
      "learning_rate": 2.6043458702381462e-06,
      "loss": 0.5929,
      "step": 15004
    },
    {
      "epoch": 1.34,
      "grad_norm": 8.510718689661477,
      "learning_rate": 2.6037118385555364e-06,
      "loss": 0.6163,
      "step": 15005
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.135706669147801,
      "learning_rate": 2.603077856890194e-06,
      "loss": 0.5388,
      "step": 15006
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.626182083814477,
      "learning_rate": 2.6024439252553523e-06,
      "loss": 0.5041,
      "step": 15007
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.522127726227082,
      "learning_rate": 2.6018100436642457e-06,
      "loss": 0.6189,
      "step": 15008
    },
    {
      "epoch": 1.34,
      "grad_norm": 7.822150041899604,
      "learning_rate": 2.6011762121301e-06,
      "loss": 0.5773,
      "step": 15009
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.791055078446454,
      "learning_rate": 2.6005424306661477e-06,
      "loss": 0.6113,
      "step": 15010
    },
    {
      "epoch": 1.34,
      "grad_norm": 7.969259672791818,
      "learning_rate": 2.599908699285617e-06,
      "loss": 0.6555,
      "step": 15011
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.471539674647872,
      "learning_rate": 2.599275018001736e-06,
      "loss": 0.5977,
      "step": 15012
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.271652143690832,
      "learning_rate": 2.5986413868277316e-06,
      "loss": 0.5622,
      "step": 15013
    },
    {
      "epoch": 1.34,
      "grad_norm": 4.630729607107719,
      "learning_rate": 2.598007805776832e-06,
      "loss": 0.5439,
      "step": 15014
    },
    {
      "epoch": 1.34,
      "grad_norm": 7.911740575164361,
      "learning_rate": 2.5973742748622588e-06,
      "loss": 0.5849,
      "step": 15015
    },
    {
      "epoch": 1.34,
      "grad_norm": 7.213653158262906,
      "learning_rate": 2.5967407940972335e-06,
      "loss": 0.6521,
      "step": 15016
    },
    {
      "epoch": 1.34,
      "grad_norm": 9.219568881890458,
      "learning_rate": 2.5961073634949816e-06,
      "loss": 0.6145,
      "step": 15017
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.647279436926166,
      "learning_rate": 2.5954739830687237e-06,
      "loss": 0.5669,
      "step": 15018
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.276974969354231,
      "learning_rate": 2.594840652831681e-06,
      "loss": 0.5998,
      "step": 15019
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.28815274887912,
      "learning_rate": 2.5942073727970725e-06,
      "loss": 0.5527,
      "step": 15020
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.973062460926717,
      "learning_rate": 2.593574142978118e-06,
      "loss": 0.5813,
      "step": 15021
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.01861791763774,
      "learning_rate": 2.592940963388032e-06,
      "loss": 0.5817,
      "step": 15022
    },
    {
      "epoch": 1.34,
      "grad_norm": 7.6031039078446865,
      "learning_rate": 2.5923078340400316e-06,
      "loss": 0.623,
      "step": 15023
    },
    {
      "epoch": 1.34,
      "grad_norm": 7.862935133592644,
      "learning_rate": 2.591674754947332e-06,
      "loss": 0.5438,
      "step": 15024
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.470287336260344,
      "learning_rate": 2.591041726123148e-06,
      "loss": 0.5638,
      "step": 15025
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.136090904155844,
      "learning_rate": 2.5904087475806947e-06,
      "loss": 0.5561,
      "step": 15026
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.712659601776069,
      "learning_rate": 2.5897758193331802e-06,
      "loss": 0.5315,
      "step": 15027
    },
    {
      "epoch": 1.34,
      "grad_norm": 7.470438683292529,
      "learning_rate": 2.5891429413938184e-06,
      "loss": 0.5974,
      "step": 15028
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.411983130879211,
      "learning_rate": 2.588510113775816e-06,
      "loss": 0.6085,
      "step": 15029
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.471130299673768,
      "learning_rate": 2.5878773364923847e-06,
      "loss": 0.5814,
      "step": 15030
    },
    {
      "epoch": 1.34,
      "grad_norm": 7.759775125645258,
      "learning_rate": 2.587244609556731e-06,
      "loss": 0.5847,
      "step": 15031
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.08368902753578,
      "learning_rate": 2.5866119329820627e-06,
      "loss": 0.5834,
      "step": 15032
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.707579446550206,
      "learning_rate": 2.585979306781584e-06,
      "loss": 0.5824,
      "step": 15033
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.434191994820986,
      "learning_rate": 2.5853467309685038e-06,
      "loss": 0.5691,
      "step": 15034
    },
    {
      "epoch": 1.34,
      "grad_norm": 7.445112291513857,
      "learning_rate": 2.5847142055560206e-06,
      "loss": 0.6737,
      "step": 15035
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.7538354251162,
      "learning_rate": 2.584081730557338e-06,
      "loss": 0.6161,
      "step": 15036
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.7107724638505,
      "learning_rate": 2.583449305985659e-06,
      "loss": 0.6046,
      "step": 15037
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.011769211858943,
      "learning_rate": 2.582816931854184e-06,
      "loss": 0.637,
      "step": 15038
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.7605259103256765,
      "learning_rate": 2.582184608176113e-06,
      "loss": 0.5333,
      "step": 15039
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.645792384446202,
      "learning_rate": 2.5815523349646423e-06,
      "loss": 0.6127,
      "step": 15040
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.623099241794625,
      "learning_rate": 2.5809201122329695e-06,
      "loss": 0.6204,
      "step": 15041
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.6186152664203375,
      "learning_rate": 2.5802879399942927e-06,
      "loss": 0.6105,
      "step": 15042
    },
    {
      "epoch": 1.34,
      "grad_norm": 7.271568505140595,
      "learning_rate": 2.579655818261807e-06,
      "loss": 0.5445,
      "step": 15043
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.394076599609442,
      "learning_rate": 2.5790237470487045e-06,
      "loss": 0.5355,
      "step": 15044
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.882987094403889,
      "learning_rate": 2.578391726368179e-06,
      "loss": 0.5768,
      "step": 15045
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.551187680016304,
      "learning_rate": 2.577759756233425e-06,
      "loss": 0.5514,
      "step": 15046
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.39151686690782,
      "learning_rate": 2.5771278366576296e-06,
      "loss": 0.5616,
      "step": 15047
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.519002045097481,
      "learning_rate": 2.5764959676539845e-06,
      "loss": 0.5801,
      "step": 15048
    },
    {
      "epoch": 1.34,
      "grad_norm": 8.648177831576234,
      "learning_rate": 2.5758641492356785e-06,
      "loss": 0.652,
      "step": 15049
    },
    {
      "epoch": 1.34,
      "grad_norm": 8.580263468383123,
      "learning_rate": 2.5752323814159003e-06,
      "loss": 0.6266,
      "step": 15050
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.2734986679626905,
      "learning_rate": 2.5746006642078354e-06,
      "loss": 0.5817,
      "step": 15051
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.37196163340586,
      "learning_rate": 2.573968997624672e-06,
      "loss": 0.5801,
      "step": 15052
    },
    {
      "epoch": 1.34,
      "grad_norm": 9.473849009977808,
      "learning_rate": 2.5733373816795914e-06,
      "loss": 0.6621,
      "step": 15053
    },
    {
      "epoch": 1.34,
      "grad_norm": 7.221721391068999,
      "learning_rate": 2.572705816385778e-06,
      "loss": 0.6416,
      "step": 15054
    },
    {
      "epoch": 1.34,
      "grad_norm": 7.991568160015231,
      "learning_rate": 2.5720743017564154e-06,
      "loss": 0.6245,
      "step": 15055
    },
    {
      "epoch": 1.34,
      "grad_norm": 8.121957352219544,
      "learning_rate": 2.571442837804685e-06,
      "loss": 0.5569,
      "step": 15056
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.876673064140517,
      "learning_rate": 2.570811424543769e-06,
      "loss": 0.5641,
      "step": 15057
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.143619272000288,
      "learning_rate": 2.5701800619868416e-06,
      "loss": 0.6047,
      "step": 15058
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.466920829384702,
      "learning_rate": 2.5695487501470866e-06,
      "loss": 0.5494,
      "step": 15059
    },
    {
      "epoch": 1.34,
      "grad_norm": 10.73799244132306,
      "learning_rate": 2.568917489037677e-06,
      "loss": 0.6117,
      "step": 15060
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.209925804057708,
      "learning_rate": 2.568286278671791e-06,
      "loss": 0.5323,
      "step": 15061
    },
    {
      "epoch": 1.34,
      "grad_norm": 7.318872821892992,
      "learning_rate": 2.5676551190626037e-06,
      "loss": 0.5978,
      "step": 15062
    },
    {
      "epoch": 1.34,
      "grad_norm": 7.421086855477212,
      "learning_rate": 2.567024010223289e-06,
      "loss": 0.5872,
      "step": 15063
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.5260175789479575,
      "learning_rate": 2.5663929521670196e-06,
      "loss": 0.5416,
      "step": 15064
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.534256245182314,
      "learning_rate": 2.56576194490697e-06,
      "loss": 0.5644,
      "step": 15065
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.950004489488865,
      "learning_rate": 2.565130988456307e-06,
      "loss": 0.6144,
      "step": 15066
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.323918232200568,
      "learning_rate": 2.564500082828202e-06,
      "loss": 0.6149,
      "step": 15067
    },
    {
      "epoch": 1.34,
      "grad_norm": 12.454712290364682,
      "learning_rate": 2.5638692280358247e-06,
      "loss": 0.6069,
      "step": 15068
    },
    {
      "epoch": 1.34,
      "grad_norm": 4.8890141665384625,
      "learning_rate": 2.5632384240923413e-06,
      "loss": 0.6155,
      "step": 15069
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.423825434499414,
      "learning_rate": 2.5626076710109206e-06,
      "loss": 0.5684,
      "step": 15070
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.555141525436559,
      "learning_rate": 2.561976968804728e-06,
      "loss": 0.5258,
      "step": 15071
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.557472569521265,
      "learning_rate": 2.5613463174869268e-06,
      "loss": 0.572,
      "step": 15072
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.942653198855258,
      "learning_rate": 2.56071571707068e-06,
      "loss": 0.5883,
      "step": 15073
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.066228497680753,
      "learning_rate": 2.5600851675691497e-06,
      "loss": 0.5295,
      "step": 15074
    },
    {
      "epoch": 1.34,
      "grad_norm": 7.257482539952881,
      "learning_rate": 2.559454668995498e-06,
      "loss": 0.6305,
      "step": 15075
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.963618885494495,
      "learning_rate": 2.558824221362886e-06,
      "loss": 0.5803,
      "step": 15076
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.867750515126046,
      "learning_rate": 2.5581938246844716e-06,
      "loss": 0.6215,
      "step": 15077
    },
    {
      "epoch": 1.35,
      "grad_norm": 7.707095374082219,
      "learning_rate": 2.557563478973416e-06,
      "loss": 0.6318,
      "step": 15078
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.258612707104679,
      "learning_rate": 2.5569331842428724e-06,
      "loss": 0.51,
      "step": 15079
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.975260764979795,
      "learning_rate": 2.556302940505998e-06,
      "loss": 0.6015,
      "step": 15080
    },
    {
      "epoch": 1.35,
      "grad_norm": 4.602050949471063,
      "learning_rate": 2.555672747775948e-06,
      "loss": 0.6496,
      "step": 15081
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.201855613614339,
      "learning_rate": 2.5550426060658766e-06,
      "loss": 0.6006,
      "step": 15082
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.03237458022645,
      "learning_rate": 2.5544125153889366e-06,
      "loss": 0.5478,
      "step": 15083
    },
    {
      "epoch": 1.35,
      "grad_norm": 7.4307416463136,
      "learning_rate": 2.553782475758281e-06,
      "loss": 0.5596,
      "step": 15084
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.045971506143133,
      "learning_rate": 2.553152487187057e-06,
      "loss": 0.5829,
      "step": 15085
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.3383291558589905,
      "learning_rate": 2.5525225496884193e-06,
      "loss": 0.5977,
      "step": 15086
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.667618549117838,
      "learning_rate": 2.551892663275511e-06,
      "loss": 0.5491,
      "step": 15087
    },
    {
      "epoch": 1.35,
      "grad_norm": 7.890037413334313,
      "learning_rate": 2.5512628279614815e-06,
      "loss": 0.5819,
      "step": 15088
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.972036499354017,
      "learning_rate": 2.5506330437594784e-06,
      "loss": 0.6112,
      "step": 15089
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.3076585249205905,
      "learning_rate": 2.5500033106826472e-06,
      "loss": 0.5939,
      "step": 15090
    },
    {
      "epoch": 1.35,
      "grad_norm": 7.362123936571728,
      "learning_rate": 2.549373628744132e-06,
      "loss": 0.6547,
      "step": 15091
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.1706013853349635,
      "learning_rate": 2.5487439979570745e-06,
      "loss": 0.6082,
      "step": 15092
    },
    {
      "epoch": 1.35,
      "grad_norm": 7.792239582298513,
      "learning_rate": 2.548114418334617e-06,
      "loss": 0.5981,
      "step": 15093
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.433942714046304,
      "learning_rate": 2.547484889889903e-06,
      "loss": 0.5994,
      "step": 15094
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.355135321617326,
      "learning_rate": 2.5468554126360702e-06,
      "loss": 0.5672,
      "step": 15095
    },
    {
      "epoch": 1.35,
      "grad_norm": 8.181964188121881,
      "learning_rate": 2.546225986586258e-06,
      "loss": 0.6128,
      "step": 15096
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.855861004094792,
      "learning_rate": 2.545596611753607e-06,
      "loss": 0.5797,
      "step": 15097
    },
    {
      "epoch": 1.35,
      "grad_norm": 8.008918690247997,
      "learning_rate": 2.5449672881512493e-06,
      "loss": 0.544,
      "step": 15098
    },
    {
      "epoch": 1.35,
      "grad_norm": 8.053494888471418,
      "learning_rate": 2.544338015792324e-06,
      "loss": 0.6366,
      "step": 15099
    },
    {
      "epoch": 1.35,
      "grad_norm": 4.346565553391315,
      "learning_rate": 2.5437087946899653e-06,
      "loss": 0.5466,
      "step": 15100
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.399941209298324,
      "learning_rate": 2.5430796248573054e-06,
      "loss": 0.5994,
      "step": 15101
    },
    {
      "epoch": 1.35,
      "grad_norm": 9.62306302035983,
      "learning_rate": 2.5424505063074777e-06,
      "loss": 0.6684,
      "step": 15102
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.536070846245072,
      "learning_rate": 2.5418214390536133e-06,
      "loss": 0.591,
      "step": 15103
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.217676320582467,
      "learning_rate": 2.541192423108846e-06,
      "loss": 0.5675,
      "step": 15104
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.6656228281171215,
      "learning_rate": 2.540563458486299e-06,
      "loss": 0.5889,
      "step": 15105
    },
    {
      "epoch": 1.35,
      "grad_norm": 8.456834594170966,
      "learning_rate": 2.5399345451991043e-06,
      "loss": 0.6139,
      "step": 15106
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.745081997571854,
      "learning_rate": 2.5393056832603885e-06,
      "loss": 0.5957,
      "step": 15107
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.803786018850342,
      "learning_rate": 2.5386768726832777e-06,
      "loss": 0.5781,
      "step": 15108
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.579312862233151,
      "learning_rate": 2.5380481134808966e-06,
      "loss": 0.6159,
      "step": 15109
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.373786471134496,
      "learning_rate": 2.537419405666372e-06,
      "loss": 0.528,
      "step": 15110
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.71752242110261,
      "learning_rate": 2.536790749252822e-06,
      "loss": 0.5588,
      "step": 15111
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.807602277259198,
      "learning_rate": 2.536162144253371e-06,
      "loss": 0.5882,
      "step": 15112
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.380798698086225,
      "learning_rate": 2.5355335906811396e-06,
      "loss": 0.5457,
      "step": 15113
    },
    {
      "epoch": 1.35,
      "grad_norm": 7.117094659304448,
      "learning_rate": 2.534905088549248e-06,
      "loss": 0.5977,
      "step": 15114
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.029045901726208,
      "learning_rate": 2.534276637870815e-06,
      "loss": 0.6189,
      "step": 15115
    },
    {
      "epoch": 1.35,
      "grad_norm": 8.688055846022806,
      "learning_rate": 2.533648238658956e-06,
      "loss": 0.581,
      "step": 15116
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.285153308987967,
      "learning_rate": 2.533019890926791e-06,
      "loss": 0.5975,
      "step": 15117
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.60133590891748,
      "learning_rate": 2.532391594687431e-06,
      "loss": 0.5346,
      "step": 15118
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.091296015613695,
      "learning_rate": 2.5317633499539916e-06,
      "loss": 0.5729,
      "step": 15119
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.542983164513425,
      "learning_rate": 2.531135156739588e-06,
      "loss": 0.6016,
      "step": 15120
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.439800004104856,
      "learning_rate": 2.5305070150573313e-06,
      "loss": 0.566,
      "step": 15121
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.4001016891442255,
      "learning_rate": 2.529878924920334e-06,
      "loss": 0.6408,
      "step": 15122
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.771024439109968,
      "learning_rate": 2.5292508863417033e-06,
      "loss": 0.6063,
      "step": 15123
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.76660501500935,
      "learning_rate": 2.5286228993345483e-06,
      "loss": 0.6127,
      "step": 15124
    },
    {
      "epoch": 1.35,
      "grad_norm": 7.495259273041217,
      "learning_rate": 2.5279949639119794e-06,
      "loss": 0.7046,
      "step": 15125
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.182695056895717,
      "learning_rate": 2.527367080087102e-06,
      "loss": 0.5675,
      "step": 15126
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.5407137440528045,
      "learning_rate": 2.5267392478730212e-06,
      "loss": 0.573,
      "step": 15127
    },
    {
      "epoch": 1.35,
      "grad_norm": 7.7989864279859145,
      "learning_rate": 2.5261114672828442e-06,
      "loss": 0.5471,
      "step": 15128
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.82333478376396,
      "learning_rate": 2.5254837383296727e-06,
      "loss": 0.5472,
      "step": 15129
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.206859513353213,
      "learning_rate": 2.524856061026607e-06,
      "loss": 0.5635,
      "step": 15130
    },
    {
      "epoch": 1.35,
      "grad_norm": 9.13842755804041,
      "learning_rate": 2.524228435386751e-06,
      "loss": 0.5927,
      "step": 15131
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.841479525263,
      "learning_rate": 2.5236008614232043e-06,
      "loss": 0.5755,
      "step": 15132
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.268236230277615,
      "learning_rate": 2.522973339149066e-06,
      "loss": 0.5755,
      "step": 15133
    },
    {
      "epoch": 1.35,
      "grad_norm": 7.838313837026516,
      "learning_rate": 2.522345868577435e-06,
      "loss": 0.5983,
      "step": 15134
    },
    {
      "epoch": 1.35,
      "grad_norm": 4.890199752954614,
      "learning_rate": 2.5217184497214097e-06,
      "loss": 0.5774,
      "step": 15135
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.416211462041913,
      "learning_rate": 2.521091082594083e-06,
      "loss": 0.566,
      "step": 15136
    },
    {
      "epoch": 1.35,
      "grad_norm": 8.711531816201989,
      "learning_rate": 2.5204637672085507e-06,
      "loss": 0.5838,
      "step": 15137
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.036535642758754,
      "learning_rate": 2.519836503577908e-06,
      "loss": 0.5484,
      "step": 15138
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.529757021022951,
      "learning_rate": 2.519209291715246e-06,
      "loss": 0.5635,
      "step": 15139
    },
    {
      "epoch": 1.35,
      "grad_norm": 7.963986755646742,
      "learning_rate": 2.5185821316336577e-06,
      "loss": 0.5796,
      "step": 15140
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.942239394637693,
      "learning_rate": 2.5179550233462348e-06,
      "loss": 0.5846,
      "step": 15141
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.598240777475695,
      "learning_rate": 2.5173279668660633e-06,
      "loss": 0.5711,
      "step": 15142
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.876855292699237,
      "learning_rate": 2.5167009622062333e-06,
      "loss": 0.5743,
      "step": 15143
    },
    {
      "epoch": 1.35,
      "grad_norm": 8.185952094628254,
      "learning_rate": 2.5160740093798354e-06,
      "loss": 0.5828,
      "step": 15144
    },
    {
      "epoch": 1.35,
      "grad_norm": 7.790928437022327,
      "learning_rate": 2.5154471083999498e-06,
      "loss": 0.5672,
      "step": 15145
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.924347638542674,
      "learning_rate": 2.514820259279665e-06,
      "loss": 0.6623,
      "step": 15146
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.609111447317407,
      "learning_rate": 2.514193462032065e-06,
      "loss": 0.6353,
      "step": 15147
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.353823645698439,
      "learning_rate": 2.5135667166702342e-06,
      "loss": 0.6047,
      "step": 15148
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.6942909410068365,
      "learning_rate": 2.5129400232072513e-06,
      "loss": 0.6168,
      "step": 15149
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.6659350574055125,
      "learning_rate": 2.512313381656199e-06,
      "loss": 0.584,
      "step": 15150
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.815683885935327,
      "learning_rate": 2.511686792030157e-06,
      "loss": 0.6168,
      "step": 15151
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.494416073640225,
      "learning_rate": 2.5110602543422034e-06,
      "loss": 0.6166,
      "step": 15152
    },
    {
      "epoch": 1.35,
      "grad_norm": 7.181288755098564,
      "learning_rate": 2.510433768605417e-06,
      "loss": 0.5479,
      "step": 15153
    },
    {
      "epoch": 1.35,
      "grad_norm": 7.427296642082458,
      "learning_rate": 2.509807334832875e-06,
      "loss": 0.6153,
      "step": 15154
    },
    {
      "epoch": 1.35,
      "grad_norm": 11.037165942144263,
      "learning_rate": 2.5091809530376497e-06,
      "loss": 0.6499,
      "step": 15155
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.4383086707381585,
      "learning_rate": 2.5085546232328174e-06,
      "loss": 0.5703,
      "step": 15156
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.984884620471906,
      "learning_rate": 2.5079283454314506e-06,
      "loss": 0.6478,
      "step": 15157
    },
    {
      "epoch": 1.35,
      "grad_norm": 10.699711657513838,
      "learning_rate": 2.507302119646625e-06,
      "loss": 0.5949,
      "step": 15158
    },
    {
      "epoch": 1.35,
      "grad_norm": 7.059173498608641,
      "learning_rate": 2.506675945891406e-06,
      "loss": 0.5159,
      "step": 15159
    },
    {
      "epoch": 1.35,
      "grad_norm": 4.887986805152354,
      "learning_rate": 2.506049824178866e-06,
      "loss": 0.5972,
      "step": 15160
    },
    {
      "epoch": 1.35,
      "grad_norm": 7.5221516682917695,
      "learning_rate": 2.5054237545220774e-06,
      "loss": 0.5813,
      "step": 15161
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.8126096708860935,
      "learning_rate": 2.5047977369341016e-06,
      "loss": 0.603,
      "step": 15162
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.639600276297569,
      "learning_rate": 2.5041717714280094e-06,
      "loss": 0.6313,
      "step": 15163
    },
    {
      "epoch": 1.35,
      "grad_norm": 7.528552371191764,
      "learning_rate": 2.503545858016866e-06,
      "loss": 0.5725,
      "step": 15164
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.503253588545533,
      "learning_rate": 2.5029199967137353e-06,
      "loss": 0.5446,
      "step": 15165
    },
    {
      "epoch": 1.35,
      "grad_norm": 7.683887283700775,
      "learning_rate": 2.502294187531681e-06,
      "loss": 0.5603,
      "step": 15166
    },
    {
      "epoch": 1.35,
      "grad_norm": 8.70973911465323,
      "learning_rate": 2.5016684304837674e-06,
      "loss": 0.6029,
      "step": 15167
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.695495681459116,
      "learning_rate": 2.5010427255830527e-06,
      "loss": 0.557,
      "step": 15168
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.874592663506318,
      "learning_rate": 2.5004170728425982e-06,
      "loss": 0.5972,
      "step": 15169
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.083628078625756,
      "learning_rate": 2.4997914722754633e-06,
      "loss": 0.5943,
      "step": 15170
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.920930501418725,
      "learning_rate": 2.499165923894706e-06,
      "loss": 0.5646,
      "step": 15171
    },
    {
      "epoch": 1.35,
      "grad_norm": 8.716693435086077,
      "learning_rate": 2.4985404277133835e-06,
      "loss": 0.6141,
      "step": 15172
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.129511780228218,
      "learning_rate": 2.497914983744553e-06,
      "loss": 0.5668,
      "step": 15173
    },
    {
      "epoch": 1.35,
      "grad_norm": 7.041043788159,
      "learning_rate": 2.497289592001268e-06,
      "loss": 0.6588,
      "step": 15174
    },
    {
      "epoch": 1.35,
      "grad_norm": 8.621278151386406,
      "learning_rate": 2.49666425249658e-06,
      "loss": 0.6242,
      "step": 15175
    },
    {
      "epoch": 1.35,
      "grad_norm": 8.452270963741837,
      "learning_rate": 2.496038965243544e-06,
      "loss": 0.6267,
      "step": 15176
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.896095680912172,
      "learning_rate": 2.4954137302552105e-06,
      "loss": 0.5302,
      "step": 15177
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.102096453575739,
      "learning_rate": 2.49478854754463e-06,
      "loss": 0.5762,
      "step": 15178
    },
    {
      "epoch": 1.35,
      "grad_norm": 4.7743832690306975,
      "learning_rate": 2.494163417124853e-06,
      "loss": 0.5853,
      "step": 15179
    },
    {
      "epoch": 1.35,
      "grad_norm": 7.804274324619159,
      "learning_rate": 2.4935383390089283e-06,
      "loss": 0.5973,
      "step": 15180
    },
    {
      "epoch": 1.35,
      "grad_norm": 4.9791009615456545,
      "learning_rate": 2.4929133132099005e-06,
      "loss": 0.5577,
      "step": 15181
    },
    {
      "epoch": 1.35,
      "grad_norm": 7.18995043662529,
      "learning_rate": 2.4922883397408167e-06,
      "loss": 0.6005,
      "step": 15182
    },
    {
      "epoch": 1.35,
      "grad_norm": 7.960896437134166,
      "learning_rate": 2.4916634186147213e-06,
      "loss": 0.54,
      "step": 15183
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.203238766005372,
      "learning_rate": 2.49103854984466e-06,
      "loss": 0.5711,
      "step": 15184
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.089274484808255,
      "learning_rate": 2.4904137334436734e-06,
      "loss": 0.5719,
      "step": 15185
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.966501377946458,
      "learning_rate": 2.489788969424807e-06,
      "loss": 0.5899,
      "step": 15186
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.06810843167128,
      "learning_rate": 2.4891642578010976e-06,
      "loss": 0.5741,
      "step": 15187
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.014840194527695,
      "learning_rate": 2.4885395985855846e-06,
      "loss": 0.6053,
      "step": 15188
    },
    {
      "epoch": 1.36,
      "grad_norm": 9.091533799435869,
      "learning_rate": 2.487914991791307e-06,
      "loss": 0.6676,
      "step": 15189
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.480514013306307,
      "learning_rate": 2.4872904374313026e-06,
      "loss": 0.5958,
      "step": 15190
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.832313768401261,
      "learning_rate": 2.4866659355186087e-06,
      "loss": 0.5966,
      "step": 15191
    },
    {
      "epoch": 1.36,
      "grad_norm": 8.43902600327396,
      "learning_rate": 2.486041486066258e-06,
      "loss": 0.5907,
      "step": 15192
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.218295270294223,
      "learning_rate": 2.485417089087288e-06,
      "loss": 0.6536,
      "step": 15193
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.725686499224136,
      "learning_rate": 2.4847927445947275e-06,
      "loss": 0.5745,
      "step": 15194
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.53792531384599,
      "learning_rate": 2.48416845260161e-06,
      "loss": 0.6466,
      "step": 15195
    },
    {
      "epoch": 1.36,
      "grad_norm": 12.290536011196703,
      "learning_rate": 2.483544213120967e-06,
      "loss": 0.649,
      "step": 15196
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.471442704289469,
      "learning_rate": 2.4829200261658274e-06,
      "loss": 0.5693,
      "step": 15197
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.619410033674413,
      "learning_rate": 2.4822958917492218e-06,
      "loss": 0.6406,
      "step": 15198
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.472663103434341,
      "learning_rate": 2.481671809884174e-06,
      "loss": 0.5819,
      "step": 15199
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.675441312725132,
      "learning_rate": 2.4810477805837117e-06,
      "loss": 0.5836,
      "step": 15200
    },
    {
      "epoch": 1.36,
      "grad_norm": 8.63605281070082,
      "learning_rate": 2.480423803860863e-06,
      "loss": 0.6051,
      "step": 15201
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.8649793669594485,
      "learning_rate": 2.479799879728648e-06,
      "loss": 0.6207,
      "step": 15202
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.779396014500936,
      "learning_rate": 2.4791760082000905e-06,
      "loss": 0.4816,
      "step": 15203
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.560770714312319,
      "learning_rate": 2.4785521892882135e-06,
      "loss": 0.5826,
      "step": 15204
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.328733051245722,
      "learning_rate": 2.47792842300604e-06,
      "loss": 0.5984,
      "step": 15205
    },
    {
      "epoch": 1.36,
      "grad_norm": 9.010489201927035,
      "learning_rate": 2.4773047093665853e-06,
      "loss": 0.6032,
      "step": 15206
    },
    {
      "epoch": 1.36,
      "grad_norm": 9.295950973413936,
      "learning_rate": 2.47668104838287e-06,
      "loss": 0.6276,
      "step": 15207
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.715660040938863,
      "learning_rate": 2.4760574400679122e-06,
      "loss": 0.5292,
      "step": 15208
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.787075531924868,
      "learning_rate": 2.475433884434728e-06,
      "loss": 0.5754,
      "step": 15209
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.64869189676575,
      "learning_rate": 2.4748103814963324e-06,
      "loss": 0.5973,
      "step": 15210
    },
    {
      "epoch": 1.36,
      "grad_norm": 8.073917222566672,
      "learning_rate": 2.4741869312657423e-06,
      "loss": 0.6225,
      "step": 15211
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.275015107285138,
      "learning_rate": 2.4735635337559664e-06,
      "loss": 0.5623,
      "step": 15212
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.930327227803697,
      "learning_rate": 2.4729401889800194e-06,
      "loss": 0.5907,
      "step": 15213
    },
    {
      "epoch": 1.36,
      "grad_norm": 8.343836862037112,
      "learning_rate": 2.472316896950911e-06,
      "loss": 0.5944,
      "step": 15214
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.738602962459229,
      "learning_rate": 2.4716936576816525e-06,
      "loss": 0.5953,
      "step": 15215
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.423956381703024,
      "learning_rate": 2.4710704711852534e-06,
      "loss": 0.6121,
      "step": 15216
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.133202722020561,
      "learning_rate": 2.4704473374747185e-06,
      "loss": 0.6042,
      "step": 15217
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.67852738993027,
      "learning_rate": 2.4698242565630575e-06,
      "loss": 0.6193,
      "step": 15218
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.114323389538373,
      "learning_rate": 2.469201228463272e-06,
      "loss": 0.5675,
      "step": 15219
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.842902571651335,
      "learning_rate": 2.468578253188369e-06,
      "loss": 0.6039,
      "step": 15220
    },
    {
      "epoch": 1.36,
      "grad_norm": 8.52332422652029,
      "learning_rate": 2.467955330751351e-06,
      "loss": 0.6525,
      "step": 15221
    },
    {
      "epoch": 1.36,
      "grad_norm": 8.195641671901301,
      "learning_rate": 2.4673324611652205e-06,
      "loss": 0.6076,
      "step": 15222
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.114136118415405,
      "learning_rate": 2.466709644442978e-06,
      "loss": 0.603,
      "step": 15223
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.153566871628214,
      "learning_rate": 2.466086880597626e-06,
      "loss": 0.5995,
      "step": 15224
    },
    {
      "epoch": 1.36,
      "grad_norm": 10.027772010815477,
      "learning_rate": 2.4654641696421593e-06,
      "loss": 0.6109,
      "step": 15225
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.55950447701592,
      "learning_rate": 2.464841511589578e-06,
      "loss": 0.5821,
      "step": 15226
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.220449669566721,
      "learning_rate": 2.464218906452878e-06,
      "loss": 0.5819,
      "step": 15227
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.6840067506996945,
      "learning_rate": 2.463596354245055e-06,
      "loss": 0.5775,
      "step": 15228
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.3292263928144825,
      "learning_rate": 2.4629738549791025e-06,
      "loss": 0.5985,
      "step": 15229
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.7968545978432795,
      "learning_rate": 2.462351408668018e-06,
      "loss": 0.5179,
      "step": 15230
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.7149575788978435,
      "learning_rate": 2.46172901532479e-06,
      "loss": 0.5668,
      "step": 15231
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.463578986003413,
      "learning_rate": 2.4611066749624085e-06,
      "loss": 0.5288,
      "step": 15232
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.660676913955222,
      "learning_rate": 2.4604843875938643e-06,
      "loss": 0.6047,
      "step": 15233
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.971889363626915,
      "learning_rate": 2.4598621532321478e-06,
      "loss": 0.5953,
      "step": 15234
    },
    {
      "epoch": 1.36,
      "grad_norm": 8.996489692496208,
      "learning_rate": 2.459239971890246e-06,
      "loss": 0.6341,
      "step": 15235
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.391706010322378,
      "learning_rate": 2.458617843581145e-06,
      "loss": 0.5676,
      "step": 15236
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.64904920549135,
      "learning_rate": 2.4579957683178334e-06,
      "loss": 0.6645,
      "step": 15237
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.101607846011128,
      "learning_rate": 2.4573737461132913e-06,
      "loss": 0.6068,
      "step": 15238
    },
    {
      "epoch": 1.36,
      "grad_norm": 8.472974958439975,
      "learning_rate": 2.4567517769805045e-06,
      "loss": 0.5837,
      "step": 15239
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.002727672750484,
      "learning_rate": 2.4561298609324537e-06,
      "loss": 0.5965,
      "step": 15240
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.238351714939946,
      "learning_rate": 2.4555079979821217e-06,
      "loss": 0.5909,
      "step": 15241
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.573192174320548,
      "learning_rate": 2.4548861881424877e-06,
      "loss": 0.5696,
      "step": 15242
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.974907494694809,
      "learning_rate": 2.454264431426533e-06,
      "loss": 0.5528,
      "step": 15243
    },
    {
      "epoch": 1.36,
      "grad_norm": 8.920151130774679,
      "learning_rate": 2.453642727847231e-06,
      "loss": 0.6124,
      "step": 15244
    },
    {
      "epoch": 1.36,
      "grad_norm": 8.282453134534977,
      "learning_rate": 2.453021077417563e-06,
      "loss": 0.6464,
      "step": 15245
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.386621769911644,
      "learning_rate": 2.4523994801505e-06,
      "loss": 0.5899,
      "step": 15246
    },
    {
      "epoch": 1.36,
      "grad_norm": 8.739283991891114,
      "learning_rate": 2.4517779360590183e-06,
      "loss": 0.5903,
      "step": 15247
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.7529296515303425,
      "learning_rate": 2.451156445156092e-06,
      "loss": 0.5722,
      "step": 15248
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.828249325743333,
      "learning_rate": 2.450535007454693e-06,
      "loss": 0.5434,
      "step": 15249
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.566982739083851,
      "learning_rate": 2.4499136229677945e-06,
      "loss": 0.5937,
      "step": 15250
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.191293094276491,
      "learning_rate": 2.4492922917083623e-06,
      "loss": 0.5916,
      "step": 15251
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.803469660215911,
      "learning_rate": 2.448671013689368e-06,
      "loss": 0.5584,
      "step": 15252
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.567467562685244,
      "learning_rate": 2.4480497889237787e-06,
      "loss": 0.6092,
      "step": 15253
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.497264268951367,
      "learning_rate": 2.4474286174245613e-06,
      "loss": 0.5474,
      "step": 15254
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.621883618952956,
      "learning_rate": 2.446807499204682e-06,
      "loss": 0.6175,
      "step": 15255
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.799288479778182,
      "learning_rate": 2.4461864342771065e-06,
      "loss": 0.5911,
      "step": 15256
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.7732591686074,
      "learning_rate": 2.4455654226547942e-06,
      "loss": 0.5354,
      "step": 15257
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.923831892757163,
      "learning_rate": 2.4449444643507098e-06,
      "loss": 0.5753,
      "step": 15258
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.004480048193925,
      "learning_rate": 2.4443235593778157e-06,
      "loss": 0.5734,
      "step": 15259
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.222556416330085,
      "learning_rate": 2.4437027077490695e-06,
      "loss": 0.5286,
      "step": 15260
    },
    {
      "epoch": 1.36,
      "grad_norm": 8.804315118268526,
      "learning_rate": 2.44308190947743e-06,
      "loss": 0.6087,
      "step": 15261
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.468055085722266,
      "learning_rate": 2.4424611645758567e-06,
      "loss": 0.6433,
      "step": 15262
    },
    {
      "epoch": 1.36,
      "grad_norm": 8.145688183376379,
      "learning_rate": 2.4418404730573075e-06,
      "loss": 0.6547,
      "step": 15263
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.728588872775224,
      "learning_rate": 2.4412198349347345e-06,
      "loss": 0.6374,
      "step": 15264
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.085884271225814,
      "learning_rate": 2.4405992502210934e-06,
      "loss": 0.6224,
      "step": 15265
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.957724731769276,
      "learning_rate": 2.439978718929338e-06,
      "loss": 0.5621,
      "step": 15266
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.290612868809058,
      "learning_rate": 2.4393582410724204e-06,
      "loss": 0.5823,
      "step": 15267
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.914417888654169,
      "learning_rate": 2.4387378166632924e-06,
      "loss": 0.6193,
      "step": 15268
    },
    {
      "epoch": 1.36,
      "grad_norm": 8.451548941398585,
      "learning_rate": 2.4381174457149048e-06,
      "loss": 0.5952,
      "step": 15269
    },
    {
      "epoch": 1.36,
      "grad_norm": 8.775926480569979,
      "learning_rate": 2.4374971282402034e-06,
      "loss": 0.569,
      "step": 15270
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.277017450134407,
      "learning_rate": 2.4368768642521374e-06,
      "loss": 0.6224,
      "step": 15271
    },
    {
      "epoch": 1.36,
      "grad_norm": 4.72541649284393,
      "learning_rate": 2.436256653763654e-06,
      "loss": 0.6092,
      "step": 15272
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.848091676641214,
      "learning_rate": 2.435636496787698e-06,
      "loss": 0.6359,
      "step": 15273
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.216685639002915,
      "learning_rate": 2.4350163933372166e-06,
      "loss": 0.5863,
      "step": 15274
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.142185281277388,
      "learning_rate": 2.434396343425149e-06,
      "loss": 0.5911,
      "step": 15275
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.268741500318552,
      "learning_rate": 2.4337763470644403e-06,
      "loss": 0.6007,
      "step": 15276
    },
    {
      "epoch": 1.36,
      "grad_norm": 8.054974349139382,
      "learning_rate": 2.4331564042680293e-06,
      "loss": 0.5895,
      "step": 15277
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.995676114512794,
      "learning_rate": 2.432536515048857e-06,
      "loss": 0.5892,
      "step": 15278
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.597081701361043,
      "learning_rate": 2.431916679419862e-06,
      "loss": 0.5568,
      "step": 15279
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.871830852221906,
      "learning_rate": 2.431296897393982e-06,
      "loss": 0.6071,
      "step": 15280
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.080159345129762,
      "learning_rate": 2.430677168984157e-06,
      "loss": 0.5899,
      "step": 15281
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.662499679559327,
      "learning_rate": 2.4300574942033163e-06,
      "loss": 0.5914,
      "step": 15282
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.760886752179844,
      "learning_rate": 2.4294378730643974e-06,
      "loss": 0.5604,
      "step": 15283
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.806902976894375,
      "learning_rate": 2.428818305580334e-06,
      "loss": 0.5353,
      "step": 15284
    },
    {
      "epoch": 1.36,
      "grad_norm": 8.68591667083678,
      "learning_rate": 2.4281987917640574e-06,
      "loss": 0.5891,
      "step": 15285
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.510382642913864,
      "learning_rate": 2.427579331628499e-06,
      "loss": 0.6546,
      "step": 15286
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.063981989818763,
      "learning_rate": 2.4269599251865904e-06,
      "loss": 0.5908,
      "step": 15287
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.810431593444579,
      "learning_rate": 2.4263405724512566e-06,
      "loss": 0.6079,
      "step": 15288
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.955311218971023,
      "learning_rate": 2.425721273435429e-06,
      "loss": 0.6037,
      "step": 15289
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.917594407147409,
      "learning_rate": 2.4251020281520306e-06,
      "loss": 0.6191,
      "step": 15290
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.875170304448491,
      "learning_rate": 2.424482836613988e-06,
      "loss": 0.5912,
      "step": 15291
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.763043923981057,
      "learning_rate": 2.4238636988342263e-06,
      "loss": 0.5806,
      "step": 15292
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.640736941970092,
      "learning_rate": 2.4232446148256676e-06,
      "loss": 0.5799,
      "step": 15293
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.705269967101258,
      "learning_rate": 2.422625584601237e-06,
      "loss": 0.5952,
      "step": 15294
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.366691802345707,
      "learning_rate": 2.4220066081738515e-06,
      "loss": 0.5836,
      "step": 15295
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.723394442898744,
      "learning_rate": 2.4213876855564315e-06,
      "loss": 0.5633,
      "step": 15296
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.898577385562662,
      "learning_rate": 2.420768816761897e-06,
      "loss": 0.5816,
      "step": 15297
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.7546627884783215,
      "learning_rate": 2.420150001803165e-06,
      "loss": 0.5665,
      "step": 15298
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.468498094482104,
      "learning_rate": 2.4195312406931513e-06,
      "loss": 0.5902,
      "step": 15299
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.135838946098599,
      "learning_rate": 2.4189125334447743e-06,
      "loss": 0.5823,
      "step": 15300
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.935080294162843,
      "learning_rate": 2.4182938800709437e-06,
      "loss": 0.5759,
      "step": 15301
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.922522100626762,
      "learning_rate": 2.4176752805845767e-06,
      "loss": 0.6347,
      "step": 15302
    },
    {
      "epoch": 1.37,
      "grad_norm": 8.98889242134265,
      "learning_rate": 2.4170567349985808e-06,
      "loss": 0.6486,
      "step": 15303
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.18932132608457,
      "learning_rate": 2.4164382433258686e-06,
      "loss": 0.5088,
      "step": 15304
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.593226351502218,
      "learning_rate": 2.4158198055793506e-06,
      "loss": 0.5722,
      "step": 15305
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.739175493310133,
      "learning_rate": 2.415201421771935e-06,
      "loss": 0.5824,
      "step": 15306
    },
    {
      "epoch": 1.37,
      "grad_norm": 4.939610527057863,
      "learning_rate": 2.414583091916531e-06,
      "loss": 0.5902,
      "step": 15307
    },
    {
      "epoch": 1.37,
      "grad_norm": 9.752602447881896,
      "learning_rate": 2.413964816026041e-06,
      "loss": 0.5649,
      "step": 15308
    },
    {
      "epoch": 1.37,
      "grad_norm": 4.605156245825359,
      "learning_rate": 2.413346594113372e-06,
      "loss": 0.6028,
      "step": 15309
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.045975814198473,
      "learning_rate": 2.4127284261914284e-06,
      "loss": 0.5892,
      "step": 15310
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.4949813152881175,
      "learning_rate": 2.4121103122731125e-06,
      "loss": 0.533,
      "step": 15311
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.2894929665333805,
      "learning_rate": 2.411492252371327e-06,
      "loss": 0.5828,
      "step": 15312
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.857416106503852,
      "learning_rate": 2.4108742464989736e-06,
      "loss": 0.605,
      "step": 15313
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.777432805594984,
      "learning_rate": 2.4102562946689486e-06,
      "loss": 0.6133,
      "step": 15314
    },
    {
      "epoch": 1.37,
      "grad_norm": 8.357018710650772,
      "learning_rate": 2.409638396894152e-06,
      "loss": 0.5693,
      "step": 15315
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.6441708361764675,
      "learning_rate": 2.4090205531874806e-06,
      "loss": 0.5922,
      "step": 15316
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.914952731825939,
      "learning_rate": 2.4084027635618328e-06,
      "loss": 0.5463,
      "step": 15317
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.209730033277364,
      "learning_rate": 2.4077850280301e-06,
      "loss": 0.596,
      "step": 15318
    },
    {
      "epoch": 1.37,
      "grad_norm": 7.168982101730439,
      "learning_rate": 2.4071673466051787e-06,
      "loss": 0.623,
      "step": 15319
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.661290221556705,
      "learning_rate": 2.406549719299962e-06,
      "loss": 0.4822,
      "step": 15320
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.1348165361027265,
      "learning_rate": 2.4059321461273388e-06,
      "loss": 0.665,
      "step": 15321
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.694043489547387,
      "learning_rate": 2.405314627100201e-06,
      "loss": 0.5743,
      "step": 15322
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.693953094275328,
      "learning_rate": 2.404697162231438e-06,
      "loss": 0.5383,
      "step": 15323
    },
    {
      "epoch": 1.37,
      "grad_norm": 8.232656787234943,
      "learning_rate": 2.404079751533938e-06,
      "loss": 0.5917,
      "step": 15324
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.64696223722908,
      "learning_rate": 2.403462395020588e-06,
      "loss": 0.5658,
      "step": 15325
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.734303504960375,
      "learning_rate": 2.4028450927042768e-06,
      "loss": 0.5848,
      "step": 15326
    },
    {
      "epoch": 1.37,
      "grad_norm": 4.853871438434748,
      "learning_rate": 2.402227844597884e-06,
      "loss": 0.5918,
      "step": 15327
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.897760847704565,
      "learning_rate": 2.401610650714297e-06,
      "loss": 0.554,
      "step": 15328
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.717017386011283,
      "learning_rate": 2.400993511066396e-06,
      "loss": 0.5972,
      "step": 15329
    },
    {
      "epoch": 1.37,
      "grad_norm": 8.57546605546565,
      "learning_rate": 2.4003764256670646e-06,
      "loss": 0.5501,
      "step": 15330
    },
    {
      "epoch": 1.37,
      "grad_norm": 7.181540412844928,
      "learning_rate": 2.399759394529184e-06,
      "loss": 0.6282,
      "step": 15331
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.682201990519442,
      "learning_rate": 2.39914241766563e-06,
      "loss": 0.6225,
      "step": 15332
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.648979245810481,
      "learning_rate": 2.3985254950892835e-06,
      "loss": 0.5732,
      "step": 15333
    },
    {
      "epoch": 1.37,
      "grad_norm": 7.092905332202179,
      "learning_rate": 2.397908626813018e-06,
      "loss": 0.5655,
      "step": 15334
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.763964220157918,
      "learning_rate": 2.397291812849712e-06,
      "loss": 0.5635,
      "step": 15335
    },
    {
      "epoch": 1.37,
      "grad_norm": 7.2968269273116135,
      "learning_rate": 2.3966750532122397e-06,
      "loss": 0.5558,
      "step": 15336
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.088252382147744,
      "learning_rate": 2.3960583479134742e-06,
      "loss": 0.6008,
      "step": 15337
    },
    {
      "epoch": 1.37,
      "grad_norm": 4.704613330760909,
      "learning_rate": 2.395441696966288e-06,
      "loss": 0.6051,
      "step": 15338
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.650957710992998,
      "learning_rate": 2.394825100383555e-06,
      "loss": 0.5274,
      "step": 15339
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.3329988663875945,
      "learning_rate": 2.3942085581781406e-06,
      "loss": 0.5693,
      "step": 15340
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.635462567603978,
      "learning_rate": 2.3935920703629156e-06,
      "loss": 0.6006,
      "step": 15341
    },
    {
      "epoch": 1.37,
      "grad_norm": 7.061540863697198,
      "learning_rate": 2.3929756369507482e-06,
      "loss": 0.5899,
      "step": 15342
    },
    {
      "epoch": 1.37,
      "grad_norm": 10.372057109582956,
      "learning_rate": 2.392359257954506e-06,
      "loss": 0.5984,
      "step": 15343
    },
    {
      "epoch": 1.37,
      "grad_norm": 7.347546018514494,
      "learning_rate": 2.3917429333870522e-06,
      "loss": 0.597,
      "step": 15344
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.412587892037886,
      "learning_rate": 2.3911266632612556e-06,
      "loss": 0.6606,
      "step": 15345
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.699438119522221,
      "learning_rate": 2.3905104475899764e-06,
      "loss": 0.609,
      "step": 15346
    },
    {
      "epoch": 1.37,
      "grad_norm": 7.554015537193053,
      "learning_rate": 2.389894286386075e-06,
      "loss": 0.6177,
      "step": 15347
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.998396420404499,
      "learning_rate": 2.3892781796624144e-06,
      "loss": 0.6221,
      "step": 15348
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.414518164591421,
      "learning_rate": 2.3886621274318543e-06,
      "loss": 0.6129,
      "step": 15349
    },
    {
      "epoch": 1.37,
      "grad_norm": 9.034411970747223,
      "learning_rate": 2.3880461297072537e-06,
      "loss": 0.5759,
      "step": 15350
    },
    {
      "epoch": 1.37,
      "grad_norm": 8.011339051190925,
      "learning_rate": 2.38743018650147e-06,
      "loss": 0.5414,
      "step": 15351
    },
    {
      "epoch": 1.37,
      "grad_norm": 7.298176755807752,
      "learning_rate": 2.3868142978273613e-06,
      "loss": 0.5665,
      "step": 15352
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.95029335361457,
      "learning_rate": 2.3861984636977796e-06,
      "loss": 0.5129,
      "step": 15353
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.139278003435558,
      "learning_rate": 2.3855826841255815e-06,
      "loss": 0.5801,
      "step": 15354
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.481914587899809,
      "learning_rate": 2.3849669591236187e-06,
      "loss": 0.5433,
      "step": 15355
    },
    {
      "epoch": 1.37,
      "grad_norm": 7.003329937576162,
      "learning_rate": 2.3843512887047436e-06,
      "loss": 0.6016,
      "step": 15356
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.278789392069724,
      "learning_rate": 2.3837356728818095e-06,
      "loss": 0.5791,
      "step": 15357
    },
    {
      "epoch": 1.37,
      "grad_norm": 8.305890622215259,
      "learning_rate": 2.3831201116676618e-06,
      "loss": 0.5725,
      "step": 15358
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.71723928232758,
      "learning_rate": 2.382504605075151e-06,
      "loss": 0.5518,
      "step": 15359
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.620527180203994,
      "learning_rate": 2.3818891531171264e-06,
      "loss": 0.6403,
      "step": 15360
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.694182238476861,
      "learning_rate": 2.3812737558064297e-06,
      "loss": 0.6112,
      "step": 15361
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.610122408878612,
      "learning_rate": 2.3806584131559084e-06,
      "loss": 0.586,
      "step": 15362
    },
    {
      "epoch": 1.37,
      "grad_norm": 7.97366642863119,
      "learning_rate": 2.380043125178407e-06,
      "loss": 0.5209,
      "step": 15363
    },
    {
      "epoch": 1.37,
      "grad_norm": 7.098692881674012,
      "learning_rate": 2.3794278918867674e-06,
      "loss": 0.5646,
      "step": 15364
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.119840763162265,
      "learning_rate": 2.378812713293834e-06,
      "loss": 0.592,
      "step": 15365
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.043464338340651,
      "learning_rate": 2.378197589412442e-06,
      "loss": 0.6183,
      "step": 15366
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.357130276274201,
      "learning_rate": 2.3775825202554347e-06,
      "loss": 0.5587,
      "step": 15367
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.898285931740217,
      "learning_rate": 2.376967505835649e-06,
      "loss": 0.613,
      "step": 15368
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.717366665939535,
      "learning_rate": 2.376352546165922e-06,
      "loss": 0.5705,
      "step": 15369
    },
    {
      "epoch": 1.37,
      "grad_norm": 7.633657966154586,
      "learning_rate": 2.375737641259092e-06,
      "loss": 0.5725,
      "step": 15370
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.482106801020406,
      "learning_rate": 2.3751227911279893e-06,
      "loss": 0.5965,
      "step": 15371
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.030511184776018,
      "learning_rate": 2.3745079957854505e-06,
      "loss": 0.6123,
      "step": 15372
    },
    {
      "epoch": 1.37,
      "grad_norm": 9.12259580983422,
      "learning_rate": 2.3738932552443076e-06,
      "loss": 0.6157,
      "step": 15373
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.7840970701545364,
      "learning_rate": 2.373278569517394e-06,
      "loss": 0.5876,
      "step": 15374
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.540579859244908,
      "learning_rate": 2.3726639386175353e-06,
      "loss": 0.5639,
      "step": 15375
    },
    {
      "epoch": 1.37,
      "grad_norm": 7.690702428764871,
      "learning_rate": 2.3720493625575637e-06,
      "loss": 0.6178,
      "step": 15376
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.474259315580586,
      "learning_rate": 2.371434841350308e-06,
      "loss": 0.6163,
      "step": 15377
    },
    {
      "epoch": 1.37,
      "grad_norm": 4.473906071088057,
      "learning_rate": 2.3708203750085924e-06,
      "loss": 0.5333,
      "step": 15378
    },
    {
      "epoch": 1.37,
      "grad_norm": 8.765730001708302,
      "learning_rate": 2.370205963545243e-06,
      "loss": 0.5838,
      "step": 15379
    },
    {
      "epoch": 1.37,
      "grad_norm": 7.688165296720368,
      "learning_rate": 2.3695916069730846e-06,
      "loss": 0.6125,
      "step": 15380
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.757312549881616,
      "learning_rate": 2.3689773053049414e-06,
      "loss": 0.5546,
      "step": 15381
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.426371603490911,
      "learning_rate": 2.3683630585536357e-06,
      "loss": 0.5691,
      "step": 15382
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.866529135455081,
      "learning_rate": 2.3677488667319887e-06,
      "loss": 0.6132,
      "step": 15383
    },
    {
      "epoch": 1.37,
      "grad_norm": 7.060161617542974,
      "learning_rate": 2.367134729852818e-06,
      "loss": 0.5181,
      "step": 15384
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.846192040728541,
      "learning_rate": 2.366520647928944e-06,
      "loss": 0.5334,
      "step": 15385
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.288635724959641,
      "learning_rate": 2.3659066209731845e-06,
      "loss": 0.5525,
      "step": 15386
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.111857908871401,
      "learning_rate": 2.365292648998356e-06,
      "loss": 0.5982,
      "step": 15387
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.917817172827464,
      "learning_rate": 2.364678732017273e-06,
      "loss": 0.5808,
      "step": 15388
    },
    {
      "epoch": 1.37,
      "grad_norm": 4.954370851938825,
      "learning_rate": 2.3640648700427525e-06,
      "loss": 0.5286,
      "step": 15389
    },
    {
      "epoch": 1.37,
      "grad_norm": 7.20788418123953,
      "learning_rate": 2.363451063087605e-06,
      "loss": 0.5972,
      "step": 15390
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.34621200864893,
      "learning_rate": 2.3628373111646407e-06,
      "loss": 0.5744,
      "step": 15391
    },
    {
      "epoch": 1.37,
      "grad_norm": 7.00789610826379,
      "learning_rate": 2.3622236142866723e-06,
      "loss": 0.6713,
      "step": 15392
    },
    {
      "epoch": 1.37,
      "grad_norm": 8.431630774032294,
      "learning_rate": 2.361609972466509e-06,
      "loss": 0.552,
      "step": 15393
    },
    {
      "epoch": 1.37,
      "grad_norm": 7.244941611732809,
      "learning_rate": 2.36099638571696e-06,
      "loss": 0.6355,
      "step": 15394
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.634512206414605,
      "learning_rate": 2.3603828540508323e-06,
      "loss": 0.6332,
      "step": 15395
    },
    {
      "epoch": 1.37,
      "grad_norm": 7.568264080847426,
      "learning_rate": 2.359769377480933e-06,
      "loss": 0.558,
      "step": 15396
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.487139159934689,
      "learning_rate": 2.3591559560200646e-06,
      "loss": 0.5964,
      "step": 15397
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.413770980047048,
      "learning_rate": 2.358542589681032e-06,
      "loss": 0.5145,
      "step": 15398
    },
    {
      "epoch": 1.37,
      "grad_norm": 7.5327009441905615,
      "learning_rate": 2.357929278476638e-06,
      "loss": 0.5774,
      "step": 15399
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.167687249599947,
      "learning_rate": 2.357316022419685e-06,
      "loss": 0.6247,
      "step": 15400
    },
    {
      "epoch": 1.37,
      "grad_norm": 4.6291125412159095,
      "learning_rate": 2.356702821522972e-06,
      "loss": 0.5862,
      "step": 15401
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.660640464945697,
      "learning_rate": 2.356089675799301e-06,
      "loss": 0.5985,
      "step": 15402
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.559025505666036,
      "learning_rate": 2.355476585261468e-06,
      "loss": 0.5415,
      "step": 15403
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.18613491776588,
      "learning_rate": 2.3548635499222684e-06,
      "loss": 0.5704,
      "step": 15404
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.3575616301316975,
      "learning_rate": 2.354250569794499e-06,
      "loss": 0.5824,
      "step": 15405
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.911251681858195,
      "learning_rate": 2.353637644890954e-06,
      "loss": 0.5046,
      "step": 15406
    },
    {
      "epoch": 1.37,
      "grad_norm": 7.586111893535961,
      "learning_rate": 2.353024775224429e-06,
      "loss": 0.5441,
      "step": 15407
    },
    {
      "epoch": 1.37,
      "grad_norm": 8.360442644779864,
      "learning_rate": 2.3524119608077144e-06,
      "loss": 0.5954,
      "step": 15408
    },
    {
      "epoch": 1.37,
      "grad_norm": 8.984293197908606,
      "learning_rate": 2.3517992016536045e-06,
      "loss": 0.623,
      "step": 15409
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.834046757998613,
      "learning_rate": 2.3511864977748845e-06,
      "loss": 0.5828,
      "step": 15410
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.579401963396583,
      "learning_rate": 2.3505738491843456e-06,
      "loss": 0.576,
      "step": 15411
    },
    {
      "epoch": 1.37,
      "grad_norm": 8.324619459382383,
      "learning_rate": 2.3499612558947752e-06,
      "loss": 0.6408,
      "step": 15412
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.456028171105316,
      "learning_rate": 2.3493487179189606e-06,
      "loss": 0.5855,
      "step": 15413
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.705556437707586,
      "learning_rate": 2.3487362352696863e-06,
      "loss": 0.5492,
      "step": 15414
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.052153392578449,
      "learning_rate": 2.3481238079597392e-06,
      "loss": 0.6022,
      "step": 15415
    },
    {
      "epoch": 1.38,
      "grad_norm": 6.370720128488505,
      "learning_rate": 2.3475114360018973e-06,
      "loss": 0.5709,
      "step": 15416
    },
    {
      "epoch": 1.38,
      "grad_norm": 4.288700833671227,
      "learning_rate": 2.3468991194089465e-06,
      "loss": 0.593,
      "step": 15417
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.6337256027231675,
      "learning_rate": 2.346286858193667e-06,
      "loss": 0.6327,
      "step": 15418
    },
    {
      "epoch": 1.38,
      "grad_norm": 6.443995094169895,
      "learning_rate": 2.345674652368837e-06,
      "loss": 0.5472,
      "step": 15419
    },
    {
      "epoch": 1.38,
      "grad_norm": 9.276278782886475,
      "learning_rate": 2.3450625019472357e-06,
      "loss": 0.6142,
      "step": 15420
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.243813205968432,
      "learning_rate": 2.34445040694164e-06,
      "loss": 0.5773,
      "step": 15421
    },
    {
      "epoch": 1.38,
      "grad_norm": 6.626470699259227,
      "learning_rate": 2.3438383673648284e-06,
      "loss": 0.5973,
      "step": 15422
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.73756272025452,
      "learning_rate": 2.343226383229572e-06,
      "loss": 0.5508,
      "step": 15423
    },
    {
      "epoch": 1.38,
      "grad_norm": 6.210751207171804,
      "learning_rate": 2.342614454548647e-06,
      "loss": 0.6161,
      "step": 15424
    },
    {
      "epoch": 1.38,
      "grad_norm": 6.258004050686239,
      "learning_rate": 2.342002581334826e-06,
      "loss": 0.6745,
      "step": 15425
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.1210533913697684,
      "learning_rate": 2.3413907636008794e-06,
      "loss": 0.6031,
      "step": 15426
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.771086723638093,
      "learning_rate": 2.3407790013595787e-06,
      "loss": 0.5893,
      "step": 15427
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.556515788048686,
      "learning_rate": 2.3401672946236947e-06,
      "loss": 0.5073,
      "step": 15428
    },
    {
      "epoch": 1.38,
      "grad_norm": 8.801979019466977,
      "learning_rate": 2.339555643405991e-06,
      "loss": 0.5676,
      "step": 15429
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.194484814837417,
      "learning_rate": 2.3389440477192377e-06,
      "loss": 0.6192,
      "step": 15430
    },
    {
      "epoch": 1.38,
      "grad_norm": 6.9410878290323215,
      "learning_rate": 2.3383325075761997e-06,
      "loss": 0.5944,
      "step": 15431
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.858146362543744,
      "learning_rate": 2.3377210229896435e-06,
      "loss": 0.6843,
      "step": 15432
    },
    {
      "epoch": 1.38,
      "grad_norm": 6.2521525322407445,
      "learning_rate": 2.3371095939723286e-06,
      "loss": 0.5454,
      "step": 15433
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.791849977493365,
      "learning_rate": 2.3364982205370196e-06,
      "loss": 0.4897,
      "step": 15434
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.483243298924606,
      "learning_rate": 2.335886902696479e-06,
      "loss": 0.5098,
      "step": 15435
    },
    {
      "epoch": 1.38,
      "grad_norm": 6.978916666531775,
      "learning_rate": 2.335275640463463e-06,
      "loss": 0.5629,
      "step": 15436
    },
    {
      "epoch": 1.38,
      "grad_norm": 4.806107106827214,
      "learning_rate": 2.3346644338507323e-06,
      "loss": 0.6272,
      "step": 15437
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.955109389342841,
      "learning_rate": 2.334053282871045e-06,
      "loss": 0.5806,
      "step": 15438
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.455806445737048,
      "learning_rate": 2.3334421875371568e-06,
      "loss": 0.5499,
      "step": 15439
    },
    {
      "epoch": 1.38,
      "grad_norm": 9.427897433079394,
      "learning_rate": 2.3328311478618234e-06,
      "loss": 0.5387,
      "step": 15440
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.241899467477099,
      "learning_rate": 2.3322201638578003e-06,
      "loss": 0.6177,
      "step": 15441
    },
    {
      "epoch": 1.38,
      "grad_norm": 6.8667195142477775,
      "learning_rate": 2.331609235537837e-06,
      "loss": 0.556,
      "step": 15442
    },
    {
      "epoch": 1.38,
      "grad_norm": 9.004489979490959,
      "learning_rate": 2.3309983629146876e-06,
      "loss": 0.6683,
      "step": 15443
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.856774800132853,
      "learning_rate": 2.330387546001102e-06,
      "loss": 0.5441,
      "step": 15444
    },
    {
      "epoch": 1.38,
      "grad_norm": 6.064553393196938,
      "learning_rate": 2.3297767848098306e-06,
      "loss": 0.5901,
      "step": 15445
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.18250206131216,
      "learning_rate": 2.3291660793536227e-06,
      "loss": 0.6375,
      "step": 15446
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.40808340597432,
      "learning_rate": 2.3285554296452223e-06,
      "loss": 0.5688,
      "step": 15447
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.8829404380098165,
      "learning_rate": 2.327944835697378e-06,
      "loss": 0.6282,
      "step": 15448
    },
    {
      "epoch": 1.38,
      "grad_norm": 6.556913145435115,
      "learning_rate": 2.3273342975228323e-06,
      "loss": 0.5684,
      "step": 15449
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.335244578939409,
      "learning_rate": 2.3267238151343303e-06,
      "loss": 0.6024,
      "step": 15450
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.256785032683414,
      "learning_rate": 2.326113388544614e-06,
      "loss": 0.5477,
      "step": 15451
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.605041216451519,
      "learning_rate": 2.325503017766425e-06,
      "loss": 0.596,
      "step": 15452
    },
    {
      "epoch": 1.38,
      "grad_norm": 6.674511088449798,
      "learning_rate": 2.3248927028125053e-06,
      "loss": 0.5341,
      "step": 15453
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.8282309775047345,
      "learning_rate": 2.32428244369559e-06,
      "loss": 0.6159,
      "step": 15454
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.423667341313421,
      "learning_rate": 2.323672240428419e-06,
      "loss": 0.5927,
      "step": 15455
    },
    {
      "epoch": 1.38,
      "grad_norm": 8.088741147254344,
      "learning_rate": 2.323062093023729e-06,
      "loss": 0.5774,
      "step": 15456
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.33109055356537,
      "learning_rate": 2.322452001494256e-06,
      "loss": 0.5137,
      "step": 15457
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.705276471148506,
      "learning_rate": 2.321841965852733e-06,
      "loss": 0.5588,
      "step": 15458
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.523881187960127,
      "learning_rate": 2.321231986111896e-06,
      "loss": 0.5561,
      "step": 15459
    },
    {
      "epoch": 1.38,
      "grad_norm": 6.243026912272245,
      "learning_rate": 2.3206220622844727e-06,
      "loss": 0.5677,
      "step": 15460
    },
    {
      "epoch": 1.38,
      "grad_norm": 8.906139166198802,
      "learning_rate": 2.3200121943831988e-06,
      "loss": 0.5872,
      "step": 15461
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.086733877569432,
      "learning_rate": 2.3194023824207985e-06,
      "loss": 0.6274,
      "step": 15462
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.180909379395099,
      "learning_rate": 2.3187926264100036e-06,
      "loss": 0.5607,
      "step": 15463
    },
    {
      "epoch": 1.38,
      "grad_norm": 6.082115094380091,
      "learning_rate": 2.318182926363541e-06,
      "loss": 0.5505,
      "step": 15464
    },
    {
      "epoch": 1.38,
      "grad_norm": 9.929785148549657,
      "learning_rate": 2.317573282294136e-06,
      "loss": 0.5879,
      "step": 15465
    },
    {
      "epoch": 1.38,
      "grad_norm": 8.342091304168124,
      "learning_rate": 2.316963694214517e-06,
      "loss": 0.6361,
      "step": 15466
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.565773379201985,
      "learning_rate": 2.3163541621374026e-06,
      "loss": 0.6039,
      "step": 15467
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.9420630646200205,
      "learning_rate": 2.3157446860755183e-06,
      "loss": 0.5898,
      "step": 15468
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.131885212553709,
      "learning_rate": 2.3151352660415854e-06,
      "loss": 0.555,
      "step": 15469
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.544152499224465,
      "learning_rate": 2.3145259020483243e-06,
      "loss": 0.5753,
      "step": 15470
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.296129048031609,
      "learning_rate": 2.313916594108453e-06,
      "loss": 0.5749,
      "step": 15471
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.133887827002375,
      "learning_rate": 2.313307342234693e-06,
      "loss": 0.5811,
      "step": 15472
    },
    {
      "epoch": 1.38,
      "grad_norm": 6.972358817196699,
      "learning_rate": 2.3126981464397564e-06,
      "loss": 0.5918,
      "step": 15473
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.093790736937827,
      "learning_rate": 2.312089006736361e-06,
      "loss": 0.6051,
      "step": 15474
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.166572416557928,
      "learning_rate": 2.3114799231372227e-06,
      "loss": 0.5943,
      "step": 15475
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.36832515016384,
      "learning_rate": 2.310870895655052e-06,
      "loss": 0.5975,
      "step": 15476
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.3408516501876395,
      "learning_rate": 2.3102619243025615e-06,
      "loss": 0.5397,
      "step": 15477
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.587408409996009,
      "learning_rate": 2.3096530090924636e-06,
      "loss": 0.5767,
      "step": 15478
    },
    {
      "epoch": 1.38,
      "grad_norm": 8.602740051885991,
      "learning_rate": 2.309044150037468e-06,
      "loss": 0.5706,
      "step": 15479
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.9944107133441635,
      "learning_rate": 2.3084353471502825e-06,
      "loss": 0.5715,
      "step": 15480
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.78972806386249,
      "learning_rate": 2.3078266004436133e-06,
      "loss": 0.6091,
      "step": 15481
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.1773874081713185,
      "learning_rate": 2.307217909930169e-06,
      "loss": 0.5428,
      "step": 15482
    },
    {
      "epoch": 1.38,
      "grad_norm": 6.1573500306151745,
      "learning_rate": 2.3066092756226534e-06,
      "loss": 0.537,
      "step": 15483
    },
    {
      "epoch": 1.38,
      "grad_norm": 8.219978077509818,
      "learning_rate": 2.306000697533771e-06,
      "loss": 0.5551,
      "step": 15484
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.537824774983657,
      "learning_rate": 2.3053921756762254e-06,
      "loss": 0.5328,
      "step": 15485
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.878449655424195,
      "learning_rate": 2.3047837100627155e-06,
      "loss": 0.5875,
      "step": 15486
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.614249164376737,
      "learning_rate": 2.3041753007059436e-06,
      "loss": 0.5809,
      "step": 15487
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.143201321680394,
      "learning_rate": 2.3035669476186084e-06,
      "loss": 0.5902,
      "step": 15488
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.728390276737316,
      "learning_rate": 2.3029586508134082e-06,
      "loss": 0.5822,
      "step": 15489
    },
    {
      "epoch": 1.38,
      "grad_norm": 6.7365649591582635,
      "learning_rate": 2.3023504103030414e-06,
      "loss": 0.5933,
      "step": 15490
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.3811986695413365,
      "learning_rate": 2.3017422261002e-06,
      "loss": 0.6053,
      "step": 15491
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.339735703851821,
      "learning_rate": 2.301134098217583e-06,
      "loss": 0.6474,
      "step": 15492
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.560044057227534,
      "learning_rate": 2.3005260266678787e-06,
      "loss": 0.5446,
      "step": 15493
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.738776697734147,
      "learning_rate": 2.2999180114637827e-06,
      "loss": 0.488,
      "step": 15494
    },
    {
      "epoch": 1.38,
      "grad_norm": 6.092076979202961,
      "learning_rate": 2.299310052617984e-06,
      "loss": 0.633,
      "step": 15495
    },
    {
      "epoch": 1.38,
      "grad_norm": 6.517811745656809,
      "learning_rate": 2.2987021501431745e-06,
      "loss": 0.5775,
      "step": 15496
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.287424039074518,
      "learning_rate": 2.2980943040520414e-06,
      "loss": 0.5544,
      "step": 15497
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.3444802632842405,
      "learning_rate": 2.2974865143572743e-06,
      "loss": 0.5928,
      "step": 15498
    },
    {
      "epoch": 1.38,
      "grad_norm": 6.617330863732741,
      "learning_rate": 2.2968787810715565e-06,
      "loss": 0.6096,
      "step": 15499
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.64000918107434,
      "learning_rate": 2.296271104207574e-06,
      "loss": 0.5604,
      "step": 15500
    },
    {
      "epoch": 1.38,
      "grad_norm": 6.88836362968792,
      "learning_rate": 2.2956634837780117e-06,
      "loss": 0.5182,
      "step": 15501
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.972708404050875,
      "learning_rate": 2.2950559197955513e-06,
      "loss": 0.6206,
      "step": 15502
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.396613974311872,
      "learning_rate": 2.294448412272875e-06,
      "loss": 0.5324,
      "step": 15503
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.990108974993849,
      "learning_rate": 2.293840961222665e-06,
      "loss": 0.6175,
      "step": 15504
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.562270066050555,
      "learning_rate": 2.293233566657598e-06,
      "loss": 0.6519,
      "step": 15505
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.478272529548018,
      "learning_rate": 2.292626228590351e-06,
      "loss": 0.5372,
      "step": 15506
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.691622446186954,
      "learning_rate": 2.292018947033603e-06,
      "loss": 0.5814,
      "step": 15507
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.126294082687872,
      "learning_rate": 2.291411722000028e-06,
      "loss": 0.5747,
      "step": 15508
    },
    {
      "epoch": 1.38,
      "grad_norm": 6.242530931900307,
      "learning_rate": 2.290804553502303e-06,
      "loss": 0.5796,
      "step": 15509
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.431727217148969,
      "learning_rate": 2.2901974415530986e-06,
      "loss": 0.5723,
      "step": 15510
    },
    {
      "epoch": 1.38,
      "grad_norm": 6.129591634919501,
      "learning_rate": 2.289590386165091e-06,
      "loss": 0.6224,
      "step": 15511
    },
    {
      "epoch": 1.38,
      "grad_norm": 9.256229210162052,
      "learning_rate": 2.288983387350946e-06,
      "loss": 0.5536,
      "step": 15512
    },
    {
      "epoch": 1.38,
      "grad_norm": 9.766328352293497,
      "learning_rate": 2.288376445123336e-06,
      "loss": 0.5686,
      "step": 15513
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.274481291683681,
      "learning_rate": 2.287769559494929e-06,
      "loss": 0.6169,
      "step": 15514
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.756244146264532,
      "learning_rate": 2.287162730478393e-06,
      "loss": 0.6844,
      "step": 15515
    },
    {
      "epoch": 1.38,
      "grad_norm": 8.662147766753808,
      "learning_rate": 2.2865559580863944e-06,
      "loss": 0.5539,
      "step": 15516
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.354784450713886,
      "learning_rate": 2.285949242331599e-06,
      "loss": 0.6123,
      "step": 15517
    },
    {
      "epoch": 1.38,
      "grad_norm": 6.821915308067003,
      "learning_rate": 2.2853425832266685e-06,
      "loss": 0.5976,
      "step": 15518
    },
    {
      "epoch": 1.38,
      "grad_norm": 9.159308143302507,
      "learning_rate": 2.284735980784268e-06,
      "loss": 0.6295,
      "step": 15519
    },
    {
      "epoch": 1.38,
      "grad_norm": 8.019737960911044,
      "learning_rate": 2.284129435017055e-06,
      "loss": 0.6273,
      "step": 15520
    },
    {
      "epoch": 1.38,
      "grad_norm": 6.802354130475889,
      "learning_rate": 2.2835229459376934e-06,
      "loss": 0.5827,
      "step": 15521
    },
    {
      "epoch": 1.38,
      "grad_norm": 6.5077442296588455,
      "learning_rate": 2.282916513558841e-06,
      "loss": 0.6386,
      "step": 15522
    },
    {
      "epoch": 1.38,
      "grad_norm": 6.8833312564281295,
      "learning_rate": 2.2823101378931556e-06,
      "loss": 0.5878,
      "step": 15523
    },
    {
      "epoch": 1.38,
      "grad_norm": 6.692726915114494,
      "learning_rate": 2.281703818953297e-06,
      "loss": 0.589,
      "step": 15524
    },
    {
      "epoch": 1.38,
      "grad_norm": 8.3904139688218,
      "learning_rate": 2.2810975567519163e-06,
      "loss": 0.6323,
      "step": 15525
    },
    {
      "epoch": 1.39,
      "grad_norm": 5.0787891151632065,
      "learning_rate": 2.280491351301669e-06,
      "loss": 0.5446,
      "step": 15526
    },
    {
      "epoch": 1.39,
      "grad_norm": 5.1552074738629985,
      "learning_rate": 2.2798852026152096e-06,
      "loss": 0.6238,
      "step": 15527
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.930723641617397,
      "learning_rate": 2.27927911070519e-06,
      "loss": 0.5932,
      "step": 15528
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.635212212948513,
      "learning_rate": 2.2786730755842623e-06,
      "loss": 0.5469,
      "step": 15529
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.179465081279692,
      "learning_rate": 2.2780670972650724e-06,
      "loss": 0.5811,
      "step": 15530
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.240229625981796,
      "learning_rate": 2.2774611757602715e-06,
      "loss": 0.5524,
      "step": 15531
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.692640059855339,
      "learning_rate": 2.276855311082506e-06,
      "loss": 0.5279,
      "step": 15532
    },
    {
      "epoch": 1.39,
      "grad_norm": 8.260172160997833,
      "learning_rate": 2.2762495032444237e-06,
      "loss": 0.6005,
      "step": 15533
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.905338256063175,
      "learning_rate": 2.2756437522586667e-06,
      "loss": 0.5705,
      "step": 15534
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.325098392747677,
      "learning_rate": 2.27503805813788e-06,
      "loss": 0.6001,
      "step": 15535
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.61194935433905,
      "learning_rate": 2.2744324208947083e-06,
      "loss": 0.6119,
      "step": 15536
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.051318931948064,
      "learning_rate": 2.273826840541789e-06,
      "loss": 0.6012,
      "step": 15537
    },
    {
      "epoch": 1.39,
      "grad_norm": 8.494731887442004,
      "learning_rate": 2.2732213170917644e-06,
      "loss": 0.5907,
      "step": 15538
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.381620715517347,
      "learning_rate": 2.2726158505572733e-06,
      "loss": 0.5676,
      "step": 15539
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.044214625921926,
      "learning_rate": 2.272010440950953e-06,
      "loss": 0.6268,
      "step": 15540
    },
    {
      "epoch": 1.39,
      "grad_norm": 5.217967044999442,
      "learning_rate": 2.2714050882854406e-06,
      "loss": 0.6475,
      "step": 15541
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.495022078253916,
      "learning_rate": 2.270799792573373e-06,
      "loss": 0.5363,
      "step": 15542
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.691000764878836,
      "learning_rate": 2.2701945538273817e-06,
      "loss": 0.5599,
      "step": 15543
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.639502100495015,
      "learning_rate": 2.269589372060101e-06,
      "loss": 0.595,
      "step": 15544
    },
    {
      "epoch": 1.39,
      "grad_norm": 5.432846793432708,
      "learning_rate": 2.268984247284162e-06,
      "loss": 0.5957,
      "step": 15545
    },
    {
      "epoch": 1.39,
      "grad_norm": 8.150157268118116,
      "learning_rate": 2.2683791795121967e-06,
      "loss": 0.6025,
      "step": 15546
    },
    {
      "epoch": 1.39,
      "grad_norm": 5.609159670886531,
      "learning_rate": 2.2677741687568354e-06,
      "loss": 0.5458,
      "step": 15547
    },
    {
      "epoch": 1.39,
      "grad_norm": 5.939084940215274,
      "learning_rate": 2.2671692150307027e-06,
      "loss": 0.6128,
      "step": 15548
    },
    {
      "epoch": 1.39,
      "grad_norm": 5.60007127048123,
      "learning_rate": 2.26656431834643e-06,
      "loss": 0.5781,
      "step": 15549
    },
    {
      "epoch": 1.39,
      "grad_norm": 5.533569136144716,
      "learning_rate": 2.2659594787166396e-06,
      "loss": 0.5655,
      "step": 15550
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.062001160902101,
      "learning_rate": 2.2653546961539575e-06,
      "loss": 0.6179,
      "step": 15551
    },
    {
      "epoch": 1.39,
      "grad_norm": 5.584899312538942,
      "learning_rate": 2.2647499706710067e-06,
      "loss": 0.5854,
      "step": 15552
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.6641901824816365,
      "learning_rate": 2.2641453022804105e-06,
      "loss": 0.5886,
      "step": 15553
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.580473794486593,
      "learning_rate": 2.26354069099479e-06,
      "loss": 0.5737,
      "step": 15554
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.523667506723249,
      "learning_rate": 2.262936136826766e-06,
      "loss": 0.5482,
      "step": 15555
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.1032255706817224,
      "learning_rate": 2.262331639788955e-06,
      "loss": 0.6308,
      "step": 15556
    },
    {
      "epoch": 1.39,
      "grad_norm": 13.606743325320394,
      "learning_rate": 2.261727199893975e-06,
      "loss": 0.5148,
      "step": 15557
    },
    {
      "epoch": 1.39,
      "grad_norm": 5.163359274854474,
      "learning_rate": 2.2611228171544437e-06,
      "loss": 0.5952,
      "step": 15558
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.480968484136979,
      "learning_rate": 2.2605184915829747e-06,
      "loss": 0.5989,
      "step": 15559
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.140179800106157,
      "learning_rate": 2.2599142231921835e-06,
      "loss": 0.5913,
      "step": 15560
    },
    {
      "epoch": 1.39,
      "grad_norm": 9.196090875471244,
      "learning_rate": 2.2593100119946836e-06,
      "loss": 0.6099,
      "step": 15561
    },
    {
      "epoch": 1.39,
      "grad_norm": 8.128399659222282,
      "learning_rate": 2.2587058580030853e-06,
      "loss": 0.6052,
      "step": 15562
    },
    {
      "epoch": 1.39,
      "grad_norm": 8.205802365533584,
      "learning_rate": 2.258101761229997e-06,
      "loss": 0.613,
      "step": 15563
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.0746849393358575,
      "learning_rate": 2.25749772168803e-06,
      "loss": 0.5996,
      "step": 15564
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.444651498365336,
      "learning_rate": 2.256893739389792e-06,
      "loss": 0.6233,
      "step": 15565
    },
    {
      "epoch": 1.39,
      "grad_norm": 12.49912525461129,
      "learning_rate": 2.25628981434789e-06,
      "loss": 0.5772,
      "step": 15566
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.062196474824254,
      "learning_rate": 2.255685946574929e-06,
      "loss": 0.5803,
      "step": 15567
    },
    {
      "epoch": 1.39,
      "grad_norm": 5.004051699993318,
      "learning_rate": 2.2550821360835164e-06,
      "loss": 0.5405,
      "step": 15568
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.034073059435542,
      "learning_rate": 2.254478382886251e-06,
      "loss": 0.5926,
      "step": 15569
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.657798285503411,
      "learning_rate": 2.2538746869957364e-06,
      "loss": 0.5834,
      "step": 15570
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.394421058167707,
      "learning_rate": 2.253271048424574e-06,
      "loss": 0.634,
      "step": 15571
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.401578132079515,
      "learning_rate": 2.2526674671853633e-06,
      "loss": 0.5897,
      "step": 15572
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.8648145691949685,
      "learning_rate": 2.2520639432907027e-06,
      "loss": 0.6078,
      "step": 15573
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.762580664121214,
      "learning_rate": 2.251460476753191e-06,
      "loss": 0.6206,
      "step": 15574
    },
    {
      "epoch": 1.39,
      "grad_norm": 5.203870606746879,
      "learning_rate": 2.2508570675854206e-06,
      "loss": 0.5376,
      "step": 15575
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.5414261675266,
      "learning_rate": 2.250253715799991e-06,
      "loss": 0.5998,
      "step": 15576
    },
    {
      "epoch": 1.39,
      "grad_norm": 5.344924594060138,
      "learning_rate": 2.2496504214094903e-06,
      "loss": 0.5372,
      "step": 15577
    },
    {
      "epoch": 1.39,
      "grad_norm": 5.3800127822715496,
      "learning_rate": 2.249047184426515e-06,
      "loss": 0.6079,
      "step": 15578
    },
    {
      "epoch": 1.39,
      "grad_norm": 5.565643012796541,
      "learning_rate": 2.2484440048636546e-06,
      "loss": 0.6011,
      "step": 15579
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.65188988426823,
      "learning_rate": 2.2478408827335e-06,
      "loss": 0.5752,
      "step": 15580
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.512822461868368,
      "learning_rate": 2.247237818048642e-06,
      "loss": 0.5969,
      "step": 15581
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.401452121864045,
      "learning_rate": 2.246634810821664e-06,
      "loss": 0.6224,
      "step": 15582
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.966459529198738,
      "learning_rate": 2.2460318610651542e-06,
      "loss": 0.5779,
      "step": 15583
    },
    {
      "epoch": 1.39,
      "grad_norm": 10.063822823480027,
      "learning_rate": 2.2454289687916986e-06,
      "loss": 0.6116,
      "step": 15584
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.7861253753645,
      "learning_rate": 2.244826134013881e-06,
      "loss": 0.5043,
      "step": 15585
    },
    {
      "epoch": 1.39,
      "grad_norm": 10.234652875259922,
      "learning_rate": 2.244223356744284e-06,
      "loss": 0.5442,
      "step": 15586
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.7388166435294385,
      "learning_rate": 2.243620636995491e-06,
      "loss": 0.602,
      "step": 15587
    },
    {
      "epoch": 1.39,
      "grad_norm": 5.817899256573841,
      "learning_rate": 2.243017974780079e-06,
      "loss": 0.5816,
      "step": 15588
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.411744056849063,
      "learning_rate": 2.2424153701106293e-06,
      "loss": 0.607,
      "step": 15589
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.071273668068562,
      "learning_rate": 2.24181282299972e-06,
      "loss": 0.5579,
      "step": 15590
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.318971582486893,
      "learning_rate": 2.241210333459929e-06,
      "loss": 0.5208,
      "step": 15591
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.503929681814502,
      "learning_rate": 2.240607901503829e-06,
      "loss": 0.6085,
      "step": 15592
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.257879264064039,
      "learning_rate": 2.2400055271439956e-06,
      "loss": 0.5918,
      "step": 15593
    },
    {
      "epoch": 1.39,
      "grad_norm": 8.02938744965514,
      "learning_rate": 2.239403210393005e-06,
      "loss": 0.5917,
      "step": 15594
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.677149893825906,
      "learning_rate": 2.238800951263425e-06,
      "loss": 0.6166,
      "step": 15595
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.945114378936415,
      "learning_rate": 2.238198749767828e-06,
      "loss": 0.5913,
      "step": 15596
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.40008784439519,
      "learning_rate": 2.2375966059187837e-06,
      "loss": 0.5265,
      "step": 15597
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.207602497082454,
      "learning_rate": 2.2369945197288613e-06,
      "loss": 0.5914,
      "step": 15598
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.536810123387598,
      "learning_rate": 2.2363924912106265e-06,
      "loss": 0.5993,
      "step": 15599
    },
    {
      "epoch": 1.39,
      "grad_norm": 8.317339480881117,
      "learning_rate": 2.235790520376649e-06,
      "loss": 0.5609,
      "step": 15600
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.059107588038489,
      "learning_rate": 2.2351886072394885e-06,
      "loss": 0.6361,
      "step": 15601
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.545937163873598,
      "learning_rate": 2.234586751811711e-06,
      "loss": 0.628,
      "step": 15602
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.096821089593124,
      "learning_rate": 2.233984954105879e-06,
      "loss": 0.6058,
      "step": 15603
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.062418190807811,
      "learning_rate": 2.2333832141345537e-06,
      "loss": 0.5527,
      "step": 15604
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.795201408707934,
      "learning_rate": 2.232781531910297e-06,
      "loss": 0.523,
      "step": 15605
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.602840990275533,
      "learning_rate": 2.2321799074456635e-06,
      "loss": 0.6211,
      "step": 15606
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.00375411876351,
      "learning_rate": 2.231578340753215e-06,
      "loss": 0.5745,
      "step": 15607
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.669499703899708,
      "learning_rate": 2.230976831845504e-06,
      "loss": 0.547,
      "step": 15608
    },
    {
      "epoch": 1.39,
      "grad_norm": 5.951029857483052,
      "learning_rate": 2.2303753807350877e-06,
      "loss": 0.5479,
      "step": 15609
    },
    {
      "epoch": 1.39,
      "grad_norm": 5.845023928682716,
      "learning_rate": 2.22977398743452e-06,
      "loss": 0.5294,
      "step": 15610
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.723177730187189,
      "learning_rate": 2.2291726519563533e-06,
      "loss": 0.5794,
      "step": 15611
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.656420293056503,
      "learning_rate": 2.2285713743131416e-06,
      "loss": 0.5806,
      "step": 15612
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.4384890700339135,
      "learning_rate": 2.227970154517431e-06,
      "loss": 0.6101,
      "step": 15613
    },
    {
      "epoch": 1.39,
      "grad_norm": 5.299444198511948,
      "learning_rate": 2.2273689925817734e-06,
      "loss": 0.5397,
      "step": 15614
    },
    {
      "epoch": 1.39,
      "grad_norm": 8.226051667954323,
      "learning_rate": 2.2267678885187153e-06,
      "loss": 0.5945,
      "step": 15615
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.186395273686995,
      "learning_rate": 2.226166842340805e-06,
      "loss": 0.5707,
      "step": 15616
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.513929340521346,
      "learning_rate": 2.225565854060587e-06,
      "loss": 0.5704,
      "step": 15617
    },
    {
      "epoch": 1.39,
      "grad_norm": 5.155406688131777,
      "learning_rate": 2.2249649236906074e-06,
      "loss": 0.5975,
      "step": 15618
    },
    {
      "epoch": 1.39,
      "grad_norm": 5.939788142019107,
      "learning_rate": 2.2243640512434063e-06,
      "loss": 0.55,
      "step": 15619
    },
    {
      "epoch": 1.39,
      "grad_norm": 9.952942731327514,
      "learning_rate": 2.223763236731529e-06,
      "loss": 0.553,
      "step": 15620
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.791200638794214,
      "learning_rate": 2.223162480167512e-06,
      "loss": 0.6532,
      "step": 15621
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.674017611742752,
      "learning_rate": 2.2225617815638976e-06,
      "loss": 0.6046,
      "step": 15622
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.003003869390359,
      "learning_rate": 2.221961140933223e-06,
      "loss": 0.5808,
      "step": 15623
    },
    {
      "epoch": 1.39,
      "grad_norm": 5.577460646500969,
      "learning_rate": 2.2213605582880267e-06,
      "loss": 0.5306,
      "step": 15624
    },
    {
      "epoch": 1.39,
      "grad_norm": 5.9164954037989705,
      "learning_rate": 2.2207600336408445e-06,
      "loss": 0.5701,
      "step": 15625
    },
    {
      "epoch": 1.39,
      "grad_norm": 8.175039937353574,
      "learning_rate": 2.220159567004209e-06,
      "loss": 0.598,
      "step": 15626
    },
    {
      "epoch": 1.39,
      "grad_norm": 8.40589863637329,
      "learning_rate": 2.219559158390654e-06,
      "loss": 0.5665,
      "step": 15627
    },
    {
      "epoch": 1.39,
      "grad_norm": 8.289927870697271,
      "learning_rate": 2.2189588078127133e-06,
      "loss": 0.6047,
      "step": 15628
    },
    {
      "epoch": 1.39,
      "grad_norm": 5.843331258542825,
      "learning_rate": 2.218358515282917e-06,
      "loss": 0.6215,
      "step": 15629
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.230380913435165,
      "learning_rate": 2.217758280813794e-06,
      "loss": 0.6042,
      "step": 15630
    },
    {
      "epoch": 1.39,
      "grad_norm": 5.978618259296307,
      "learning_rate": 2.217158104417877e-06,
      "loss": 0.5401,
      "step": 15631
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.91382112469569,
      "learning_rate": 2.216557986107688e-06,
      "loss": 0.5482,
      "step": 15632
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.967820062097949,
      "learning_rate": 2.2159579258957553e-06,
      "loss": 0.6014,
      "step": 15633
    },
    {
      "epoch": 1.39,
      "grad_norm": 5.812314975275016,
      "learning_rate": 2.2153579237946054e-06,
      "loss": 0.5818,
      "step": 15634
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.286475745042007,
      "learning_rate": 2.2147579798167597e-06,
      "loss": 0.6168,
      "step": 15635
    },
    {
      "epoch": 1.39,
      "grad_norm": 8.513088922788546,
      "learning_rate": 2.2141580939747408e-06,
      "loss": 0.5796,
      "step": 15636
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.942028879466706,
      "learning_rate": 2.2135582662810712e-06,
      "loss": 0.5637,
      "step": 15637
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.812554770176907,
      "learning_rate": 2.212958496748272e-06,
      "loss": 0.5527,
      "step": 15638
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.440702737846951,
      "learning_rate": 2.212358785388859e-06,
      "loss": 0.5712,
      "step": 15639
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.194360386412732,
      "learning_rate": 2.211759132215352e-06,
      "loss": 0.5831,
      "step": 15640
    },
    {
      "epoch": 1.4,
      "grad_norm": 7.852375292816661,
      "learning_rate": 2.211159537240266e-06,
      "loss": 0.5321,
      "step": 15641
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.552701099561052,
      "learning_rate": 2.210560000476118e-06,
      "loss": 0.5771,
      "step": 15642
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.615719633723244,
      "learning_rate": 2.2099605219354207e-06,
      "loss": 0.559,
      "step": 15643
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.379730959696422,
      "learning_rate": 2.20936110163069e-06,
      "loss": 0.5472,
      "step": 15644
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.750688988713783,
      "learning_rate": 2.2087617395744325e-06,
      "loss": 0.574,
      "step": 15645
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.485735847937765,
      "learning_rate": 2.2081624357791616e-06,
      "loss": 0.5905,
      "step": 15646
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.02474309702848,
      "learning_rate": 2.2075631902573853e-06,
      "loss": 0.6219,
      "step": 15647
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.003567980708661,
      "learning_rate": 2.206964003021614e-06,
      "loss": 0.5279,
      "step": 15648
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.570689065339225,
      "learning_rate": 2.20636487408435e-06,
      "loss": 0.6038,
      "step": 15649
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.8324203635522,
      "learning_rate": 2.205765803458102e-06,
      "loss": 0.6162,
      "step": 15650
    },
    {
      "epoch": 1.4,
      "grad_norm": 8.251205354420003,
      "learning_rate": 2.2051667911553755e-06,
      "loss": 0.5866,
      "step": 15651
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.708706482612838,
      "learning_rate": 2.2045678371886695e-06,
      "loss": 0.5754,
      "step": 15652
    },
    {
      "epoch": 1.4,
      "grad_norm": 7.044540190275258,
      "learning_rate": 2.2039689415704884e-06,
      "loss": 0.5853,
      "step": 15653
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.513298737234314,
      "learning_rate": 2.2033701043133322e-06,
      "loss": 0.6157,
      "step": 15654
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.251246736442857,
      "learning_rate": 2.2027713254297006e-06,
      "loss": 0.5777,
      "step": 15655
    },
    {
      "epoch": 1.4,
      "grad_norm": 8.60103473466808,
      "learning_rate": 2.202172604932092e-06,
      "loss": 0.5561,
      "step": 15656
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.23671225315828,
      "learning_rate": 2.2015739428330048e-06,
      "loss": 0.5985,
      "step": 15657
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.50414140293372,
      "learning_rate": 2.2009753391449305e-06,
      "loss": 0.6258,
      "step": 15658
    },
    {
      "epoch": 1.4,
      "grad_norm": 8.583250011230986,
      "learning_rate": 2.2003767938803672e-06,
      "loss": 0.519,
      "step": 15659
    },
    {
      "epoch": 1.4,
      "grad_norm": 7.941374611238873,
      "learning_rate": 2.199778307051807e-06,
      "loss": 0.5896,
      "step": 15660
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.337216847467996,
      "learning_rate": 2.199179878671742e-06,
      "loss": 0.5852,
      "step": 15661
    },
    {
      "epoch": 1.4,
      "grad_norm": 7.574492363303739,
      "learning_rate": 2.1985815087526636e-06,
      "loss": 0.5817,
      "step": 15662
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.641587735515801,
      "learning_rate": 2.1979831973070633e-06,
      "loss": 0.5903,
      "step": 15663
    },
    {
      "epoch": 1.4,
      "grad_norm": 8.154703977153485,
      "learning_rate": 2.1973849443474266e-06,
      "loss": 0.5262,
      "step": 15664
    },
    {
      "epoch": 1.4,
      "grad_norm": 7.243722191711865,
      "learning_rate": 2.1967867498862404e-06,
      "loss": 0.6406,
      "step": 15665
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.451266639995774,
      "learning_rate": 2.1961886139359917e-06,
      "loss": 0.6095,
      "step": 15666
    },
    {
      "epoch": 1.4,
      "grad_norm": 7.715373580611428,
      "learning_rate": 2.1955905365091657e-06,
      "loss": 0.571,
      "step": 15667
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.306867963714762,
      "learning_rate": 2.194992517618245e-06,
      "loss": 0.5747,
      "step": 15668
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.643358437955661,
      "learning_rate": 2.194394557275713e-06,
      "loss": 0.5979,
      "step": 15669
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.790702019157372,
      "learning_rate": 2.1937966554940526e-06,
      "loss": 0.6099,
      "step": 15670
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.363658459039709,
      "learning_rate": 2.1931988122857393e-06,
      "loss": 0.5417,
      "step": 15671
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.033870747203033,
      "learning_rate": 2.1926010276632543e-06,
      "loss": 0.5721,
      "step": 15672
    },
    {
      "epoch": 1.4,
      "grad_norm": 7.7262675474074936,
      "learning_rate": 2.1920033016390744e-06,
      "loss": 0.5745,
      "step": 15673
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.842109755115991,
      "learning_rate": 2.191405634225677e-06,
      "loss": 0.5761,
      "step": 15674
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.4228387236894,
      "learning_rate": 2.190808025435535e-06,
      "loss": 0.5568,
      "step": 15675
    },
    {
      "epoch": 1.4,
      "grad_norm": 8.043262038896328,
      "learning_rate": 2.1902104752811266e-06,
      "loss": 0.558,
      "step": 15676
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.907952635859913,
      "learning_rate": 2.1896129837749203e-06,
      "loss": 0.5754,
      "step": 15677
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.720722234861124,
      "learning_rate": 2.1890155509293866e-06,
      "loss": 0.5545,
      "step": 15678
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.4826126466908285,
      "learning_rate": 2.188418176756997e-06,
      "loss": 0.5783,
      "step": 15679
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.306921953174648,
      "learning_rate": 2.187820861270222e-06,
      "loss": 0.6009,
      "step": 15680
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.9531255252659125,
      "learning_rate": 2.1872236044815264e-06,
      "loss": 0.5809,
      "step": 15681
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.397331859769235,
      "learning_rate": 2.1866264064033793e-06,
      "loss": 0.691,
      "step": 15682
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.666996358758681,
      "learning_rate": 2.1860292670482464e-06,
      "loss": 0.5479,
      "step": 15683
    },
    {
      "epoch": 1.4,
      "grad_norm": 7.005446876921415,
      "learning_rate": 2.1854321864285876e-06,
      "loss": 0.59,
      "step": 15684
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.199544261099525,
      "learning_rate": 2.184835164556869e-06,
      "loss": 0.6147,
      "step": 15685
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.510860699884743,
      "learning_rate": 2.1842382014455505e-06,
      "loss": 0.5674,
      "step": 15686
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.624098520666755,
      "learning_rate": 2.1836412971070937e-06,
      "loss": 0.5796,
      "step": 15687
    },
    {
      "epoch": 1.4,
      "grad_norm": 7.091728329368284,
      "learning_rate": 2.1830444515539584e-06,
      "loss": 0.5826,
      "step": 15688
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.829096403026924,
      "learning_rate": 2.1824476647985993e-06,
      "loss": 0.5337,
      "step": 15689
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.1442032222669,
      "learning_rate": 2.181850936853475e-06,
      "loss": 0.5498,
      "step": 15690
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.170694821264278,
      "learning_rate": 2.181254267731041e-06,
      "loss": 0.598,
      "step": 15691
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.5386660279352045,
      "learning_rate": 2.180657657443753e-06,
      "loss": 0.5603,
      "step": 15692
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.726610193244645,
      "learning_rate": 2.18006110600406e-06,
      "loss": 0.6021,
      "step": 15693
    },
    {
      "epoch": 1.4,
      "grad_norm": 4.8886733240856195,
      "learning_rate": 2.1794646134244164e-06,
      "loss": 0.5615,
      "step": 15694
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.96813292034928,
      "learning_rate": 2.1788681797172733e-06,
      "loss": 0.5935,
      "step": 15695
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.291938071584457,
      "learning_rate": 2.1782718048950773e-06,
      "loss": 0.5906,
      "step": 15696
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.690068570967789,
      "learning_rate": 2.177675488970278e-06,
      "loss": 0.5583,
      "step": 15697
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.239412876760367,
      "learning_rate": 2.177079231955322e-06,
      "loss": 0.5767,
      "step": 15698
    },
    {
      "epoch": 1.4,
      "grad_norm": 8.434531571779699,
      "learning_rate": 2.1764830338626548e-06,
      "loss": 0.6725,
      "step": 15699
    },
    {
      "epoch": 1.4,
      "grad_norm": 4.701132057919579,
      "learning_rate": 2.175886894704721e-06,
      "loss": 0.5344,
      "step": 15700
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.1935513105814,
      "learning_rate": 2.1752908144939656e-06,
      "loss": 0.6033,
      "step": 15701
    },
    {
      "epoch": 1.4,
      "grad_norm": 4.9551594759325,
      "learning_rate": 2.1746947932428263e-06,
      "loss": 0.5203,
      "step": 15702
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.614104059658775,
      "learning_rate": 2.1740988309637455e-06,
      "loss": 0.6517,
      "step": 15703
    },
    {
      "epoch": 1.4,
      "grad_norm": 7.683478447472192,
      "learning_rate": 2.173502927669164e-06,
      "loss": 0.5855,
      "step": 15704
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.663120613195911,
      "learning_rate": 2.1729070833715183e-06,
      "loss": 0.5707,
      "step": 15705
    },
    {
      "epoch": 1.4,
      "grad_norm": 7.2468490273477295,
      "learning_rate": 2.1723112980832477e-06,
      "loss": 0.6345,
      "step": 15706
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.652671145829867,
      "learning_rate": 2.1717155718167847e-06,
      "loss": 0.5894,
      "step": 15707
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.017025107059707,
      "learning_rate": 2.171119904584567e-06,
      "loss": 0.5864,
      "step": 15708
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.822947915971672,
      "learning_rate": 2.1705242963990248e-06,
      "loss": 0.6026,
      "step": 15709
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.04697909237207,
      "learning_rate": 2.1699287472725904e-06,
      "loss": 0.5976,
      "step": 15710
    },
    {
      "epoch": 1.4,
      "grad_norm": 8.428641140303187,
      "learning_rate": 2.1693332572176968e-06,
      "loss": 0.6328,
      "step": 15711
    },
    {
      "epoch": 1.4,
      "grad_norm": 7.002559924725337,
      "learning_rate": 2.168737826246772e-06,
      "loss": 0.5618,
      "step": 15712
    },
    {
      "epoch": 1.4,
      "grad_norm": 7.354016262926374,
      "learning_rate": 2.1681424543722444e-06,
      "loss": 0.601,
      "step": 15713
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.777716702022312,
      "learning_rate": 2.1675471416065436e-06,
      "loss": 0.5572,
      "step": 15714
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.4652321295823345,
      "learning_rate": 2.166951887962091e-06,
      "loss": 0.5866,
      "step": 15715
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.269019590306064,
      "learning_rate": 2.1663566934513136e-06,
      "loss": 0.5804,
      "step": 15716
    },
    {
      "epoch": 1.4,
      "grad_norm": 8.501924083466992,
      "learning_rate": 2.1657615580866347e-06,
      "loss": 0.5824,
      "step": 15717
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.169769342951194,
      "learning_rate": 2.1651664818804767e-06,
      "loss": 0.6052,
      "step": 15718
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.734643315845236,
      "learning_rate": 2.16457146484526e-06,
      "loss": 0.5928,
      "step": 15719
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.456288897012107,
      "learning_rate": 2.1639765069934067e-06,
      "loss": 0.6104,
      "step": 15720
    },
    {
      "epoch": 1.4,
      "grad_norm": 7.537893145432655,
      "learning_rate": 2.1633816083373322e-06,
      "loss": 0.5696,
      "step": 15721
    },
    {
      "epoch": 1.4,
      "grad_norm": 8.406328837021045,
      "learning_rate": 2.1627867688894533e-06,
      "loss": 0.5779,
      "step": 15722
    },
    {
      "epoch": 1.4,
      "grad_norm": 7.074076771825386,
      "learning_rate": 2.1621919886621875e-06,
      "loss": 0.5929,
      "step": 15723
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.2061612098053995,
      "learning_rate": 2.1615972676679483e-06,
      "loss": 0.5827,
      "step": 15724
    },
    {
      "epoch": 1.4,
      "grad_norm": 7.0053521727125085,
      "learning_rate": 2.1610026059191508e-06,
      "loss": 0.6337,
      "step": 15725
    },
    {
      "epoch": 1.4,
      "grad_norm": 9.460106959150751,
      "learning_rate": 2.1604080034282065e-06,
      "loss": 0.5734,
      "step": 15726
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.717029827975725,
      "learning_rate": 2.1598134602075283e-06,
      "loss": 0.6244,
      "step": 15727
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.917060068517886,
      "learning_rate": 2.159218976269522e-06,
      "loss": 0.5334,
      "step": 15728
    },
    {
      "epoch": 1.4,
      "grad_norm": 7.242488479037595,
      "learning_rate": 2.158624551626599e-06,
      "loss": 0.581,
      "step": 15729
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.915452496399931,
      "learning_rate": 2.158030186291166e-06,
      "loss": 0.5827,
      "step": 15730
    },
    {
      "epoch": 1.4,
      "grad_norm": 9.728676399644009,
      "learning_rate": 2.1574358802756283e-06,
      "loss": 0.5773,
      "step": 15731
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.909038911851871,
      "learning_rate": 2.1568416335923924e-06,
      "loss": 0.6236,
      "step": 15732
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.800502901551602,
      "learning_rate": 2.1562474462538623e-06,
      "loss": 0.5354,
      "step": 15733
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.583221313467861,
      "learning_rate": 2.155653318272437e-06,
      "loss": 0.5954,
      "step": 15734
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.36200548500084,
      "learning_rate": 2.155059249660522e-06,
      "loss": 0.5587,
      "step": 15735
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.4362806567823085,
      "learning_rate": 2.154465240430513e-06,
      "loss": 0.6326,
      "step": 15736
    },
    {
      "epoch": 1.4,
      "grad_norm": 7.716879021726436,
      "learning_rate": 2.15387129059481e-06,
      "loss": 0.5881,
      "step": 15737
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.177335491118514,
      "learning_rate": 2.1532774001658113e-06,
      "loss": 0.5905,
      "step": 15738
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.420446460076768,
      "learning_rate": 2.1526835691559127e-06,
      "loss": 0.6147,
      "step": 15739
    },
    {
      "epoch": 1.4,
      "grad_norm": 8.857395157657413,
      "learning_rate": 2.1520897975775107e-06,
      "loss": 0.5849,
      "step": 15740
    },
    {
      "epoch": 1.4,
      "grad_norm": 7.173872445865954,
      "learning_rate": 2.1514960854429954e-06,
      "loss": 0.5526,
      "step": 15741
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.246084242185098,
      "learning_rate": 2.1509024327647614e-06,
      "loss": 0.5936,
      "step": 15742
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.401917464956223,
      "learning_rate": 2.1503088395552004e-06,
      "loss": 0.5736,
      "step": 15743
    },
    {
      "epoch": 1.4,
      "grad_norm": 7.169704479552531,
      "learning_rate": 2.1497153058267006e-06,
      "loss": 0.5781,
      "step": 15744
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.215289216012471,
      "learning_rate": 2.1491218315916525e-06,
      "loss": 0.6347,
      "step": 15745
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.709593121073899,
      "learning_rate": 2.148528416862444e-06,
      "loss": 0.4937,
      "step": 15746
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.212548973915593,
      "learning_rate": 2.147935061651459e-06,
      "loss": 0.6048,
      "step": 15747
    },
    {
      "epoch": 1.4,
      "grad_norm": 7.358257068542974,
      "learning_rate": 2.1473417659710834e-06,
      "loss": 0.5595,
      "step": 15748
    },
    {
      "epoch": 1.4,
      "grad_norm": 8.237022647264794,
      "learning_rate": 2.1467485298337026e-06,
      "loss": 0.5312,
      "step": 15749
    },
    {
      "epoch": 1.41,
      "grad_norm": 4.61067774633747,
      "learning_rate": 2.1461553532516967e-06,
      "loss": 0.536,
      "step": 15750
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.477099037610732,
      "learning_rate": 2.145562236237447e-06,
      "loss": 0.5704,
      "step": 15751
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.753480261026587,
      "learning_rate": 2.144969178803335e-06,
      "loss": 0.5764,
      "step": 15752
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.711612695378747,
      "learning_rate": 2.1443761809617413e-06,
      "loss": 0.5793,
      "step": 15753
    },
    {
      "epoch": 1.41,
      "grad_norm": 5.9213645101858035,
      "learning_rate": 2.1437832427250387e-06,
      "loss": 0.5378,
      "step": 15754
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.250016055935236,
      "learning_rate": 2.1431903641056062e-06,
      "loss": 0.6086,
      "step": 15755
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.343369522076429,
      "learning_rate": 2.142597545115818e-06,
      "loss": 0.6113,
      "step": 15756
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.240743741962293,
      "learning_rate": 2.142004785768049e-06,
      "loss": 0.574,
      "step": 15757
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.6337058723541125,
      "learning_rate": 2.1414120860746714e-06,
      "loss": 0.6237,
      "step": 15758
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.455075468246401,
      "learning_rate": 2.1408194460480576e-06,
      "loss": 0.5599,
      "step": 15759
    },
    {
      "epoch": 1.41,
      "grad_norm": 5.215469976761498,
      "learning_rate": 2.140226865700575e-06,
      "loss": 0.6488,
      "step": 15760
    },
    {
      "epoch": 1.41,
      "grad_norm": 8.861684532726306,
      "learning_rate": 2.1396343450445935e-06,
      "loss": 0.6567,
      "step": 15761
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.056689436230939,
      "learning_rate": 2.139041884092481e-06,
      "loss": 0.5968,
      "step": 15762
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.552096246201852,
      "learning_rate": 2.1384494828566034e-06,
      "loss": 0.5721,
      "step": 15763
    },
    {
      "epoch": 1.41,
      "grad_norm": 4.482743736304597,
      "learning_rate": 2.137857141349328e-06,
      "loss": 0.5939,
      "step": 15764
    },
    {
      "epoch": 1.41,
      "grad_norm": 8.20623344835906,
      "learning_rate": 2.1372648595830157e-06,
      "loss": 0.6445,
      "step": 15765
    },
    {
      "epoch": 1.41,
      "grad_norm": 8.14451085174631,
      "learning_rate": 2.1366726375700315e-06,
      "loss": 0.5775,
      "step": 15766
    },
    {
      "epoch": 1.41,
      "grad_norm": 9.369208756980273,
      "learning_rate": 2.1360804753227337e-06,
      "loss": 0.6083,
      "step": 15767
    },
    {
      "epoch": 1.41,
      "grad_norm": 8.722570906548366,
      "learning_rate": 2.1354883728534843e-06,
      "loss": 0.5685,
      "step": 15768
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.435184692637387,
      "learning_rate": 2.134896330174642e-06,
      "loss": 0.599,
      "step": 15769
    },
    {
      "epoch": 1.41,
      "grad_norm": 5.777383324033865,
      "learning_rate": 2.1343043472985637e-06,
      "loss": 0.6173,
      "step": 15770
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.7529198598204045,
      "learning_rate": 2.1337124242376086e-06,
      "loss": 0.5856,
      "step": 15771
    },
    {
      "epoch": 1.41,
      "grad_norm": 8.696326008727045,
      "learning_rate": 2.133120561004127e-06,
      "loss": 0.642,
      "step": 15772
    },
    {
      "epoch": 1.41,
      "grad_norm": 5.904343140770151,
      "learning_rate": 2.1325287576104765e-06,
      "loss": 0.5695,
      "step": 15773
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.802145694173473,
      "learning_rate": 2.1319370140690083e-06,
      "loss": 0.5748,
      "step": 15774
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.543510335563226,
      "learning_rate": 2.1313453303920735e-06,
      "loss": 0.5624,
      "step": 15775
    },
    {
      "epoch": 1.41,
      "grad_norm": 5.874950462129129,
      "learning_rate": 2.130753706592023e-06,
      "loss": 0.5592,
      "step": 15776
    },
    {
      "epoch": 1.41,
      "grad_norm": 5.928576561718683,
      "learning_rate": 2.130162142681207e-06,
      "loss": 0.5953,
      "step": 15777
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.462084499062974,
      "learning_rate": 2.12957063867197e-06,
      "loss": 0.6209,
      "step": 15778
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.031200910093054,
      "learning_rate": 2.1289791945766613e-06,
      "loss": 0.5937,
      "step": 15779
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.686387350906905,
      "learning_rate": 2.128387810407622e-06,
      "loss": 0.5429,
      "step": 15780
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.1027745264469955,
      "learning_rate": 2.1277964861771992e-06,
      "loss": 0.5793,
      "step": 15781
    },
    {
      "epoch": 1.41,
      "grad_norm": 5.949842846990863,
      "learning_rate": 2.127205221897735e-06,
      "loss": 0.5458,
      "step": 15782
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.525160864325934,
      "learning_rate": 2.1266140175815703e-06,
      "loss": 0.6098,
      "step": 15783
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.5970218306002595,
      "learning_rate": 2.126022873241047e-06,
      "loss": 0.5805,
      "step": 15784
    },
    {
      "epoch": 1.41,
      "grad_norm": 8.450251062276902,
      "learning_rate": 2.1254317888885e-06,
      "loss": 0.6077,
      "step": 15785
    },
    {
      "epoch": 1.41,
      "grad_norm": 8.133292069756514,
      "learning_rate": 2.12484076453627e-06,
      "loss": 0.6403,
      "step": 15786
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.153124663258722,
      "learning_rate": 2.1242498001966922e-06,
      "loss": 0.576,
      "step": 15787
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.546884909319679,
      "learning_rate": 2.1236588958821024e-06,
      "loss": 0.5899,
      "step": 15788
    },
    {
      "epoch": 1.41,
      "grad_norm": 8.78339075164476,
      "learning_rate": 2.1230680516048335e-06,
      "loss": 0.6482,
      "step": 15789
    },
    {
      "epoch": 1.41,
      "grad_norm": 10.126058513530888,
      "learning_rate": 2.122477267377221e-06,
      "loss": 0.5765,
      "step": 15790
    },
    {
      "epoch": 1.41,
      "grad_norm": 4.503840680562158,
      "learning_rate": 2.1218865432115913e-06,
      "loss": 0.5681,
      "step": 15791
    },
    {
      "epoch": 1.41,
      "grad_norm": 5.878986778858247,
      "learning_rate": 2.121295879120278e-06,
      "loss": 0.5194,
      "step": 15792
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.538482664872857,
      "learning_rate": 2.1207052751156103e-06,
      "loss": 0.5882,
      "step": 15793
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.290667997470873,
      "learning_rate": 2.1201147312099126e-06,
      "loss": 0.5591,
      "step": 15794
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.50402363550051,
      "learning_rate": 2.119524247415513e-06,
      "loss": 0.5148,
      "step": 15795
    },
    {
      "epoch": 1.41,
      "grad_norm": 8.515651450989647,
      "learning_rate": 2.1189338237447365e-06,
      "loss": 0.5599,
      "step": 15796
    },
    {
      "epoch": 1.41,
      "grad_norm": 11.101668510674969,
      "learning_rate": 2.1183434602099094e-06,
      "loss": 0.5111,
      "step": 15797
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.718096241499279,
      "learning_rate": 2.1177531568233497e-06,
      "loss": 0.6009,
      "step": 15798
    },
    {
      "epoch": 1.41,
      "grad_norm": 5.612136264877197,
      "learning_rate": 2.117162913597381e-06,
      "loss": 0.5588,
      "step": 15799
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.631296961227163,
      "learning_rate": 2.116572730544323e-06,
      "loss": 0.6159,
      "step": 15800
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.143538361540989,
      "learning_rate": 2.1159826076764946e-06,
      "loss": 0.6211,
      "step": 15801
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.1813333258444425,
      "learning_rate": 2.115392545006213e-06,
      "loss": 0.5916,
      "step": 15802
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.846940547652701,
      "learning_rate": 2.114802542545797e-06,
      "loss": 0.561,
      "step": 15803
    },
    {
      "epoch": 1.41,
      "grad_norm": 4.803596312450302,
      "learning_rate": 2.1142126003075576e-06,
      "loss": 0.5494,
      "step": 15804
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.344427521234288,
      "learning_rate": 2.1136227183038104e-06,
      "loss": 0.5925,
      "step": 15805
    },
    {
      "epoch": 1.41,
      "grad_norm": 5.98306049768175,
      "learning_rate": 2.1130328965468684e-06,
      "loss": 0.5766,
      "step": 15806
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.358778063776489,
      "learning_rate": 2.112443135049044e-06,
      "loss": 0.5855,
      "step": 15807
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.112862458615103,
      "learning_rate": 2.111853433822643e-06,
      "loss": 0.5819,
      "step": 15808
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.0109680034700705,
      "learning_rate": 2.1112637928799766e-06,
      "loss": 0.5587,
      "step": 15809
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.251478111528706,
      "learning_rate": 2.110674212233355e-06,
      "loss": 0.6339,
      "step": 15810
    },
    {
      "epoch": 1.41,
      "grad_norm": 5.451209272611212,
      "learning_rate": 2.1100846918950796e-06,
      "loss": 0.5748,
      "step": 15811
    },
    {
      "epoch": 1.41,
      "grad_norm": 4.989117766534107,
      "learning_rate": 2.109495231877457e-06,
      "loss": 0.6017,
      "step": 15812
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.54673761074574,
      "learning_rate": 2.1089058321927917e-06,
      "loss": 0.6382,
      "step": 15813
    },
    {
      "epoch": 1.41,
      "grad_norm": 5.499827680918088,
      "learning_rate": 2.1083164928533856e-06,
      "loss": 0.5724,
      "step": 15814
    },
    {
      "epoch": 1.41,
      "grad_norm": 5.628861307391039,
      "learning_rate": 2.1077272138715402e-06,
      "loss": 0.5601,
      "step": 15815
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.55191049189998,
      "learning_rate": 2.107137995259557e-06,
      "loss": 0.6074,
      "step": 15816
    },
    {
      "epoch": 1.41,
      "grad_norm": 9.54037532679063,
      "learning_rate": 2.106548837029731e-06,
      "loss": 0.6261,
      "step": 15817
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.1551083746490605,
      "learning_rate": 2.105959739194362e-06,
      "loss": 0.6214,
      "step": 15818
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.991594501125826,
      "learning_rate": 2.105370701765745e-06,
      "loss": 0.5824,
      "step": 15819
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.119155241654609,
      "learning_rate": 2.104781724756176e-06,
      "loss": 0.6237,
      "step": 15820
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.70986421733844,
      "learning_rate": 2.1041928081779488e-06,
      "loss": 0.5637,
      "step": 15821
    },
    {
      "epoch": 1.41,
      "grad_norm": 4.527250557326591,
      "learning_rate": 2.103603952043354e-06,
      "loss": 0.5685,
      "step": 15822
    },
    {
      "epoch": 1.41,
      "grad_norm": 8.863754844414323,
      "learning_rate": 2.1030151563646857e-06,
      "loss": 0.5964,
      "step": 15823
    },
    {
      "epoch": 1.41,
      "grad_norm": 4.810719152340684,
      "learning_rate": 2.1024264211542295e-06,
      "loss": 0.5651,
      "step": 15824
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.191539446976396,
      "learning_rate": 2.101837746424276e-06,
      "loss": 0.6497,
      "step": 15825
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.574389521844845,
      "learning_rate": 2.1012491321871136e-06,
      "loss": 0.5399,
      "step": 15826
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.4411305070226215,
      "learning_rate": 2.1006605784550265e-06,
      "loss": 0.5617,
      "step": 15827
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.69950060157889,
      "learning_rate": 2.1000720852403006e-06,
      "loss": 0.6183,
      "step": 15828
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.233064871417518,
      "learning_rate": 2.0994836525552216e-06,
      "loss": 0.5475,
      "step": 15829
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.178073218576332,
      "learning_rate": 2.0988952804120665e-06,
      "loss": 0.6401,
      "step": 15830
    },
    {
      "epoch": 1.41,
      "grad_norm": 11.48091220370895,
      "learning_rate": 2.09830696882312e-06,
      "loss": 0.5924,
      "step": 15831
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.726544970210392,
      "learning_rate": 2.0977187178006603e-06,
      "loss": 0.6108,
      "step": 15832
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.493024074615426,
      "learning_rate": 2.097130527356967e-06,
      "loss": 0.6887,
      "step": 15833
    },
    {
      "epoch": 1.41,
      "grad_norm": 5.77517073545966,
      "learning_rate": 2.0965423975043163e-06,
      "loss": 0.5874,
      "step": 15834
    },
    {
      "epoch": 1.41,
      "grad_norm": 9.376542625033617,
      "learning_rate": 2.0959543282549864e-06,
      "loss": 0.6239,
      "step": 15835
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.084768083218725,
      "learning_rate": 2.09536631962125e-06,
      "loss": 0.6112,
      "step": 15836
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.663233268856237,
      "learning_rate": 2.09477837161538e-06,
      "loss": 0.542,
      "step": 15837
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.462507541565511,
      "learning_rate": 2.094190484249648e-06,
      "loss": 0.4877,
      "step": 15838
    },
    {
      "epoch": 1.41,
      "grad_norm": 5.4968173694307145,
      "learning_rate": 2.093602657536327e-06,
      "loss": 0.5857,
      "step": 15839
    },
    {
      "epoch": 1.41,
      "grad_norm": 5.5866405970213,
      "learning_rate": 2.0930148914876848e-06,
      "loss": 0.6428,
      "step": 15840
    },
    {
      "epoch": 1.41,
      "grad_norm": 5.578651769847526,
      "learning_rate": 2.092427186115991e-06,
      "loss": 0.5947,
      "step": 15841
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.345894580000176,
      "learning_rate": 2.091839541433513e-06,
      "loss": 0.5434,
      "step": 15842
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.009414613587293,
      "learning_rate": 2.091251957452515e-06,
      "loss": 0.6036,
      "step": 15843
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.089714893731891,
      "learning_rate": 2.0906644341852623e-06,
      "loss": 0.5901,
      "step": 15844
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.7328367371378866,
      "learning_rate": 2.090076971644018e-06,
      "loss": 0.586,
      "step": 15845
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.306922764053833,
      "learning_rate": 2.089489569841044e-06,
      "loss": 0.6063,
      "step": 15846
    },
    {
      "epoch": 1.41,
      "grad_norm": 5.967776470254952,
      "learning_rate": 2.0889022287886034e-06,
      "loss": 0.5729,
      "step": 15847
    },
    {
      "epoch": 1.41,
      "grad_norm": 8.349808263351427,
      "learning_rate": 2.088314948498952e-06,
      "loss": 0.5579,
      "step": 15848
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.530259061203962,
      "learning_rate": 2.087727728984349e-06,
      "loss": 0.6075,
      "step": 15849
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.466231216346293,
      "learning_rate": 2.087140570257054e-06,
      "loss": 0.6223,
      "step": 15850
    },
    {
      "epoch": 1.41,
      "grad_norm": 8.006262937775748,
      "learning_rate": 2.086553472329319e-06,
      "loss": 0.5263,
      "step": 15851
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.406007823940519,
      "learning_rate": 2.0859664352133997e-06,
      "loss": 0.5913,
      "step": 15852
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.6262602915417,
      "learning_rate": 2.085379458921549e-06,
      "loss": 0.5992,
      "step": 15853
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.97501010009488,
      "learning_rate": 2.08479254346602e-06,
      "loss": 0.6272,
      "step": 15854
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.007084030741908,
      "learning_rate": 2.084205688859064e-06,
      "loss": 0.632,
      "step": 15855
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.804467956836428,
      "learning_rate": 2.0836188951129267e-06,
      "loss": 0.6072,
      "step": 15856
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.946640947023971,
      "learning_rate": 2.0830321622398588e-06,
      "loss": 0.6186,
      "step": 15857
    },
    {
      "epoch": 1.41,
      "grad_norm": 4.944131918069959,
      "learning_rate": 2.0824454902521056e-06,
      "loss": 0.5271,
      "step": 15858
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.445833340203318,
      "learning_rate": 2.0818588791619137e-06,
      "loss": 0.5788,
      "step": 15859
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.832684914665286,
      "learning_rate": 2.0812723289815294e-06,
      "loss": 0.5802,
      "step": 15860
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.605933292814659,
      "learning_rate": 2.080685839723191e-06,
      "loss": 0.5335,
      "step": 15861
    },
    {
      "epoch": 1.42,
      "grad_norm": 7.127773116154133,
      "learning_rate": 2.080099411399143e-06,
      "loss": 0.5977,
      "step": 15862
    },
    {
      "epoch": 1.42,
      "grad_norm": 7.568861836182093,
      "learning_rate": 2.079513044021625e-06,
      "loss": 0.5692,
      "step": 15863
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.477152709754503,
      "learning_rate": 2.0789267376028766e-06,
      "loss": 0.619,
      "step": 15864
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.287732142187681,
      "learning_rate": 2.078340492155137e-06,
      "loss": 0.6034,
      "step": 15865
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.1855103442060155,
      "learning_rate": 2.07775430769064e-06,
      "loss": 0.6194,
      "step": 15866
    },
    {
      "epoch": 1.42,
      "grad_norm": 7.487852517070933,
      "learning_rate": 2.0771681842216236e-06,
      "loss": 0.6228,
      "step": 15867
    },
    {
      "epoch": 1.42,
      "grad_norm": 8.697018057539774,
      "learning_rate": 2.0765821217603184e-06,
      "loss": 0.5573,
      "step": 15868
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.717883430251962,
      "learning_rate": 2.075996120318959e-06,
      "loss": 0.5498,
      "step": 15869
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.17180001263375,
      "learning_rate": 2.075410179909778e-06,
      "loss": 0.5621,
      "step": 15870
    },
    {
      "epoch": 1.42,
      "grad_norm": 8.368616124983477,
      "learning_rate": 2.074824300545005e-06,
      "loss": 0.5997,
      "step": 15871
    },
    {
      "epoch": 1.42,
      "grad_norm": 7.6142468108445325,
      "learning_rate": 2.0742384822368677e-06,
      "loss": 0.6142,
      "step": 15872
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.969723460376286,
      "learning_rate": 2.0736527249975975e-06,
      "loss": 0.6245,
      "step": 15873
    },
    {
      "epoch": 1.42,
      "grad_norm": 9.426794375040764,
      "learning_rate": 2.0730670288394155e-06,
      "loss": 0.5236,
      "step": 15874
    },
    {
      "epoch": 1.42,
      "grad_norm": 4.641289138735613,
      "learning_rate": 2.0724813937745496e-06,
      "loss": 0.5406,
      "step": 15875
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.830927903281725,
      "learning_rate": 2.071895819815224e-06,
      "loss": 0.5971,
      "step": 15876
    },
    {
      "epoch": 1.42,
      "grad_norm": 8.147245355764186,
      "learning_rate": 2.07131030697366e-06,
      "loss": 0.6186,
      "step": 15877
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.860390016140075,
      "learning_rate": 2.07072485526208e-06,
      "loss": 0.5645,
      "step": 15878
    },
    {
      "epoch": 1.42,
      "grad_norm": 7.524945605625171,
      "learning_rate": 2.070139464692705e-06,
      "loss": 0.5161,
      "step": 15879
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.482882574671355,
      "learning_rate": 2.0695541352777533e-06,
      "loss": 0.6248,
      "step": 15880
    },
    {
      "epoch": 1.42,
      "grad_norm": 9.498791917131,
      "learning_rate": 2.068968867029439e-06,
      "loss": 0.5968,
      "step": 15881
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.3990774763065605,
      "learning_rate": 2.0683836599599806e-06,
      "loss": 0.6186,
      "step": 15882
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.734927371576587,
      "learning_rate": 2.0677985140815932e-06,
      "loss": 0.593,
      "step": 15883
    },
    {
      "epoch": 1.42,
      "grad_norm": 7.91964649850289,
      "learning_rate": 2.067213429406491e-06,
      "loss": 0.6225,
      "step": 15884
    },
    {
      "epoch": 1.42,
      "grad_norm": 4.999450455976784,
      "learning_rate": 2.066628405946885e-06,
      "loss": 0.5479,
      "step": 15885
    },
    {
      "epoch": 1.42,
      "grad_norm": 7.340778814117683,
      "learning_rate": 2.0660434437149898e-06,
      "loss": 0.6317,
      "step": 15886
    },
    {
      "epoch": 1.42,
      "grad_norm": 7.495256066293773,
      "learning_rate": 2.0654585427230097e-06,
      "loss": 0.5821,
      "step": 15887
    },
    {
      "epoch": 1.42,
      "grad_norm": 8.328512374538857,
      "learning_rate": 2.0648737029831567e-06,
      "loss": 0.5731,
      "step": 15888
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.829210866076961,
      "learning_rate": 2.064288924507637e-06,
      "loss": 0.5736,
      "step": 15889
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.824611374698113,
      "learning_rate": 2.0637042073086566e-06,
      "loss": 0.558,
      "step": 15890
    },
    {
      "epoch": 1.42,
      "grad_norm": 8.109226521166494,
      "learning_rate": 2.063119551398421e-06,
      "loss": 0.5705,
      "step": 15891
    },
    {
      "epoch": 1.42,
      "grad_norm": 9.519451920797252,
      "learning_rate": 2.0625349567891343e-06,
      "loss": 0.5557,
      "step": 15892
    },
    {
      "epoch": 1.42,
      "grad_norm": 9.871335058618977,
      "learning_rate": 2.0619504234929966e-06,
      "loss": 0.6642,
      "step": 15893
    },
    {
      "epoch": 1.42,
      "grad_norm": 8.956002981923469,
      "learning_rate": 2.0613659515222108e-06,
      "loss": 0.6153,
      "step": 15894
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.289504458340173,
      "learning_rate": 2.060781540888973e-06,
      "loss": 0.5858,
      "step": 15895
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.265328717777523,
      "learning_rate": 2.060197191605485e-06,
      "loss": 0.5732,
      "step": 15896
    },
    {
      "epoch": 1.42,
      "grad_norm": 12.009683445188044,
      "learning_rate": 2.0596129036839415e-06,
      "loss": 0.6305,
      "step": 15897
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.377841796975436,
      "learning_rate": 2.05902867713654e-06,
      "loss": 0.5482,
      "step": 15898
    },
    {
      "epoch": 1.42,
      "grad_norm": 4.925834181287161,
      "learning_rate": 2.058444511975475e-06,
      "loss": 0.5595,
      "step": 15899
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.140999447968709,
      "learning_rate": 2.0578604082129383e-06,
      "loss": 0.616,
      "step": 15900
    },
    {
      "epoch": 1.42,
      "grad_norm": 7.26401068514973,
      "learning_rate": 2.0572763658611215e-06,
      "loss": 0.5854,
      "step": 15901
    },
    {
      "epoch": 1.42,
      "grad_norm": 10.416149798527723,
      "learning_rate": 2.056692384932217e-06,
      "loss": 0.6126,
      "step": 15902
    },
    {
      "epoch": 1.42,
      "grad_norm": 7.350137118790361,
      "learning_rate": 2.056108465438412e-06,
      "loss": 0.6033,
      "step": 15903
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.260592519656983,
      "learning_rate": 2.055524607391897e-06,
      "loss": 0.5771,
      "step": 15904
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.112199928615497,
      "learning_rate": 2.0549408108048584e-06,
      "loss": 0.5537,
      "step": 15905
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.122640962536826,
      "learning_rate": 2.0543570756894797e-06,
      "loss": 0.56,
      "step": 15906
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.075638964881355,
      "learning_rate": 2.053773402057946e-06,
      "loss": 0.5917,
      "step": 15907
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.97282009949862,
      "learning_rate": 2.0531897899224424e-06,
      "loss": 0.6115,
      "step": 15908
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.334086882654917,
      "learning_rate": 2.0526062392951464e-06,
      "loss": 0.5881,
      "step": 15909
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.290725868814526,
      "learning_rate": 2.0520227501882405e-06,
      "loss": 0.5839,
      "step": 15910
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.131939786175758,
      "learning_rate": 2.0514393226139033e-06,
      "loss": 0.6402,
      "step": 15911
    },
    {
      "epoch": 1.42,
      "grad_norm": 8.939074274030288,
      "learning_rate": 2.0508559565843155e-06,
      "loss": 0.5939,
      "step": 15912
    },
    {
      "epoch": 1.42,
      "grad_norm": 8.724167599367004,
      "learning_rate": 2.050272652111649e-06,
      "loss": 0.5389,
      "step": 15913
    },
    {
      "epoch": 1.42,
      "grad_norm": 7.938776331629604,
      "learning_rate": 2.0496894092080817e-06,
      "loss": 0.5883,
      "step": 15914
    },
    {
      "epoch": 1.42,
      "grad_norm": 8.662818609701498,
      "learning_rate": 2.049106227885786e-06,
      "loss": 0.5688,
      "step": 15915
    },
    {
      "epoch": 1.42,
      "grad_norm": 7.705934058168152,
      "learning_rate": 2.0485231081569356e-06,
      "loss": 0.5698,
      "step": 15916
    },
    {
      "epoch": 1.42,
      "grad_norm": 8.7342496087874,
      "learning_rate": 2.0479400500337023e-06,
      "loss": 0.5903,
      "step": 15917
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.207447837349036,
      "learning_rate": 2.047357053528257e-06,
      "loss": 0.6287,
      "step": 15918
    },
    {
      "epoch": 1.42,
      "grad_norm": 7.153135841280834,
      "learning_rate": 2.046774118652765e-06,
      "loss": 0.6111,
      "step": 15919
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.909478709812742,
      "learning_rate": 2.046191245419396e-06,
      "loss": 0.5348,
      "step": 15920
    },
    {
      "epoch": 1.42,
      "grad_norm": 9.073960735227457,
      "learning_rate": 2.045608433840316e-06,
      "loss": 0.611,
      "step": 15921
    },
    {
      "epoch": 1.42,
      "grad_norm": 8.344778417138222,
      "learning_rate": 2.0450256839276917e-06,
      "loss": 0.636,
      "step": 15922
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.8580318172095724,
      "learning_rate": 2.0444429956936836e-06,
      "loss": 0.5987,
      "step": 15923
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.281737178797946,
      "learning_rate": 2.043860369150455e-06,
      "loss": 0.543,
      "step": 15924
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.164453545096212,
      "learning_rate": 2.043277804310169e-06,
      "loss": 0.5762,
      "step": 15925
    },
    {
      "epoch": 1.42,
      "grad_norm": 7.487062003135259,
      "learning_rate": 2.042695301184982e-06,
      "loss": 0.6334,
      "step": 15926
    },
    {
      "epoch": 1.42,
      "grad_norm": 7.499123742038753,
      "learning_rate": 2.042112859787054e-06,
      "loss": 0.5892,
      "step": 15927
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.770798266330652,
      "learning_rate": 2.0415304801285435e-06,
      "loss": 0.6055,
      "step": 15928
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.15318418995858,
      "learning_rate": 2.0409481622216044e-06,
      "loss": 0.6245,
      "step": 15929
    },
    {
      "epoch": 1.42,
      "grad_norm": 7.013526019282697,
      "learning_rate": 2.040365906078392e-06,
      "loss": 0.6079,
      "step": 15930
    },
    {
      "epoch": 1.42,
      "grad_norm": 8.26226499231029,
      "learning_rate": 2.039783711711062e-06,
      "loss": 0.5882,
      "step": 15931
    },
    {
      "epoch": 1.42,
      "grad_norm": 8.320676343796011,
      "learning_rate": 2.0392015791317625e-06,
      "loss": 0.6587,
      "step": 15932
    },
    {
      "epoch": 1.42,
      "grad_norm": 7.276092686077966,
      "learning_rate": 2.038619508352646e-06,
      "loss": 0.5758,
      "step": 15933
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.22071042921743,
      "learning_rate": 2.0380374993858616e-06,
      "loss": 0.6238,
      "step": 15934
    },
    {
      "epoch": 1.42,
      "grad_norm": 8.794324270514013,
      "learning_rate": 2.037455552243558e-06,
      "loss": 0.5867,
      "step": 15935
    },
    {
      "epoch": 1.42,
      "grad_norm": 8.366755859389485,
      "learning_rate": 2.0368736669378843e-06,
      "loss": 0.6461,
      "step": 15936
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.981158329554912,
      "learning_rate": 2.036291843480981e-06,
      "loss": 0.5854,
      "step": 15937
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.574797131935246,
      "learning_rate": 2.035710081884997e-06,
      "loss": 0.5977,
      "step": 15938
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.170404655659058,
      "learning_rate": 2.0351283821620713e-06,
      "loss": 0.5625,
      "step": 15939
    },
    {
      "epoch": 1.42,
      "grad_norm": 8.191045342917489,
      "learning_rate": 2.0345467443243483e-06,
      "loss": 0.621,
      "step": 15940
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.255107766918525,
      "learning_rate": 2.033965168383968e-06,
      "loss": 0.6033,
      "step": 15941
    },
    {
      "epoch": 1.42,
      "grad_norm": 7.967790026151575,
      "learning_rate": 2.033383654353069e-06,
      "loss": 0.5858,
      "step": 15942
    },
    {
      "epoch": 1.42,
      "grad_norm": 7.412993899795627,
      "learning_rate": 2.0328022022437905e-06,
      "loss": 0.5763,
      "step": 15943
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.285440517820598,
      "learning_rate": 2.032220812068267e-06,
      "loss": 0.5607,
      "step": 15944
    },
    {
      "epoch": 1.42,
      "grad_norm": 7.531616418562098,
      "learning_rate": 2.0316394838386345e-06,
      "loss": 0.647,
      "step": 15945
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.443811596534067,
      "learning_rate": 2.0310582175670274e-06,
      "loss": 0.5986,
      "step": 15946
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.464872370033625,
      "learning_rate": 2.0304770132655777e-06,
      "loss": 0.5271,
      "step": 15947
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.5396999885186595,
      "learning_rate": 2.0298958709464173e-06,
      "loss": 0.564,
      "step": 15948
    },
    {
      "epoch": 1.42,
      "grad_norm": 7.301182256868482,
      "learning_rate": 2.029314790621678e-06,
      "loss": 0.6193,
      "step": 15949
    },
    {
      "epoch": 1.42,
      "grad_norm": 8.062049731388118,
      "learning_rate": 2.028733772303485e-06,
      "loss": 0.597,
      "step": 15950
    },
    {
      "epoch": 1.42,
      "grad_norm": 8.862842157304733,
      "learning_rate": 2.02815281600397e-06,
      "loss": 0.5723,
      "step": 15951
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.4835716760137565,
      "learning_rate": 2.0275719217352542e-06,
      "loss": 0.5954,
      "step": 15952
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.535705313679449,
      "learning_rate": 2.0269910895094645e-06,
      "loss": 0.5529,
      "step": 15953
    },
    {
      "epoch": 1.42,
      "grad_norm": 4.670363611489401,
      "learning_rate": 2.026410319338726e-06,
      "loss": 0.6731,
      "step": 15954
    },
    {
      "epoch": 1.42,
      "grad_norm": 9.311778748298462,
      "learning_rate": 2.02582961123516e-06,
      "loss": 0.6234,
      "step": 15955
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.7934270309874885,
      "learning_rate": 2.0252489652108885e-06,
      "loss": 0.5647,
      "step": 15956
    },
    {
      "epoch": 1.42,
      "grad_norm": 7.5466940418928194,
      "learning_rate": 2.0246683812780295e-06,
      "loss": 0.544,
      "step": 15957
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.498004374543092,
      "learning_rate": 2.0240878594487012e-06,
      "loss": 0.4943,
      "step": 15958
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.362091421214674,
      "learning_rate": 2.0235073997350226e-06,
      "loss": 0.5554,
      "step": 15959
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.362735782348668,
      "learning_rate": 2.0229270021491083e-06,
      "loss": 0.6098,
      "step": 15960
    },
    {
      "epoch": 1.42,
      "grad_norm": 7.797016668315229,
      "learning_rate": 2.0223466667030727e-06,
      "loss": 0.561,
      "step": 15961
    },
    {
      "epoch": 1.42,
      "grad_norm": 8.22249383342666,
      "learning_rate": 2.021766393409032e-06,
      "loss": 0.5865,
      "step": 15962
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.5792888205845275,
      "learning_rate": 2.0211861822790934e-06,
      "loss": 0.5735,
      "step": 15963
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.908960075784189,
      "learning_rate": 2.0206060333253694e-06,
      "loss": 0.5397,
      "step": 15964
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.5144034065023435,
      "learning_rate": 2.02002594655997e-06,
      "loss": 0.5681,
      "step": 15965
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.329578421810544,
      "learning_rate": 2.019445921995005e-06,
      "loss": 0.5775,
      "step": 15966
    },
    {
      "epoch": 1.42,
      "grad_norm": 7.554309050930799,
      "learning_rate": 2.0188659596425775e-06,
      "loss": 0.5543,
      "step": 15967
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.163947760793578,
      "learning_rate": 2.018286059514794e-06,
      "loss": 0.614,
      "step": 15968
    },
    {
      "epoch": 1.42,
      "grad_norm": 4.5336397464142575,
      "learning_rate": 2.0177062216237608e-06,
      "loss": 0.623,
      "step": 15969
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.829518955528446,
      "learning_rate": 2.0171264459815776e-06,
      "loss": 0.5652,
      "step": 15970
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.63283050343638,
      "learning_rate": 2.0165467326003475e-06,
      "loss": 0.5572,
      "step": 15971
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.35501040940004,
      "learning_rate": 2.0159670814921704e-06,
      "loss": 0.5861,
      "step": 15972
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.389063714099731,
      "learning_rate": 2.0153874926691465e-06,
      "loss": 0.56,
      "step": 15973
    },
    {
      "epoch": 1.43,
      "grad_norm": 8.011939506165056,
      "learning_rate": 2.014807966143372e-06,
      "loss": 0.5884,
      "step": 15974
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.411127141636999,
      "learning_rate": 2.0142285019269446e-06,
      "loss": 0.5778,
      "step": 15975
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.454164134292326,
      "learning_rate": 2.013649100031958e-06,
      "loss": 0.5369,
      "step": 15976
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.77670483070315,
      "learning_rate": 2.013069760470506e-06,
      "loss": 0.5849,
      "step": 15977
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.444345996877541,
      "learning_rate": 2.012490483254681e-06,
      "loss": 0.6043,
      "step": 15978
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.955383615343555,
      "learning_rate": 2.0119112683965753e-06,
      "loss": 0.6303,
      "step": 15979
    },
    {
      "epoch": 1.43,
      "grad_norm": 9.93953554011719,
      "learning_rate": 2.011332115908279e-06,
      "loss": 0.6254,
      "step": 15980
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.715957344939878,
      "learning_rate": 2.0107530258018785e-06,
      "loss": 0.5797,
      "step": 15981
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.906450882151817,
      "learning_rate": 2.0101739980894646e-06,
      "loss": 0.5731,
      "step": 15982
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.761568198542431,
      "learning_rate": 2.0095950327831184e-06,
      "loss": 0.609,
      "step": 15983
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.139236744289577,
      "learning_rate": 2.0090161298949273e-06,
      "loss": 0.6449,
      "step": 15984
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.734790040946267,
      "learning_rate": 2.0084372894369743e-06,
      "loss": 0.6186,
      "step": 15985
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.060260444738768,
      "learning_rate": 2.007858511421341e-06,
      "loss": 0.5252,
      "step": 15986
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.471837914167092,
      "learning_rate": 2.0072797958601093e-06,
      "loss": 0.5645,
      "step": 15987
    },
    {
      "epoch": 1.43,
      "grad_norm": 9.889079945000988,
      "learning_rate": 2.0067011427653594e-06,
      "loss": 0.5702,
      "step": 15988
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.580752441840207,
      "learning_rate": 2.0061225521491657e-06,
      "loss": 0.6035,
      "step": 15989
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.852854827390154,
      "learning_rate": 2.005544024023608e-06,
      "loss": 0.6082,
      "step": 15990
    },
    {
      "epoch": 1.43,
      "grad_norm": 8.090219504366223,
      "learning_rate": 2.0049655584007606e-06,
      "loss": 0.6227,
      "step": 15991
    },
    {
      "epoch": 1.43,
      "grad_norm": 10.490579929952798,
      "learning_rate": 2.004387155292698e-06,
      "loss": 0.5784,
      "step": 15992
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.943457950422283,
      "learning_rate": 2.003808814711494e-06,
      "loss": 0.5362,
      "step": 15993
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.591457854030872,
      "learning_rate": 2.0032305366692208e-06,
      "loss": 0.6673,
      "step": 15994
    },
    {
      "epoch": 1.43,
      "grad_norm": 4.47664965051732,
      "learning_rate": 2.002652321177947e-06,
      "loss": 0.6692,
      "step": 15995
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.05004000002149,
      "learning_rate": 2.0020741682497402e-06,
      "loss": 0.5771,
      "step": 15996
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.969202451221911,
      "learning_rate": 2.0014960778966702e-06,
      "loss": 0.5766,
      "step": 15997
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.694087904360678,
      "learning_rate": 2.000918050130803e-06,
      "loss": 0.6168,
      "step": 15998
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.181265260013639,
      "learning_rate": 2.000340084964203e-06,
      "loss": 0.6091,
      "step": 15999
    },
    {
      "epoch": 1.43,
      "grad_norm": 8.354382593445214,
      "learning_rate": 1.9997621824089354e-06,
      "loss": 0.6587,
      "step": 16000
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.814919070473513,
      "learning_rate": 1.999184342477063e-06,
      "loss": 0.6367,
      "step": 16001
    },
    {
      "epoch": 1.43,
      "grad_norm": 8.178997957531067,
      "learning_rate": 1.998606565180644e-06,
      "loss": 0.609,
      "step": 16002
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.1142017619619855,
      "learning_rate": 1.99802885053174e-06,
      "loss": 0.5667,
      "step": 16003
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.1510973925132255,
      "learning_rate": 1.99745119854241e-06,
      "loss": 0.5763,
      "step": 16004
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.415211382322082,
      "learning_rate": 1.9968736092247104e-06,
      "loss": 0.6202,
      "step": 16005
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.617842435271332,
      "learning_rate": 1.9962960825906973e-06,
      "loss": 0.5722,
      "step": 16006
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.425534766451619,
      "learning_rate": 1.995718618652428e-06,
      "loss": 0.5958,
      "step": 16007
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.422413930611282,
      "learning_rate": 1.995141217421951e-06,
      "loss": 0.544,
      "step": 16008
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.908846958222902,
      "learning_rate": 1.994563878911322e-06,
      "loss": 0.5651,
      "step": 16009
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.526125178781459,
      "learning_rate": 1.9939866031325887e-06,
      "loss": 0.5526,
      "step": 16010
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.688337989263283,
      "learning_rate": 1.9934093900978025e-06,
      "loss": 0.5786,
      "step": 16011
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.8149991922480675,
      "learning_rate": 1.9928322398190105e-06,
      "loss": 0.5981,
      "step": 16012
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.979665016270829,
      "learning_rate": 1.99225515230826e-06,
      "loss": 0.634,
      "step": 16013
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.7753638466426835,
      "learning_rate": 1.991678127577598e-06,
      "loss": 0.5941,
      "step": 16014
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.927203650455647,
      "learning_rate": 1.9911011656390654e-06,
      "loss": 0.5612,
      "step": 16015
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.201180940526621,
      "learning_rate": 1.9905242665047063e-06,
      "loss": 0.5753,
      "step": 16016
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.340636232779513,
      "learning_rate": 1.989947430186563e-06,
      "loss": 0.5866,
      "step": 16017
    },
    {
      "epoch": 1.43,
      "grad_norm": 8.068618936842679,
      "learning_rate": 1.9893706566966747e-06,
      "loss": 0.6116,
      "step": 16018
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.001688415482067,
      "learning_rate": 1.988793946047083e-06,
      "loss": 0.5993,
      "step": 16019
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.427248547755618,
      "learning_rate": 1.988217298249821e-06,
      "loss": 0.5758,
      "step": 16020
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.004449054998432,
      "learning_rate": 1.987640713316927e-06,
      "loss": 0.5712,
      "step": 16021
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.648007732663391,
      "learning_rate": 1.987064191260437e-06,
      "loss": 0.5927,
      "step": 16022
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.476487092090985,
      "learning_rate": 1.9864877320923854e-06,
      "loss": 0.5639,
      "step": 16023
    },
    {
      "epoch": 1.43,
      "grad_norm": 11.165538250763875,
      "learning_rate": 1.9859113358248005e-06,
      "loss": 0.6691,
      "step": 16024
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.070824792300607,
      "learning_rate": 1.985335002469716e-06,
      "loss": 0.5576,
      "step": 16025
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.758979443702664,
      "learning_rate": 1.9847587320391633e-06,
      "loss": 0.5754,
      "step": 16026
    },
    {
      "epoch": 1.43,
      "grad_norm": 9.136558313719993,
      "learning_rate": 1.9841825245451666e-06,
      "loss": 0.6132,
      "step": 16027
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.8633714638441825,
      "learning_rate": 1.983606379999756e-06,
      "loss": 0.5656,
      "step": 16028
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.469485251449335,
      "learning_rate": 1.983030298414956e-06,
      "loss": 0.611,
      "step": 16029
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.018749936711844,
      "learning_rate": 1.982454279802791e-06,
      "loss": 0.6068,
      "step": 16030
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.730677162802577,
      "learning_rate": 1.981878324175285e-06,
      "loss": 0.5593,
      "step": 16031
    },
    {
      "epoch": 1.43,
      "grad_norm": 8.506726032742797,
      "learning_rate": 1.9813024315444614e-06,
      "loss": 0.6043,
      "step": 16032
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.172257752684266,
      "learning_rate": 1.9807266019223365e-06,
      "loss": 0.6218,
      "step": 16033
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.587829837895418,
      "learning_rate": 1.980150835320932e-06,
      "loss": 0.5282,
      "step": 16034
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.811627431119983,
      "learning_rate": 1.9795751317522654e-06,
      "loss": 0.606,
      "step": 16035
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.646944549902293,
      "learning_rate": 1.9789994912283533e-06,
      "loss": 0.5311,
      "step": 16036
    },
    {
      "epoch": 1.43,
      "grad_norm": 9.077002930116182,
      "learning_rate": 1.978423913761211e-06,
      "loss": 0.5695,
      "step": 16037
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.081621325272147,
      "learning_rate": 1.977848399362854e-06,
      "loss": 0.499,
      "step": 16038
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.995064415461952,
      "learning_rate": 1.9772729480452928e-06,
      "loss": 0.5853,
      "step": 16039
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.704191357641765,
      "learning_rate": 1.9766975598205384e-06,
      "loss": 0.631,
      "step": 16040
    },
    {
      "epoch": 1.43,
      "grad_norm": 9.318486079708535,
      "learning_rate": 1.976122234700601e-06,
      "loss": 0.577,
      "step": 16041
    },
    {
      "epoch": 1.43,
      "grad_norm": 8.745813370487864,
      "learning_rate": 1.97554697269749e-06,
      "loss": 0.6303,
      "step": 16042
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.259947703411182,
      "learning_rate": 1.9749717738232116e-06,
      "loss": 0.62,
      "step": 16043
    },
    {
      "epoch": 1.43,
      "grad_norm": 8.078871578087316,
      "learning_rate": 1.974396638089773e-06,
      "loss": 0.6131,
      "step": 16044
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.877564616694175,
      "learning_rate": 1.9738215655091814e-06,
      "loss": 0.5971,
      "step": 16045
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.292474836258553,
      "learning_rate": 1.973246556093435e-06,
      "loss": 0.564,
      "step": 16046
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.587989434649652,
      "learning_rate": 1.9726716098545385e-06,
      "loss": 0.5672,
      "step": 16047
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.952588146031541,
      "learning_rate": 1.9720967268044923e-06,
      "loss": 0.5326,
      "step": 16048
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.476657519544199,
      "learning_rate": 1.9715219069552953e-06,
      "loss": 0.5513,
      "step": 16049
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.156090464323989,
      "learning_rate": 1.970947150318947e-06,
      "loss": 0.5781,
      "step": 16050
    },
    {
      "epoch": 1.43,
      "grad_norm": 4.071208452803583,
      "learning_rate": 1.970372456907445e-06,
      "loss": 0.5469,
      "step": 16051
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.610561592928258,
      "learning_rate": 1.969797826732783e-06,
      "loss": 0.5579,
      "step": 16052
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.420699724964787,
      "learning_rate": 1.969223259806954e-06,
      "loss": 0.5577,
      "step": 16053
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.130735590464181,
      "learning_rate": 1.9686487561419516e-06,
      "loss": 0.5012,
      "step": 16054
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.8722203621543,
      "learning_rate": 1.9680743157497685e-06,
      "loss": 0.5495,
      "step": 16055
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.5312657384502435,
      "learning_rate": 1.967499938642394e-06,
      "loss": 0.5716,
      "step": 16056
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.854095566334479,
      "learning_rate": 1.966925624831817e-06,
      "loss": 0.6487,
      "step": 16057
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.962514852774252,
      "learning_rate": 1.966351374330027e-06,
      "loss": 0.5631,
      "step": 16058
    },
    {
      "epoch": 1.43,
      "grad_norm": 4.527788897008529,
      "learning_rate": 1.9657771871490065e-06,
      "loss": 0.5794,
      "step": 16059
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.336361422058109,
      "learning_rate": 1.965203063300743e-06,
      "loss": 0.5561,
      "step": 16060
    },
    {
      "epoch": 1.43,
      "grad_norm": 8.128854984108214,
      "learning_rate": 1.964629002797219e-06,
      "loss": 0.5839,
      "step": 16061
    },
    {
      "epoch": 1.43,
      "grad_norm": 9.191564117125191,
      "learning_rate": 1.9640550056504177e-06,
      "loss": 0.5628,
      "step": 16062
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.902842210308377,
      "learning_rate": 1.963481071872319e-06,
      "loss": 0.6322,
      "step": 16063
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.138003758710454,
      "learning_rate": 1.962907201474905e-06,
      "loss": 0.5696,
      "step": 16064
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.755455774113239,
      "learning_rate": 1.9623333944701508e-06,
      "loss": 0.5611,
      "step": 16065
    },
    {
      "epoch": 1.43,
      "grad_norm": 4.922964259380893,
      "learning_rate": 1.961759650870034e-06,
      "loss": 0.6112,
      "step": 16066
    },
    {
      "epoch": 1.43,
      "grad_norm": 8.33289626854815,
      "learning_rate": 1.9611859706865326e-06,
      "loss": 0.5797,
      "step": 16067
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.652115177844114,
      "learning_rate": 1.960612353931618e-06,
      "loss": 0.5427,
      "step": 16068
    },
    {
      "epoch": 1.43,
      "grad_norm": 8.119048144619292,
      "learning_rate": 1.960038800617264e-06,
      "loss": 0.616,
      "step": 16069
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.7644584434076735,
      "learning_rate": 1.959465310755442e-06,
      "loss": 0.6118,
      "step": 16070
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.912114724017251,
      "learning_rate": 1.958891884358126e-06,
      "loss": 0.5262,
      "step": 16071
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.337982170333464,
      "learning_rate": 1.9583185214372796e-06,
      "loss": 0.6201,
      "step": 16072
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.73587871124126,
      "learning_rate": 1.9577452220048724e-06,
      "loss": 0.5678,
      "step": 16073
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.201379808943579,
      "learning_rate": 1.9571719860728723e-06,
      "loss": 0.549,
      "step": 16074
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.520800806080823,
      "learning_rate": 1.9565988136532427e-06,
      "loss": 0.5347,
      "step": 16075
    },
    {
      "epoch": 1.43,
      "grad_norm": 9.368952313020728,
      "learning_rate": 1.956025704757948e-06,
      "loss": 0.6021,
      "step": 16076
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.558342988822617,
      "learning_rate": 1.955452659398952e-06,
      "loss": 0.5796,
      "step": 16077
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.804483840174626,
      "learning_rate": 1.9548796775882127e-06,
      "loss": 0.6016,
      "step": 16078
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.552701660224498,
      "learning_rate": 1.954306759337692e-06,
      "loss": 0.5762,
      "step": 16079
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.648580824996596,
      "learning_rate": 1.953733904659347e-06,
      "loss": 0.5666,
      "step": 16080
    },
    {
      "epoch": 1.43,
      "grad_norm": 8.245924133529442,
      "learning_rate": 1.953161113565138e-06,
      "loss": 0.5816,
      "step": 16081
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.3165315247797,
      "learning_rate": 1.9525883860670157e-06,
      "loss": 0.6286,
      "step": 16082
    },
    {
      "epoch": 1.43,
      "grad_norm": 4.753637184215156,
      "learning_rate": 1.9520157221769377e-06,
      "loss": 0.5689,
      "step": 16083
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.154671314036131,
      "learning_rate": 1.951443121906858e-06,
      "loss": 0.5419,
      "step": 16084
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.6268725941740785,
      "learning_rate": 1.9508705852687256e-06,
      "loss": 0.5936,
      "step": 16085
    },
    {
      "epoch": 1.44,
      "grad_norm": 7.8525686908671055,
      "learning_rate": 1.9502981122744917e-06,
      "loss": 0.6145,
      "step": 16086
    },
    {
      "epoch": 1.44,
      "grad_norm": 7.118698582413041,
      "learning_rate": 1.9497257029361065e-06,
      "loss": 0.5497,
      "step": 16087
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.50782890728477,
      "learning_rate": 1.949153357265517e-06,
      "loss": 0.5827,
      "step": 16088
    },
    {
      "epoch": 1.44,
      "grad_norm": 7.952054012458648,
      "learning_rate": 1.94858107527467e-06,
      "loss": 0.5881,
      "step": 16089
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.457346033026266,
      "learning_rate": 1.948008856975512e-06,
      "loss": 0.5563,
      "step": 16090
    },
    {
      "epoch": 1.44,
      "grad_norm": 8.159302629396956,
      "learning_rate": 1.9474367023799838e-06,
      "loss": 0.5735,
      "step": 16091
    },
    {
      "epoch": 1.44,
      "grad_norm": 9.278627118449968,
      "learning_rate": 1.9468646115000294e-06,
      "loss": 0.6478,
      "step": 16092
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.722274537841317,
      "learning_rate": 1.94629258434759e-06,
      "loss": 0.6644,
      "step": 16093
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.409989286112338,
      "learning_rate": 1.9457206209346057e-06,
      "loss": 0.6096,
      "step": 16094
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.507476980964579,
      "learning_rate": 1.945148721273016e-06,
      "loss": 0.6325,
      "step": 16095
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.093631343276634,
      "learning_rate": 1.944576885374755e-06,
      "loss": 0.5825,
      "step": 16096
    },
    {
      "epoch": 1.44,
      "grad_norm": 8.418405980313318,
      "learning_rate": 1.9440051132517625e-06,
      "loss": 0.6497,
      "step": 16097
    },
    {
      "epoch": 1.44,
      "grad_norm": 8.087790991613335,
      "learning_rate": 1.9434334049159687e-06,
      "loss": 0.5786,
      "step": 16098
    },
    {
      "epoch": 1.44,
      "grad_norm": 8.029228342610708,
      "learning_rate": 1.942861760379309e-06,
      "loss": 0.6018,
      "step": 16099
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.354159848802613,
      "learning_rate": 1.9422901796537143e-06,
      "loss": 0.5942,
      "step": 16100
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.143911781557417,
      "learning_rate": 1.9417186627511163e-06,
      "loss": 0.6158,
      "step": 16101
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.5863437366695905,
      "learning_rate": 1.9411472096834454e-06,
      "loss": 0.6127,
      "step": 16102
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.57437514664594,
      "learning_rate": 1.940575820462626e-06,
      "loss": 0.6055,
      "step": 16103
    },
    {
      "epoch": 1.44,
      "grad_norm": 7.381167007821189,
      "learning_rate": 1.9400044951005864e-06,
      "loss": 0.5259,
      "step": 16104
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.646824438647992,
      "learning_rate": 1.939433233609252e-06,
      "loss": 0.541,
      "step": 16105
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.310518859822356,
      "learning_rate": 1.9388620360005456e-06,
      "loss": 0.6307,
      "step": 16106
    },
    {
      "epoch": 1.44,
      "grad_norm": 9.302631898596294,
      "learning_rate": 1.938290902286391e-06,
      "loss": 0.5204,
      "step": 16107
    },
    {
      "epoch": 1.44,
      "grad_norm": 7.349506267685692,
      "learning_rate": 1.9377198324787105e-06,
      "loss": 0.5459,
      "step": 16108
    },
    {
      "epoch": 1.44,
      "grad_norm": 7.589995819995555,
      "learning_rate": 1.93714882658942e-06,
      "loss": 0.5735,
      "step": 16109
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.729408798623771,
      "learning_rate": 1.9365778846304424e-06,
      "loss": 0.5196,
      "step": 16110
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.483295237758713,
      "learning_rate": 1.936007006613691e-06,
      "loss": 0.6272,
      "step": 16111
    },
    {
      "epoch": 1.44,
      "grad_norm": 7.027916435351266,
      "learning_rate": 1.9354361925510827e-06,
      "loss": 0.5847,
      "step": 16112
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.829169306150323,
      "learning_rate": 1.9348654424545332e-06,
      "loss": 0.5829,
      "step": 16113
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.095464800969377,
      "learning_rate": 1.9342947563359547e-06,
      "loss": 0.5868,
      "step": 16114
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.01007475858308,
      "learning_rate": 1.9337241342072617e-06,
      "loss": 0.6197,
      "step": 16115
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.314778740530741,
      "learning_rate": 1.93315357608036e-06,
      "loss": 0.6107,
      "step": 16116
    },
    {
      "epoch": 1.44,
      "grad_norm": 7.9648242602804435,
      "learning_rate": 1.9325830819671616e-06,
      "loss": 0.5541,
      "step": 16117
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.83805105346306,
      "learning_rate": 1.9320126518795736e-06,
      "loss": 0.5969,
      "step": 16118
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.433223203516213,
      "learning_rate": 1.931442285829502e-06,
      "loss": 0.5595,
      "step": 16119
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.128100605828463,
      "learning_rate": 1.9308719838288536e-06,
      "loss": 0.5469,
      "step": 16120
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.351078538534099,
      "learning_rate": 1.930301745889533e-06,
      "loss": 0.5726,
      "step": 16121
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.71133699185647,
      "learning_rate": 1.9297315720234388e-06,
      "loss": 0.5841,
      "step": 16122
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.9756983007548135,
      "learning_rate": 1.929161462242474e-06,
      "loss": 0.5981,
      "step": 16123
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.055352926546926,
      "learning_rate": 1.9285914165585414e-06,
      "loss": 0.5546,
      "step": 16124
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.299120709407728,
      "learning_rate": 1.928021434983534e-06,
      "loss": 0.5849,
      "step": 16125
    },
    {
      "epoch": 1.44,
      "grad_norm": 7.6291138094335365,
      "learning_rate": 1.9274515175293524e-06,
      "loss": 0.5898,
      "step": 16126
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.665711381671933,
      "learning_rate": 1.926881664207891e-06,
      "loss": 0.5218,
      "step": 16127
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.582968666889471,
      "learning_rate": 1.926311875031047e-06,
      "loss": 0.5204,
      "step": 16128
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.752641129260513,
      "learning_rate": 1.9257421500107093e-06,
      "loss": 0.5791,
      "step": 16129
    },
    {
      "epoch": 1.44,
      "grad_norm": 7.5674538284703905,
      "learning_rate": 1.9251724891587715e-06,
      "loss": 0.6003,
      "step": 16130
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.5272792370344455,
      "learning_rate": 1.9246028924871245e-06,
      "loss": 0.5628,
      "step": 16131
    },
    {
      "epoch": 1.44,
      "grad_norm": 8.821078697654732,
      "learning_rate": 1.924033360007657e-06,
      "loss": 0.618,
      "step": 16132
    },
    {
      "epoch": 1.44,
      "grad_norm": 9.583921055265614,
      "learning_rate": 1.9234638917322573e-06,
      "loss": 0.5472,
      "step": 16133
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.3151988710740765,
      "learning_rate": 1.922894487672813e-06,
      "loss": 0.581,
      "step": 16134
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.9087635294888985,
      "learning_rate": 1.9223251478412054e-06,
      "loss": 0.5424,
      "step": 16135
    },
    {
      "epoch": 1.44,
      "grad_norm": 7.957929665797019,
      "learning_rate": 1.9217558722493203e-06,
      "loss": 0.5851,
      "step": 16136
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.4928496719962805,
      "learning_rate": 1.9211866609090402e-06,
      "loss": 0.518,
      "step": 16137
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.627151272642593,
      "learning_rate": 1.9206175138322465e-06,
      "loss": 0.5874,
      "step": 16138
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.976620516620107,
      "learning_rate": 1.92004843103082e-06,
      "loss": 0.5755,
      "step": 16139
    },
    {
      "epoch": 1.44,
      "grad_norm": 7.399909894858208,
      "learning_rate": 1.9194794125166353e-06,
      "loss": 0.5857,
      "step": 16140
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.631469273126346,
      "learning_rate": 1.9189104583015738e-06,
      "loss": 0.642,
      "step": 16141
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.349292417263737,
      "learning_rate": 1.918341568397508e-06,
      "loss": 0.6745,
      "step": 16142
    },
    {
      "epoch": 1.44,
      "grad_norm": 9.287650691685396,
      "learning_rate": 1.9177727428163116e-06,
      "loss": 0.5674,
      "step": 16143
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.535127069190486,
      "learning_rate": 1.9172039815698605e-06,
      "loss": 0.5596,
      "step": 16144
    },
    {
      "epoch": 1.44,
      "grad_norm": 7.3608598400116385,
      "learning_rate": 1.9166352846700247e-06,
      "loss": 0.594,
      "step": 16145
    },
    {
      "epoch": 1.44,
      "grad_norm": 7.2952687776653224,
      "learning_rate": 1.916066652128675e-06,
      "loss": 0.614,
      "step": 16146
    },
    {
      "epoch": 1.44,
      "grad_norm": 7.030531196666821,
      "learning_rate": 1.915498083957682e-06,
      "loss": 0.5804,
      "step": 16147
    },
    {
      "epoch": 1.44,
      "grad_norm": 8.831853092974606,
      "learning_rate": 1.914929580168909e-06,
      "loss": 0.5958,
      "step": 16148
    },
    {
      "epoch": 1.44,
      "grad_norm": 9.132433755438306,
      "learning_rate": 1.9143611407742257e-06,
      "loss": 0.6192,
      "step": 16149
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.943563687327087,
      "learning_rate": 1.913792765785496e-06,
      "loss": 0.5069,
      "step": 16150
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.304090945343582,
      "learning_rate": 1.913224455214584e-06,
      "loss": 0.5834,
      "step": 16151
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.586419279145824,
      "learning_rate": 1.9126562090733507e-06,
      "loss": 0.5971,
      "step": 16152
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.050162035846773,
      "learning_rate": 1.9120880273736603e-06,
      "loss": 0.5938,
      "step": 16153
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.563650414849799,
      "learning_rate": 1.9115199101273695e-06,
      "loss": 0.5616,
      "step": 16154
    },
    {
      "epoch": 1.44,
      "grad_norm": 7.636284291413535,
      "learning_rate": 1.9109518573463354e-06,
      "loss": 0.6362,
      "step": 16155
    },
    {
      "epoch": 1.44,
      "grad_norm": 7.198221775188881,
      "learning_rate": 1.9103838690424166e-06,
      "loss": 0.5977,
      "step": 16156
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.273368016075817,
      "learning_rate": 1.9098159452274678e-06,
      "loss": 0.5343,
      "step": 16157
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.93322889929172,
      "learning_rate": 1.9092480859133443e-06,
      "loss": 0.5941,
      "step": 16158
    },
    {
      "epoch": 1.44,
      "grad_norm": 8.30716907413396,
      "learning_rate": 1.908680291111898e-06,
      "loss": 0.6152,
      "step": 16159
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.549750602962268,
      "learning_rate": 1.9081125608349824e-06,
      "loss": 0.6336,
      "step": 16160
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.8441774489818235,
      "learning_rate": 1.907544895094445e-06,
      "loss": 0.4872,
      "step": 16161
    },
    {
      "epoch": 1.44,
      "grad_norm": 7.109103954558313,
      "learning_rate": 1.9069772939021348e-06,
      "loss": 0.6046,
      "step": 16162
    },
    {
      "epoch": 1.44,
      "grad_norm": 8.430142114166337,
      "learning_rate": 1.9064097572699003e-06,
      "loss": 0.6049,
      "step": 16163
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.107053845148975,
      "learning_rate": 1.905842285209587e-06,
      "loss": 0.5462,
      "step": 16164
    },
    {
      "epoch": 1.44,
      "grad_norm": 7.043834041107527,
      "learning_rate": 1.9052748777330403e-06,
      "loss": 0.612,
      "step": 16165
    },
    {
      "epoch": 1.44,
      "grad_norm": 7.526871826125952,
      "learning_rate": 1.904707534852105e-06,
      "loss": 0.596,
      "step": 16166
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.513545024009426,
      "learning_rate": 1.9041402565786192e-06,
      "loss": 0.6133,
      "step": 16167
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.470141525193471,
      "learning_rate": 1.9035730429244275e-06,
      "loss": 0.5266,
      "step": 16168
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.8113304184394,
      "learning_rate": 1.9030058939013658e-06,
      "loss": 0.6146,
      "step": 16169
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.401025910459513,
      "learning_rate": 1.9024388095212737e-06,
      "loss": 0.6084,
      "step": 16170
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.920439319037354,
      "learning_rate": 1.9018717897959882e-06,
      "loss": 0.5585,
      "step": 16171
    },
    {
      "epoch": 1.44,
      "grad_norm": 7.000302358078976,
      "learning_rate": 1.9013048347373442e-06,
      "loss": 0.6036,
      "step": 16172
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.811952167226684,
      "learning_rate": 1.9007379443571771e-06,
      "loss": 0.5648,
      "step": 16173
    },
    {
      "epoch": 1.44,
      "grad_norm": 9.13308017724825,
      "learning_rate": 1.9001711186673166e-06,
      "loss": 0.6274,
      "step": 16174
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.302559392857365,
      "learning_rate": 1.8996043576795957e-06,
      "loss": 0.5865,
      "step": 16175
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.208870529901284,
      "learning_rate": 1.8990376614058443e-06,
      "loss": 0.5988,
      "step": 16176
    },
    {
      "epoch": 1.44,
      "grad_norm": 9.45466974802684,
      "learning_rate": 1.8984710298578902e-06,
      "loss": 0.596,
      "step": 16177
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.253790904557484,
      "learning_rate": 1.8979044630475629e-06,
      "loss": 0.6321,
      "step": 16178
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.139125840940434,
      "learning_rate": 1.8973379609866848e-06,
      "loss": 0.5439,
      "step": 16179
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.641224591318603,
      "learning_rate": 1.8967715236870826e-06,
      "loss": 0.5296,
      "step": 16180
    },
    {
      "epoch": 1.44,
      "grad_norm": 7.346929218555322,
      "learning_rate": 1.8962051511605783e-06,
      "loss": 0.5714,
      "step": 16181
    },
    {
      "epoch": 1.44,
      "grad_norm": 9.506959720583549,
      "learning_rate": 1.895638843418996e-06,
      "loss": 0.5658,
      "step": 16182
    },
    {
      "epoch": 1.44,
      "grad_norm": 7.648380937874914,
      "learning_rate": 1.8950726004741527e-06,
      "loss": 0.6261,
      "step": 16183
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.639583586032304,
      "learning_rate": 1.8945064223378696e-06,
      "loss": 0.5467,
      "step": 16184
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.9475973502144015,
      "learning_rate": 1.8939403090219633e-06,
      "loss": 0.5319,
      "step": 16185
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.505884030895042,
      "learning_rate": 1.8933742605382532e-06,
      "loss": 0.6138,
      "step": 16186
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.6293405090921,
      "learning_rate": 1.89280827689855e-06,
      "loss": 0.5402,
      "step": 16187
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.980753885725871,
      "learning_rate": 1.8922423581146698e-06,
      "loss": 0.5944,
      "step": 16188
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.881411848390724,
      "learning_rate": 1.8916765041984242e-06,
      "loss": 0.605,
      "step": 16189
    },
    {
      "epoch": 1.44,
      "grad_norm": 8.620240360614416,
      "learning_rate": 1.8911107151616243e-06,
      "loss": 0.5987,
      "step": 16190
    },
    {
      "epoch": 1.44,
      "grad_norm": 8.128255522567805,
      "learning_rate": 1.8905449910160818e-06,
      "loss": 0.6345,
      "step": 16191
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.153289749998234,
      "learning_rate": 1.8899793317736015e-06,
      "loss": 0.6097,
      "step": 16192
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.548421330610204,
      "learning_rate": 1.8894137374459915e-06,
      "loss": 0.5874,
      "step": 16193
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.369668652509313,
      "learning_rate": 1.8888482080450577e-06,
      "loss": 0.5899,
      "step": 16194
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.09086283052267,
      "learning_rate": 1.888282743582604e-06,
      "loss": 0.6241,
      "step": 16195
    },
    {
      "epoch": 1.44,
      "grad_norm": 7.302410372012042,
      "learning_rate": 1.8877173440704354e-06,
      "loss": 0.5566,
      "step": 16196
    },
    {
      "epoch": 1.44,
      "grad_norm": 8.219202120431495,
      "learning_rate": 1.88715200952035e-06,
      "loss": 0.5274,
      "step": 16197
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.013474409535862,
      "learning_rate": 1.8865867399441511e-06,
      "loss": 0.5986,
      "step": 16198
    },
    {
      "epoch": 1.45,
      "grad_norm": 7.1722464301011035,
      "learning_rate": 1.8860215353536338e-06,
      "loss": 0.5787,
      "step": 16199
    },
    {
      "epoch": 1.45,
      "grad_norm": 7.844395309159375,
      "learning_rate": 1.8854563957605976e-06,
      "loss": 0.62,
      "step": 16200
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.554476638939635,
      "learning_rate": 1.8848913211768384e-06,
      "loss": 0.5497,
      "step": 16201
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.459544156790797,
      "learning_rate": 1.8843263116141507e-06,
      "loss": 0.6117,
      "step": 16202
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.808625329366765,
      "learning_rate": 1.883761367084328e-06,
      "loss": 0.6122,
      "step": 16203
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.14999665907887,
      "learning_rate": 1.8831964875991643e-06,
      "loss": 0.5561,
      "step": 16204
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.174330758748959,
      "learning_rate": 1.8826316731704463e-06,
      "loss": 0.5617,
      "step": 16205
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.439614200723388,
      "learning_rate": 1.8820669238099654e-06,
      "loss": 0.538,
      "step": 16206
    },
    {
      "epoch": 1.45,
      "grad_norm": 9.89087284063784,
      "learning_rate": 1.881502239529509e-06,
      "loss": 0.5783,
      "step": 16207
    },
    {
      "epoch": 1.45,
      "grad_norm": 7.668635109264112,
      "learning_rate": 1.8809376203408642e-06,
      "loss": 0.5666,
      "step": 16208
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.365650680015462,
      "learning_rate": 1.8803730662558162e-06,
      "loss": 0.5833,
      "step": 16209
    },
    {
      "epoch": 1.45,
      "grad_norm": 8.224612509540579,
      "learning_rate": 1.8798085772861496e-06,
      "loss": 0.5793,
      "step": 16210
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.539691044032317,
      "learning_rate": 1.8792441534436468e-06,
      "loss": 0.5974,
      "step": 16211
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.66181352234482,
      "learning_rate": 1.8786797947400854e-06,
      "loss": 0.5775,
      "step": 16212
    },
    {
      "epoch": 1.45,
      "grad_norm": 7.767900977976966,
      "learning_rate": 1.8781155011872481e-06,
      "loss": 0.6676,
      "step": 16213
    },
    {
      "epoch": 1.45,
      "grad_norm": 8.10846891093178,
      "learning_rate": 1.8775512727969126e-06,
      "loss": 0.53,
      "step": 16214
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.843303839791726,
      "learning_rate": 1.8769871095808568e-06,
      "loss": 0.6477,
      "step": 16215
    },
    {
      "epoch": 1.45,
      "grad_norm": 11.116242735815137,
      "learning_rate": 1.8764230115508553e-06,
      "loss": 0.5565,
      "step": 16216
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.984750152159757,
      "learning_rate": 1.8758589787186848e-06,
      "loss": 0.5834,
      "step": 16217
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.798392813309136,
      "learning_rate": 1.8752950110961144e-06,
      "loss": 0.6079,
      "step": 16218
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.974095943196885,
      "learning_rate": 1.8747311086949183e-06,
      "loss": 0.5875,
      "step": 16219
    },
    {
      "epoch": 1.45,
      "grad_norm": 8.064280177558413,
      "learning_rate": 1.8741672715268655e-06,
      "loss": 0.5885,
      "step": 16220
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.408675590704521,
      "learning_rate": 1.8736034996037255e-06,
      "loss": 0.5702,
      "step": 16221
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.620684192192458,
      "learning_rate": 1.8730397929372661e-06,
      "loss": 0.545,
      "step": 16222
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.715497175690145,
      "learning_rate": 1.8724761515392543e-06,
      "loss": 0.5321,
      "step": 16223
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.924352900239795,
      "learning_rate": 1.8719125754214523e-06,
      "loss": 0.5875,
      "step": 16224
    },
    {
      "epoch": 1.45,
      "grad_norm": 7.657967848338932,
      "learning_rate": 1.8713490645956267e-06,
      "loss": 0.5774,
      "step": 16225
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.6716698882609675,
      "learning_rate": 1.8707856190735358e-06,
      "loss": 0.6347,
      "step": 16226
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.650040207102774,
      "learning_rate": 1.8702222388669421e-06,
      "loss": 0.5769,
      "step": 16227
    },
    {
      "epoch": 1.45,
      "grad_norm": 9.792710190568242,
      "learning_rate": 1.8696589239876057e-06,
      "loss": 0.5057,
      "step": 16228
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.762259665247584,
      "learning_rate": 1.8690956744472832e-06,
      "loss": 0.6208,
      "step": 16229
    },
    {
      "epoch": 1.45,
      "grad_norm": 7.278840283391858,
      "learning_rate": 1.868532490257734e-06,
      "loss": 0.5691,
      "step": 16230
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.285164777848194,
      "learning_rate": 1.8679693714307096e-06,
      "loss": 0.5073,
      "step": 16231
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.828805017336723,
      "learning_rate": 1.8674063179779657e-06,
      "loss": 0.5729,
      "step": 16232
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.066172970964223,
      "learning_rate": 1.8668433299112544e-06,
      "loss": 0.5507,
      "step": 16233
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.665400072061841,
      "learning_rate": 1.8662804072423268e-06,
      "loss": 0.5363,
      "step": 16234
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.870575368774762,
      "learning_rate": 1.8657175499829333e-06,
      "loss": 0.6173,
      "step": 16235
    },
    {
      "epoch": 1.45,
      "grad_norm": 7.252490698573941,
      "learning_rate": 1.8651547581448237e-06,
      "loss": 0.6264,
      "step": 16236
    },
    {
      "epoch": 1.45,
      "grad_norm": 7.638946225178121,
      "learning_rate": 1.864592031739742e-06,
      "loss": 0.5423,
      "step": 16237
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.88989171890227,
      "learning_rate": 1.864029370779435e-06,
      "loss": 0.6124,
      "step": 16238
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.868171815225098,
      "learning_rate": 1.8634667752756469e-06,
      "loss": 0.6207,
      "step": 16239
    },
    {
      "epoch": 1.45,
      "grad_norm": 7.3799433398891185,
      "learning_rate": 1.8629042452401224e-06,
      "loss": 0.6282,
      "step": 16240
    },
    {
      "epoch": 1.45,
      "grad_norm": 9.238081442560404,
      "learning_rate": 1.8623417806846005e-06,
      "loss": 0.5808,
      "step": 16241
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.832227534513327,
      "learning_rate": 1.8617793816208224e-06,
      "loss": 0.5961,
      "step": 16242
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.6071395814177025,
      "learning_rate": 1.861217048060529e-06,
      "loss": 0.598,
      "step": 16243
    },
    {
      "epoch": 1.45,
      "grad_norm": 7.3718730982186855,
      "learning_rate": 1.860654780015454e-06,
      "loss": 0.5493,
      "step": 16244
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.914270501679444,
      "learning_rate": 1.8600925774973355e-06,
      "loss": 0.5434,
      "step": 16245
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.226563034693283,
      "learning_rate": 1.8595304405179082e-06,
      "loss": 0.65,
      "step": 16246
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.953254218395929,
      "learning_rate": 1.8589683690889055e-06,
      "loss": 0.5843,
      "step": 16247
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.9379835556223455,
      "learning_rate": 1.858406363222059e-06,
      "loss": 0.5348,
      "step": 16248
    },
    {
      "epoch": 1.45,
      "grad_norm": 7.907646086263747,
      "learning_rate": 1.857844422929102e-06,
      "loss": 0.5619,
      "step": 16249
    },
    {
      "epoch": 1.45,
      "grad_norm": 7.955682883921274,
      "learning_rate": 1.8572825482217594e-06,
      "loss": 0.6393,
      "step": 16250
    },
    {
      "epoch": 1.45,
      "grad_norm": 8.841507622686544,
      "learning_rate": 1.8567207391117608e-06,
      "loss": 0.5502,
      "step": 16251
    },
    {
      "epoch": 1.45,
      "grad_norm": 7.241821554692742,
      "learning_rate": 1.856158995610834e-06,
      "loss": 0.6621,
      "step": 16252
    },
    {
      "epoch": 1.45,
      "grad_norm": 9.312578413024093,
      "learning_rate": 1.8555973177307025e-06,
      "loss": 0.5866,
      "step": 16253
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.535019360277999,
      "learning_rate": 1.8550357054830926e-06,
      "loss": 0.5086,
      "step": 16254
    },
    {
      "epoch": 1.45,
      "grad_norm": 7.117852987541589,
      "learning_rate": 1.8544741588797238e-06,
      "loss": 0.593,
      "step": 16255
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.748711752900766,
      "learning_rate": 1.8539126779323197e-06,
      "loss": 0.5803,
      "step": 16256
    },
    {
      "epoch": 1.45,
      "grad_norm": 7.967447400939709,
      "learning_rate": 1.8533512626525969e-06,
      "loss": 0.5447,
      "step": 16257
    },
    {
      "epoch": 1.45,
      "grad_norm": 7.98092216328443,
      "learning_rate": 1.852789913052276e-06,
      "loss": 0.5763,
      "step": 16258
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.5204135232556935,
      "learning_rate": 1.8522286291430723e-06,
      "loss": 0.5361,
      "step": 16259
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.9325934787780685,
      "learning_rate": 1.8516674109367034e-06,
      "loss": 0.6009,
      "step": 16260
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.794695591938233,
      "learning_rate": 1.8511062584448824e-06,
      "loss": 0.5354,
      "step": 16261
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.935311666390683,
      "learning_rate": 1.8505451716793238e-06,
      "loss": 0.584,
      "step": 16262
    },
    {
      "epoch": 1.45,
      "grad_norm": 7.791446156505,
      "learning_rate": 1.8499841506517357e-06,
      "loss": 0.57,
      "step": 16263
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.752048910148552,
      "learning_rate": 1.8494231953738306e-06,
      "loss": 0.5412,
      "step": 16264
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.25661586950429,
      "learning_rate": 1.8488623058573163e-06,
      "loss": 0.6287,
      "step": 16265
    },
    {
      "epoch": 1.45,
      "grad_norm": 7.008435066728198,
      "learning_rate": 1.8483014821139005e-06,
      "loss": 0.5814,
      "step": 16266
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.561768660655434,
      "learning_rate": 1.8477407241552908e-06,
      "loss": 0.5612,
      "step": 16267
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.767940836106502,
      "learning_rate": 1.8471800319931886e-06,
      "loss": 0.6398,
      "step": 16268
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.3772597863636085,
      "learning_rate": 1.8466194056393e-06,
      "loss": 0.601,
      "step": 16269
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.307017242632708,
      "learning_rate": 1.8460588451053246e-06,
      "loss": 0.6452,
      "step": 16270
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.140670176685203,
      "learning_rate": 1.8454983504029633e-06,
      "loss": 0.559,
      "step": 16271
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.501170088494278,
      "learning_rate": 1.8449379215439156e-06,
      "loss": 0.5778,
      "step": 16272
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.7414366170538775,
      "learning_rate": 1.8443775585398793e-06,
      "loss": 0.5814,
      "step": 16273
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.480585899924761,
      "learning_rate": 1.8438172614025523e-06,
      "loss": 0.5814,
      "step": 16274
    },
    {
      "epoch": 1.45,
      "grad_norm": 7.171936874183452,
      "learning_rate": 1.843257030143627e-06,
      "loss": 0.5545,
      "step": 16275
    },
    {
      "epoch": 1.45,
      "grad_norm": 9.313044327922912,
      "learning_rate": 1.8426968647747978e-06,
      "loss": 0.6089,
      "step": 16276
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.9524844467010105,
      "learning_rate": 1.8421367653077571e-06,
      "loss": 0.6083,
      "step": 16277
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.388680558907984,
      "learning_rate": 1.8415767317541954e-06,
      "loss": 0.4958,
      "step": 16278
    },
    {
      "epoch": 1.45,
      "grad_norm": 9.612614162241591,
      "learning_rate": 1.841016764125803e-06,
      "loss": 0.5944,
      "step": 16279
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.197071401254663,
      "learning_rate": 1.8404568624342696e-06,
      "loss": 0.6057,
      "step": 16280
    },
    {
      "epoch": 1.45,
      "grad_norm": 8.788672765885801,
      "learning_rate": 1.8398970266912775e-06,
      "loss": 0.5863,
      "step": 16281
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.34432337201427,
      "learning_rate": 1.839337256908515e-06,
      "loss": 0.5974,
      "step": 16282
    },
    {
      "epoch": 1.45,
      "grad_norm": 8.03130411353677,
      "learning_rate": 1.838777553097667e-06,
      "loss": 0.5822,
      "step": 16283
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.654015060021429,
      "learning_rate": 1.8382179152704133e-06,
      "loss": 0.5449,
      "step": 16284
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.124565549144169,
      "learning_rate": 1.8376583434384359e-06,
      "loss": 0.5697,
      "step": 16285
    },
    {
      "epoch": 1.45,
      "grad_norm": 9.105481997250518,
      "learning_rate": 1.8370988376134152e-06,
      "loss": 0.6279,
      "step": 16286
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.6802535292513054,
      "learning_rate": 1.8365393978070318e-06,
      "loss": 0.5565,
      "step": 16287
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.46991484681146,
      "learning_rate": 1.8359800240309589e-06,
      "loss": 0.6244,
      "step": 16288
    },
    {
      "epoch": 1.45,
      "grad_norm": 7.281604875624274,
      "learning_rate": 1.8354207162968734e-06,
      "loss": 0.5676,
      "step": 16289
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.87015468704604,
      "learning_rate": 1.8348614746164507e-06,
      "loss": 0.5689,
      "step": 16290
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.279443504218136,
      "learning_rate": 1.8343022990013626e-06,
      "loss": 0.5643,
      "step": 16291
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.119794713215712,
      "learning_rate": 1.8337431894632813e-06,
      "loss": 0.5389,
      "step": 16292
    },
    {
      "epoch": 1.45,
      "grad_norm": 7.1096547347079975,
      "learning_rate": 1.8331841460138788e-06,
      "loss": 0.5856,
      "step": 16293
    },
    {
      "epoch": 1.45,
      "grad_norm": 7.568462879772909,
      "learning_rate": 1.8326251686648206e-06,
      "loss": 0.565,
      "step": 16294
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.471584211937221,
      "learning_rate": 1.8320662574277752e-06,
      "loss": 0.6153,
      "step": 16295
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.9926222575426875,
      "learning_rate": 1.8315074123144095e-06,
      "loss": 0.615,
      "step": 16296
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.9283244648809745,
      "learning_rate": 1.830948633336389e-06,
      "loss": 0.5751,
      "step": 16297
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.498145594968313,
      "learning_rate": 1.8303899205053737e-06,
      "loss": 0.5784,
      "step": 16298
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.365005395258117,
      "learning_rate": 1.8298312738330282e-06,
      "loss": 0.5806,
      "step": 16299
    },
    {
      "epoch": 1.45,
      "grad_norm": 7.208699418245424,
      "learning_rate": 1.8292726933310134e-06,
      "loss": 0.5831,
      "step": 16300
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.86267833441473,
      "learning_rate": 1.8287141790109863e-06,
      "loss": 0.5744,
      "step": 16301
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.9391496907998285,
      "learning_rate": 1.8281557308846048e-06,
      "loss": 0.5898,
      "step": 16302
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.727900753601115,
      "learning_rate": 1.8275973489635269e-06,
      "loss": 0.6217,
      "step": 16303
    },
    {
      "epoch": 1.45,
      "grad_norm": 7.776533182953649,
      "learning_rate": 1.8270390332594069e-06,
      "loss": 0.5477,
      "step": 16304
    },
    {
      "epoch": 1.45,
      "grad_norm": 8.739776611514607,
      "learning_rate": 1.8264807837838984e-06,
      "loss": 0.5997,
      "step": 16305
    },
    {
      "epoch": 1.45,
      "grad_norm": 7.881110880699467,
      "learning_rate": 1.8259226005486546e-06,
      "loss": 0.6152,
      "step": 16306
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.79267813934058,
      "learning_rate": 1.8253644835653246e-06,
      "loss": 0.6278,
      "step": 16307
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.963118010119635,
      "learning_rate": 1.8248064328455584e-06,
      "loss": 0.5867,
      "step": 16308
    },
    {
      "epoch": 1.45,
      "grad_norm": 7.509784420871031,
      "learning_rate": 1.8242484484010043e-06,
      "loss": 0.5875,
      "step": 16309
    },
    {
      "epoch": 1.46,
      "grad_norm": 8.761721715297359,
      "learning_rate": 1.823690530243309e-06,
      "loss": 0.6384,
      "step": 16310
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.8677910886859275,
      "learning_rate": 1.823132678384118e-06,
      "loss": 0.6004,
      "step": 16311
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.787450160647995,
      "learning_rate": 1.8225748928350768e-06,
      "loss": 0.5489,
      "step": 16312
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.620884479461966,
      "learning_rate": 1.8220171736078262e-06,
      "loss": 0.5606,
      "step": 16313
    },
    {
      "epoch": 1.46,
      "grad_norm": 8.224192459143556,
      "learning_rate": 1.8214595207140057e-06,
      "loss": 0.607,
      "step": 16314
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.3780925978028895,
      "learning_rate": 1.8209019341652568e-06,
      "loss": 0.5179,
      "step": 16315
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.35100096296534,
      "learning_rate": 1.820344413973218e-06,
      "loss": 0.5825,
      "step": 16316
    },
    {
      "epoch": 1.46,
      "grad_norm": 8.018503118362428,
      "learning_rate": 1.8197869601495266e-06,
      "loss": 0.5751,
      "step": 16317
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.205054550624041,
      "learning_rate": 1.8192295727058173e-06,
      "loss": 0.6155,
      "step": 16318
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.170218966387842,
      "learning_rate": 1.818672251653727e-06,
      "loss": 0.5774,
      "step": 16319
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.319257654567645,
      "learning_rate": 1.8181149970048845e-06,
      "loss": 0.5992,
      "step": 16320
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.99747917922349,
      "learning_rate": 1.8175578087709235e-06,
      "loss": 0.6031,
      "step": 16321
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.180349885510405,
      "learning_rate": 1.8170006869634732e-06,
      "loss": 0.6224,
      "step": 16322
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.889343604751983,
      "learning_rate": 1.8164436315941636e-06,
      "loss": 0.6623,
      "step": 16323
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.35165286231498,
      "learning_rate": 1.8158866426746208e-06,
      "loss": 0.5968,
      "step": 16324
    },
    {
      "epoch": 1.46,
      "grad_norm": 4.676287046207306,
      "learning_rate": 1.8153297202164733e-06,
      "loss": 0.5593,
      "step": 16325
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.742830516677,
      "learning_rate": 1.8147728642313433e-06,
      "loss": 0.6331,
      "step": 16326
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.480131439749104,
      "learning_rate": 1.8142160747308529e-06,
      "loss": 0.5986,
      "step": 16327
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.129823294366671,
      "learning_rate": 1.8136593517266248e-06,
      "loss": 0.5818,
      "step": 16328
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.09663871287527,
      "learning_rate": 1.8131026952302799e-06,
      "loss": 0.5814,
      "step": 16329
    },
    {
      "epoch": 1.46,
      "grad_norm": 10.536229061324756,
      "learning_rate": 1.8125461052534372e-06,
      "loss": 0.5919,
      "step": 16330
    },
    {
      "epoch": 1.46,
      "grad_norm": 9.298875508631372,
      "learning_rate": 1.8119895818077139e-06,
      "loss": 0.6211,
      "step": 16331
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.407934867337465,
      "learning_rate": 1.8114331249047284e-06,
      "loss": 0.5521,
      "step": 16332
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.049368171249713,
      "learning_rate": 1.8108767345560917e-06,
      "loss": 0.6387,
      "step": 16333
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.777152072826537,
      "learning_rate": 1.810320410773419e-06,
      "loss": 0.5921,
      "step": 16334
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.915295872546212,
      "learning_rate": 1.809764153568323e-06,
      "loss": 0.5879,
      "step": 16335
    },
    {
      "epoch": 1.46,
      "grad_norm": 8.24097664423097,
      "learning_rate": 1.8092079629524134e-06,
      "loss": 0.5875,
      "step": 16336
    },
    {
      "epoch": 1.46,
      "grad_norm": 8.522546151274087,
      "learning_rate": 1.8086518389372997e-06,
      "loss": 0.6018,
      "step": 16337
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.9793097194203435,
      "learning_rate": 1.8080957815345917e-06,
      "loss": 0.6312,
      "step": 16338
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.3268046494974355,
      "learning_rate": 1.8075397907558928e-06,
      "loss": 0.6065,
      "step": 16339
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.996413440278465,
      "learning_rate": 1.8069838666128092e-06,
      "loss": 0.6328,
      "step": 16340
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.0023634398639105,
      "learning_rate": 1.806428009116946e-06,
      "loss": 0.5738,
      "step": 16341
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.2591221938597625,
      "learning_rate": 1.8058722182799033e-06,
      "loss": 0.5711,
      "step": 16342
    },
    {
      "epoch": 1.46,
      "grad_norm": 8.819464184656283,
      "learning_rate": 1.8053164941132828e-06,
      "loss": 0.6165,
      "step": 16343
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.005922404526672,
      "learning_rate": 1.804760836628684e-06,
      "loss": 0.5655,
      "step": 16344
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.4774953453816915,
      "learning_rate": 1.8042052458377074e-06,
      "loss": 0.5583,
      "step": 16345
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.357042728700943,
      "learning_rate": 1.8036497217519455e-06,
      "loss": 0.5772,
      "step": 16346
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.054178568447883,
      "learning_rate": 1.8030942643829962e-06,
      "loss": 0.5189,
      "step": 16347
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.416366900940011,
      "learning_rate": 1.8025388737424525e-06,
      "loss": 0.677,
      "step": 16348
    },
    {
      "epoch": 1.46,
      "grad_norm": 8.270732316699222,
      "learning_rate": 1.801983549841908e-06,
      "loss": 0.6107,
      "step": 16349
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.4013031746739335,
      "learning_rate": 1.8014282926929549e-06,
      "loss": 0.5902,
      "step": 16350
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.866427259821172,
      "learning_rate": 1.8008731023071795e-06,
      "loss": 0.568,
      "step": 16351
    },
    {
      "epoch": 1.46,
      "grad_norm": 4.7346255860178825,
      "learning_rate": 1.8003179786961723e-06,
      "loss": 0.5683,
      "step": 16352
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.613088524420837,
      "learning_rate": 1.7997629218715202e-06,
      "loss": 0.6077,
      "step": 16353
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.65503981856747,
      "learning_rate": 1.7992079318448086e-06,
      "loss": 0.5411,
      "step": 16354
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.822945030232632,
      "learning_rate": 1.7986530086276233e-06,
      "loss": 0.6039,
      "step": 16355
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.384875719425806,
      "learning_rate": 1.7980981522315444e-06,
      "loss": 0.6603,
      "step": 16356
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.484024236540993,
      "learning_rate": 1.797543362668156e-06,
      "loss": 0.5426,
      "step": 16357
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.887113584972544,
      "learning_rate": 1.7969886399490349e-06,
      "loss": 0.5579,
      "step": 16358
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.014255035696754,
      "learning_rate": 1.7964339840857613e-06,
      "loss": 0.6259,
      "step": 16359
    },
    {
      "epoch": 1.46,
      "grad_norm": 8.542543382145876,
      "learning_rate": 1.795879395089913e-06,
      "loss": 0.6329,
      "step": 16360
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.798775541274467,
      "learning_rate": 1.7953248729730649e-06,
      "loss": 0.6272,
      "step": 16361
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.5989667883147956,
      "learning_rate": 1.7947704177467922e-06,
      "loss": 0.5784,
      "step": 16362
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.907383672694986,
      "learning_rate": 1.7942160294226696e-06,
      "loss": 0.6281,
      "step": 16363
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.857906127203873,
      "learning_rate": 1.7936617080122647e-06,
      "loss": 0.5778,
      "step": 16364
    },
    {
      "epoch": 1.46,
      "grad_norm": 9.107736716023513,
      "learning_rate": 1.7931074535271504e-06,
      "loss": 0.5489,
      "step": 16365
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.090924376955901,
      "learning_rate": 1.7925532659788946e-06,
      "loss": 0.5684,
      "step": 16366
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.130619380261054,
      "learning_rate": 1.7919991453790658e-06,
      "loss": 0.6047,
      "step": 16367
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.131546780775847,
      "learning_rate": 1.791445091739229e-06,
      "loss": 0.6121,
      "step": 16368
    },
    {
      "epoch": 1.46,
      "grad_norm": 4.7780971017423655,
      "learning_rate": 1.790891105070951e-06,
      "loss": 0.5838,
      "step": 16369
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.172190985316813,
      "learning_rate": 1.7903371853857932e-06,
      "loss": 0.6455,
      "step": 16370
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.550905781183034,
      "learning_rate": 1.7897833326953163e-06,
      "loss": 0.6034,
      "step": 16371
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.159074538253677,
      "learning_rate": 1.7892295470110816e-06,
      "loss": 0.6053,
      "step": 16372
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.778393793574685,
      "learning_rate": 1.7886758283446493e-06,
      "loss": 0.5864,
      "step": 16373
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.40541597988544,
      "learning_rate": 1.7881221767075758e-06,
      "loss": 0.5494,
      "step": 16374
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.985553059936738,
      "learning_rate": 1.7875685921114183e-06,
      "loss": 0.5458,
      "step": 16375
    },
    {
      "epoch": 1.46,
      "grad_norm": 8.403783854063095,
      "learning_rate": 1.7870150745677328e-06,
      "loss": 0.5148,
      "step": 16376
    },
    {
      "epoch": 1.46,
      "grad_norm": 8.257542271760794,
      "learning_rate": 1.7864616240880695e-06,
      "loss": 0.6432,
      "step": 16377
    },
    {
      "epoch": 1.46,
      "grad_norm": 10.090677851824813,
      "learning_rate": 1.7859082406839829e-06,
      "loss": 0.6054,
      "step": 16378
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.963683683681956,
      "learning_rate": 1.7853549243670226e-06,
      "loss": 0.6161,
      "step": 16379
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.34587301430023,
      "learning_rate": 1.784801675148738e-06,
      "loss": 0.6177,
      "step": 16380
    },
    {
      "epoch": 1.46,
      "grad_norm": 8.502210959767844,
      "learning_rate": 1.7842484930406777e-06,
      "loss": 0.5891,
      "step": 16381
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.345200328663422,
      "learning_rate": 1.7836953780543892e-06,
      "loss": 0.5884,
      "step": 16382
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.162294791378467,
      "learning_rate": 1.7831423302014144e-06,
      "loss": 0.5762,
      "step": 16383
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.757664762506362,
      "learning_rate": 1.7825893494933004e-06,
      "loss": 0.6098,
      "step": 16384
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.5982109921185685,
      "learning_rate": 1.7820364359415866e-06,
      "loss": 0.6148,
      "step": 16385
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.4861150503482525,
      "learning_rate": 1.7814835895578143e-06,
      "loss": 0.5619,
      "step": 16386
    },
    {
      "epoch": 1.46,
      "grad_norm": 9.010386869416386,
      "learning_rate": 1.780930810353524e-06,
      "loss": 0.5517,
      "step": 16387
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.488620912388788,
      "learning_rate": 1.780378098340253e-06,
      "loss": 0.5748,
      "step": 16388
    },
    {
      "epoch": 1.46,
      "grad_norm": 9.37634697268703,
      "learning_rate": 1.77982545352954e-06,
      "loss": 0.5997,
      "step": 16389
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.747776695134638,
      "learning_rate": 1.7792728759329175e-06,
      "loss": 0.5372,
      "step": 16390
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.74506048532081,
      "learning_rate": 1.77872036556192e-06,
      "loss": 0.6079,
      "step": 16391
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.552135228030642,
      "learning_rate": 1.7781679224280802e-06,
      "loss": 0.5907,
      "step": 16392
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.350695114090226,
      "learning_rate": 1.7776155465429295e-06,
      "loss": 0.5805,
      "step": 16393
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.917067070581846,
      "learning_rate": 1.7770632379179974e-06,
      "loss": 0.5582,
      "step": 16394
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.370524421998532,
      "learning_rate": 1.7765109965648142e-06,
      "loss": 0.582,
      "step": 16395
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.2700719100010165,
      "learning_rate": 1.7759588224949027e-06,
      "loss": 0.5602,
      "step": 16396
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.855570517977772,
      "learning_rate": 1.7754067157197903e-06,
      "loss": 0.5916,
      "step": 16397
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.285386008060553,
      "learning_rate": 1.7748546762510022e-06,
      "loss": 0.5792,
      "step": 16398
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.296600356846596,
      "learning_rate": 1.7743027041000583e-06,
      "loss": 0.6275,
      "step": 16399
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.478797359030748,
      "learning_rate": 1.7737507992784813e-06,
      "loss": 0.5353,
      "step": 16400
    },
    {
      "epoch": 1.46,
      "grad_norm": 8.339159481991633,
      "learning_rate": 1.7731989617977913e-06,
      "loss": 0.5457,
      "step": 16401
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.755782852412119,
      "learning_rate": 1.7726471916695075e-06,
      "loss": 0.5983,
      "step": 16402
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.488101470458473,
      "learning_rate": 1.7720954889051446e-06,
      "loss": 0.5713,
      "step": 16403
    },
    {
      "epoch": 1.46,
      "grad_norm": 8.369793803829351,
      "learning_rate": 1.7715438535162194e-06,
      "loss": 0.5355,
      "step": 16404
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.696974310062657,
      "learning_rate": 1.770992285514246e-06,
      "loss": 0.5587,
      "step": 16405
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.2362624015491805,
      "learning_rate": 1.770440784910737e-06,
      "loss": 0.5982,
      "step": 16406
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.959145627490885,
      "learning_rate": 1.7698893517172044e-06,
      "loss": 0.5825,
      "step": 16407
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.4254572471326235,
      "learning_rate": 1.7693379859451594e-06,
      "loss": 0.5254,
      "step": 16408
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.312904735368366,
      "learning_rate": 1.7687866876061067e-06,
      "loss": 0.598,
      "step": 16409
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.942059761584586,
      "learning_rate": 1.7682354567115561e-06,
      "loss": 0.5775,
      "step": 16410
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.508029409927698,
      "learning_rate": 1.7676842932730126e-06,
      "loss": 0.6087,
      "step": 16411
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.82634930122898,
      "learning_rate": 1.7671331973019812e-06,
      "loss": 0.5299,
      "step": 16412
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.908623143678013,
      "learning_rate": 1.766582168809966e-06,
      "loss": 0.6611,
      "step": 16413
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.988879632618105,
      "learning_rate": 1.7660312078084651e-06,
      "loss": 0.5802,
      "step": 16414
    },
    {
      "epoch": 1.46,
      "grad_norm": 8.511844104203053,
      "learning_rate": 1.7654803143089822e-06,
      "loss": 0.6115,
      "step": 16415
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.095299196093812,
      "learning_rate": 1.764929488323013e-06,
      "loss": 0.6084,
      "step": 16416
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.884386276805763,
      "learning_rate": 1.764378729862055e-06,
      "loss": 0.5469,
      "step": 16417
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.265833856796682,
      "learning_rate": 1.7638280389376062e-06,
      "loss": 0.5899,
      "step": 16418
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.113083319442705,
      "learning_rate": 1.7632774155611594e-06,
      "loss": 0.6116,
      "step": 16419
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.526903764825625,
      "learning_rate": 1.7627268597442089e-06,
      "loss": 0.5557,
      "step": 16420
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.449843735274143,
      "learning_rate": 1.7621763714982466e-06,
      "loss": 0.5136,
      "step": 16421
    },
    {
      "epoch": 1.47,
      "grad_norm": 4.935520975617091,
      "learning_rate": 1.7616259508347606e-06,
      "loss": 0.6259,
      "step": 16422
    },
    {
      "epoch": 1.47,
      "grad_norm": 8.446641740268513,
      "learning_rate": 1.7610755977652405e-06,
      "loss": 0.5741,
      "step": 16423
    },
    {
      "epoch": 1.47,
      "grad_norm": 4.523522043225614,
      "learning_rate": 1.7605253123011745e-06,
      "loss": 0.5812,
      "step": 16424
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.100239845336078,
      "learning_rate": 1.7599750944540484e-06,
      "loss": 0.6504,
      "step": 16425
    },
    {
      "epoch": 1.47,
      "grad_norm": 10.508854736111482,
      "learning_rate": 1.7594249442353484e-06,
      "loss": 0.5553,
      "step": 16426
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.95883471653413,
      "learning_rate": 1.7588748616565543e-06,
      "loss": 0.5536,
      "step": 16427
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.4658683026330745,
      "learning_rate": 1.7583248467291508e-06,
      "loss": 0.6096,
      "step": 16428
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.391304583632329,
      "learning_rate": 1.7577748994646154e-06,
      "loss": 0.6084,
      "step": 16429
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.868654446102574,
      "learning_rate": 1.7572250198744289e-06,
      "loss": 0.5564,
      "step": 16430
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.159560403993126,
      "learning_rate": 1.7566752079700683e-06,
      "loss": 0.6009,
      "step": 16431
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.848815766143723,
      "learning_rate": 1.7561254637630104e-06,
      "loss": 0.5567,
      "step": 16432
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.564617998226331,
      "learning_rate": 1.7555757872647305e-06,
      "loss": 0.5348,
      "step": 16433
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.0063722694897645,
      "learning_rate": 1.7550261784866996e-06,
      "loss": 0.5491,
      "step": 16434
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.641869635127833,
      "learning_rate": 1.754476637440391e-06,
      "loss": 0.5988,
      "step": 16435
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.3912221846705295,
      "learning_rate": 1.7539271641372745e-06,
      "loss": 0.5293,
      "step": 16436
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.379154189457372,
      "learning_rate": 1.75337775858882e-06,
      "loss": 0.6103,
      "step": 16437
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.747307379007315,
      "learning_rate": 1.752828420806495e-06,
      "loss": 0.6146,
      "step": 16438
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.141754531769357,
      "learning_rate": 1.7522791508017668e-06,
      "loss": 0.5775,
      "step": 16439
    },
    {
      "epoch": 1.47,
      "grad_norm": 8.071004271213777,
      "learning_rate": 1.7517299485860971e-06,
      "loss": 0.486,
      "step": 16440
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.82141363190007,
      "learning_rate": 1.751180814170952e-06,
      "loss": 0.6774,
      "step": 16441
    },
    {
      "epoch": 1.47,
      "grad_norm": 8.419992518976178,
      "learning_rate": 1.7506317475677936e-06,
      "loss": 0.6162,
      "step": 16442
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.605130950790347,
      "learning_rate": 1.75008274878808e-06,
      "loss": 0.6248,
      "step": 16443
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.237289316305557,
      "learning_rate": 1.7495338178432714e-06,
      "loss": 0.5815,
      "step": 16444
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.6334659407380725,
      "learning_rate": 1.7489849547448267e-06,
      "loss": 0.5891,
      "step": 16445
    },
    {
      "epoch": 1.47,
      "grad_norm": 8.116689993732846,
      "learning_rate": 1.7484361595042026e-06,
      "loss": 0.5295,
      "step": 16446
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.211655853298133,
      "learning_rate": 1.7478874321328516e-06,
      "loss": 0.6426,
      "step": 16447
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.115320588904381,
      "learning_rate": 1.747338772642228e-06,
      "loss": 0.6178,
      "step": 16448
    },
    {
      "epoch": 1.47,
      "grad_norm": 12.072628340157333,
      "learning_rate": 1.7467901810437842e-06,
      "loss": 0.6188,
      "step": 16449
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.0148729756478065,
      "learning_rate": 1.7462416573489716e-06,
      "loss": 0.5996,
      "step": 16450
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.552429292404611,
      "learning_rate": 1.7456932015692384e-06,
      "loss": 0.5699,
      "step": 16451
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.783157919566073,
      "learning_rate": 1.745144813716035e-06,
      "loss": 0.5847,
      "step": 16452
    },
    {
      "epoch": 1.47,
      "grad_norm": 8.889362861991899,
      "learning_rate": 1.744596493800803e-06,
      "loss": 0.6019,
      "step": 16453
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.9732792788612254,
      "learning_rate": 1.7440482418349908e-06,
      "loss": 0.5664,
      "step": 16454
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.450408625638357,
      "learning_rate": 1.7435000578300404e-06,
      "loss": 0.6158,
      "step": 16455
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.840335244916685,
      "learning_rate": 1.7429519417973972e-06,
      "loss": 0.5827,
      "step": 16456
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.983676486315571,
      "learning_rate": 1.7424038937484973e-06,
      "loss": 0.5873,
      "step": 16457
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.182682482052349,
      "learning_rate": 1.7418559136947822e-06,
      "loss": 0.5376,
      "step": 16458
    },
    {
      "epoch": 1.47,
      "grad_norm": 9.228246701456108,
      "learning_rate": 1.7413080016476908e-06,
      "loss": 0.5265,
      "step": 16459
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.542901373354848,
      "learning_rate": 1.7407601576186573e-06,
      "loss": 0.659,
      "step": 16460
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.100189893437488,
      "learning_rate": 1.740212381619118e-06,
      "loss": 0.5765,
      "step": 16461
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.2278794175283245,
      "learning_rate": 1.739664673660506e-06,
      "loss": 0.5939,
      "step": 16462
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.624046773275481,
      "learning_rate": 1.7391170337542545e-06,
      "loss": 0.6329,
      "step": 16463
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.5140356523056155,
      "learning_rate": 1.7385694619117933e-06,
      "loss": 0.588,
      "step": 16464
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.5590733528292775,
      "learning_rate": 1.738021958144554e-06,
      "loss": 0.6249,
      "step": 16465
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.768323764133486,
      "learning_rate": 1.737474522463961e-06,
      "loss": 0.5638,
      "step": 16466
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.3637754242454925,
      "learning_rate": 1.7369271548814427e-06,
      "loss": 0.5967,
      "step": 16467
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.9021058065854755,
      "learning_rate": 1.7363798554084233e-06,
      "loss": 0.5742,
      "step": 16468
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.388257866318392,
      "learning_rate": 1.7358326240563278e-06,
      "loss": 0.5997,
      "step": 16469
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.920003303595473,
      "learning_rate": 1.7352854608365798e-06,
      "loss": 0.6088,
      "step": 16470
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.591527304914028,
      "learning_rate": 1.734738365760596e-06,
      "loss": 0.5296,
      "step": 16471
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.042667818418202,
      "learning_rate": 1.7341913388398001e-06,
      "loss": 0.5629,
      "step": 16472
    },
    {
      "epoch": 1.47,
      "grad_norm": 9.987595367452592,
      "learning_rate": 1.7336443800856063e-06,
      "loss": 0.5593,
      "step": 16473
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.408967023544354,
      "learning_rate": 1.7330974895094326e-06,
      "loss": 0.5314,
      "step": 16474
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.927844212267987,
      "learning_rate": 1.7325506671226944e-06,
      "loss": 0.6658,
      "step": 16475
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.662738532041472,
      "learning_rate": 1.7320039129368054e-06,
      "loss": 0.5508,
      "step": 16476
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.366333622631743,
      "learning_rate": 1.731457226963178e-06,
      "loss": 0.5661,
      "step": 16477
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.4623796979354635,
      "learning_rate": 1.730910609213225e-06,
      "loss": 0.5793,
      "step": 16478
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.251000645718733,
      "learning_rate": 1.7303640596983517e-06,
      "loss": 0.5397,
      "step": 16479
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.894155544410825,
      "learning_rate": 1.7298175784299687e-06,
      "loss": 0.5972,
      "step": 16480
    },
    {
      "epoch": 1.47,
      "grad_norm": 8.253088409171935,
      "learning_rate": 1.729271165419482e-06,
      "loss": 0.5873,
      "step": 16481
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.826955125219654,
      "learning_rate": 1.7287248206782969e-06,
      "loss": 0.5556,
      "step": 16482
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.705706379338986,
      "learning_rate": 1.7281785442178172e-06,
      "loss": 0.5436,
      "step": 16483
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.229898458422859,
      "learning_rate": 1.7276323360494468e-06,
      "loss": 0.5525,
      "step": 16484
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.911040823939299,
      "learning_rate": 1.7270861961845853e-06,
      "loss": 0.6267,
      "step": 16485
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.364408725427549,
      "learning_rate": 1.7265401246346297e-06,
      "loss": 0.5544,
      "step": 16486
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.660698259834166,
      "learning_rate": 1.725994121410981e-06,
      "loss": 0.5935,
      "step": 16487
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.0981224498678115,
      "learning_rate": 1.7254481865250343e-06,
      "loss": 0.5861,
      "step": 16488
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.594231949013769,
      "learning_rate": 1.724902319988186e-06,
      "loss": 0.5656,
      "step": 16489
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.446473556869346,
      "learning_rate": 1.7243565218118297e-06,
      "loss": 0.5833,
      "step": 16490
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.892282604342074,
      "learning_rate": 1.7238107920073587e-06,
      "loss": 0.655,
      "step": 16491
    },
    {
      "epoch": 1.47,
      "grad_norm": 8.417659055164213,
      "learning_rate": 1.7232651305861608e-06,
      "loss": 0.567,
      "step": 16492
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.6612810327057455,
      "learning_rate": 1.7227195375596278e-06,
      "loss": 0.5808,
      "step": 16493
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.398299180595786,
      "learning_rate": 1.7221740129391473e-06,
      "loss": 0.532,
      "step": 16494
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.89502126144822,
      "learning_rate": 1.7216285567361057e-06,
      "loss": 0.5547,
      "step": 16495
    },
    {
      "epoch": 1.47,
      "grad_norm": 9.069613885552089,
      "learning_rate": 1.721083168961889e-06,
      "loss": 0.5336,
      "step": 16496
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.326908206658528,
      "learning_rate": 1.7205378496278819e-06,
      "loss": 0.5878,
      "step": 16497
    },
    {
      "epoch": 1.47,
      "grad_norm": 8.195943823514401,
      "learning_rate": 1.719992598745463e-06,
      "loss": 0.5581,
      "step": 16498
    },
    {
      "epoch": 1.47,
      "grad_norm": 4.890217353093632,
      "learning_rate": 1.7194474163260177e-06,
      "loss": 0.5668,
      "step": 16499
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.2022225192996,
      "learning_rate": 1.718902302380922e-06,
      "loss": 0.5778,
      "step": 16500
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.0303982239117975,
      "learning_rate": 1.7183572569215546e-06,
      "loss": 0.5649,
      "step": 16501
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.243568532151577,
      "learning_rate": 1.7178122799592928e-06,
      "loss": 0.5626,
      "step": 16502
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.7177653505792065,
      "learning_rate": 1.7172673715055122e-06,
      "loss": 0.6472,
      "step": 16503
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.585908549472535,
      "learning_rate": 1.7167225315715875e-06,
      "loss": 0.5777,
      "step": 16504
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.201407942730368,
      "learning_rate": 1.7161777601688884e-06,
      "loss": 0.593,
      "step": 16505
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.551146527073849,
      "learning_rate": 1.715633057308787e-06,
      "loss": 0.564,
      "step": 16506
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.726890711438248,
      "learning_rate": 1.7150884230026526e-06,
      "loss": 0.5585,
      "step": 16507
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.84556524854544,
      "learning_rate": 1.714543857261854e-06,
      "loss": 0.5713,
      "step": 16508
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.341648573196326,
      "learning_rate": 1.7139993600977584e-06,
      "loss": 0.5381,
      "step": 16509
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.320066142562166,
      "learning_rate": 1.7134549315217285e-06,
      "loss": 0.5895,
      "step": 16510
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.920735426355344,
      "learning_rate": 1.71291057154513e-06,
      "loss": 0.6122,
      "step": 16511
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.448360109151868,
      "learning_rate": 1.7123662801793238e-06,
      "loss": 0.5792,
      "step": 16512
    },
    {
      "epoch": 1.47,
      "grad_norm": 10.719274782276246,
      "learning_rate": 1.7118220574356726e-06,
      "loss": 0.6349,
      "step": 16513
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.767291162002003,
      "learning_rate": 1.7112779033255362e-06,
      "loss": 0.6315,
      "step": 16514
    },
    {
      "epoch": 1.47,
      "grad_norm": 8.247966498288287,
      "learning_rate": 1.7107338178602695e-06,
      "loss": 0.6481,
      "step": 16515
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.334886885675697,
      "learning_rate": 1.7101898010512314e-06,
      "loss": 0.5758,
      "step": 16516
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.10570135098028,
      "learning_rate": 1.7096458529097782e-06,
      "loss": 0.644,
      "step": 16517
    },
    {
      "epoch": 1.47,
      "grad_norm": 8.067835330079424,
      "learning_rate": 1.7091019734472602e-06,
      "loss": 0.5651,
      "step": 16518
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.261835544281759,
      "learning_rate": 1.708558162675032e-06,
      "loss": 0.5476,
      "step": 16519
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.093236247934998,
      "learning_rate": 1.7080144206044436e-06,
      "loss": 0.5982,
      "step": 16520
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.434780478211395,
      "learning_rate": 1.707470747246845e-06,
      "loss": 0.6341,
      "step": 16521
    },
    {
      "epoch": 1.47,
      "grad_norm": 10.396918319603516,
      "learning_rate": 1.706927142613586e-06,
      "loss": 0.6313,
      "step": 16522
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.9451153009657185,
      "learning_rate": 1.706383606716009e-06,
      "loss": 0.575,
      "step": 16523
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.990815018571126,
      "learning_rate": 1.705840139565461e-06,
      "loss": 0.5477,
      "step": 16524
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.336118001988609,
      "learning_rate": 1.7052967411732868e-06,
      "loss": 0.6527,
      "step": 16525
    },
    {
      "epoch": 1.47,
      "grad_norm": 8.381307056280331,
      "learning_rate": 1.7047534115508275e-06,
      "loss": 0.6262,
      "step": 16526
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.39208435460862,
      "learning_rate": 1.7042101507094245e-06,
      "loss": 0.5986,
      "step": 16527
    },
    {
      "epoch": 1.47,
      "grad_norm": 11.190441279039538,
      "learning_rate": 1.7036669586604181e-06,
      "loss": 0.5429,
      "step": 16528
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.83192644969971,
      "learning_rate": 1.7031238354151459e-06,
      "loss": 0.6681,
      "step": 16529
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.951067847134404,
      "learning_rate": 1.7025807809849415e-06,
      "loss": 0.5555,
      "step": 16530
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.559167036066518,
      "learning_rate": 1.7020377953811417e-06,
      "loss": 0.5874,
      "step": 16531
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.37285094076611,
      "learning_rate": 1.7014948786150808e-06,
      "loss": 0.5652,
      "step": 16532
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.91981596470154,
      "learning_rate": 1.7009520306980908e-06,
      "loss": 0.6155,
      "step": 16533
    },
    {
      "epoch": 1.47,
      "grad_norm": 8.937347893790458,
      "learning_rate": 1.7004092516415028e-06,
      "loss": 0.6173,
      "step": 16534
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.0091233593499345,
      "learning_rate": 1.6998665414566468e-06,
      "loss": 0.6022,
      "step": 16535
    },
    {
      "epoch": 1.48,
      "grad_norm": 4.636789076231314,
      "learning_rate": 1.6993239001548479e-06,
      "loss": 0.5725,
      "step": 16536
    },
    {
      "epoch": 1.48,
      "grad_norm": 10.253770329494172,
      "learning_rate": 1.698781327747434e-06,
      "loss": 0.5997,
      "step": 16537
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.428425380032173,
      "learning_rate": 1.698238824245731e-06,
      "loss": 0.5878,
      "step": 16538
    },
    {
      "epoch": 1.48,
      "grad_norm": 4.260750680893849,
      "learning_rate": 1.6976963896610616e-06,
      "loss": 0.5474,
      "step": 16539
    },
    {
      "epoch": 1.48,
      "grad_norm": 7.594576995943618,
      "learning_rate": 1.697154024004748e-06,
      "loss": 0.6043,
      "step": 16540
    },
    {
      "epoch": 1.48,
      "grad_norm": 7.663287193126755,
      "learning_rate": 1.6966117272881128e-06,
      "loss": 0.5747,
      "step": 16541
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.361909326069752,
      "learning_rate": 1.696069499522473e-06,
      "loss": 0.5878,
      "step": 16542
    },
    {
      "epoch": 1.48,
      "grad_norm": 9.697076854712439,
      "learning_rate": 1.6955273407191458e-06,
      "loss": 0.5627,
      "step": 16543
    },
    {
      "epoch": 1.48,
      "grad_norm": 7.80678395347635,
      "learning_rate": 1.6949852508894487e-06,
      "loss": 0.5632,
      "step": 16544
    },
    {
      "epoch": 1.48,
      "grad_norm": 7.486827688184406,
      "learning_rate": 1.6944432300446962e-06,
      "loss": 0.5889,
      "step": 16545
    },
    {
      "epoch": 1.48,
      "grad_norm": 7.02809799268602,
      "learning_rate": 1.6939012781962027e-06,
      "loss": 0.5344,
      "step": 16546
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.378808999105505,
      "learning_rate": 1.6933593953552797e-06,
      "loss": 0.5255,
      "step": 16547
    },
    {
      "epoch": 1.48,
      "grad_norm": 7.431992336574399,
      "learning_rate": 1.6928175815332393e-06,
      "loss": 0.5724,
      "step": 16548
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.152550001983407,
      "learning_rate": 1.6922758367413877e-06,
      "loss": 0.5773,
      "step": 16549
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.393385165646159,
      "learning_rate": 1.6917341609910343e-06,
      "loss": 0.5188,
      "step": 16550
    },
    {
      "epoch": 1.48,
      "grad_norm": 8.068840934993005,
      "learning_rate": 1.6911925542934854e-06,
      "loss": 0.5988,
      "step": 16551
    },
    {
      "epoch": 1.48,
      "grad_norm": 8.280324837267758,
      "learning_rate": 1.6906510166600453e-06,
      "loss": 0.5786,
      "step": 16552
    },
    {
      "epoch": 1.48,
      "grad_norm": 8.207147491305431,
      "learning_rate": 1.6901095481020186e-06,
      "loss": 0.5687,
      "step": 16553
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.527103629119033,
      "learning_rate": 1.689568148630708e-06,
      "loss": 0.5611,
      "step": 16554
    },
    {
      "epoch": 1.48,
      "grad_norm": 7.7652518075511,
      "learning_rate": 1.6890268182574105e-06,
      "loss": 0.5898,
      "step": 16555
    },
    {
      "epoch": 1.48,
      "grad_norm": 7.994604716928091,
      "learning_rate": 1.6884855569934277e-06,
      "loss": 0.628,
      "step": 16556
    },
    {
      "epoch": 1.48,
      "grad_norm": 8.896071386189432,
      "learning_rate": 1.6879443648500583e-06,
      "loss": 0.6175,
      "step": 16557
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.963064498233573,
      "learning_rate": 1.6874032418385956e-06,
      "loss": 0.5787,
      "step": 16558
    },
    {
      "epoch": 1.48,
      "grad_norm": 7.594275067964992,
      "learning_rate": 1.6868621879703352e-06,
      "loss": 0.5915,
      "step": 16559
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.9111448819211185,
      "learning_rate": 1.6863212032565711e-06,
      "loss": 0.5795,
      "step": 16560
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.497332272768373,
      "learning_rate": 1.685780287708597e-06,
      "loss": 0.5584,
      "step": 16561
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.393054669006064,
      "learning_rate": 1.6852394413376993e-06,
      "loss": 0.6107,
      "step": 16562
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.186714646085623,
      "learning_rate": 1.6846986641551698e-06,
      "loss": 0.6104,
      "step": 16563
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.281449912520924,
      "learning_rate": 1.6841579561722948e-06,
      "loss": 0.611,
      "step": 16564
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.992975683782048,
      "learning_rate": 1.6836173174003612e-06,
      "loss": 0.6178,
      "step": 16565
    },
    {
      "epoch": 1.48,
      "grad_norm": 7.38173420698355,
      "learning_rate": 1.6830767478506526e-06,
      "loss": 0.6153,
      "step": 16566
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.013868596119787,
      "learning_rate": 1.6825362475344554e-06,
      "loss": 0.5972,
      "step": 16567
    },
    {
      "epoch": 1.48,
      "grad_norm": 4.969636074282615,
      "learning_rate": 1.6819958164630467e-06,
      "loss": 0.5447,
      "step": 16568
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.234929506250406,
      "learning_rate": 1.6814554546477097e-06,
      "loss": 0.5399,
      "step": 16569
    },
    {
      "epoch": 1.48,
      "grad_norm": 4.983584075996516,
      "learning_rate": 1.6809151620997223e-06,
      "loss": 0.5587,
      "step": 16570
    },
    {
      "epoch": 1.48,
      "grad_norm": 12.099099127040672,
      "learning_rate": 1.6803749388303635e-06,
      "loss": 0.578,
      "step": 16571
    },
    {
      "epoch": 1.48,
      "grad_norm": 9.660101057504896,
      "learning_rate": 1.6798347848509066e-06,
      "loss": 0.5615,
      "step": 16572
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.513756195581352,
      "learning_rate": 1.6792947001726272e-06,
      "loss": 0.5342,
      "step": 16573
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.406667380212826,
      "learning_rate": 1.6787546848068009e-06,
      "loss": 0.5942,
      "step": 16574
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.471590197973633,
      "learning_rate": 1.6782147387646952e-06,
      "loss": 0.5446,
      "step": 16575
    },
    {
      "epoch": 1.48,
      "grad_norm": 7.306045701849128,
      "learning_rate": 1.6776748620575827e-06,
      "loss": 0.5858,
      "step": 16576
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.199609514298581,
      "learning_rate": 1.6771350546967313e-06,
      "loss": 0.6053,
      "step": 16577
    },
    {
      "epoch": 1.48,
      "grad_norm": 4.936926547546596,
      "learning_rate": 1.6765953166934084e-06,
      "loss": 0.5273,
      "step": 16578
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.049296351796455,
      "learning_rate": 1.6760556480588808e-06,
      "loss": 0.5647,
      "step": 16579
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.824906192656742,
      "learning_rate": 1.675516048804413e-06,
      "loss": 0.5832,
      "step": 16580
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.627082339631345,
      "learning_rate": 1.6749765189412664e-06,
      "loss": 0.5755,
      "step": 16581
    },
    {
      "epoch": 1.48,
      "grad_norm": 7.630146473809049,
      "learning_rate": 1.674437058480703e-06,
      "loss": 0.6432,
      "step": 16582
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.651120908544123,
      "learning_rate": 1.673897667433983e-06,
      "loss": 0.5461,
      "step": 16583
    },
    {
      "epoch": 1.48,
      "grad_norm": 4.914929660480168,
      "learning_rate": 1.6733583458123653e-06,
      "loss": 0.5926,
      "step": 16584
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.539824555046824,
      "learning_rate": 1.6728190936271083e-06,
      "loss": 0.5693,
      "step": 16585
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.8287051149684475,
      "learning_rate": 1.6722799108894649e-06,
      "loss": 0.5877,
      "step": 16586
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.955510709179059,
      "learning_rate": 1.6717407976106924e-06,
      "loss": 0.5121,
      "step": 16587
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.663675724114788,
      "learning_rate": 1.6712017538020403e-06,
      "loss": 0.5384,
      "step": 16588
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.750613381587288,
      "learning_rate": 1.670662779474762e-06,
      "loss": 0.5577,
      "step": 16589
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.186775837948213,
      "learning_rate": 1.6701238746401067e-06,
      "loss": 0.5424,
      "step": 16590
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.654051989346489,
      "learning_rate": 1.6695850393093232e-06,
      "loss": 0.5128,
      "step": 16591
    },
    {
      "epoch": 1.48,
      "grad_norm": 7.916845528115406,
      "learning_rate": 1.6690462734936585e-06,
      "loss": 0.6194,
      "step": 16592
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.873349880905711,
      "learning_rate": 1.66850757720436e-06,
      "loss": 0.5328,
      "step": 16593
    },
    {
      "epoch": 1.48,
      "grad_norm": 10.05275849358468,
      "learning_rate": 1.6679689504526681e-06,
      "loss": 0.577,
      "step": 16594
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.759958750937419,
      "learning_rate": 1.6674303932498276e-06,
      "loss": 0.5411,
      "step": 16595
    },
    {
      "epoch": 1.48,
      "grad_norm": 10.178021153286334,
      "learning_rate": 1.6668919056070791e-06,
      "loss": 0.622,
      "step": 16596
    },
    {
      "epoch": 1.48,
      "grad_norm": 9.991715223919822,
      "learning_rate": 1.666353487535663e-06,
      "loss": 0.6455,
      "step": 16597
    },
    {
      "epoch": 1.48,
      "grad_norm": 8.087950414974,
      "learning_rate": 1.665815139046819e-06,
      "loss": 0.604,
      "step": 16598
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.033741306683741,
      "learning_rate": 1.6652768601517806e-06,
      "loss": 0.5777,
      "step": 16599
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.439815337910645,
      "learning_rate": 1.6647386508617859e-06,
      "loss": 0.6402,
      "step": 16600
    },
    {
      "epoch": 1.48,
      "grad_norm": 8.494904963878362,
      "learning_rate": 1.6642005111880667e-06,
      "loss": 0.592,
      "step": 16601
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.457864431489162,
      "learning_rate": 1.663662441141856e-06,
      "loss": 0.5947,
      "step": 16602
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.056866316047348,
      "learning_rate": 1.663124440734386e-06,
      "loss": 0.5652,
      "step": 16603
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.995928368157356,
      "learning_rate": 1.6625865099768857e-06,
      "loss": 0.5752,
      "step": 16604
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.747724772196537,
      "learning_rate": 1.6620486488805843e-06,
      "loss": 0.6337,
      "step": 16605
    },
    {
      "epoch": 1.48,
      "grad_norm": 9.35551256479892,
      "learning_rate": 1.6615108574567058e-06,
      "loss": 0.57,
      "step": 16606
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.103698483276872,
      "learning_rate": 1.6609731357164771e-06,
      "loss": 0.5674,
      "step": 16607
    },
    {
      "epoch": 1.48,
      "grad_norm": 7.79342806235292,
      "learning_rate": 1.6604354836711223e-06,
      "loss": 0.5899,
      "step": 16608
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.62488845513608,
      "learning_rate": 1.6598979013318627e-06,
      "loss": 0.5268,
      "step": 16609
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.910075097509002,
      "learning_rate": 1.6593603887099197e-06,
      "loss": 0.611,
      "step": 16610
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.7007034101346425,
      "learning_rate": 1.6588229458165146e-06,
      "loss": 0.5504,
      "step": 16611
    },
    {
      "epoch": 1.48,
      "grad_norm": 8.599594331958338,
      "learning_rate": 1.6582855726628616e-06,
      "loss": 0.5729,
      "step": 16612
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.1839356863591,
      "learning_rate": 1.6577482692601792e-06,
      "loss": 0.6002,
      "step": 16613
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.704475860980167,
      "learning_rate": 1.6572110356196824e-06,
      "loss": 0.6151,
      "step": 16614
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.201011643137781,
      "learning_rate": 1.656673871752586e-06,
      "loss": 0.5659,
      "step": 16615
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.193811455059978,
      "learning_rate": 1.6561367776700992e-06,
      "loss": 0.4949,
      "step": 16616
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.322774419402176,
      "learning_rate": 1.6555997533834344e-06,
      "loss": 0.5575,
      "step": 16617
    },
    {
      "epoch": 1.48,
      "grad_norm": 7.898296678091713,
      "learning_rate": 1.6550627989038021e-06,
      "loss": 0.5205,
      "step": 16618
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.375989766115637,
      "learning_rate": 1.6545259142424068e-06,
      "loss": 0.5927,
      "step": 16619
    },
    {
      "epoch": 1.48,
      "grad_norm": 7.0032702986742095,
      "learning_rate": 1.6539890994104575e-06,
      "loss": 0.5731,
      "step": 16620
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.14675028719798,
      "learning_rate": 1.653452354419157e-06,
      "loss": 0.5696,
      "step": 16621
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.5899674622885644,
      "learning_rate": 1.6529156792797103e-06,
      "loss": 0.558,
      "step": 16622
    },
    {
      "epoch": 1.48,
      "grad_norm": 7.542103532816081,
      "learning_rate": 1.6523790740033186e-06,
      "loss": 0.5681,
      "step": 16623
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.912798220405645,
      "learning_rate": 1.6518425386011843e-06,
      "loss": 0.5484,
      "step": 16624
    },
    {
      "epoch": 1.48,
      "grad_norm": 7.615110354113473,
      "learning_rate": 1.6513060730845033e-06,
      "loss": 0.6139,
      "step": 16625
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.181576852344464,
      "learning_rate": 1.6507696774644738e-06,
      "loss": 0.5209,
      "step": 16626
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.8786776000173315,
      "learning_rate": 1.6502333517522934e-06,
      "loss": 0.5269,
      "step": 16627
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.1049147940846185,
      "learning_rate": 1.6496970959591551e-06,
      "loss": 0.6305,
      "step": 16628
    },
    {
      "epoch": 1.48,
      "grad_norm": 7.552948581576313,
      "learning_rate": 1.6491609100962552e-06,
      "loss": 0.5346,
      "step": 16629
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.447048407610653,
      "learning_rate": 1.6486247941747814e-06,
      "loss": 0.5712,
      "step": 16630
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.47248307326096,
      "learning_rate": 1.648088748205927e-06,
      "loss": 0.5357,
      "step": 16631
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.0205452795778776,
      "learning_rate": 1.6475527722008778e-06,
      "loss": 0.562,
      "step": 16632
    },
    {
      "epoch": 1.48,
      "grad_norm": 8.03459093078818,
      "learning_rate": 1.647016866170823e-06,
      "loss": 0.5862,
      "step": 16633
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.164269649813392,
      "learning_rate": 1.6464810301269478e-06,
      "loss": 0.5868,
      "step": 16634
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.507291188247103,
      "learning_rate": 1.6459452640804375e-06,
      "loss": 0.527,
      "step": 16635
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.965831526796126,
      "learning_rate": 1.6454095680424742e-06,
      "loss": 0.5659,
      "step": 16636
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.493163489805926,
      "learning_rate": 1.6448739420242415e-06,
      "loss": 0.5626,
      "step": 16637
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.712547380098561,
      "learning_rate": 1.644338386036916e-06,
      "loss": 0.6207,
      "step": 16638
    },
    {
      "epoch": 1.48,
      "grad_norm": 7.490833915331839,
      "learning_rate": 1.6438029000916777e-06,
      "loss": 0.6342,
      "step": 16639
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.854992469299814,
      "learning_rate": 1.6432674841997043e-06,
      "loss": 0.6371,
      "step": 16640
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.902218775789666,
      "learning_rate": 1.642732138372171e-06,
      "loss": 0.5702,
      "step": 16641
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.9455572731226685,
      "learning_rate": 1.6421968626202517e-06,
      "loss": 0.6081,
      "step": 16642
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.689781124202494,
      "learning_rate": 1.6416616569551215e-06,
      "loss": 0.5497,
      "step": 16643
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.960609559904213,
      "learning_rate": 1.641126521387949e-06,
      "loss": 0.5892,
      "step": 16644
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.140138300341864,
      "learning_rate": 1.6405914559299035e-06,
      "loss": 0.6215,
      "step": 16645
    },
    {
      "epoch": 1.48,
      "grad_norm": 9.515282672561462,
      "learning_rate": 1.6400564605921544e-06,
      "loss": 0.6107,
      "step": 16646
    },
    {
      "epoch": 1.49,
      "grad_norm": 7.604401659373851,
      "learning_rate": 1.639521535385868e-06,
      "loss": 0.5802,
      "step": 16647
    },
    {
      "epoch": 1.49,
      "grad_norm": 7.001151857977251,
      "learning_rate": 1.6389866803222103e-06,
      "loss": 0.5914,
      "step": 16648
    },
    {
      "epoch": 1.49,
      "grad_norm": 7.0211447620198095,
      "learning_rate": 1.6384518954123458e-06,
      "loss": 0.5712,
      "step": 16649
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.485660832722303,
      "learning_rate": 1.637917180667437e-06,
      "loss": 0.6246,
      "step": 16650
    },
    {
      "epoch": 1.49,
      "grad_norm": 8.518229386585967,
      "learning_rate": 1.6373825360986433e-06,
      "loss": 0.5964,
      "step": 16651
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.284903333346108,
      "learning_rate": 1.6368479617171252e-06,
      "loss": 0.532,
      "step": 16652
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.0209026568469675,
      "learning_rate": 1.6363134575340406e-06,
      "loss": 0.6041,
      "step": 16653
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.708289448200882,
      "learning_rate": 1.6357790235605458e-06,
      "loss": 0.6003,
      "step": 16654
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.290959956570718,
      "learning_rate": 1.635244659807797e-06,
      "loss": 0.6078,
      "step": 16655
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.72734090756184,
      "learning_rate": 1.6347103662869484e-06,
      "loss": 0.594,
      "step": 16656
    },
    {
      "epoch": 1.49,
      "grad_norm": 4.965067709899691,
      "learning_rate": 1.6341761430091496e-06,
      "loss": 0.5733,
      "step": 16657
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.366230254486659,
      "learning_rate": 1.6336419899855543e-06,
      "loss": 0.563,
      "step": 16658
    },
    {
      "epoch": 1.49,
      "grad_norm": 7.909160635289137,
      "learning_rate": 1.6331079072273088e-06,
      "loss": 0.6554,
      "step": 16659
    },
    {
      "epoch": 1.49,
      "grad_norm": 7.033414196553512,
      "learning_rate": 1.6325738947455622e-06,
      "loss": 0.5714,
      "step": 16660
    },
    {
      "epoch": 1.49,
      "grad_norm": 7.635448458332921,
      "learning_rate": 1.6320399525514607e-06,
      "loss": 0.6046,
      "step": 16661
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.713828785374766,
      "learning_rate": 1.63150608065615e-06,
      "loss": 0.5734,
      "step": 16662
    },
    {
      "epoch": 1.49,
      "grad_norm": 7.271150437677039,
      "learning_rate": 1.6309722790707743e-06,
      "loss": 0.6403,
      "step": 16663
    },
    {
      "epoch": 1.49,
      "grad_norm": 7.025148751273399,
      "learning_rate": 1.6304385478064733e-06,
      "loss": 0.5598,
      "step": 16664
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.808731394193037,
      "learning_rate": 1.6299048868743877e-06,
      "loss": 0.5412,
      "step": 16665
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.229872536117584,
      "learning_rate": 1.6293712962856573e-06,
      "loss": 0.5343,
      "step": 16666
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.336061732774519,
      "learning_rate": 1.6288377760514196e-06,
      "loss": 0.6017,
      "step": 16667
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.610966991128474,
      "learning_rate": 1.6283043261828108e-06,
      "loss": 0.5718,
      "step": 16668
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.2715599096960215,
      "learning_rate": 1.6277709466909665e-06,
      "loss": 0.6031,
      "step": 16669
    },
    {
      "epoch": 1.49,
      "grad_norm": 4.806732465137839,
      "learning_rate": 1.6272376375870175e-06,
      "loss": 0.5427,
      "step": 16670
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.625003950590108,
      "learning_rate": 1.6267043988820964e-06,
      "loss": 0.5204,
      "step": 16671
    },
    {
      "epoch": 1.49,
      "grad_norm": 7.74521689137516,
      "learning_rate": 1.626171230587335e-06,
      "loss": 0.5901,
      "step": 16672
    },
    {
      "epoch": 1.49,
      "grad_norm": 11.117689826062975,
      "learning_rate": 1.6256381327138592e-06,
      "loss": 0.6322,
      "step": 16673
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.797582290533975,
      "learning_rate": 1.6251051052727972e-06,
      "loss": 0.622,
      "step": 16674
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.351792263753973,
      "learning_rate": 1.6245721482752758e-06,
      "loss": 0.6261,
      "step": 16675
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.303232590996054,
      "learning_rate": 1.6240392617324201e-06,
      "loss": 0.565,
      "step": 16676
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.05355647029909,
      "learning_rate": 1.62350644565535e-06,
      "loss": 0.555,
      "step": 16677
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.650389682371294,
      "learning_rate": 1.6229737000551887e-06,
      "loss": 0.5439,
      "step": 16678
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.378009632278634,
      "learning_rate": 1.6224410249430556e-06,
      "loss": 0.5732,
      "step": 16679
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.793531313277027,
      "learning_rate": 1.6219084203300694e-06,
      "loss": 0.5668,
      "step": 16680
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.790090437937187,
      "learning_rate": 1.6213758862273488e-06,
      "loss": 0.5826,
      "step": 16681
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.837965475276673,
      "learning_rate": 1.6208434226460057e-06,
      "loss": 0.6326,
      "step": 16682
    },
    {
      "epoch": 1.49,
      "grad_norm": 8.443416630272074,
      "learning_rate": 1.6203110295971564e-06,
      "loss": 0.6027,
      "step": 16683
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.528876909275584,
      "learning_rate": 1.6197787070919125e-06,
      "loss": 0.5794,
      "step": 16684
    },
    {
      "epoch": 1.49,
      "grad_norm": 7.676104133551258,
      "learning_rate": 1.6192464551413856e-06,
      "loss": 0.5879,
      "step": 16685
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.814349724983146,
      "learning_rate": 1.6187142737566853e-06,
      "loss": 0.5529,
      "step": 16686
    },
    {
      "epoch": 1.49,
      "grad_norm": 8.292522033463866,
      "learning_rate": 1.6181821629489207e-06,
      "loss": 0.5873,
      "step": 16687
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.261182769541119,
      "learning_rate": 1.617650122729198e-06,
      "loss": 0.5623,
      "step": 16688
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.772272625769619,
      "learning_rate": 1.61711815310862e-06,
      "loss": 0.6192,
      "step": 16689
    },
    {
      "epoch": 1.49,
      "grad_norm": 7.407282572254293,
      "learning_rate": 1.6165862540982925e-06,
      "loss": 0.4841,
      "step": 16690
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.67164465366453,
      "learning_rate": 1.6160544257093169e-06,
      "loss": 0.6261,
      "step": 16691
    },
    {
      "epoch": 1.49,
      "grad_norm": 7.991257380542349,
      "learning_rate": 1.6155226679527947e-06,
      "loss": 0.5292,
      "step": 16692
    },
    {
      "epoch": 1.49,
      "grad_norm": 4.951781130407027,
      "learning_rate": 1.614990980839825e-06,
      "loss": 0.5387,
      "step": 16693
    },
    {
      "epoch": 1.49,
      "grad_norm": 10.14783282544295,
      "learning_rate": 1.6144593643815072e-06,
      "loss": 0.5465,
      "step": 16694
    },
    {
      "epoch": 1.49,
      "grad_norm": 7.539908575903127,
      "learning_rate": 1.613927818588934e-06,
      "loss": 0.5492,
      "step": 16695
    },
    {
      "epoch": 1.49,
      "grad_norm": 8.872591144314589,
      "learning_rate": 1.613396343473202e-06,
      "loss": 0.5717,
      "step": 16696
    },
    {
      "epoch": 1.49,
      "grad_norm": 8.507985598624865,
      "learning_rate": 1.6128649390454054e-06,
      "loss": 0.6166,
      "step": 16697
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.872565254484306,
      "learning_rate": 1.6123336053166356e-06,
      "loss": 0.6033,
      "step": 16698
    },
    {
      "epoch": 1.49,
      "grad_norm": 8.072640520381217,
      "learning_rate": 1.6118023422979824e-06,
      "loss": 0.6203,
      "step": 16699
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.614923398617723,
      "learning_rate": 1.611271150000537e-06,
      "loss": 0.6769,
      "step": 16700
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.820166654772655,
      "learning_rate": 1.6107400284353853e-06,
      "loss": 0.5871,
      "step": 16701
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.984714869319991,
      "learning_rate": 1.6102089776136115e-06,
      "loss": 0.5826,
      "step": 16702
    },
    {
      "epoch": 1.49,
      "grad_norm": 4.5894678041458015,
      "learning_rate": 1.6096779975463017e-06,
      "loss": 0.5909,
      "step": 16703
    },
    {
      "epoch": 1.49,
      "grad_norm": 4.909938221599264,
      "learning_rate": 1.6091470882445392e-06,
      "loss": 0.5375,
      "step": 16704
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.898843697070322,
      "learning_rate": 1.6086162497194053e-06,
      "loss": 0.5818,
      "step": 16705
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.794170215688313,
      "learning_rate": 1.6080854819819802e-06,
      "loss": 0.5788,
      "step": 16706
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.79958408572142,
      "learning_rate": 1.6075547850433438e-06,
      "loss": 0.6074,
      "step": 16707
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.464420432732051,
      "learning_rate": 1.6070241589145707e-06,
      "loss": 0.6008,
      "step": 16708
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.914940611028755,
      "learning_rate": 1.6064936036067375e-06,
      "loss": 0.6076,
      "step": 16709
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.458014272574896,
      "learning_rate": 1.6059631191309188e-06,
      "loss": 0.5346,
      "step": 16710
    },
    {
      "epoch": 1.49,
      "grad_norm": 7.355617067042456,
      "learning_rate": 1.6054327054981872e-06,
      "loss": 0.5485,
      "step": 16711
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.566484226785939,
      "learning_rate": 1.6049023627196142e-06,
      "loss": 0.5595,
      "step": 16712
    },
    {
      "epoch": 1.49,
      "grad_norm": 7.597167938616306,
      "learning_rate": 1.6043720908062704e-06,
      "loss": 0.6605,
      "step": 16713
    },
    {
      "epoch": 1.49,
      "grad_norm": 8.263576046207257,
      "learning_rate": 1.6038418897692216e-06,
      "loss": 0.5941,
      "step": 16714
    },
    {
      "epoch": 1.49,
      "grad_norm": 9.39715580968337,
      "learning_rate": 1.6033117596195374e-06,
      "loss": 0.5778,
      "step": 16715
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.848323229540237,
      "learning_rate": 1.60278170036828e-06,
      "loss": 0.5315,
      "step": 16716
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.310809064454719,
      "learning_rate": 1.6022517120265147e-06,
      "loss": 0.6096,
      "step": 16717
    },
    {
      "epoch": 1.49,
      "grad_norm": 7.366407359242103,
      "learning_rate": 1.6017217946053043e-06,
      "loss": 0.5357,
      "step": 16718
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.382373595327632,
      "learning_rate": 1.601191948115709e-06,
      "loss": 0.5756,
      "step": 16719
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.74537348680575,
      "learning_rate": 1.6006621725687898e-06,
      "loss": 0.6326,
      "step": 16720
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.545372676706489,
      "learning_rate": 1.6001324679756025e-06,
      "loss": 0.5688,
      "step": 16721
    },
    {
      "epoch": 1.49,
      "grad_norm": 7.361894963171339,
      "learning_rate": 1.5996028343472037e-06,
      "loss": 0.5889,
      "step": 16722
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.925789878060734,
      "learning_rate": 1.5990732716946494e-06,
      "loss": 0.5602,
      "step": 16723
    },
    {
      "epoch": 1.49,
      "grad_norm": 7.660571515877035,
      "learning_rate": 1.5985437800289927e-06,
      "loss": 0.594,
      "step": 16724
    },
    {
      "epoch": 1.49,
      "grad_norm": 7.053704217982891,
      "learning_rate": 1.598014359361285e-06,
      "loss": 0.6119,
      "step": 16725
    },
    {
      "epoch": 1.49,
      "grad_norm": 8.808648241283187,
      "learning_rate": 1.5974850097025796e-06,
      "loss": 0.589,
      "step": 16726
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.834943724720598,
      "learning_rate": 1.596955731063921e-06,
      "loss": 0.5673,
      "step": 16727
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.826934484310981,
      "learning_rate": 1.5964265234563598e-06,
      "loss": 0.5569,
      "step": 16728
    },
    {
      "epoch": 1.49,
      "grad_norm": 8.48301202766483,
      "learning_rate": 1.5958973868909405e-06,
      "loss": 0.535,
      "step": 16729
    },
    {
      "epoch": 1.49,
      "grad_norm": 7.5721790617803855,
      "learning_rate": 1.5953683213787103e-06,
      "loss": 0.588,
      "step": 16730
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.8207179456687514,
      "learning_rate": 1.5948393269307089e-06,
      "loss": 0.5193,
      "step": 16731
    },
    {
      "epoch": 1.49,
      "grad_norm": 11.28320685927629,
      "learning_rate": 1.5943104035579788e-06,
      "loss": 0.663,
      "step": 16732
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.35600836946967,
      "learning_rate": 1.5937815512715626e-06,
      "loss": 0.6383,
      "step": 16733
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.968123952905001,
      "learning_rate": 1.5932527700824958e-06,
      "loss": 0.5437,
      "step": 16734
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.988569404127991,
      "learning_rate": 1.5927240600018163e-06,
      "loss": 0.5514,
      "step": 16735
    },
    {
      "epoch": 1.49,
      "grad_norm": 4.912546840163095,
      "learning_rate": 1.5921954210405604e-06,
      "loss": 0.575,
      "step": 16736
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.693930891212401,
      "learning_rate": 1.5916668532097623e-06,
      "loss": 0.6015,
      "step": 16737
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.557722772014418,
      "learning_rate": 1.5911383565204542e-06,
      "loss": 0.5781,
      "step": 16738
    },
    {
      "epoch": 1.49,
      "grad_norm": 4.637071915738045,
      "learning_rate": 1.5906099309836692e-06,
      "loss": 0.5912,
      "step": 16739
    },
    {
      "epoch": 1.49,
      "grad_norm": 8.222988690762703,
      "learning_rate": 1.5900815766104343e-06,
      "loss": 0.5736,
      "step": 16740
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.643031930790192,
      "learning_rate": 1.5895532934117786e-06,
      "loss": 0.6055,
      "step": 16741
    },
    {
      "epoch": 1.49,
      "grad_norm": 7.526633741889608,
      "learning_rate": 1.589025081398729e-06,
      "loss": 0.6608,
      "step": 16742
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.92228043208061,
      "learning_rate": 1.5884969405823108e-06,
      "loss": 0.5515,
      "step": 16743
    },
    {
      "epoch": 1.49,
      "grad_norm": 37.175921219515786,
      "learning_rate": 1.5879688709735503e-06,
      "loss": 0.5627,
      "step": 16744
    },
    {
      "epoch": 1.49,
      "grad_norm": 7.545507744545335,
      "learning_rate": 1.587440872583465e-06,
      "loss": 0.6245,
      "step": 16745
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.139268806700995,
      "learning_rate": 1.5869129454230803e-06,
      "loss": 0.5421,
      "step": 16746
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.557413508936121,
      "learning_rate": 1.5863850895034111e-06,
      "loss": 0.5396,
      "step": 16747
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.237603577264708,
      "learning_rate": 1.585857304835478e-06,
      "loss": 0.5217,
      "step": 16748
    },
    {
      "epoch": 1.49,
      "grad_norm": 7.0468979522732464,
      "learning_rate": 1.5853295914302963e-06,
      "loss": 0.5956,
      "step": 16749
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.805988833876837,
      "learning_rate": 1.584801949298882e-06,
      "loss": 0.534,
      "step": 16750
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.318620256762924,
      "learning_rate": 1.5842743784522474e-06,
      "loss": 0.5526,
      "step": 16751
    },
    {
      "epoch": 1.49,
      "grad_norm": 8.470532115060395,
      "learning_rate": 1.5837468789014065e-06,
      "loss": 0.5403,
      "step": 16752
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.220646207366034,
      "learning_rate": 1.5832194506573668e-06,
      "loss": 0.5842,
      "step": 16753
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.692356559529226,
      "learning_rate": 1.582692093731138e-06,
      "loss": 0.6114,
      "step": 16754
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.399612801301306,
      "learning_rate": 1.5821648081337282e-06,
      "loss": 0.569,
      "step": 16755
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.086331036861599,
      "learning_rate": 1.581637593876143e-06,
      "loss": 0.5774,
      "step": 16756
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.699185694144533,
      "learning_rate": 1.5811104509693885e-06,
      "loss": 0.6002,
      "step": 16757
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.816872706272658,
      "learning_rate": 1.5805833794244646e-06,
      "loss": 0.6098,
      "step": 16758
    },
    {
      "epoch": 1.5,
      "grad_norm": 4.935027128919735,
      "learning_rate": 1.5800563792523753e-06,
      "loss": 0.5373,
      "step": 16759
    },
    {
      "epoch": 1.5,
      "grad_norm": 7.711214428057946,
      "learning_rate": 1.579529450464119e-06,
      "loss": 0.5794,
      "step": 16760
    },
    {
      "epoch": 1.5,
      "grad_norm": 8.951590348349548,
      "learning_rate": 1.5790025930706942e-06,
      "loss": 0.5302,
      "step": 16761
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.6799708879233926,
      "learning_rate": 1.578475807083098e-06,
      "loss": 0.6207,
      "step": 16762
    },
    {
      "epoch": 1.5,
      "grad_norm": 7.324378560309559,
      "learning_rate": 1.577949092512327e-06,
      "loss": 0.6075,
      "step": 16763
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.292971885625318,
      "learning_rate": 1.5774224493693758e-06,
      "loss": 0.6078,
      "step": 16764
    },
    {
      "epoch": 1.5,
      "grad_norm": 11.434302272053369,
      "learning_rate": 1.5768958776652337e-06,
      "loss": 0.6097,
      "step": 16765
    },
    {
      "epoch": 1.5,
      "grad_norm": 7.4537356989544605,
      "learning_rate": 1.5763693774108947e-06,
      "loss": 0.6137,
      "step": 16766
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.692422017733832,
      "learning_rate": 1.5758429486173466e-06,
      "loss": 0.5395,
      "step": 16767
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.984120391552226,
      "learning_rate": 1.5753165912955782e-06,
      "loss": 0.5974,
      "step": 16768
    },
    {
      "epoch": 1.5,
      "grad_norm": 7.194015864942682,
      "learning_rate": 1.574790305456576e-06,
      "loss": 0.5895,
      "step": 16769
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.466290861714776,
      "learning_rate": 1.5742640911113266e-06,
      "loss": 0.5717,
      "step": 16770
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.505166767675522,
      "learning_rate": 1.573737948270811e-06,
      "loss": 0.5893,
      "step": 16771
    },
    {
      "epoch": 1.5,
      "grad_norm": 8.956972197127385,
      "learning_rate": 1.5732118769460114e-06,
      "loss": 0.6324,
      "step": 16772
    },
    {
      "epoch": 1.5,
      "grad_norm": 7.81632473599714,
      "learning_rate": 1.572685877147912e-06,
      "loss": 0.609,
      "step": 16773
    },
    {
      "epoch": 1.5,
      "grad_norm": 8.032868449432764,
      "learning_rate": 1.5721599488874868e-06,
      "loss": 0.5293,
      "step": 16774
    },
    {
      "epoch": 1.5,
      "grad_norm": 4.848094488963424,
      "learning_rate": 1.5716340921757163e-06,
      "loss": 0.6037,
      "step": 16775
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.365854374609732,
      "learning_rate": 1.5711083070235761e-06,
      "loss": 0.6153,
      "step": 16776
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.5475550147255355,
      "learning_rate": 1.570582593442042e-06,
      "loss": 0.4942,
      "step": 16777
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.5538132619482665,
      "learning_rate": 1.570056951442085e-06,
      "loss": 0.5585,
      "step": 16778
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.753281795377747,
      "learning_rate": 1.5695313810346773e-06,
      "loss": 0.6307,
      "step": 16779
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.122768767233399,
      "learning_rate": 1.5690058822307897e-06,
      "loss": 0.6047,
      "step": 16780
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.236408478936888,
      "learning_rate": 1.56848045504139e-06,
      "loss": 0.5134,
      "step": 16781
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.8478781119709815,
      "learning_rate": 1.5679550994774468e-06,
      "loss": 0.5676,
      "step": 16782
    },
    {
      "epoch": 1.5,
      "grad_norm": 8.642244717617755,
      "learning_rate": 1.5674298155499256e-06,
      "loss": 0.5981,
      "step": 16783
    },
    {
      "epoch": 1.5,
      "grad_norm": 7.427131849180458,
      "learning_rate": 1.5669046032697887e-06,
      "loss": 0.6602,
      "step": 16784
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.973882235941195,
      "learning_rate": 1.5663794626479995e-06,
      "loss": 0.5977,
      "step": 16785
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.306449840687734,
      "learning_rate": 1.56585439369552e-06,
      "loss": 0.5363,
      "step": 16786
    },
    {
      "epoch": 1.5,
      "grad_norm": 8.490969417576634,
      "learning_rate": 1.5653293964233091e-06,
      "loss": 0.4922,
      "step": 16787
    },
    {
      "epoch": 1.5,
      "grad_norm": 7.21398665990198,
      "learning_rate": 1.5648044708423266e-06,
      "loss": 0.6041,
      "step": 16788
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.725159605917228,
      "learning_rate": 1.5642796169635266e-06,
      "loss": 0.5739,
      "step": 16789
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.434902109849927,
      "learning_rate": 1.5637548347978672e-06,
      "loss": 0.5281,
      "step": 16790
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.611414180750037,
      "learning_rate": 1.563230124356298e-06,
      "loss": 0.501,
      "step": 16791
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.053918409348015,
      "learning_rate": 1.5627054856497742e-06,
      "loss": 0.6053,
      "step": 16792
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.852498829476014,
      "learning_rate": 1.5621809186892456e-06,
      "loss": 0.5942,
      "step": 16793
    },
    {
      "epoch": 1.5,
      "grad_norm": 8.031463005049226,
      "learning_rate": 1.5616564234856618e-06,
      "loss": 0.5631,
      "step": 16794
    },
    {
      "epoch": 1.5,
      "grad_norm": 8.704780300603673,
      "learning_rate": 1.5611320000499703e-06,
      "loss": 0.5471,
      "step": 16795
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.296510254166477,
      "learning_rate": 1.5606076483931194e-06,
      "loss": 0.6254,
      "step": 16796
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.153200659118216,
      "learning_rate": 1.560083368526049e-06,
      "loss": 0.6022,
      "step": 16797
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.8459699849027595,
      "learning_rate": 1.5595591604597061e-06,
      "loss": 0.6264,
      "step": 16798
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.752436644413176,
      "learning_rate": 1.5590350242050307e-06,
      "loss": 0.5972,
      "step": 16799
    },
    {
      "epoch": 1.5,
      "grad_norm": 8.622995591578157,
      "learning_rate": 1.5585109597729636e-06,
      "loss": 0.5799,
      "step": 16800
    },
    {
      "epoch": 1.5,
      "grad_norm": 7.003510220025189,
      "learning_rate": 1.5579869671744435e-06,
      "loss": 0.5526,
      "step": 16801
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.109543097393736,
      "learning_rate": 1.5574630464204088e-06,
      "loss": 0.5868,
      "step": 16802
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.390315666806166,
      "learning_rate": 1.5569391975217946e-06,
      "loss": 0.5862,
      "step": 16803
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.400884180037942,
      "learning_rate": 1.5564154204895326e-06,
      "loss": 0.6366,
      "step": 16804
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.949500842996548,
      "learning_rate": 1.5558917153345576e-06,
      "loss": 0.5849,
      "step": 16805
    },
    {
      "epoch": 1.5,
      "grad_norm": 9.531653189461284,
      "learning_rate": 1.5553680820678001e-06,
      "loss": 0.624,
      "step": 16806
    },
    {
      "epoch": 1.5,
      "grad_norm": 10.105981370431945,
      "learning_rate": 1.5548445207001905e-06,
      "loss": 0.5829,
      "step": 16807
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.409710555357583,
      "learning_rate": 1.5543210312426572e-06,
      "loss": 0.6064,
      "step": 16808
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.204022142372398,
      "learning_rate": 1.5537976137061277e-06,
      "loss": 0.6145,
      "step": 16809
    },
    {
      "epoch": 1.5,
      "grad_norm": 4.610417126751667,
      "learning_rate": 1.5532742681015246e-06,
      "loss": 0.6273,
      "step": 16810
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.666492543800809,
      "learning_rate": 1.5527509944397722e-06,
      "loss": 0.5669,
      "step": 16811
    },
    {
      "epoch": 1.5,
      "grad_norm": 7.014163606065578,
      "learning_rate": 1.552227792731794e-06,
      "loss": 0.5074,
      "step": 16812
    },
    {
      "epoch": 1.5,
      "grad_norm": 7.578866835006075,
      "learning_rate": 1.5517046629885102e-06,
      "loss": 0.5974,
      "step": 16813
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.755308615221877,
      "learning_rate": 1.55118160522084e-06,
      "loss": 0.5292,
      "step": 16814
    },
    {
      "epoch": 1.5,
      "grad_norm": 7.581632136474802,
      "learning_rate": 1.5506586194397022e-06,
      "loss": 0.5992,
      "step": 16815
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.7000648947954335,
      "learning_rate": 1.5501357056560124e-06,
      "loss": 0.5871,
      "step": 16816
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.017812880946282,
      "learning_rate": 1.5496128638806823e-06,
      "loss": 0.5247,
      "step": 16817
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.9238154317379,
      "learning_rate": 1.5490900941246277e-06,
      "loss": 0.5859,
      "step": 16818
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.881708604251993,
      "learning_rate": 1.5485673963987602e-06,
      "loss": 0.5535,
      "step": 16819
    },
    {
      "epoch": 1.5,
      "grad_norm": 7.56207547671571,
      "learning_rate": 1.5480447707139901e-06,
      "loss": 0.5455,
      "step": 16820
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.502301929375096,
      "learning_rate": 1.5475222170812254e-06,
      "loss": 0.5373,
      "step": 16821
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.781356036280376,
      "learning_rate": 1.5469997355113754e-06,
      "loss": 0.5741,
      "step": 16822
    },
    {
      "epoch": 1.5,
      "grad_norm": 7.957798746290355,
      "learning_rate": 1.5464773260153422e-06,
      "loss": 0.5484,
      "step": 16823
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.330582846339879,
      "learning_rate": 1.5459549886040316e-06,
      "loss": 0.5426,
      "step": 16824
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.219110400666478,
      "learning_rate": 1.5454327232883464e-06,
      "loss": 0.5879,
      "step": 16825
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.690894005412815,
      "learning_rate": 1.5449105300791877e-06,
      "loss": 0.5791,
      "step": 16826
    },
    {
      "epoch": 1.5,
      "grad_norm": 7.094725398301818,
      "learning_rate": 1.5443884089874556e-06,
      "loss": 0.5417,
      "step": 16827
    },
    {
      "epoch": 1.5,
      "grad_norm": 7.859573947025127,
      "learning_rate": 1.5438663600240489e-06,
      "loss": 0.5899,
      "step": 16828
    },
    {
      "epoch": 1.5,
      "grad_norm": 7.198966971163168,
      "learning_rate": 1.5433443831998617e-06,
      "loss": 0.5603,
      "step": 16829
    },
    {
      "epoch": 1.5,
      "grad_norm": 7.5852795176589165,
      "learning_rate": 1.5428224785257905e-06,
      "loss": 0.5737,
      "step": 16830
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.850092883001039,
      "learning_rate": 1.542300646012731e-06,
      "loss": 0.5121,
      "step": 16831
    },
    {
      "epoch": 1.5,
      "grad_norm": 7.131816184152208,
      "learning_rate": 1.5417788856715715e-06,
      "loss": 0.5183,
      "step": 16832
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.129416243612443,
      "learning_rate": 1.541257197513204e-06,
      "loss": 0.5946,
      "step": 16833
    },
    {
      "epoch": 1.5,
      "grad_norm": 4.914848007828148,
      "learning_rate": 1.5407355815485186e-06,
      "loss": 0.5603,
      "step": 16834
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.808754844607723,
      "learning_rate": 1.5402140377884033e-06,
      "loss": 0.6444,
      "step": 16835
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.360562096805939,
      "learning_rate": 1.5396925662437418e-06,
      "loss": 0.6203,
      "step": 16836
    },
    {
      "epoch": 1.5,
      "grad_norm": 7.416345933163767,
      "learning_rate": 1.53917116692542e-06,
      "loss": 0.5844,
      "step": 16837
    },
    {
      "epoch": 1.5,
      "grad_norm": 8.215498326348715,
      "learning_rate": 1.5386498398443212e-06,
      "loss": 0.6025,
      "step": 16838
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.457286907001019,
      "learning_rate": 1.5381285850113265e-06,
      "loss": 0.5717,
      "step": 16839
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.929342921416541,
      "learning_rate": 1.5376074024373177e-06,
      "loss": 0.5899,
      "step": 16840
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.6516565144397735,
      "learning_rate": 1.5370862921331702e-06,
      "loss": 0.6176,
      "step": 16841
    },
    {
      "epoch": 1.5,
      "grad_norm": 7.63403378821846,
      "learning_rate": 1.5365652541097632e-06,
      "loss": 0.5112,
      "step": 16842
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.619954474913986,
      "learning_rate": 1.5360442883779714e-06,
      "loss": 0.5737,
      "step": 16843
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.496357411073156,
      "learning_rate": 1.5355233949486692e-06,
      "loss": 0.5969,
      "step": 16844
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.6564859816153,
      "learning_rate": 1.5350025738327306e-06,
      "loss": 0.6549,
      "step": 16845
    },
    {
      "epoch": 1.5,
      "grad_norm": 4.953769198471918,
      "learning_rate": 1.5344818250410236e-06,
      "loss": 0.5996,
      "step": 16846
    },
    {
      "epoch": 1.5,
      "grad_norm": 7.081171506946415,
      "learning_rate": 1.533961148584421e-06,
      "loss": 0.5547,
      "step": 16847
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.497737228076618,
      "learning_rate": 1.5334405444737872e-06,
      "loss": 0.5209,
      "step": 16848
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.530693774119447,
      "learning_rate": 1.53292001271999e-06,
      "loss": 0.554,
      "step": 16849
    },
    {
      "epoch": 1.5,
      "grad_norm": 7.3617373942971,
      "learning_rate": 1.5323995533338948e-06,
      "loss": 0.5859,
      "step": 16850
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.566887394395408,
      "learning_rate": 1.5318791663263655e-06,
      "loss": 0.653,
      "step": 16851
    },
    {
      "epoch": 1.5,
      "grad_norm": 8.126462117665236,
      "learning_rate": 1.5313588517082629e-06,
      "loss": 0.5672,
      "step": 16852
    },
    {
      "epoch": 1.5,
      "grad_norm": 8.362663880686503,
      "learning_rate": 1.5308386094904503e-06,
      "loss": 0.643,
      "step": 16853
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.804395003636447,
      "learning_rate": 1.5303184396837822e-06,
      "loss": 0.5192,
      "step": 16854
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.792810022007835,
      "learning_rate": 1.5297983422991185e-06,
      "loss": 0.5922,
      "step": 16855
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.208169827099368,
      "learning_rate": 1.5292783173473148e-06,
      "loss": 0.5378,
      "step": 16856
    },
    {
      "epoch": 1.5,
      "grad_norm": 14.911308930877057,
      "learning_rate": 1.5287583648392257e-06,
      "loss": 0.5516,
      "step": 16857
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.567176430498715,
      "learning_rate": 1.5282384847857034e-06,
      "loss": 0.5936,
      "step": 16858
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.226493358083275,
      "learning_rate": 1.5277186771976017e-06,
      "loss": 0.685,
      "step": 16859
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.703301127168551,
      "learning_rate": 1.527198942085768e-06,
      "loss": 0.5947,
      "step": 16860
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.00481229006696,
      "learning_rate": 1.5266792794610502e-06,
      "loss": 0.5947,
      "step": 16861
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.339412458804203,
      "learning_rate": 1.526159689334295e-06,
      "loss": 0.5903,
      "step": 16862
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.2605824636914695,
      "learning_rate": 1.5256401717163493e-06,
      "loss": 0.5978,
      "step": 16863
    },
    {
      "epoch": 1.5,
      "grad_norm": 8.589631169591014,
      "learning_rate": 1.5251207266180563e-06,
      "loss": 0.5784,
      "step": 16864
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.845554417266622,
      "learning_rate": 1.5246013540502585e-06,
      "loss": 0.5875,
      "step": 16865
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.320865328208996,
      "learning_rate": 1.5240820540237977e-06,
      "loss": 0.5528,
      "step": 16866
    },
    {
      "epoch": 1.5,
      "grad_norm": 7.946965172803159,
      "learning_rate": 1.52356282654951e-06,
      "loss": 0.6102,
      "step": 16867
    },
    {
      "epoch": 1.5,
      "grad_norm": 7.816021703274899,
      "learning_rate": 1.5230436716382357e-06,
      "loss": 0.5848,
      "step": 16868
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.143598880538589,
      "learning_rate": 1.52252458930081e-06,
      "loss": 0.5734,
      "step": 16869
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.1002160057754375,
      "learning_rate": 1.5220055795480682e-06,
      "loss": 0.5517,
      "step": 16870
    },
    {
      "epoch": 1.51,
      "grad_norm": 5.533570892571663,
      "learning_rate": 1.5214866423908426e-06,
      "loss": 0.5923,
      "step": 16871
    },
    {
      "epoch": 1.51,
      "grad_norm": 8.638158150679322,
      "learning_rate": 1.5209677778399679e-06,
      "loss": 0.6162,
      "step": 16872
    },
    {
      "epoch": 1.51,
      "grad_norm": 7.7771281823540654,
      "learning_rate": 1.5204489859062698e-06,
      "loss": 0.6016,
      "step": 16873
    },
    {
      "epoch": 1.51,
      "grad_norm": 5.137086332278969,
      "learning_rate": 1.5199302666005806e-06,
      "loss": 0.5689,
      "step": 16874
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.236746757028715,
      "learning_rate": 1.5194116199337244e-06,
      "loss": 0.6159,
      "step": 16875
    },
    {
      "epoch": 1.51,
      "grad_norm": 7.421301297833723,
      "learning_rate": 1.5188930459165285e-06,
      "loss": 0.516,
      "step": 16876
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.369528047401714,
      "learning_rate": 1.5183745445598163e-06,
      "loss": 0.5652,
      "step": 16877
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.268908387895806,
      "learning_rate": 1.5178561158744114e-06,
      "loss": 0.5954,
      "step": 16878
    },
    {
      "epoch": 1.51,
      "grad_norm": 5.785568696390738,
      "learning_rate": 1.5173377598711358e-06,
      "loss": 0.5547,
      "step": 16879
    },
    {
      "epoch": 1.51,
      "grad_norm": 4.963414360977335,
      "learning_rate": 1.5168194765608056e-06,
      "loss": 0.551,
      "step": 16880
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.096847788836213,
      "learning_rate": 1.5163012659542415e-06,
      "loss": 0.6148,
      "step": 16881
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.475800259595204,
      "learning_rate": 1.5157831280622588e-06,
      "loss": 0.598,
      "step": 16882
    },
    {
      "epoch": 1.51,
      "grad_norm": 5.516214648830509,
      "learning_rate": 1.5152650628956728e-06,
      "loss": 0.5884,
      "step": 16883
    },
    {
      "epoch": 1.51,
      "grad_norm": 7.4927402721726475,
      "learning_rate": 1.5147470704652972e-06,
      "loss": 0.6516,
      "step": 16884
    },
    {
      "epoch": 1.51,
      "grad_norm": 5.312731820842023,
      "learning_rate": 1.5142291507819456e-06,
      "loss": 0.6066,
      "step": 16885
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.698428053854762,
      "learning_rate": 1.5137113038564255e-06,
      "loss": 0.5768,
      "step": 16886
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.380225913351903,
      "learning_rate": 1.513193529699547e-06,
      "loss": 0.6028,
      "step": 16887
    },
    {
      "epoch": 1.51,
      "grad_norm": 5.970754808949401,
      "learning_rate": 1.512675828322117e-06,
      "loss": 0.5253,
      "step": 16888
    },
    {
      "epoch": 1.51,
      "grad_norm": 8.068637685352222,
      "learning_rate": 1.5121581997349444e-06,
      "loss": 0.6,
      "step": 16889
    },
    {
      "epoch": 1.51,
      "grad_norm": 7.303789554567294,
      "learning_rate": 1.511640643948829e-06,
      "loss": 0.5473,
      "step": 16890
    },
    {
      "epoch": 1.51,
      "grad_norm": 8.353584205687582,
      "learning_rate": 1.5111231609745757e-06,
      "loss": 0.6044,
      "step": 16891
    },
    {
      "epoch": 1.51,
      "grad_norm": 5.94678239147186,
      "learning_rate": 1.5106057508229883e-06,
      "loss": 0.5684,
      "step": 16892
    },
    {
      "epoch": 1.51,
      "grad_norm": 13.28060723767068,
      "learning_rate": 1.510088413504862e-06,
      "loss": 0.5193,
      "step": 16893
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.941891450640458,
      "learning_rate": 1.5095711490309971e-06,
      "loss": 0.5588,
      "step": 16894
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.573108442517075,
      "learning_rate": 1.509053957412191e-06,
      "loss": 0.5752,
      "step": 16895
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.322290896729642,
      "learning_rate": 1.508536838659238e-06,
      "loss": 0.5518,
      "step": 16896
    },
    {
      "epoch": 1.51,
      "grad_norm": 5.698157130322034,
      "learning_rate": 1.5080197927829327e-06,
      "loss": 0.6051,
      "step": 16897
    },
    {
      "epoch": 1.51,
      "grad_norm": 4.949685492741059,
      "learning_rate": 1.507502819794069e-06,
      "loss": 0.6003,
      "step": 16898
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.625093219866395,
      "learning_rate": 1.5069859197034331e-06,
      "loss": 0.5506,
      "step": 16899
    },
    {
      "epoch": 1.51,
      "grad_norm": 7.065878393403962,
      "learning_rate": 1.5064690925218168e-06,
      "loss": 0.5771,
      "step": 16900
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.831908400791802,
      "learning_rate": 1.505952338260007e-06,
      "loss": 0.522,
      "step": 16901
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.368798533246033,
      "learning_rate": 1.5054356569287915e-06,
      "loss": 0.5834,
      "step": 16902
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.587107382856981,
      "learning_rate": 1.5049190485389542e-06,
      "loss": 0.5537,
      "step": 16903
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.1794080740734065,
      "learning_rate": 1.5044025131012768e-06,
      "loss": 0.5512,
      "step": 16904
    },
    {
      "epoch": 1.51,
      "grad_norm": 7.234934825915811,
      "learning_rate": 1.503886050626543e-06,
      "loss": 0.6082,
      "step": 16905
    },
    {
      "epoch": 1.51,
      "grad_norm": 5.058257044554382,
      "learning_rate": 1.5033696611255295e-06,
      "loss": 0.5819,
      "step": 16906
    },
    {
      "epoch": 1.51,
      "grad_norm": 8.229679049709825,
      "learning_rate": 1.5028533446090177e-06,
      "loss": 0.5977,
      "step": 16907
    },
    {
      "epoch": 1.51,
      "grad_norm": 4.947598913005139,
      "learning_rate": 1.502337101087783e-06,
      "loss": 0.5931,
      "step": 16908
    },
    {
      "epoch": 1.51,
      "grad_norm": 7.108274904674371,
      "learning_rate": 1.5018209305726018e-06,
      "loss": 0.5713,
      "step": 16909
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.579044993432209,
      "learning_rate": 1.501304833074248e-06,
      "loss": 0.5767,
      "step": 16910
    },
    {
      "epoch": 1.51,
      "grad_norm": 5.5350739747985465,
      "learning_rate": 1.500788808603495e-06,
      "loss": 0.5453,
      "step": 16911
    },
    {
      "epoch": 1.51,
      "grad_norm": 8.186964577244913,
      "learning_rate": 1.5002728571711112e-06,
      "loss": 0.5772,
      "step": 16912
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.649075751012441,
      "learning_rate": 1.4997569787878669e-06,
      "loss": 0.5231,
      "step": 16913
    },
    {
      "epoch": 1.51,
      "grad_norm": 9.922576033936998,
      "learning_rate": 1.4992411734645306e-06,
      "loss": 0.5571,
      "step": 16914
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.122536903201537,
      "learning_rate": 1.498725441211868e-06,
      "loss": 0.6606,
      "step": 16915
    },
    {
      "epoch": 1.51,
      "grad_norm": 11.676300874363852,
      "learning_rate": 1.498209782040646e-06,
      "loss": 0.5837,
      "step": 16916
    },
    {
      "epoch": 1.51,
      "grad_norm": 7.401980000517567,
      "learning_rate": 1.4976941959616243e-06,
      "loss": 0.5333,
      "step": 16917
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.505126893853667,
      "learning_rate": 1.4971786829855677e-06,
      "loss": 0.5397,
      "step": 16918
    },
    {
      "epoch": 1.51,
      "grad_norm": 5.121363167108908,
      "learning_rate": 1.4966632431232342e-06,
      "loss": 0.5616,
      "step": 16919
    },
    {
      "epoch": 1.51,
      "grad_norm": 8.371938913630846,
      "learning_rate": 1.4961478763853826e-06,
      "loss": 0.5583,
      "step": 16920
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.032093087233527,
      "learning_rate": 1.495632582782771e-06,
      "loss": 0.6029,
      "step": 16921
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.167540307889497,
      "learning_rate": 1.495117362326155e-06,
      "loss": 0.6352,
      "step": 16922
    },
    {
      "epoch": 1.51,
      "grad_norm": 9.281804431728423,
      "learning_rate": 1.4946022150262895e-06,
      "loss": 0.6379,
      "step": 16923
    },
    {
      "epoch": 1.51,
      "grad_norm": 5.6998891308402575,
      "learning_rate": 1.494087140893925e-06,
      "loss": 0.5748,
      "step": 16924
    },
    {
      "epoch": 1.51,
      "grad_norm": 11.021861221179147,
      "learning_rate": 1.4935721399398134e-06,
      "loss": 0.5667,
      "step": 16925
    },
    {
      "epoch": 1.51,
      "grad_norm": 8.205758654106678,
      "learning_rate": 1.493057212174705e-06,
      "loss": 0.5488,
      "step": 16926
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.673298460605711,
      "learning_rate": 1.4925423576093467e-06,
      "loss": 0.5189,
      "step": 16927
    },
    {
      "epoch": 1.51,
      "grad_norm": 5.550696210393093,
      "learning_rate": 1.4920275762544855e-06,
      "loss": 0.6093,
      "step": 16928
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.259836067506893,
      "learning_rate": 1.491512868120868e-06,
      "loss": 0.5832,
      "step": 16929
    },
    {
      "epoch": 1.51,
      "grad_norm": 5.432706063009192,
      "learning_rate": 1.4909982332192342e-06,
      "loss": 0.5367,
      "step": 16930
    },
    {
      "epoch": 1.51,
      "grad_norm": 7.337540694619478,
      "learning_rate": 1.4904836715603277e-06,
      "loss": 0.5521,
      "step": 16931
    },
    {
      "epoch": 1.51,
      "grad_norm": 8.155671258102808,
      "learning_rate": 1.4899691831548906e-06,
      "loss": 0.6556,
      "step": 16932
    },
    {
      "epoch": 1.51,
      "grad_norm": 8.125246501037363,
      "learning_rate": 1.4894547680136583e-06,
      "loss": 0.5465,
      "step": 16933
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.331457651777294,
      "learning_rate": 1.4889404261473695e-06,
      "loss": 0.5679,
      "step": 16934
    },
    {
      "epoch": 1.51,
      "grad_norm": 5.675738305890865,
      "learning_rate": 1.4884261575667609e-06,
      "loss": 0.5328,
      "step": 16935
    },
    {
      "epoch": 1.51,
      "grad_norm": 8.919153645497104,
      "learning_rate": 1.4879119622825665e-06,
      "loss": 0.6,
      "step": 16936
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.928387492889682,
      "learning_rate": 1.4873978403055177e-06,
      "loss": 0.604,
      "step": 16937
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.800001362905879,
      "learning_rate": 1.486883791646347e-06,
      "loss": 0.605,
      "step": 16938
    },
    {
      "epoch": 1.51,
      "grad_norm": 13.127859140706317,
      "learning_rate": 1.4863698163157824e-06,
      "loss": 0.6081,
      "step": 16939
    },
    {
      "epoch": 1.51,
      "grad_norm": 5.54511341384218,
      "learning_rate": 1.4858559143245544e-06,
      "loss": 0.5413,
      "step": 16940
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.357791716089701,
      "learning_rate": 1.4853420856833873e-06,
      "loss": 0.5996,
      "step": 16941
    },
    {
      "epoch": 1.51,
      "grad_norm": 5.999622745628324,
      "learning_rate": 1.4848283304030092e-06,
      "loss": 0.5738,
      "step": 16942
    },
    {
      "epoch": 1.51,
      "grad_norm": 5.81557122094534,
      "learning_rate": 1.48431464849414e-06,
      "loss": 0.6007,
      "step": 16943
    },
    {
      "epoch": 1.51,
      "grad_norm": 5.976561085509542,
      "learning_rate": 1.4838010399675034e-06,
      "loss": 0.563,
      "step": 16944
    },
    {
      "epoch": 1.51,
      "grad_norm": 8.007905199390978,
      "learning_rate": 1.4832875048338197e-06,
      "loss": 0.6301,
      "step": 16945
    },
    {
      "epoch": 1.51,
      "grad_norm": 8.187648323041252,
      "learning_rate": 1.4827740431038096e-06,
      "loss": 0.5219,
      "step": 16946
    },
    {
      "epoch": 1.51,
      "grad_norm": 7.613614816891229,
      "learning_rate": 1.4822606547881868e-06,
      "loss": 0.662,
      "step": 16947
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.135455813000128,
      "learning_rate": 1.4817473398976694e-06,
      "loss": 0.5469,
      "step": 16948
    },
    {
      "epoch": 1.51,
      "grad_norm": 7.154659253153712,
      "learning_rate": 1.481234098442973e-06,
      "loss": 0.5628,
      "step": 16949
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.954271222745732,
      "learning_rate": 1.480720930434807e-06,
      "loss": 0.5569,
      "step": 16950
    },
    {
      "epoch": 1.51,
      "grad_norm": 4.970741520091181,
      "learning_rate": 1.4802078358838846e-06,
      "loss": 0.5359,
      "step": 16951
    },
    {
      "epoch": 1.51,
      "grad_norm": 10.556978732077475,
      "learning_rate": 1.4796948148009154e-06,
      "loss": 0.6334,
      "step": 16952
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.148771128113096,
      "learning_rate": 1.479181867196608e-06,
      "loss": 0.6177,
      "step": 16953
    },
    {
      "epoch": 1.51,
      "grad_norm": 9.529060178580059,
      "learning_rate": 1.4786689930816684e-06,
      "loss": 0.5555,
      "step": 16954
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.2290759468001955,
      "learning_rate": 1.478156192466803e-06,
      "loss": 0.6116,
      "step": 16955
    },
    {
      "epoch": 1.51,
      "grad_norm": 10.441380037035078,
      "learning_rate": 1.4776434653627135e-06,
      "loss": 0.6153,
      "step": 16956
    },
    {
      "epoch": 1.51,
      "grad_norm": 7.374788467635253,
      "learning_rate": 1.4771308117801026e-06,
      "loss": 0.6498,
      "step": 16957
    },
    {
      "epoch": 1.51,
      "grad_norm": 8.40346301075115,
      "learning_rate": 1.4766182317296712e-06,
      "loss": 0.575,
      "step": 16958
    },
    {
      "epoch": 1.51,
      "grad_norm": 8.06572885937927,
      "learning_rate": 1.4761057252221177e-06,
      "loss": 0.5665,
      "step": 16959
    },
    {
      "epoch": 1.51,
      "grad_norm": 5.017676081238112,
      "learning_rate": 1.4755932922681405e-06,
      "loss": 0.5842,
      "step": 16960
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.9089317805024075,
      "learning_rate": 1.4750809328784366e-06,
      "loss": 0.5624,
      "step": 16961
    },
    {
      "epoch": 1.51,
      "grad_norm": 8.170103075492344,
      "learning_rate": 1.4745686470636983e-06,
      "loss": 0.5871,
      "step": 16962
    },
    {
      "epoch": 1.51,
      "grad_norm": 9.35976995624796,
      "learning_rate": 1.4740564348346175e-06,
      "loss": 0.5351,
      "step": 16963
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.474340321253286,
      "learning_rate": 1.4735442962018875e-06,
      "loss": 0.5579,
      "step": 16964
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.77185154503553,
      "learning_rate": 1.4730322311761969e-06,
      "loss": 0.5737,
      "step": 16965
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.330677686238066,
      "learning_rate": 1.4725202397682349e-06,
      "loss": 0.5854,
      "step": 16966
    },
    {
      "epoch": 1.51,
      "grad_norm": 8.833550313847681,
      "learning_rate": 1.4720083219886876e-06,
      "loss": 0.6223,
      "step": 16967
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.901323390193607,
      "learning_rate": 1.4714964778482421e-06,
      "loss": 0.613,
      "step": 16968
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.134137593856659,
      "learning_rate": 1.4709847073575789e-06,
      "loss": 0.5695,
      "step": 16969
    },
    {
      "epoch": 1.51,
      "grad_norm": 4.269011154381314,
      "learning_rate": 1.4704730105273812e-06,
      "loss": 0.5585,
      "step": 16970
    },
    {
      "epoch": 1.51,
      "grad_norm": 8.726669617280312,
      "learning_rate": 1.4699613873683299e-06,
      "loss": 0.5795,
      "step": 16971
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.454584544124313,
      "learning_rate": 1.4694498378911042e-06,
      "loss": 0.5705,
      "step": 16972
    },
    {
      "epoch": 1.51,
      "grad_norm": 7.826911093720265,
      "learning_rate": 1.4689383621063813e-06,
      "loss": 0.5251,
      "step": 16973
    },
    {
      "epoch": 1.51,
      "grad_norm": 7.989212286133505,
      "learning_rate": 1.4684269600248386e-06,
      "loss": 0.5691,
      "step": 16974
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.72731095518736,
      "learning_rate": 1.4679156316571491e-06,
      "loss": 0.563,
      "step": 16975
    },
    {
      "epoch": 1.51,
      "grad_norm": 7.127027033232882,
      "learning_rate": 1.4674043770139845e-06,
      "loss": 0.5999,
      "step": 16976
    },
    {
      "epoch": 1.51,
      "grad_norm": 7.8670026110484255,
      "learning_rate": 1.4668931961060167e-06,
      "loss": 0.5319,
      "step": 16977
    },
    {
      "epoch": 1.51,
      "grad_norm": 7.966683172180466,
      "learning_rate": 1.4663820889439162e-06,
      "loss": 0.5767,
      "step": 16978
    },
    {
      "epoch": 1.51,
      "grad_norm": 7.208040738568394,
      "learning_rate": 1.4658710555383514e-06,
      "loss": 0.5613,
      "step": 16979
    },
    {
      "epoch": 1.51,
      "grad_norm": 7.65608397514103,
      "learning_rate": 1.4653600958999886e-06,
      "loss": 0.5928,
      "step": 16980
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.730181792705907,
      "learning_rate": 1.4648492100394945e-06,
      "loss": 0.5538,
      "step": 16981
    },
    {
      "epoch": 1.51,
      "grad_norm": 8.956717601004273,
      "learning_rate": 1.4643383979675301e-06,
      "loss": 0.5375,
      "step": 16982
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.711230511663297,
      "learning_rate": 1.4638276596947582e-06,
      "loss": 0.6045,
      "step": 16983
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.716785830959988,
      "learning_rate": 1.4633169952318399e-06,
      "loss": 0.5689,
      "step": 16984
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.865416214991,
      "learning_rate": 1.462806404589434e-06,
      "loss": 0.554,
      "step": 16985
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.223765876311126,
      "learning_rate": 1.4622958877781984e-06,
      "loss": 0.5815,
      "step": 16986
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.651384479422669,
      "learning_rate": 1.4617854448087903e-06,
      "loss": 0.5987,
      "step": 16987
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.732375905213915,
      "learning_rate": 1.461275075691861e-06,
      "loss": 0.5575,
      "step": 16988
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.568258258778996,
      "learning_rate": 1.460764780438066e-06,
      "loss": 0.5907,
      "step": 16989
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.772168713986043,
      "learning_rate": 1.4602545590580542e-06,
      "loss": 0.5847,
      "step": 16990
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.903571049075014,
      "learning_rate": 1.4597444115624765e-06,
      "loss": 0.5781,
      "step": 16991
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.697963815901953,
      "learning_rate": 1.4592343379619811e-06,
      "loss": 0.5401,
      "step": 16992
    },
    {
      "epoch": 1.52,
      "grad_norm": 8.207606374399852,
      "learning_rate": 1.4587243382672144e-06,
      "loss": 0.5883,
      "step": 16993
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.010681293929883,
      "learning_rate": 1.4582144124888236e-06,
      "loss": 0.5601,
      "step": 16994
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.924733992958024,
      "learning_rate": 1.4577045606374495e-06,
      "loss": 0.569,
      "step": 16995
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.149419950985524,
      "learning_rate": 1.4571947827237347e-06,
      "loss": 0.6002,
      "step": 16996
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.548513618143878,
      "learning_rate": 1.4566850787583203e-06,
      "loss": 0.6148,
      "step": 16997
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.441147614603664,
      "learning_rate": 1.4561754487518453e-06,
      "loss": 0.5637,
      "step": 16998
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.2704946933655155,
      "learning_rate": 1.4556658927149486e-06,
      "loss": 0.556,
      "step": 16999
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.423241182910366,
      "learning_rate": 1.4551564106582627e-06,
      "loss": 0.5969,
      "step": 17000
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.9610497748127065,
      "learning_rate": 1.4546470025924236e-06,
      "loss": 0.6348,
      "step": 17001
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.374804204856671,
      "learning_rate": 1.4541376685280645e-06,
      "loss": 0.5783,
      "step": 17002
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.287040592895309,
      "learning_rate": 1.4536284084758158e-06,
      "loss": 0.5929,
      "step": 17003
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.9459225754592016,
      "learning_rate": 1.4531192224463092e-06,
      "loss": 0.584,
      "step": 17004
    },
    {
      "epoch": 1.52,
      "grad_norm": 9.381562277913252,
      "learning_rate": 1.45261011045017e-06,
      "loss": 0.5811,
      "step": 17005
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.229638833004457,
      "learning_rate": 1.452101072498026e-06,
      "loss": 0.5529,
      "step": 17006
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.610994267315945,
      "learning_rate": 1.4515921086005042e-06,
      "loss": 0.5768,
      "step": 17007
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.466925359609324,
      "learning_rate": 1.4510832187682238e-06,
      "loss": 0.5811,
      "step": 17008
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.32255262784391,
      "learning_rate": 1.4505744030118101e-06,
      "loss": 0.5812,
      "step": 17009
    },
    {
      "epoch": 1.52,
      "grad_norm": 4.924745742089377,
      "learning_rate": 1.4500656613418823e-06,
      "loss": 0.5679,
      "step": 17010
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.675467232968973,
      "learning_rate": 1.4495569937690596e-06,
      "loss": 0.5792,
      "step": 17011
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.999430014820665,
      "learning_rate": 1.4490484003039612e-06,
      "loss": 0.5404,
      "step": 17012
    },
    {
      "epoch": 1.52,
      "grad_norm": 4.401867955412797,
      "learning_rate": 1.4485398809571987e-06,
      "loss": 0.5566,
      "step": 17013
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.037650000839489,
      "learning_rate": 1.4480314357393894e-06,
      "loss": 0.6018,
      "step": 17014
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.79481737185491,
      "learning_rate": 1.447523064661145e-06,
      "loss": 0.5828,
      "step": 17015
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.246310364582602,
      "learning_rate": 1.4470147677330765e-06,
      "loss": 0.6104,
      "step": 17016
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.248819128479334,
      "learning_rate": 1.4465065449657938e-06,
      "loss": 0.6043,
      "step": 17017
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.37401185373268,
      "learning_rate": 1.445998396369907e-06,
      "loss": 0.5951,
      "step": 17018
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.965007713527907,
      "learning_rate": 1.4454903219560202e-06,
      "loss": 0.5866,
      "step": 17019
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.730367440462978,
      "learning_rate": 1.4449823217347374e-06,
      "loss": 0.6012,
      "step": 17020
    },
    {
      "epoch": 1.52,
      "grad_norm": 9.497564765854772,
      "learning_rate": 1.444474395716663e-06,
      "loss": 0.6141,
      "step": 17021
    },
    {
      "epoch": 1.52,
      "grad_norm": 8.212334247427338,
      "learning_rate": 1.443966543912399e-06,
      "loss": 0.5242,
      "step": 17022
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.866404223139299,
      "learning_rate": 1.4434587663325461e-06,
      "loss": 0.5551,
      "step": 17023
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.111070802165409,
      "learning_rate": 1.4429510629877024e-06,
      "loss": 0.577,
      "step": 17024
    },
    {
      "epoch": 1.52,
      "grad_norm": 8.659516591616146,
      "learning_rate": 1.4424434338884673e-06,
      "loss": 0.597,
      "step": 17025
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.1816492716599205,
      "learning_rate": 1.4419358790454335e-06,
      "loss": 0.5434,
      "step": 17026
    },
    {
      "epoch": 1.52,
      "grad_norm": 8.814873084403002,
      "learning_rate": 1.4414283984691957e-06,
      "loss": 0.5728,
      "step": 17027
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.921360581994144,
      "learning_rate": 1.4409209921703472e-06,
      "loss": 0.5589,
      "step": 17028
    },
    {
      "epoch": 1.52,
      "grad_norm": 8.066792840849494,
      "learning_rate": 1.440413660159478e-06,
      "loss": 0.5547,
      "step": 17029
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.00201394203805,
      "learning_rate": 1.4399064024471787e-06,
      "loss": 0.5737,
      "step": 17030
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.317986570868885,
      "learning_rate": 1.4393992190440387e-06,
      "loss": 0.6148,
      "step": 17031
    },
    {
      "epoch": 1.52,
      "grad_norm": 4.907747874005524,
      "learning_rate": 1.4388921099606406e-06,
      "loss": 0.58,
      "step": 17032
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.510114527045778,
      "learning_rate": 1.4383850752075722e-06,
      "loss": 0.5938,
      "step": 17033
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.330826189766111,
      "learning_rate": 1.4378781147954146e-06,
      "loss": 0.6149,
      "step": 17034
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.533918618248367,
      "learning_rate": 1.4373712287347503e-06,
      "loss": 0.5753,
      "step": 17035
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.501711238818725,
      "learning_rate": 1.4368644170361596e-06,
      "loss": 0.5855,
      "step": 17036
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.731019116024797,
      "learning_rate": 1.4363576797102209e-06,
      "loss": 0.5519,
      "step": 17037
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.399919282005908,
      "learning_rate": 1.4358510167675127e-06,
      "loss": 0.5264,
      "step": 17038
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.7046073530896555,
      "learning_rate": 1.435344428218608e-06,
      "loss": 0.5698,
      "step": 17039
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.659277033431452,
      "learning_rate": 1.434837914074082e-06,
      "loss": 0.6135,
      "step": 17040
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.028677435708255,
      "learning_rate": 1.434331474344507e-06,
      "loss": 0.5683,
      "step": 17041
    },
    {
      "epoch": 1.52,
      "grad_norm": 8.281433100661124,
      "learning_rate": 1.433825109040453e-06,
      "loss": 0.627,
      "step": 17042
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.859346543878757,
      "learning_rate": 1.433318818172491e-06,
      "loss": 0.5969,
      "step": 17043
    },
    {
      "epoch": 1.52,
      "grad_norm": 10.09264460304895,
      "learning_rate": 1.432812601751189e-06,
      "loss": 0.5606,
      "step": 17044
    },
    {
      "epoch": 1.52,
      "grad_norm": 9.056394973278627,
      "learning_rate": 1.43230645978711e-06,
      "loss": 0.6208,
      "step": 17045
    },
    {
      "epoch": 1.52,
      "grad_norm": 8.656508658506924,
      "learning_rate": 1.4318003922908209e-06,
      "loss": 0.5709,
      "step": 17046
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.388810885660052,
      "learning_rate": 1.4312943992728861e-06,
      "loss": 0.6635,
      "step": 17047
    },
    {
      "epoch": 1.52,
      "grad_norm": 4.997425791485858,
      "learning_rate": 1.4307884807438638e-06,
      "loss": 0.5217,
      "step": 17048
    },
    {
      "epoch": 1.52,
      "grad_norm": 4.689121616601459,
      "learning_rate": 1.4302826367143152e-06,
      "loss": 0.5607,
      "step": 17049
    },
    {
      "epoch": 1.52,
      "grad_norm": 8.57077540295167,
      "learning_rate": 1.4297768671947993e-06,
      "loss": 0.539,
      "step": 17050
    },
    {
      "epoch": 1.52,
      "grad_norm": 9.9097316221064,
      "learning_rate": 1.429271172195874e-06,
      "loss": 0.5385,
      "step": 17051
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.644919609914317,
      "learning_rate": 1.4287655517280914e-06,
      "loss": 0.5836,
      "step": 17052
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.89454848095521,
      "learning_rate": 1.4282600058020075e-06,
      "loss": 0.5733,
      "step": 17053
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.425362405136173,
      "learning_rate": 1.427754534428174e-06,
      "loss": 0.5261,
      "step": 17054
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.004830530081999,
      "learning_rate": 1.427249137617141e-06,
      "loss": 0.5791,
      "step": 17055
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.528118408170565,
      "learning_rate": 1.4267438153794587e-06,
      "loss": 0.5715,
      "step": 17056
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.150775696860845,
      "learning_rate": 1.4262385677256752e-06,
      "loss": 0.5999,
      "step": 17057
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.75458757416896,
      "learning_rate": 1.4257333946663337e-06,
      "loss": 0.5496,
      "step": 17058
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.798288314986123,
      "learning_rate": 1.42522829621198e-06,
      "loss": 0.5932,
      "step": 17059
    },
    {
      "epoch": 1.52,
      "grad_norm": 8.47014171041464,
      "learning_rate": 1.4247232723731568e-06,
      "loss": 0.5657,
      "step": 17060
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.539769631878163,
      "learning_rate": 1.4242183231604062e-06,
      "loss": 0.5362,
      "step": 17061
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.090022373630785,
      "learning_rate": 1.4237134485842686e-06,
      "loss": 0.5913,
      "step": 17062
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.395845164162824,
      "learning_rate": 1.4232086486552787e-06,
      "loss": 0.5015,
      "step": 17063
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.100527999433925,
      "learning_rate": 1.4227039233839773e-06,
      "loss": 0.5565,
      "step": 17064
    },
    {
      "epoch": 1.52,
      "grad_norm": 8.364808059342145,
      "learning_rate": 1.4221992727808958e-06,
      "loss": 0.5427,
      "step": 17065
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.333190514829901,
      "learning_rate": 1.4216946968565692e-06,
      "loss": 0.6033,
      "step": 17066
    },
    {
      "epoch": 1.52,
      "grad_norm": 8.795771845250657,
      "learning_rate": 1.4211901956215296e-06,
      "loss": 0.639,
      "step": 17067
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.293848447781189,
      "learning_rate": 1.4206857690863069e-06,
      "loss": 0.6258,
      "step": 17068
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.406113685914998,
      "learning_rate": 1.4201814172614304e-06,
      "loss": 0.5444,
      "step": 17069
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.2203752536094825,
      "learning_rate": 1.4196771401574283e-06,
      "loss": 0.6146,
      "step": 17070
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.890502286589144,
      "learning_rate": 1.419172937784824e-06,
      "loss": 0.5686,
      "step": 17071
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.911029133267699,
      "learning_rate": 1.4186688101541424e-06,
      "loss": 0.5746,
      "step": 17072
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.915699627830184,
      "learning_rate": 1.4181647572759066e-06,
      "loss": 0.651,
      "step": 17073
    },
    {
      "epoch": 1.52,
      "grad_norm": 10.54165062283654,
      "learning_rate": 1.4176607791606378e-06,
      "loss": 0.5701,
      "step": 17074
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.24303961642342,
      "learning_rate": 1.4171568758188559e-06,
      "loss": 0.5378,
      "step": 17075
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.058354805581522,
      "learning_rate": 1.4166530472610767e-06,
      "loss": 0.4983,
      "step": 17076
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.787151245411657,
      "learning_rate": 1.4161492934978193e-06,
      "loss": 0.6237,
      "step": 17077
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.210216407515425,
      "learning_rate": 1.4156456145395953e-06,
      "loss": 0.5399,
      "step": 17078
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.673595822795365,
      "learning_rate": 1.41514201039692e-06,
      "loss": 0.5473,
      "step": 17079
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.11531931345151,
      "learning_rate": 1.4146384810803037e-06,
      "loss": 0.5903,
      "step": 17080
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.473145491893182,
      "learning_rate": 1.4141350266002579e-06,
      "loss": 0.617,
      "step": 17081
    },
    {
      "epoch": 1.52,
      "grad_norm": 4.81752703803079,
      "learning_rate": 1.41363164696729e-06,
      "loss": 0.5195,
      "step": 17082
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.218602758536105,
      "learning_rate": 1.4131283421919095e-06,
      "loss": 0.575,
      "step": 17083
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.949264138943726,
      "learning_rate": 1.412625112284618e-06,
      "loss": 0.5167,
      "step": 17084
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.80423636969925,
      "learning_rate": 1.4121219572559208e-06,
      "loss": 0.5598,
      "step": 17085
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.963377593054064,
      "learning_rate": 1.4116188771163209e-06,
      "loss": 0.5924,
      "step": 17086
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.282565605454972,
      "learning_rate": 1.4111158718763178e-06,
      "loss": 0.6284,
      "step": 17087
    },
    {
      "epoch": 1.52,
      "grad_norm": 8.340652634776813,
      "learning_rate": 1.4106129415464137e-06,
      "loss": 0.6034,
      "step": 17088
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.503411791150942,
      "learning_rate": 1.410110086137102e-06,
      "loss": 0.6064,
      "step": 17089
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.202567400877777,
      "learning_rate": 1.409607305658882e-06,
      "loss": 0.6036,
      "step": 17090
    },
    {
      "epoch": 1.52,
      "grad_norm": 9.54170363205557,
      "learning_rate": 1.4091046001222452e-06,
      "loss": 0.5371,
      "step": 17091
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.181845470674902,
      "learning_rate": 1.4086019695376857e-06,
      "loss": 0.5469,
      "step": 17092
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.247900563601584,
      "learning_rate": 1.4080994139156951e-06,
      "loss": 0.5956,
      "step": 17093
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.203691266473642,
      "learning_rate": 1.4075969332667634e-06,
      "loss": 0.5269,
      "step": 17094
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.292161864616709,
      "learning_rate": 1.4070945276013798e-06,
      "loss": 0.5692,
      "step": 17095
    },
    {
      "epoch": 1.53,
      "grad_norm": 8.72133611494868,
      "learning_rate": 1.4065921969300283e-06,
      "loss": 0.5994,
      "step": 17096
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.712680534444186,
      "learning_rate": 1.4060899412631952e-06,
      "loss": 0.5783,
      "step": 17097
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.3852637369112,
      "learning_rate": 1.4055877606113638e-06,
      "loss": 0.596,
      "step": 17098
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.538009025413003,
      "learning_rate": 1.4050856549850167e-06,
      "loss": 0.6015,
      "step": 17099
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.085812333683136,
      "learning_rate": 1.404583624394633e-06,
      "loss": 0.5941,
      "step": 17100
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.544788132678288,
      "learning_rate": 1.4040816688506947e-06,
      "loss": 0.5262,
      "step": 17101
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.762615506556841,
      "learning_rate": 1.4035797883636748e-06,
      "loss": 0.5381,
      "step": 17102
    },
    {
      "epoch": 1.53,
      "grad_norm": 8.341307263927458,
      "learning_rate": 1.403077982944051e-06,
      "loss": 0.5942,
      "step": 17103
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.613060596095744,
      "learning_rate": 1.4025762526022967e-06,
      "loss": 0.5247,
      "step": 17104
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.486935404739469,
      "learning_rate": 1.402074597348887e-06,
      "loss": 0.5158,
      "step": 17105
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.407385358804876,
      "learning_rate": 1.4015730171942887e-06,
      "loss": 0.5187,
      "step": 17106
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.7347560973513545,
      "learning_rate": 1.4010715121489738e-06,
      "loss": 0.5418,
      "step": 17107
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.475381948460167,
      "learning_rate": 1.4005700822234103e-06,
      "loss": 0.5765,
      "step": 17108
    },
    {
      "epoch": 1.53,
      "grad_norm": 8.123065194099311,
      "learning_rate": 1.4000687274280628e-06,
      "loss": 0.5424,
      "step": 17109
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.005833465060595,
      "learning_rate": 1.3995674477733966e-06,
      "loss": 0.5853,
      "step": 17110
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.422340420733775,
      "learning_rate": 1.3990662432698754e-06,
      "loss": 0.5706,
      "step": 17111
    },
    {
      "epoch": 1.53,
      "grad_norm": 8.434492507514486,
      "learning_rate": 1.3985651139279604e-06,
      "loss": 0.6086,
      "step": 17112
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.831721890267589,
      "learning_rate": 1.3980640597581118e-06,
      "loss": 0.5311,
      "step": 17113
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.768505560551281,
      "learning_rate": 1.397563080770789e-06,
      "loss": 0.6147,
      "step": 17114
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.554759950822776,
      "learning_rate": 1.3970621769764464e-06,
      "loss": 0.5395,
      "step": 17115
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.00255772985404,
      "learning_rate": 1.3965613483855407e-06,
      "loss": 0.6163,
      "step": 17116
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.909666508039577,
      "learning_rate": 1.396060595008525e-06,
      "loss": 0.5831,
      "step": 17117
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.895321846788557,
      "learning_rate": 1.395559916855852e-06,
      "loss": 0.6005,
      "step": 17118
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.048870029571252,
      "learning_rate": 1.3950593139379737e-06,
      "loss": 0.5612,
      "step": 17119
    },
    {
      "epoch": 1.53,
      "grad_norm": 4.923943264471194,
      "learning_rate": 1.3945587862653364e-06,
      "loss": 0.5977,
      "step": 17120
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.3913623112344675,
      "learning_rate": 1.3940583338483898e-06,
      "loss": 0.5877,
      "step": 17121
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.608899038285433,
      "learning_rate": 1.393557956697577e-06,
      "loss": 0.5143,
      "step": 17122
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.311225548446261,
      "learning_rate": 1.3930576548233444e-06,
      "loss": 0.5499,
      "step": 17123
    },
    {
      "epoch": 1.53,
      "grad_norm": 8.356160989808739,
      "learning_rate": 1.3925574282361337e-06,
      "loss": 0.5633,
      "step": 17124
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.524088672353561,
      "learning_rate": 1.3920572769463863e-06,
      "loss": 0.5766,
      "step": 17125
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.633612419111581,
      "learning_rate": 1.3915572009645424e-06,
      "loss": 0.53,
      "step": 17126
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.424412973553313,
      "learning_rate": 1.3910572003010409e-06,
      "loss": 0.61,
      "step": 17127
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.708626771784581,
      "learning_rate": 1.390557274966316e-06,
      "loss": 0.5425,
      "step": 17128
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.825640530283765,
      "learning_rate": 1.3900574249708026e-06,
      "loss": 0.5403,
      "step": 17129
    },
    {
      "epoch": 1.53,
      "grad_norm": 8.565101484358312,
      "learning_rate": 1.3895576503249353e-06,
      "loss": 0.5722,
      "step": 17130
    },
    {
      "epoch": 1.53,
      "grad_norm": 8.473971459554877,
      "learning_rate": 1.389057951039145e-06,
      "loss": 0.5888,
      "step": 17131
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.2858879451788425,
      "learning_rate": 1.3885583271238623e-06,
      "loss": 0.5836,
      "step": 17132
    },
    {
      "epoch": 1.53,
      "grad_norm": 8.183077975943158,
      "learning_rate": 1.3880587785895172e-06,
      "loss": 0.5656,
      "step": 17133
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.873234073700398,
      "learning_rate": 1.3875593054465354e-06,
      "loss": 0.5863,
      "step": 17134
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.914949148826478,
      "learning_rate": 1.3870599077053398e-06,
      "loss": 0.604,
      "step": 17135
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.960098597862141,
      "learning_rate": 1.3865605853763565e-06,
      "loss": 0.6009,
      "step": 17136
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.746782057245046,
      "learning_rate": 1.3860613384700079e-06,
      "loss": 0.5696,
      "step": 17137
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.092434252637906,
      "learning_rate": 1.3855621669967146e-06,
      "loss": 0.568,
      "step": 17138
    },
    {
      "epoch": 1.53,
      "grad_norm": 8.059536043403599,
      "learning_rate": 1.3850630709668955e-06,
      "loss": 0.6352,
      "step": 17139
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.441010218434534,
      "learning_rate": 1.3845640503909697e-06,
      "loss": 0.608,
      "step": 17140
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.52485271955215,
      "learning_rate": 1.3840651052793502e-06,
      "loss": 0.5795,
      "step": 17141
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.119131884496756,
      "learning_rate": 1.3835662356424523e-06,
      "loss": 0.544,
      "step": 17142
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.593199791405027,
      "learning_rate": 1.3830674414906903e-06,
      "loss": 0.519,
      "step": 17143
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.532707441828155,
      "learning_rate": 1.3825687228344737e-06,
      "loss": 0.5294,
      "step": 17144
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.736672632322665,
      "learning_rate": 1.3820700796842134e-06,
      "loss": 0.5289,
      "step": 17145
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.5041837512061,
      "learning_rate": 1.3815715120503187e-06,
      "loss": 0.596,
      "step": 17146
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.36567910027059,
      "learning_rate": 1.3810730199431926e-06,
      "loss": 0.5792,
      "step": 17147
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.35250356153874,
      "learning_rate": 1.3805746033732436e-06,
      "loss": 0.5448,
      "step": 17148
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.967663569395669,
      "learning_rate": 1.3800762623508718e-06,
      "loss": 0.5656,
      "step": 17149
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.128653795764353,
      "learning_rate": 1.3795779968864809e-06,
      "loss": 0.5399,
      "step": 17150
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.468037630911577,
      "learning_rate": 1.3790798069904704e-06,
      "loss": 0.5653,
      "step": 17151
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.252543886713164,
      "learning_rate": 1.3785816926732386e-06,
      "loss": 0.5323,
      "step": 17152
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.652344369408779,
      "learning_rate": 1.378083653945186e-06,
      "loss": 0.5912,
      "step": 17153
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.746934849512741,
      "learning_rate": 1.377585690816703e-06,
      "loss": 0.6201,
      "step": 17154
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.595270129921932,
      "learning_rate": 1.3770878032981855e-06,
      "loss": 0.5475,
      "step": 17155
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.458343033265828,
      "learning_rate": 1.3765899914000264e-06,
      "loss": 0.5211,
      "step": 17156
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.396505445998879,
      "learning_rate": 1.3760922551326161e-06,
      "loss": 0.5756,
      "step": 17157
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.409950084025306,
      "learning_rate": 1.3755945945063437e-06,
      "loss": 0.5732,
      "step": 17158
    },
    {
      "epoch": 1.53,
      "grad_norm": 8.50468168717987,
      "learning_rate": 1.3750970095315985e-06,
      "loss": 0.5691,
      "step": 17159
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.73658086535854,
      "learning_rate": 1.3745995002187628e-06,
      "loss": 0.5813,
      "step": 17160
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.64410403907359,
      "learning_rate": 1.374102066578224e-06,
      "loss": 0.5058,
      "step": 17161
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.6634437558626995,
      "learning_rate": 1.3736047086203646e-06,
      "loss": 0.557,
      "step": 17162
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.693474412219433,
      "learning_rate": 1.3731074263555644e-06,
      "loss": 0.599,
      "step": 17163
    },
    {
      "epoch": 1.53,
      "grad_norm": 8.829400735690452,
      "learning_rate": 1.3726102197942032e-06,
      "loss": 0.6393,
      "step": 17164
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.040320654644621,
      "learning_rate": 1.3721130889466605e-06,
      "loss": 0.5803,
      "step": 17165
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.424055474795878,
      "learning_rate": 1.371616033823313e-06,
      "loss": 0.6783,
      "step": 17166
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.325946275760738,
      "learning_rate": 1.3711190544345338e-06,
      "loss": 0.5596,
      "step": 17167
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.772113740904194,
      "learning_rate": 1.370622150790697e-06,
      "loss": 0.6217,
      "step": 17168
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.1436612950390606,
      "learning_rate": 1.3701253229021743e-06,
      "loss": 0.6003,
      "step": 17169
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.188933023333301,
      "learning_rate": 1.3696285707793367e-06,
      "loss": 0.5531,
      "step": 17170
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.660661952164542,
      "learning_rate": 1.3691318944325537e-06,
      "loss": 0.5639,
      "step": 17171
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.404268850189079,
      "learning_rate": 1.3686352938721891e-06,
      "loss": 0.562,
      "step": 17172
    },
    {
      "epoch": 1.53,
      "grad_norm": 8.356135569509364,
      "learning_rate": 1.3681387691086106e-06,
      "loss": 0.5895,
      "step": 17173
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.963103654957032,
      "learning_rate": 1.3676423201521816e-06,
      "loss": 0.5697,
      "step": 17174
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.664938026634077,
      "learning_rate": 1.3671459470132642e-06,
      "loss": 0.5716,
      "step": 17175
    },
    {
      "epoch": 1.53,
      "grad_norm": 4.934932144087231,
      "learning_rate": 1.366649649702219e-06,
      "loss": 0.5673,
      "step": 17176
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.735893107563485,
      "learning_rate": 1.3661534282294076e-06,
      "loss": 0.589,
      "step": 17177
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.922995519893428,
      "learning_rate": 1.3656572826051855e-06,
      "loss": 0.5825,
      "step": 17178
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.383885688005431,
      "learning_rate": 1.3651612128399066e-06,
      "loss": 0.5979,
      "step": 17179
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.125472813171852,
      "learning_rate": 1.3646652189439274e-06,
      "loss": 0.5675,
      "step": 17180
    },
    {
      "epoch": 1.53,
      "grad_norm": 4.871495666191913,
      "learning_rate": 1.3641693009276008e-06,
      "loss": 0.5449,
      "step": 17181
    },
    {
      "epoch": 1.53,
      "grad_norm": 8.466121559417719,
      "learning_rate": 1.3636734588012773e-06,
      "loss": 0.5667,
      "step": 17182
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.183328942381242,
      "learning_rate": 1.363177692575307e-06,
      "loss": 0.6173,
      "step": 17183
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.326729498920868,
      "learning_rate": 1.3626820022600396e-06,
      "loss": 0.6117,
      "step": 17184
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.440073519740164,
      "learning_rate": 1.3621863878658175e-06,
      "loss": 0.5382,
      "step": 17185
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.709914531512257,
      "learning_rate": 1.361690849402989e-06,
      "loss": 0.578,
      "step": 17186
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.7057942429563955,
      "learning_rate": 1.3611953868818955e-06,
      "loss": 0.6453,
      "step": 17187
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.389504478235555,
      "learning_rate": 1.3607000003128795e-06,
      "loss": 0.5591,
      "step": 17188
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.771795446396589,
      "learning_rate": 1.3602046897062816e-06,
      "loss": 0.5513,
      "step": 17189
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.689915186590739,
      "learning_rate": 1.359709455072441e-06,
      "loss": 0.5412,
      "step": 17190
    },
    {
      "epoch": 1.53,
      "grad_norm": 4.769806607016308,
      "learning_rate": 1.359214296421693e-06,
      "loss": 0.5973,
      "step": 17191
    },
    {
      "epoch": 1.53,
      "grad_norm": 4.580144854601644,
      "learning_rate": 1.3587192137643724e-06,
      "loss": 0.6545,
      "step": 17192
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.355361570421232,
      "learning_rate": 1.3582242071108137e-06,
      "loss": 0.5534,
      "step": 17193
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.442230003203185,
      "learning_rate": 1.3577292764713495e-06,
      "loss": 0.5962,
      "step": 17194
    },
    {
      "epoch": 1.53,
      "grad_norm": 8.299109219735138,
      "learning_rate": 1.35723442185631e-06,
      "loss": 0.5474,
      "step": 17195
    },
    {
      "epoch": 1.53,
      "grad_norm": 11.89131766733263,
      "learning_rate": 1.356739643276025e-06,
      "loss": 0.5739,
      "step": 17196
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.084525599281193,
      "learning_rate": 1.3562449407408224e-06,
      "loss": 0.5715,
      "step": 17197
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.651234459872414,
      "learning_rate": 1.355750314261025e-06,
      "loss": 0.5653,
      "step": 17198
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.593601304039529,
      "learning_rate": 1.3552557638469598e-06,
      "loss": 0.59,
      "step": 17199
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.391088223007171,
      "learning_rate": 1.3547612895089484e-06,
      "loss": 0.5686,
      "step": 17200
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.731818543463911,
      "learning_rate": 1.354266891257312e-06,
      "loss": 0.538,
      "step": 17201
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.941340349079574,
      "learning_rate": 1.35377256910237e-06,
      "loss": 0.6421,
      "step": 17202
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.05003289734198,
      "learning_rate": 1.3532783230544427e-06,
      "loss": 0.5449,
      "step": 17203
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.050969330137939,
      "learning_rate": 1.3527841531238424e-06,
      "loss": 0.5109,
      "step": 17204
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.251876431825372,
      "learning_rate": 1.3522900593208855e-06,
      "loss": 0.653,
      "step": 17205
    },
    {
      "epoch": 1.53,
      "grad_norm": 4.507660076440911,
      "learning_rate": 1.3517960416558867e-06,
      "loss": 0.573,
      "step": 17206
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.9492525280987705,
      "learning_rate": 1.351302100139155e-06,
      "loss": 0.5899,
      "step": 17207
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.737250786599135,
      "learning_rate": 1.3508082347810015e-06,
      "loss": 0.6003,
      "step": 17208
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.176956386321778,
      "learning_rate": 1.3503144455917338e-06,
      "loss": 0.5373,
      "step": 17209
    },
    {
      "epoch": 1.54,
      "grad_norm": 9.233353015812144,
      "learning_rate": 1.3498207325816616e-06,
      "loss": 0.621,
      "step": 17210
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.865537707031853,
      "learning_rate": 1.349327095761086e-06,
      "loss": 0.5251,
      "step": 17211
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.307457729914165,
      "learning_rate": 1.3488335351403126e-06,
      "loss": 0.5923,
      "step": 17212
    },
    {
      "epoch": 1.54,
      "grad_norm": 7.666545530329459,
      "learning_rate": 1.348340050729643e-06,
      "loss": 0.5762,
      "step": 17213
    },
    {
      "epoch": 1.54,
      "grad_norm": 8.083835936859414,
      "learning_rate": 1.3478466425393777e-06,
      "loss": 0.588,
      "step": 17214
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.723388187308514,
      "learning_rate": 1.3473533105798165e-06,
      "loss": 0.5957,
      "step": 17215
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.621775797697638,
      "learning_rate": 1.3468600548612564e-06,
      "loss": 0.5555,
      "step": 17216
    },
    {
      "epoch": 1.54,
      "grad_norm": 7.376757545061219,
      "learning_rate": 1.3463668753939912e-06,
      "loss": 0.5441,
      "step": 17217
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.345131954590711,
      "learning_rate": 1.3458737721883164e-06,
      "loss": 0.641,
      "step": 17218
    },
    {
      "epoch": 1.54,
      "grad_norm": 7.4600938979622065,
      "learning_rate": 1.3453807452545237e-06,
      "loss": 0.6177,
      "step": 17219
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.525025611048279,
      "learning_rate": 1.3448877946029066e-06,
      "loss": 0.585,
      "step": 17220
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.052636035158669,
      "learning_rate": 1.34439492024375e-06,
      "loss": 0.4968,
      "step": 17221
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.2400243982581065,
      "learning_rate": 1.3439021221873444e-06,
      "loss": 0.5576,
      "step": 17222
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.680058238085944,
      "learning_rate": 1.3434094004439763e-06,
      "loss": 0.5938,
      "step": 17223
    },
    {
      "epoch": 1.54,
      "grad_norm": 9.645817839131162,
      "learning_rate": 1.342916755023928e-06,
      "loss": 0.5308,
      "step": 17224
    },
    {
      "epoch": 1.54,
      "grad_norm": 4.3579330983813565,
      "learning_rate": 1.3424241859374838e-06,
      "loss": 0.6107,
      "step": 17225
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.51045558972247,
      "learning_rate": 1.3419316931949244e-06,
      "loss": 0.6173,
      "step": 17226
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.0635030595174415,
      "learning_rate": 1.3414392768065299e-06,
      "loss": 0.518,
      "step": 17227
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.546095559568027,
      "learning_rate": 1.3409469367825783e-06,
      "loss": 0.568,
      "step": 17228
    },
    {
      "epoch": 1.54,
      "grad_norm": 7.427457411431497,
      "learning_rate": 1.3404546731333479e-06,
      "loss": 0.5646,
      "step": 17229
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.735403751908864,
      "learning_rate": 1.33996248586911e-06,
      "loss": 0.5889,
      "step": 17230
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.521213096138269,
      "learning_rate": 1.3394703750001402e-06,
      "loss": 0.5328,
      "step": 17231
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.857847196777849,
      "learning_rate": 1.3389783405367101e-06,
      "loss": 0.5799,
      "step": 17232
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.689310098665093,
      "learning_rate": 1.3384863824890892e-06,
      "loss": 0.5833,
      "step": 17233
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.0469807765847525,
      "learning_rate": 1.3379945008675471e-06,
      "loss": 0.5446,
      "step": 17234
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.385839946422461,
      "learning_rate": 1.3375026956823511e-06,
      "loss": 0.575,
      "step": 17235
    },
    {
      "epoch": 1.54,
      "grad_norm": 8.537239846371557,
      "learning_rate": 1.3370109669437658e-06,
      "loss": 0.5768,
      "step": 17236
    },
    {
      "epoch": 1.54,
      "grad_norm": 8.74254941842663,
      "learning_rate": 1.336519314662053e-06,
      "loss": 0.5685,
      "step": 17237
    },
    {
      "epoch": 1.54,
      "grad_norm": 7.088149501599103,
      "learning_rate": 1.3360277388474775e-06,
      "loss": 0.5371,
      "step": 17238
    },
    {
      "epoch": 1.54,
      "grad_norm": 4.907752948339701,
      "learning_rate": 1.3355362395102978e-06,
      "loss": 0.5433,
      "step": 17239
    },
    {
      "epoch": 1.54,
      "grad_norm": 7.466592954661629,
      "learning_rate": 1.335044816660775e-06,
      "loss": 0.5721,
      "step": 17240
    },
    {
      "epoch": 1.54,
      "grad_norm": 8.643739334134564,
      "learning_rate": 1.334553470309165e-06,
      "loss": 0.5485,
      "step": 17241
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.293671885429159,
      "learning_rate": 1.3340622004657256e-06,
      "loss": 0.5873,
      "step": 17242
    },
    {
      "epoch": 1.54,
      "grad_norm": 8.816699485460827,
      "learning_rate": 1.3335710071407083e-06,
      "loss": 0.5576,
      "step": 17243
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.896603293774081,
      "learning_rate": 1.3330798903443664e-06,
      "loss": 0.5881,
      "step": 17244
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.2898247501867734,
      "learning_rate": 1.3325888500869521e-06,
      "loss": 0.5302,
      "step": 17245
    },
    {
      "epoch": 1.54,
      "grad_norm": 8.212186048251358,
      "learning_rate": 1.3320978863787138e-06,
      "loss": 0.6015,
      "step": 17246
    },
    {
      "epoch": 1.54,
      "grad_norm": 9.689162834221072,
      "learning_rate": 1.3316069992299008e-06,
      "loss": 0.6375,
      "step": 17247
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.776247723007912,
      "learning_rate": 1.3311161886507567e-06,
      "loss": 0.5829,
      "step": 17248
    },
    {
      "epoch": 1.54,
      "grad_norm": 8.125661946561763,
      "learning_rate": 1.330625454651529e-06,
      "loss": 0.649,
      "step": 17249
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.762690563721956,
      "learning_rate": 1.3301347972424571e-06,
      "loss": 0.5172,
      "step": 17250
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.060673459796315,
      "learning_rate": 1.3296442164337853e-06,
      "loss": 0.5862,
      "step": 17251
    },
    {
      "epoch": 1.54,
      "grad_norm": 7.400547592277442,
      "learning_rate": 1.3291537122357516e-06,
      "loss": 0.6272,
      "step": 17252
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.558162792623261,
      "learning_rate": 1.3286632846585962e-06,
      "loss": 0.5665,
      "step": 17253
    },
    {
      "epoch": 1.54,
      "grad_norm": 8.147254434823392,
      "learning_rate": 1.3281729337125555e-06,
      "loss": 0.6542,
      "step": 17254
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.914219439315344,
      "learning_rate": 1.3276826594078624e-06,
      "loss": 0.5612,
      "step": 17255
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.224201794993361,
      "learning_rate": 1.327192461754751e-06,
      "loss": 0.5783,
      "step": 17256
    },
    {
      "epoch": 1.54,
      "grad_norm": 7.477388226571542,
      "learning_rate": 1.3267023407634545e-06,
      "loss": 0.5723,
      "step": 17257
    },
    {
      "epoch": 1.54,
      "grad_norm": 7.617188723293537,
      "learning_rate": 1.3262122964442015e-06,
      "loss": 0.5966,
      "step": 17258
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.988454674425447,
      "learning_rate": 1.325722328807222e-06,
      "loss": 0.5792,
      "step": 17259
    },
    {
      "epoch": 1.54,
      "grad_norm": 7.276064495762684,
      "learning_rate": 1.3252324378627436e-06,
      "loss": 0.5763,
      "step": 17260
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.44593463128524,
      "learning_rate": 1.3247426236209888e-06,
      "loss": 0.5306,
      "step": 17261
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.997806998845987,
      "learning_rate": 1.3242528860921828e-06,
      "loss": 0.5863,
      "step": 17262
    },
    {
      "epoch": 1.54,
      "grad_norm": 8.995027915848794,
      "learning_rate": 1.32376322528655e-06,
      "loss": 0.5639,
      "step": 17263
    },
    {
      "epoch": 1.54,
      "grad_norm": 8.645436398407703,
      "learning_rate": 1.323273641214307e-06,
      "loss": 0.5578,
      "step": 17264
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.742021292393683,
      "learning_rate": 1.3227841338856756e-06,
      "loss": 0.5688,
      "step": 17265
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.391831239357793,
      "learning_rate": 1.3222947033108723e-06,
      "loss": 0.5233,
      "step": 17266
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.986138945531844,
      "learning_rate": 1.3218053495001143e-06,
      "loss": 0.5766,
      "step": 17267
    },
    {
      "epoch": 1.54,
      "grad_norm": 9.704256007936365,
      "learning_rate": 1.3213160724636131e-06,
      "loss": 0.6232,
      "step": 17268
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.060612500962514,
      "learning_rate": 1.3208268722115825e-06,
      "loss": 0.5826,
      "step": 17269
    },
    {
      "epoch": 1.54,
      "grad_norm": 9.653650272126415,
      "learning_rate": 1.320337748754234e-06,
      "loss": 0.5968,
      "step": 17270
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.840766154009693,
      "learning_rate": 1.3198487021017764e-06,
      "loss": 0.5103,
      "step": 17271
    },
    {
      "epoch": 1.54,
      "grad_norm": 7.874350103719004,
      "learning_rate": 1.3193597322644181e-06,
      "loss": 0.5664,
      "step": 17272
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.691861256692753,
      "learning_rate": 1.3188708392523658e-06,
      "loss": 0.6094,
      "step": 17273
    },
    {
      "epoch": 1.54,
      "grad_norm": 7.754711192315602,
      "learning_rate": 1.318382023075822e-06,
      "loss": 0.6109,
      "step": 17274
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.744563464408076,
      "learning_rate": 1.3178932837449915e-06,
      "loss": 0.536,
      "step": 17275
    },
    {
      "epoch": 1.54,
      "grad_norm": 7.29435858995001,
      "learning_rate": 1.3174046212700741e-06,
      "loss": 0.6039,
      "step": 17276
    },
    {
      "epoch": 1.54,
      "grad_norm": 7.574385768081241,
      "learning_rate": 1.3169160356612709e-06,
      "loss": 0.604,
      "step": 17277
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.989069496385774,
      "learning_rate": 1.3164275269287813e-06,
      "loss": 0.5794,
      "step": 17278
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.925875257193215,
      "learning_rate": 1.3159390950827983e-06,
      "loss": 0.5883,
      "step": 17279
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.443585166261667,
      "learning_rate": 1.3154507401335203e-06,
      "loss": 0.5362,
      "step": 17280
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.142624026374915,
      "learning_rate": 1.3149624620911383e-06,
      "loss": 0.5591,
      "step": 17281
    },
    {
      "epoch": 1.54,
      "grad_norm": 8.112807055874939,
      "learning_rate": 1.314474260965844e-06,
      "loss": 0.5953,
      "step": 17282
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.831955221853754,
      "learning_rate": 1.3139861367678285e-06,
      "loss": 0.6228,
      "step": 17283
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.506625451766481,
      "learning_rate": 1.3134980895072801e-06,
      "loss": 0.5525,
      "step": 17284
    },
    {
      "epoch": 1.54,
      "grad_norm": 4.744234970448051,
      "learning_rate": 1.3130101191943862e-06,
      "loss": 0.5863,
      "step": 17285
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.665854295837863,
      "learning_rate": 1.3125222258393328e-06,
      "loss": 0.5452,
      "step": 17286
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.937346679108787,
      "learning_rate": 1.3120344094523013e-06,
      "loss": 0.4993,
      "step": 17287
    },
    {
      "epoch": 1.54,
      "grad_norm": 7.543173454027063,
      "learning_rate": 1.3115466700434748e-06,
      "loss": 0.5353,
      "step": 17288
    },
    {
      "epoch": 1.54,
      "grad_norm": 9.249846991069104,
      "learning_rate": 1.311059007623034e-06,
      "loss": 0.5837,
      "step": 17289
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.607967512185825,
      "learning_rate": 1.3105714222011573e-06,
      "loss": 0.5559,
      "step": 17290
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.438346236725876,
      "learning_rate": 1.3100839137880233e-06,
      "loss": 0.5126,
      "step": 17291
    },
    {
      "epoch": 1.54,
      "grad_norm": 8.363293475057521,
      "learning_rate": 1.309596482393808e-06,
      "loss": 0.6378,
      "step": 17292
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.109516232616605,
      "learning_rate": 1.3091091280286839e-06,
      "loss": 0.5302,
      "step": 17293
    },
    {
      "epoch": 1.54,
      "grad_norm": 7.469059445326008,
      "learning_rate": 1.3086218507028225e-06,
      "loss": 0.5718,
      "step": 17294
    },
    {
      "epoch": 1.54,
      "grad_norm": 7.519727820808889,
      "learning_rate": 1.3081346504263965e-06,
      "loss": 0.5635,
      "step": 17295
    },
    {
      "epoch": 1.54,
      "grad_norm": 7.4090645105864255,
      "learning_rate": 1.307647527209574e-06,
      "loss": 0.5512,
      "step": 17296
    },
    {
      "epoch": 1.54,
      "grad_norm": 8.924462378419442,
      "learning_rate": 1.3071604810625238e-06,
      "loss": 0.5898,
      "step": 17297
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.963865259798091,
      "learning_rate": 1.306673511995411e-06,
      "loss": 0.5464,
      "step": 17298
    },
    {
      "epoch": 1.54,
      "grad_norm": 8.233148354648746,
      "learning_rate": 1.306186620018402e-06,
      "loss": 0.5831,
      "step": 17299
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.95769236680475,
      "learning_rate": 1.3056998051416564e-06,
      "loss": 0.5446,
      "step": 17300
    },
    {
      "epoch": 1.54,
      "grad_norm": 8.990008786889529,
      "learning_rate": 1.3052130673753371e-06,
      "loss": 0.5402,
      "step": 17301
    },
    {
      "epoch": 1.54,
      "grad_norm": 8.18673863882345,
      "learning_rate": 1.3047264067296033e-06,
      "loss": 0.6324,
      "step": 17302
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.862994418072173,
      "learning_rate": 1.3042398232146137e-06,
      "loss": 0.5162,
      "step": 17303
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.717723546229892,
      "learning_rate": 1.3037533168405237e-06,
      "loss": 0.5651,
      "step": 17304
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.257003965634214,
      "learning_rate": 1.3032668876174903e-06,
      "loss": 0.5323,
      "step": 17305
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.329755364792456,
      "learning_rate": 1.3027805355556638e-06,
      "loss": 0.5239,
      "step": 17306
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.556761176061359,
      "learning_rate": 1.3022942606651978e-06,
      "loss": 0.5558,
      "step": 17307
    },
    {
      "epoch": 1.54,
      "grad_norm": 8.90833858982528,
      "learning_rate": 1.3018080629562403e-06,
      "loss": 0.6134,
      "step": 17308
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.44114822285237,
      "learning_rate": 1.30132194243894e-06,
      "loss": 0.5715,
      "step": 17309
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.03933345369698,
      "learning_rate": 1.3008358991234449e-06,
      "loss": 0.5932,
      "step": 17310
    },
    {
      "epoch": 1.54,
      "grad_norm": 8.03662734901584,
      "learning_rate": 1.3003499330198988e-06,
      "loss": 0.6465,
      "step": 17311
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.007990680498799,
      "learning_rate": 1.2998640441384474e-06,
      "loss": 0.6266,
      "step": 17312
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.615670055093644,
      "learning_rate": 1.299378232489229e-06,
      "loss": 0.6115,
      "step": 17313
    },
    {
      "epoch": 1.54,
      "grad_norm": 9.639660647423137,
      "learning_rate": 1.2988924980823863e-06,
      "loss": 0.5945,
      "step": 17314
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.8583074459546856,
      "learning_rate": 1.2984068409280575e-06,
      "loss": 0.5284,
      "step": 17315
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.642721270487775,
      "learning_rate": 1.2979212610363795e-06,
      "loss": 0.5048,
      "step": 17316
    },
    {
      "epoch": 1.54,
      "grad_norm": 7.619949391458259,
      "learning_rate": 1.2974357584174875e-06,
      "loss": 0.573,
      "step": 17317
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.957207118813275,
      "learning_rate": 1.296950333081517e-06,
      "loss": 0.6212,
      "step": 17318
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.130068820172727,
      "learning_rate": 1.2964649850385973e-06,
      "loss": 0.5425,
      "step": 17319
    },
    {
      "epoch": 1.55,
      "grad_norm": 8.397878514607624,
      "learning_rate": 1.2959797142988606e-06,
      "loss": 0.5417,
      "step": 17320
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.147378636626203,
      "learning_rate": 1.2954945208724373e-06,
      "loss": 0.5606,
      "step": 17321
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.4299083388866975,
      "learning_rate": 1.2950094047694516e-06,
      "loss": 0.6266,
      "step": 17322
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.30084481447171,
      "learning_rate": 1.2945243660000306e-06,
      "loss": 0.5983,
      "step": 17323
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.74439112916879,
      "learning_rate": 1.2940394045742987e-06,
      "loss": 0.5296,
      "step": 17324
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.5314200170317225,
      "learning_rate": 1.2935545205023797e-06,
      "loss": 0.5449,
      "step": 17325
    },
    {
      "epoch": 1.55,
      "grad_norm": 7.500080433141965,
      "learning_rate": 1.293069713794392e-06,
      "loss": 0.5835,
      "step": 17326
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.464567088067988,
      "learning_rate": 1.2925849844604554e-06,
      "loss": 0.5686,
      "step": 17327
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.535663860008356,
      "learning_rate": 1.2921003325106885e-06,
      "loss": 0.5483,
      "step": 17328
    },
    {
      "epoch": 1.55,
      "grad_norm": 7.462844500440332,
      "learning_rate": 1.2916157579552068e-06,
      "loss": 0.5871,
      "step": 17329
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.701895253490909,
      "learning_rate": 1.291131260804127e-06,
      "loss": 0.5538,
      "step": 17330
    },
    {
      "epoch": 1.55,
      "grad_norm": 8.303542088568534,
      "learning_rate": 1.2906468410675576e-06,
      "loss": 0.582,
      "step": 17331
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.774927310850524,
      "learning_rate": 1.2901624987556127e-06,
      "loss": 0.6156,
      "step": 17332
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.883386806311535,
      "learning_rate": 1.289678233878401e-06,
      "loss": 0.5774,
      "step": 17333
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.347578328360446,
      "learning_rate": 1.2891940464460307e-06,
      "loss": 0.5165,
      "step": 17334
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.864645939782381,
      "learning_rate": 1.288709936468608e-06,
      "loss": 0.5579,
      "step": 17335
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.8457705574006775,
      "learning_rate": 1.2882259039562388e-06,
      "loss": 0.5749,
      "step": 17336
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.47952245941112,
      "learning_rate": 1.2877419489190247e-06,
      "loss": 0.5587,
      "step": 17337
    },
    {
      "epoch": 1.55,
      "grad_norm": 8.96728662099873,
      "learning_rate": 1.2872580713670685e-06,
      "loss": 0.6003,
      "step": 17338
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.497134620616692,
      "learning_rate": 1.2867742713104681e-06,
      "loss": 0.6048,
      "step": 17339
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.7826070877584135,
      "learning_rate": 1.286290548759323e-06,
      "loss": 0.6112,
      "step": 17340
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.275838887523173,
      "learning_rate": 1.2858069037237297e-06,
      "loss": 0.5696,
      "step": 17341
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.989755664653599,
      "learning_rate": 1.2853233362137834e-06,
      "loss": 0.5563,
      "step": 17342
    },
    {
      "epoch": 1.55,
      "grad_norm": 8.283723876098554,
      "learning_rate": 1.284839846239579e-06,
      "loss": 0.5581,
      "step": 17343
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.087608397177082,
      "learning_rate": 1.2843564338112053e-06,
      "loss": 0.5828,
      "step": 17344
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.6466926673436735,
      "learning_rate": 1.2838730989387538e-06,
      "loss": 0.6639,
      "step": 17345
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.82354726597059,
      "learning_rate": 1.2833898416323132e-06,
      "loss": 0.6134,
      "step": 17346
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.978271596064632,
      "learning_rate": 1.2829066619019704e-06,
      "loss": 0.5556,
      "step": 17347
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.183317712742663,
      "learning_rate": 1.2824235597578105e-06,
      "loss": 0.565,
      "step": 17348
    },
    {
      "epoch": 1.55,
      "grad_norm": 7.209882796896954,
      "learning_rate": 1.281940535209919e-06,
      "loss": 0.5736,
      "step": 17349
    },
    {
      "epoch": 1.55,
      "grad_norm": 9.209196233562606,
      "learning_rate": 1.2814575882683767e-06,
      "loss": 0.6018,
      "step": 17350
    },
    {
      "epoch": 1.55,
      "grad_norm": 7.077723242901754,
      "learning_rate": 1.2809747189432625e-06,
      "loss": 0.619,
      "step": 17351
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.251291602509589,
      "learning_rate": 1.2804919272446558e-06,
      "loss": 0.5588,
      "step": 17352
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.020536803162597,
      "learning_rate": 1.2800092131826352e-06,
      "loss": 0.5977,
      "step": 17353
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.309385119656159,
      "learning_rate": 1.2795265767672755e-06,
      "loss": 0.5726,
      "step": 17354
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.991265310699507,
      "learning_rate": 1.2790440180086505e-06,
      "loss": 0.6044,
      "step": 17355
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.716133921907439,
      "learning_rate": 1.2785615369168348e-06,
      "loss": 0.5331,
      "step": 17356
    },
    {
      "epoch": 1.55,
      "grad_norm": 7.028972751645588,
      "learning_rate": 1.2780791335018949e-06,
      "loss": 0.5895,
      "step": 17357
    },
    {
      "epoch": 1.55,
      "grad_norm": 8.539145636486476,
      "learning_rate": 1.277596807773903e-06,
      "loss": 0.6064,
      "step": 17358
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.226798478437806,
      "learning_rate": 1.2771145597429258e-06,
      "loss": 0.628,
      "step": 17359
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.303066508181731,
      "learning_rate": 1.2766323894190285e-06,
      "loss": 0.5412,
      "step": 17360
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.281463666536227,
      "learning_rate": 1.276150296812277e-06,
      "loss": 0.5464,
      "step": 17361
    },
    {
      "epoch": 1.55,
      "grad_norm": 11.304835011836763,
      "learning_rate": 1.2756682819327338e-06,
      "loss": 0.6251,
      "step": 17362
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.7447746020163795,
      "learning_rate": 1.2751863447904578e-06,
      "loss": 0.5236,
      "step": 17363
    },
    {
      "epoch": 1.55,
      "grad_norm": 7.634700829932257,
      "learning_rate": 1.274704485395511e-06,
      "loss": 0.5442,
      "step": 17364
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.878913949677425,
      "learning_rate": 1.274222703757948e-06,
      "loss": 0.58,
      "step": 17365
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.168336236057548,
      "learning_rate": 1.2737409998878275e-06,
      "loss": 0.6197,
      "step": 17366
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.47938954240068,
      "learning_rate": 1.2732593737952026e-06,
      "loss": 0.536,
      "step": 17367
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.120253382959296,
      "learning_rate": 1.2727778254901274e-06,
      "loss": 0.5719,
      "step": 17368
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.163106629751549,
      "learning_rate": 1.272296354982654e-06,
      "loss": 0.5501,
      "step": 17369
    },
    {
      "epoch": 1.55,
      "grad_norm": 7.076162849052849,
      "learning_rate": 1.2718149622828284e-06,
      "loss": 0.604,
      "step": 17370
    },
    {
      "epoch": 1.55,
      "grad_norm": 7.202840745181803,
      "learning_rate": 1.2713336474007016e-06,
      "loss": 0.6497,
      "step": 17371
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.869114363994082,
      "learning_rate": 1.2708524103463193e-06,
      "loss": 0.515,
      "step": 17372
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.717167747236392,
      "learning_rate": 1.2703712511297262e-06,
      "loss": 0.5056,
      "step": 17373
    },
    {
      "epoch": 1.55,
      "grad_norm": 7.167528924486516,
      "learning_rate": 1.2698901697609656e-06,
      "loss": 0.501,
      "step": 17374
    },
    {
      "epoch": 1.55,
      "grad_norm": 8.188106380164934,
      "learning_rate": 1.2694091662500802e-06,
      "loss": 0.5412,
      "step": 17375
    },
    {
      "epoch": 1.55,
      "grad_norm": 8.475440760250637,
      "learning_rate": 1.2689282406071068e-06,
      "loss": 0.5305,
      "step": 17376
    },
    {
      "epoch": 1.55,
      "grad_norm": 7.134242151850569,
      "learning_rate": 1.268447392842086e-06,
      "loss": 0.5694,
      "step": 17377
    },
    {
      "epoch": 1.55,
      "grad_norm": 7.592762297763197,
      "learning_rate": 1.2679666229650535e-06,
      "loss": 0.581,
      "step": 17378
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.9185869608611545,
      "learning_rate": 1.2674859309860467e-06,
      "loss": 0.5688,
      "step": 17379
    },
    {
      "epoch": 1.55,
      "grad_norm": 9.358404706136367,
      "learning_rate": 1.2670053169150947e-06,
      "loss": 0.6441,
      "step": 17380
    },
    {
      "epoch": 1.55,
      "grad_norm": 7.514340010592223,
      "learning_rate": 1.2665247807622322e-06,
      "loss": 0.5796,
      "step": 17381
    },
    {
      "epoch": 1.55,
      "grad_norm": 8.828243188124922,
      "learning_rate": 1.2660443225374901e-06,
      "loss": 0.5848,
      "step": 17382
    },
    {
      "epoch": 1.55,
      "grad_norm": 7.143889821302738,
      "learning_rate": 1.2655639422508942e-06,
      "loss": 0.546,
      "step": 17383
    },
    {
      "epoch": 1.55,
      "grad_norm": 7.811558151819244,
      "learning_rate": 1.2650836399124728e-06,
      "loss": 0.5389,
      "step": 17384
    },
    {
      "epoch": 1.55,
      "grad_norm": 12.259479844745151,
      "learning_rate": 1.2646034155322507e-06,
      "loss": 0.5897,
      "step": 17385
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.491060810351299,
      "learning_rate": 1.2641232691202521e-06,
      "loss": 0.5776,
      "step": 17386
    },
    {
      "epoch": 1.55,
      "grad_norm": 7.349517452618759,
      "learning_rate": 1.263643200686499e-06,
      "loss": 0.6231,
      "step": 17387
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.957422636358688,
      "learning_rate": 1.263163210241013e-06,
      "loss": 0.5766,
      "step": 17388
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.193821674054475,
      "learning_rate": 1.2626832977938097e-06,
      "loss": 0.5987,
      "step": 17389
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.732039625682774,
      "learning_rate": 1.2622034633549084e-06,
      "loss": 0.6051,
      "step": 17390
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.636372365665808,
      "learning_rate": 1.2617237069343237e-06,
      "loss": 0.5391,
      "step": 17391
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.5792845413978025,
      "learning_rate": 1.26124402854207e-06,
      "loss": 0.6774,
      "step": 17392
    },
    {
      "epoch": 1.55,
      "grad_norm": 7.26516192719665,
      "learning_rate": 1.2607644281881614e-06,
      "loss": 0.544,
      "step": 17393
    },
    {
      "epoch": 1.55,
      "grad_norm": 7.748804044321922,
      "learning_rate": 1.260284905882605e-06,
      "loss": 0.5998,
      "step": 17394
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.897225723365914,
      "learning_rate": 1.2598054616354128e-06,
      "loss": 0.6042,
      "step": 17395
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.442007542996957,
      "learning_rate": 1.259326095456589e-06,
      "loss": 0.57,
      "step": 17396
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.466936854007667,
      "learning_rate": 1.2588468073561411e-06,
      "loss": 0.5587,
      "step": 17397
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.291988331152342,
      "learning_rate": 1.2583675973440728e-06,
      "loss": 0.5595,
      "step": 17398
    },
    {
      "epoch": 1.55,
      "grad_norm": 7.697696545467632,
      "learning_rate": 1.2578884654303874e-06,
      "loss": 0.5786,
      "step": 17399
    },
    {
      "epoch": 1.55,
      "grad_norm": 7.584286456790095,
      "learning_rate": 1.257409411625085e-06,
      "loss": 0.549,
      "step": 17400
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.388720685933137,
      "learning_rate": 1.2569304359381662e-06,
      "loss": 0.5232,
      "step": 17401
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.745303609455814,
      "learning_rate": 1.2564515383796265e-06,
      "loss": 0.5561,
      "step": 17402
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.278913654171453,
      "learning_rate": 1.2559727189594622e-06,
      "loss": 0.5353,
      "step": 17403
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.744580632994773,
      "learning_rate": 1.2554939776876685e-06,
      "loss": 0.5977,
      "step": 17404
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.942346809265603,
      "learning_rate": 1.2550153145742372e-06,
      "loss": 0.5931,
      "step": 17405
    },
    {
      "epoch": 1.55,
      "grad_norm": 17.283645726848174,
      "learning_rate": 1.2545367296291616e-06,
      "loss": 0.5445,
      "step": 17406
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.996382571033899,
      "learning_rate": 1.2540582228624276e-06,
      "loss": 0.5851,
      "step": 17407
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.943684946664891,
      "learning_rate": 1.2535797942840266e-06,
      "loss": 0.643,
      "step": 17408
    },
    {
      "epoch": 1.55,
      "grad_norm": 7.330552704184243,
      "learning_rate": 1.2531014439039412e-06,
      "loss": 0.5718,
      "step": 17409
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.124795186306825,
      "learning_rate": 1.2526231717321579e-06,
      "loss": 0.5497,
      "step": 17410
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.289613483148967,
      "learning_rate": 1.2521449777786588e-06,
      "loss": 0.5712,
      "step": 17411
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.633336080303487,
      "learning_rate": 1.251666862053426e-06,
      "loss": 0.6618,
      "step": 17412
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.493377971288357,
      "learning_rate": 1.2511888245664383e-06,
      "loss": 0.5944,
      "step": 17413
    },
    {
      "epoch": 1.55,
      "grad_norm": 7.442078103266807,
      "learning_rate": 1.2507108653276763e-06,
      "loss": 0.6224,
      "step": 17414
    },
    {
      "epoch": 1.55,
      "grad_norm": 7.554689374024966,
      "learning_rate": 1.2502329843471123e-06,
      "loss": 0.5872,
      "step": 17415
    },
    {
      "epoch": 1.55,
      "grad_norm": 7.854034827507665,
      "learning_rate": 1.2497551816347225e-06,
      "loss": 0.5824,
      "step": 17416
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.933673278112818,
      "learning_rate": 1.2492774572004807e-06,
      "loss": 0.5147,
      "step": 17417
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.696584649091364,
      "learning_rate": 1.2487998110543582e-06,
      "loss": 0.568,
      "step": 17418
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.449716681928235,
      "learning_rate": 1.2483222432063263e-06,
      "loss": 0.5656,
      "step": 17419
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.967172918609351,
      "learning_rate": 1.2478447536663496e-06,
      "loss": 0.5308,
      "step": 17420
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.655498136413138,
      "learning_rate": 1.2473673424443967e-06,
      "loss": 0.5702,
      "step": 17421
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.502569824872035,
      "learning_rate": 1.2468900095504344e-06,
      "loss": 0.596,
      "step": 17422
    },
    {
      "epoch": 1.55,
      "grad_norm": 7.005621091119339,
      "learning_rate": 1.2464127549944217e-06,
      "loss": 0.5719,
      "step": 17423
    },
    {
      "epoch": 1.55,
      "grad_norm": 7.728514011822037,
      "learning_rate": 1.245935578786323e-06,
      "loss": 0.6022,
      "step": 17424
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.074323556187885,
      "learning_rate": 1.2454584809360975e-06,
      "loss": 0.5902,
      "step": 17425
    },
    {
      "epoch": 1.55,
      "grad_norm": 8.5492543447658,
      "learning_rate": 1.244981461453706e-06,
      "loss": 0.648,
      "step": 17426
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.911501679909859,
      "learning_rate": 1.2445045203491012e-06,
      "loss": 0.5763,
      "step": 17427
    },
    {
      "epoch": 1.55,
      "grad_norm": 10.95835956950578,
      "learning_rate": 1.2440276576322401e-06,
      "loss": 0.5599,
      "step": 17428
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.913698574888189,
      "learning_rate": 1.2435508733130763e-06,
      "loss": 0.5852,
      "step": 17429
    },
    {
      "epoch": 1.55,
      "grad_norm": 7.552328851650199,
      "learning_rate": 1.2430741674015618e-06,
      "loss": 0.5842,
      "step": 17430
    },
    {
      "epoch": 1.56,
      "grad_norm": 4.714268508975373,
      "learning_rate": 1.2425975399076462e-06,
      "loss": 0.5796,
      "step": 17431
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.39943832367043,
      "learning_rate": 1.2421209908412802e-06,
      "loss": 0.624,
      "step": 17432
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.40268151745162,
      "learning_rate": 1.2416445202124067e-06,
      "loss": 0.5699,
      "step": 17433
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.88948449275006,
      "learning_rate": 1.241168128030974e-06,
      "loss": 0.5188,
      "step": 17434
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.117668067626755,
      "learning_rate": 1.2406918143069247e-06,
      "loss": 0.538,
      "step": 17435
    },
    {
      "epoch": 1.56,
      "grad_norm": 9.140514177026661,
      "learning_rate": 1.2402155790502025e-06,
      "loss": 0.596,
      "step": 17436
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.986916924592516,
      "learning_rate": 1.2397394222707444e-06,
      "loss": 0.5605,
      "step": 17437
    },
    {
      "epoch": 1.56,
      "grad_norm": 8.320222467861766,
      "learning_rate": 1.2392633439784912e-06,
      "loss": 0.5507,
      "step": 17438
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.793995661779951,
      "learning_rate": 1.2387873441833815e-06,
      "loss": 0.5757,
      "step": 17439
    },
    {
      "epoch": 1.56,
      "grad_norm": 8.3288312413563,
      "learning_rate": 1.2383114228953475e-06,
      "loss": 0.6319,
      "step": 17440
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.387145179973341,
      "learning_rate": 1.2378355801243247e-06,
      "loss": 0.5471,
      "step": 17441
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.069976446105957,
      "learning_rate": 1.2373598158802451e-06,
      "loss": 0.5754,
      "step": 17442
    },
    {
      "epoch": 1.56,
      "grad_norm": 8.17768159878173,
      "learning_rate": 1.2368841301730395e-06,
      "loss": 0.6187,
      "step": 17443
    },
    {
      "epoch": 1.56,
      "grad_norm": 8.32442714186497,
      "learning_rate": 1.236408523012636e-06,
      "loss": 0.5877,
      "step": 17444
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.137038139855707,
      "learning_rate": 1.2359329944089644e-06,
      "loss": 0.5948,
      "step": 17445
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.804529743742463,
      "learning_rate": 1.2354575443719468e-06,
      "loss": 0.643,
      "step": 17446
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.115533200021045,
      "learning_rate": 1.234982172911509e-06,
      "loss": 0.5612,
      "step": 17447
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.74410593598114,
      "learning_rate": 1.234506880037573e-06,
      "loss": 0.5465,
      "step": 17448
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.7760410769887045,
      "learning_rate": 1.2340316657600593e-06,
      "loss": 0.5672,
      "step": 17449
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.362668923162735,
      "learning_rate": 1.2335565300888874e-06,
      "loss": 0.541,
      "step": 17450
    },
    {
      "epoch": 1.56,
      "grad_norm": 8.32970869856787,
      "learning_rate": 1.2330814730339763e-06,
      "loss": 0.6055,
      "step": 17451
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.177203715461129,
      "learning_rate": 1.2326064946052397e-06,
      "loss": 0.6262,
      "step": 17452
    },
    {
      "epoch": 1.56,
      "grad_norm": 8.61977247998519,
      "learning_rate": 1.2321315948125905e-06,
      "loss": 0.5412,
      "step": 17453
    },
    {
      "epoch": 1.56,
      "grad_norm": 8.971913365378189,
      "learning_rate": 1.231656773665943e-06,
      "loss": 0.5452,
      "step": 17454
    },
    {
      "epoch": 1.56,
      "grad_norm": 8.409040111201957,
      "learning_rate": 1.2311820311752076e-06,
      "loss": 0.6313,
      "step": 17455
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.284515083630078,
      "learning_rate": 1.230707367350294e-06,
      "loss": 0.5906,
      "step": 17456
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.789747707709738,
      "learning_rate": 1.2302327822011096e-06,
      "loss": 0.5443,
      "step": 17457
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.2682858013212215,
      "learning_rate": 1.2297582757375614e-06,
      "loss": 0.5602,
      "step": 17458
    },
    {
      "epoch": 1.56,
      "grad_norm": 8.185725919193953,
      "learning_rate": 1.2292838479695512e-06,
      "loss": 0.591,
      "step": 17459
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.4373965714299475,
      "learning_rate": 1.2288094989069827e-06,
      "loss": 0.5762,
      "step": 17460
    },
    {
      "epoch": 1.56,
      "grad_norm": 8.510182736549691,
      "learning_rate": 1.2283352285597578e-06,
      "loss": 0.5755,
      "step": 17461
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.678599073827171,
      "learning_rate": 1.2278610369377748e-06,
      "loss": 0.5818,
      "step": 17462
    },
    {
      "epoch": 1.56,
      "grad_norm": 11.04693049636378,
      "learning_rate": 1.2273869240509311e-06,
      "loss": 0.5812,
      "step": 17463
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.297345108869949,
      "learning_rate": 1.2269128899091259e-06,
      "loss": 0.5661,
      "step": 17464
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.0312790094692135,
      "learning_rate": 1.226438934522251e-06,
      "loss": 0.5528,
      "step": 17465
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.419629614039998,
      "learning_rate": 1.2259650579001976e-06,
      "loss": 0.5702,
      "step": 17466
    },
    {
      "epoch": 1.56,
      "grad_norm": 9.261425522074216,
      "learning_rate": 1.2254912600528591e-06,
      "loss": 0.6362,
      "step": 17467
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.643826913833802,
      "learning_rate": 1.2250175409901237e-06,
      "loss": 0.5456,
      "step": 17468
    },
    {
      "epoch": 1.56,
      "grad_norm": 11.500597059103457,
      "learning_rate": 1.2245439007218807e-06,
      "loss": 0.557,
      "step": 17469
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.146138494860286,
      "learning_rate": 1.2240703392580155e-06,
      "loss": 0.5582,
      "step": 17470
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.000505571331699,
      "learning_rate": 1.2235968566084145e-06,
      "loss": 0.609,
      "step": 17471
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.246654394445658,
      "learning_rate": 1.2231234527829572e-06,
      "loss": 0.6229,
      "step": 17472
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.188415570503029,
      "learning_rate": 1.222650127791527e-06,
      "loss": 0.5606,
      "step": 17473
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.417222533053564,
      "learning_rate": 1.2221768816440028e-06,
      "loss": 0.5509,
      "step": 17474
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.064451617865004,
      "learning_rate": 1.221703714350263e-06,
      "loss": 0.6459,
      "step": 17475
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.104306178049994,
      "learning_rate": 1.2212306259201839e-06,
      "loss": 0.5722,
      "step": 17476
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.921234408154279,
      "learning_rate": 1.2207576163636414e-06,
      "loss": 0.6274,
      "step": 17477
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.5989764362581305,
      "learning_rate": 1.220284685690506e-06,
      "loss": 0.5347,
      "step": 17478
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.808123218471105,
      "learning_rate": 1.2198118339106502e-06,
      "loss": 0.5984,
      "step": 17479
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.892510637659492,
      "learning_rate": 1.219339061033945e-06,
      "loss": 0.5075,
      "step": 17480
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.83172668768152,
      "learning_rate": 1.2188663670702567e-06,
      "loss": 0.6643,
      "step": 17481
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.828689804456462,
      "learning_rate": 1.218393752029452e-06,
      "loss": 0.5223,
      "step": 17482
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.3693019239925945,
      "learning_rate": 1.217921215921396e-06,
      "loss": 0.5822,
      "step": 17483
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.323963211257537,
      "learning_rate": 1.2174487587559536e-06,
      "loss": 0.6023,
      "step": 17484
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.28000771124247,
      "learning_rate": 1.2169763805429835e-06,
      "loss": 0.6106,
      "step": 17485
    },
    {
      "epoch": 1.56,
      "grad_norm": 8.192305334998114,
      "learning_rate": 1.2165040812923469e-06,
      "loss": 0.5892,
      "step": 17486
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.3409938334387,
      "learning_rate": 1.216031861013901e-06,
      "loss": 0.5737,
      "step": 17487
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.192748356193479,
      "learning_rate": 1.215559719717504e-06,
      "loss": 0.6072,
      "step": 17488
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.703313470632493,
      "learning_rate": 1.21508765741301e-06,
      "loss": 0.6082,
      "step": 17489
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.40457828702366,
      "learning_rate": 1.2146156741102738e-06,
      "loss": 0.5864,
      "step": 17490
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.683124187525168,
      "learning_rate": 1.2141437698191438e-06,
      "loss": 0.61,
      "step": 17491
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.5592190676857,
      "learning_rate": 1.2136719445494722e-06,
      "loss": 0.6083,
      "step": 17492
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.723956615389475,
      "learning_rate": 1.2132001983111063e-06,
      "loss": 0.5614,
      "step": 17493
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.456372449913669,
      "learning_rate": 1.2127285311138953e-06,
      "loss": 0.5535,
      "step": 17494
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.651094708277122,
      "learning_rate": 1.21225694296768e-06,
      "loss": 0.5762,
      "step": 17495
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.40011041661267,
      "learning_rate": 1.211785433882307e-06,
      "loss": 0.6046,
      "step": 17496
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.806765664935959,
      "learning_rate": 1.211314003867618e-06,
      "loss": 0.6059,
      "step": 17497
    },
    {
      "epoch": 1.56,
      "grad_norm": 9.9896509502385,
      "learning_rate": 1.2108426529334505e-06,
      "loss": 0.5961,
      "step": 17498
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.433581216076977,
      "learning_rate": 1.2103713810896444e-06,
      "loss": 0.5485,
      "step": 17499
    },
    {
      "epoch": 1.56,
      "grad_norm": 4.716122719120951,
      "learning_rate": 1.209900188346037e-06,
      "loss": 0.5595,
      "step": 17500
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.735565398442674,
      "learning_rate": 1.2094290747124625e-06,
      "loss": 0.6159,
      "step": 17501
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.077466605776167,
      "learning_rate": 1.2089580401987565e-06,
      "loss": 0.5383,
      "step": 17502
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.223327832765259,
      "learning_rate": 1.2084870848147478e-06,
      "loss": 0.5611,
      "step": 17503
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.072638303510687,
      "learning_rate": 1.2080162085702684e-06,
      "loss": 0.5714,
      "step": 17504
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.86199714333141,
      "learning_rate": 1.2075454114751456e-06,
      "loss": 0.6115,
      "step": 17505
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.563242160266035,
      "learning_rate": 1.2070746935392075e-06,
      "loss": 0.5824,
      "step": 17506
    },
    {
      "epoch": 1.56,
      "grad_norm": 8.373450207118786,
      "learning_rate": 1.2066040547722786e-06,
      "loss": 0.713,
      "step": 17507
    },
    {
      "epoch": 1.56,
      "grad_norm": 8.124507342068487,
      "learning_rate": 1.2061334951841842e-06,
      "loss": 0.5252,
      "step": 17508
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.101838980803422,
      "learning_rate": 1.2056630147847443e-06,
      "loss": 0.5963,
      "step": 17509
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.658888657759445,
      "learning_rate": 1.2051926135837782e-06,
      "loss": 0.5933,
      "step": 17510
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.852504894577199,
      "learning_rate": 1.2047222915911061e-06,
      "loss": 0.5454,
      "step": 17511
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.810147318295622,
      "learning_rate": 1.2042520488165444e-06,
      "loss": 0.5448,
      "step": 17512
    },
    {
      "epoch": 1.56,
      "grad_norm": 4.985444086672907,
      "learning_rate": 1.203781885269909e-06,
      "loss": 0.5109,
      "step": 17513
    },
    {
      "epoch": 1.56,
      "grad_norm": 8.111466516927749,
      "learning_rate": 1.2033118009610129e-06,
      "loss": 0.549,
      "step": 17514
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.253520658131106,
      "learning_rate": 1.2028417958996697e-06,
      "loss": 0.5829,
      "step": 17515
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.610722961520685,
      "learning_rate": 1.2023718700956871e-06,
      "loss": 0.5672,
      "step": 17516
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.587370140444301,
      "learning_rate": 1.2019020235588746e-06,
      "loss": 0.5798,
      "step": 17517
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.822915148311635,
      "learning_rate": 1.2014322562990399e-06,
      "loss": 0.5741,
      "step": 17518
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.676255404325774,
      "learning_rate": 1.2009625683259884e-06,
      "loss": 0.5556,
      "step": 17519
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.624171115592358,
      "learning_rate": 1.2004929596495235e-06,
      "loss": 0.5897,
      "step": 17520
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.2320123978786555,
      "learning_rate": 1.200023430279449e-06,
      "loss": 0.5597,
      "step": 17521
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.451904420178755,
      "learning_rate": 1.1995539802255613e-06,
      "loss": 0.5959,
      "step": 17522
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.187249608200956,
      "learning_rate": 1.1990846094976633e-06,
      "loss": 0.569,
      "step": 17523
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.700284464525787,
      "learning_rate": 1.1986153181055487e-06,
      "loss": 0.5978,
      "step": 17524
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.373417187084846,
      "learning_rate": 1.1981461060590149e-06,
      "loss": 0.5556,
      "step": 17525
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.83430413630013,
      "learning_rate": 1.1976769733678545e-06,
      "loss": 0.5487,
      "step": 17526
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.4292427010407,
      "learning_rate": 1.1972079200418607e-06,
      "loss": 0.5729,
      "step": 17527
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.34457904747408,
      "learning_rate": 1.1967389460908252e-06,
      "loss": 0.5698,
      "step": 17528
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.928357290386004,
      "learning_rate": 1.1962700515245334e-06,
      "loss": 0.4968,
      "step": 17529
    },
    {
      "epoch": 1.56,
      "grad_norm": 4.966534594500497,
      "learning_rate": 1.195801236352775e-06,
      "loss": 0.596,
      "step": 17530
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.049498188129611,
      "learning_rate": 1.1953325005853338e-06,
      "loss": 0.5768,
      "step": 17531
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.14840821757045,
      "learning_rate": 1.1948638442319948e-06,
      "loss": 0.5957,
      "step": 17532
    },
    {
      "epoch": 1.56,
      "grad_norm": 15.742418485475719,
      "learning_rate": 1.1943952673025405e-06,
      "loss": 0.527,
      "step": 17533
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.361481873804907,
      "learning_rate": 1.193926769806752e-06,
      "loss": 0.5711,
      "step": 17534
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.8058257334515115,
      "learning_rate": 1.1934583517544051e-06,
      "loss": 0.5481,
      "step": 17535
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.522003222664767,
      "learning_rate": 1.1929900131552797e-06,
      "loss": 0.6394,
      "step": 17536
    },
    {
      "epoch": 1.56,
      "grad_norm": 8.568055990091468,
      "learning_rate": 1.1925217540191514e-06,
      "loss": 0.5829,
      "step": 17537
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.6179381809101185,
      "learning_rate": 1.192053574355792e-06,
      "loss": 0.5401,
      "step": 17538
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.079608716629765,
      "learning_rate": 1.191585474174975e-06,
      "loss": 0.6404,
      "step": 17539
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.0144933926428905,
      "learning_rate": 1.1911174534864712e-06,
      "loss": 0.6223,
      "step": 17540
    },
    {
      "epoch": 1.56,
      "grad_norm": 8.931475641912185,
      "learning_rate": 1.19064951230005e-06,
      "loss": 0.6207,
      "step": 17541
    },
    {
      "epoch": 1.56,
      "grad_norm": 8.48372971843655,
      "learning_rate": 1.190181650625477e-06,
      "loss": 0.611,
      "step": 17542
    },
    {
      "epoch": 1.57,
      "grad_norm": 7.6976954213756885,
      "learning_rate": 1.1897138684725185e-06,
      "loss": 0.5677,
      "step": 17543
    },
    {
      "epoch": 1.57,
      "grad_norm": 8.203647708357257,
      "learning_rate": 1.189246165850939e-06,
      "loss": 0.604,
      "step": 17544
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.739744933272041,
      "learning_rate": 1.1887785427705002e-06,
      "loss": 0.5736,
      "step": 17545
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.4963304568911315,
      "learning_rate": 1.1883109992409626e-06,
      "loss": 0.5364,
      "step": 17546
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.199572128192072,
      "learning_rate": 1.1878435352720868e-06,
      "loss": 0.5595,
      "step": 17547
    },
    {
      "epoch": 1.57,
      "grad_norm": 9.619231243239213,
      "learning_rate": 1.1873761508736276e-06,
      "loss": 0.6181,
      "step": 17548
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.685566374448985,
      "learning_rate": 1.1869088460553419e-06,
      "loss": 0.5892,
      "step": 17549
    },
    {
      "epoch": 1.57,
      "grad_norm": 4.62412880939974,
      "learning_rate": 1.186441620826983e-06,
      "loss": 0.5654,
      "step": 17550
    },
    {
      "epoch": 1.57,
      "grad_norm": 4.944455648565771,
      "learning_rate": 1.1859744751983038e-06,
      "loss": 0.5602,
      "step": 17551
    },
    {
      "epoch": 1.57,
      "grad_norm": 8.046929538831908,
      "learning_rate": 1.1855074091790564e-06,
      "loss": 0.5862,
      "step": 17552
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.609677379300007,
      "learning_rate": 1.185040422778987e-06,
      "loss": 0.5675,
      "step": 17553
    },
    {
      "epoch": 1.57,
      "grad_norm": 8.178996898822414,
      "learning_rate": 1.1845735160078453e-06,
      "loss": 0.5181,
      "step": 17554
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.293241749510595,
      "learning_rate": 1.184106688875375e-06,
      "loss": 0.5253,
      "step": 17555
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.738574059653685,
      "learning_rate": 1.1836399413913202e-06,
      "loss": 0.4986,
      "step": 17556
    },
    {
      "epoch": 1.57,
      "grad_norm": 9.072559090882352,
      "learning_rate": 1.1831732735654244e-06,
      "loss": 0.5652,
      "step": 17557
    },
    {
      "epoch": 1.57,
      "grad_norm": 8.005683916670149,
      "learning_rate": 1.1827066854074276e-06,
      "loss": 0.5655,
      "step": 17558
    },
    {
      "epoch": 1.57,
      "grad_norm": 9.709550710540134,
      "learning_rate": 1.1822401769270686e-06,
      "loss": 0.5438,
      "step": 17559
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.951344911305748,
      "learning_rate": 1.1817737481340873e-06,
      "loss": 0.5661,
      "step": 17560
    },
    {
      "epoch": 1.57,
      "grad_norm": 7.75094450919303,
      "learning_rate": 1.181307399038215e-06,
      "loss": 0.5858,
      "step": 17561
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.577197074944826,
      "learning_rate": 1.180841129649189e-06,
      "loss": 0.5904,
      "step": 17562
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.664382703152952,
      "learning_rate": 1.18037493997674e-06,
      "loss": 0.541,
      "step": 17563
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.44626506938711,
      "learning_rate": 1.1799088300305988e-06,
      "loss": 0.528,
      "step": 17564
    },
    {
      "epoch": 1.57,
      "grad_norm": 11.149435392406595,
      "learning_rate": 1.179442799820496e-06,
      "loss": 0.636,
      "step": 17565
    },
    {
      "epoch": 1.57,
      "grad_norm": 10.308166721249735,
      "learning_rate": 1.1789768493561588e-06,
      "loss": 0.5913,
      "step": 17566
    },
    {
      "epoch": 1.57,
      "grad_norm": 7.762513803425698,
      "learning_rate": 1.1785109786473115e-06,
      "loss": 0.6022,
      "step": 17567
    },
    {
      "epoch": 1.57,
      "grad_norm": 7.936050500553025,
      "learning_rate": 1.178045187703678e-06,
      "loss": 0.5888,
      "step": 17568
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.545817291591111,
      "learning_rate": 1.1775794765349808e-06,
      "loss": 0.5215,
      "step": 17569
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.083466547408499,
      "learning_rate": 1.1771138451509411e-06,
      "loss": 0.5443,
      "step": 17570
    },
    {
      "epoch": 1.57,
      "grad_norm": 8.765782575109482,
      "learning_rate": 1.1766482935612777e-06,
      "loss": 0.5207,
      "step": 17571
    },
    {
      "epoch": 1.57,
      "grad_norm": 11.025180648890347,
      "learning_rate": 1.1761828217757086e-06,
      "loss": 0.5459,
      "step": 17572
    },
    {
      "epoch": 1.57,
      "grad_norm": 8.490475243845772,
      "learning_rate": 1.1757174298039504e-06,
      "loss": 0.5996,
      "step": 17573
    },
    {
      "epoch": 1.57,
      "grad_norm": 7.020848604936575,
      "learning_rate": 1.175252117655714e-06,
      "loss": 0.5472,
      "step": 17574
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.860657648245051,
      "learning_rate": 1.174786885340714e-06,
      "loss": 0.5577,
      "step": 17575
    },
    {
      "epoch": 1.57,
      "grad_norm": 8.077163190648616,
      "learning_rate": 1.1743217328686606e-06,
      "loss": 0.5521,
      "step": 17576
    },
    {
      "epoch": 1.57,
      "grad_norm": 8.586000214711108,
      "learning_rate": 1.173856660249263e-06,
      "loss": 0.5786,
      "step": 17577
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.298788892657059,
      "learning_rate": 1.1733916674922297e-06,
      "loss": 0.5813,
      "step": 17578
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.879600066371928,
      "learning_rate": 1.1729267546072643e-06,
      "loss": 0.5538,
      "step": 17579
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.0006610919841,
      "learning_rate": 1.1724619216040711e-06,
      "loss": 0.5493,
      "step": 17580
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.0419483445649895,
      "learning_rate": 1.1719971684923547e-06,
      "loss": 0.5685,
      "step": 17581
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.697016448544396,
      "learning_rate": 1.1715324952818131e-06,
      "loss": 0.558,
      "step": 17582
    },
    {
      "epoch": 1.57,
      "grad_norm": 8.934118603147919,
      "learning_rate": 1.1710679019821464e-06,
      "loss": 0.5632,
      "step": 17583
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.466383918998832,
      "learning_rate": 1.1706033886030522e-06,
      "loss": 0.5253,
      "step": 17584
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.232780734584359,
      "learning_rate": 1.170138955154227e-06,
      "loss": 0.525,
      "step": 17585
    },
    {
      "epoch": 1.57,
      "grad_norm": 7.767842190564642,
      "learning_rate": 1.1696746016453625e-06,
      "loss": 0.5371,
      "step": 17586
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.7705543079910795,
      "learning_rate": 1.1692103280861527e-06,
      "loss": 0.5996,
      "step": 17587
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.22515234073975,
      "learning_rate": 1.1687461344862877e-06,
      "loss": 0.5639,
      "step": 17588
    },
    {
      "epoch": 1.57,
      "grad_norm": 7.536203549360837,
      "learning_rate": 1.1682820208554568e-06,
      "loss": 0.6654,
      "step": 17589
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.666714806595545,
      "learning_rate": 1.167817987203348e-06,
      "loss": 0.5511,
      "step": 17590
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.553629238873237,
      "learning_rate": 1.167354033539647e-06,
      "loss": 0.5833,
      "step": 17591
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.778856454918171,
      "learning_rate": 1.1668901598740361e-06,
      "loss": 0.5411,
      "step": 17592
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.991463373254179,
      "learning_rate": 1.166426366216199e-06,
      "loss": 0.6322,
      "step": 17593
    },
    {
      "epoch": 1.57,
      "grad_norm": 4.94847358974116,
      "learning_rate": 1.1659626525758162e-06,
      "loss": 0.5848,
      "step": 17594
    },
    {
      "epoch": 1.57,
      "grad_norm": 10.455772706609835,
      "learning_rate": 1.1654990189625675e-06,
      "loss": 0.6667,
      "step": 17595
    },
    {
      "epoch": 1.57,
      "grad_norm": 8.732808264126477,
      "learning_rate": 1.1650354653861285e-06,
      "loss": 0.5502,
      "step": 17596
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.892109443238059,
      "learning_rate": 1.1645719918561754e-06,
      "loss": 0.5485,
      "step": 17597
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.623143240164863,
      "learning_rate": 1.1641085983823842e-06,
      "loss": 0.6079,
      "step": 17598
    },
    {
      "epoch": 1.57,
      "grad_norm": 7.174589788914577,
      "learning_rate": 1.1636452849744234e-06,
      "loss": 0.6394,
      "step": 17599
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.706087256025584,
      "learning_rate": 1.1631820516419666e-06,
      "loss": 0.6194,
      "step": 17600
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.6481938292610545,
      "learning_rate": 1.1627188983946818e-06,
      "loss": 0.5826,
      "step": 17601
    },
    {
      "epoch": 1.57,
      "grad_norm": 9.440743521523116,
      "learning_rate": 1.1622558252422362e-06,
      "loss": 0.5945,
      "step": 17602
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.530970971570583,
      "learning_rate": 1.161792832194296e-06,
      "loss": 0.5367,
      "step": 17603
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.134415825938056,
      "learning_rate": 1.1613299192605266e-06,
      "loss": 0.5751,
      "step": 17604
    },
    {
      "epoch": 1.57,
      "grad_norm": 7.461679353721445,
      "learning_rate": 1.1608670864505866e-06,
      "loss": 0.5266,
      "step": 17605
    },
    {
      "epoch": 1.57,
      "grad_norm": 8.441428249777484,
      "learning_rate": 1.1604043337741389e-06,
      "loss": 0.6099,
      "step": 17606
    },
    {
      "epoch": 1.57,
      "grad_norm": 7.151995179630965,
      "learning_rate": 1.159941661240842e-06,
      "loss": 0.5817,
      "step": 17607
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.509525239267581,
      "learning_rate": 1.1594790688603536e-06,
      "loss": 0.5275,
      "step": 17608
    },
    {
      "epoch": 1.57,
      "grad_norm": 7.229352783757999,
      "learning_rate": 1.159016556642329e-06,
      "loss": 0.5845,
      "step": 17609
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.703699121671413,
      "learning_rate": 1.1585541245964233e-06,
      "loss": 0.5707,
      "step": 17610
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.133714358673982,
      "learning_rate": 1.1580917727322877e-06,
      "loss": 0.5852,
      "step": 17611
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.10137159819938,
      "learning_rate": 1.157629501059571e-06,
      "loss": 0.6194,
      "step": 17612
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.401974627595709,
      "learning_rate": 1.1571673095879243e-06,
      "loss": 0.5648,
      "step": 17613
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.709783962686572,
      "learning_rate": 1.1567051983269944e-06,
      "loss": 0.6162,
      "step": 17614
    },
    {
      "epoch": 1.57,
      "grad_norm": 7.301432933364805,
      "learning_rate": 1.1562431672864266e-06,
      "loss": 0.54,
      "step": 17615
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.070898424920377,
      "learning_rate": 1.1557812164758653e-06,
      "loss": 0.5079,
      "step": 17616
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.737304784229033,
      "learning_rate": 1.1553193459049539e-06,
      "loss": 0.5557,
      "step": 17617
    },
    {
      "epoch": 1.57,
      "grad_norm": 7.868875271528416,
      "learning_rate": 1.1548575555833296e-06,
      "loss": 0.5449,
      "step": 17618
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.655474048772448,
      "learning_rate": 1.1543958455206332e-06,
      "loss": 0.6246,
      "step": 17619
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.6244002371188895,
      "learning_rate": 1.1539342157265016e-06,
      "loss": 0.5731,
      "step": 17620
    },
    {
      "epoch": 1.57,
      "grad_norm": 7.394729089621897,
      "learning_rate": 1.1534726662105711e-06,
      "loss": 0.5732,
      "step": 17621
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.166296960212958,
      "learning_rate": 1.1530111969824741e-06,
      "loss": 0.6375,
      "step": 17622
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.118623491137184,
      "learning_rate": 1.1525498080518454e-06,
      "loss": 0.5683,
      "step": 17623
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.55332377402804,
      "learning_rate": 1.1520884994283143e-06,
      "loss": 0.5322,
      "step": 17624
    },
    {
      "epoch": 1.57,
      "grad_norm": 7.436402084787508,
      "learning_rate": 1.1516272711215071e-06,
      "loss": 0.5557,
      "step": 17625
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.474515025132489,
      "learning_rate": 1.1511661231410525e-06,
      "loss": 0.5171,
      "step": 17626
    },
    {
      "epoch": 1.57,
      "grad_norm": 7.86304459797412,
      "learning_rate": 1.1507050554965765e-06,
      "loss": 0.5451,
      "step": 17627
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.889198987651089,
      "learning_rate": 1.1502440681977028e-06,
      "loss": 0.5785,
      "step": 17628
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.9069599294165265,
      "learning_rate": 1.1497831612540533e-06,
      "loss": 0.5159,
      "step": 17629
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.543425686435022,
      "learning_rate": 1.1493223346752497e-06,
      "loss": 0.5505,
      "step": 17630
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.757359246944983,
      "learning_rate": 1.1488615884709087e-06,
      "loss": 0.6159,
      "step": 17631
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.439803658969097,
      "learning_rate": 1.1484009226506476e-06,
      "loss": 0.6027,
      "step": 17632
    },
    {
      "epoch": 1.57,
      "grad_norm": 7.140376393213283,
      "learning_rate": 1.1479403372240827e-06,
      "loss": 0.6429,
      "step": 17633
    },
    {
      "epoch": 1.57,
      "grad_norm": 7.6127737813441945,
      "learning_rate": 1.1474798322008268e-06,
      "loss": 0.6145,
      "step": 17634
    },
    {
      "epoch": 1.57,
      "grad_norm": 4.636709319676939,
      "learning_rate": 1.1470194075904933e-06,
      "loss": 0.5427,
      "step": 17635
    },
    {
      "epoch": 1.57,
      "grad_norm": 7.26971096420603,
      "learning_rate": 1.146559063402693e-06,
      "loss": 0.6114,
      "step": 17636
    },
    {
      "epoch": 1.57,
      "grad_norm": 7.89730471010234,
      "learning_rate": 1.1460987996470312e-06,
      "loss": 0.6108,
      "step": 17637
    },
    {
      "epoch": 1.57,
      "grad_norm": 7.648310567035245,
      "learning_rate": 1.1456386163331196e-06,
      "loss": 0.61,
      "step": 17638
    },
    {
      "epoch": 1.57,
      "grad_norm": 8.068494283055411,
      "learning_rate": 1.145178513470559e-06,
      "loss": 0.6556,
      "step": 17639
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.305359839361662,
      "learning_rate": 1.1447184910689551e-06,
      "loss": 0.544,
      "step": 17640
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.956723305108273,
      "learning_rate": 1.1442585491379093e-06,
      "loss": 0.5757,
      "step": 17641
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.7889438822487245,
      "learning_rate": 1.1437986876870233e-06,
      "loss": 0.5868,
      "step": 17642
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.696042383043067,
      "learning_rate": 1.1433389067258954e-06,
      "loss": 0.5977,
      "step": 17643
    },
    {
      "epoch": 1.57,
      "grad_norm": 7.448711761256851,
      "learning_rate": 1.1428792062641208e-06,
      "loss": 0.6048,
      "step": 17644
    },
    {
      "epoch": 1.57,
      "grad_norm": 7.640758342254516,
      "learning_rate": 1.1424195863112958e-06,
      "loss": 0.564,
      "step": 17645
    },
    {
      "epoch": 1.57,
      "grad_norm": 10.838259614073339,
      "learning_rate": 1.1419600468770136e-06,
      "loss": 0.6264,
      "step": 17646
    },
    {
      "epoch": 1.57,
      "grad_norm": 9.124060894674399,
      "learning_rate": 1.1415005879708667e-06,
      "loss": 0.5439,
      "step": 17647
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.216321286460809,
      "learning_rate": 1.1410412096024455e-06,
      "loss": 0.573,
      "step": 17648
    },
    {
      "epoch": 1.57,
      "grad_norm": 8.351468318380226,
      "learning_rate": 1.1405819117813388e-06,
      "loss": 0.5775,
      "step": 17649
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.888094900736607,
      "learning_rate": 1.1401226945171317e-06,
      "loss": 0.7039,
      "step": 17650
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.096366194892592,
      "learning_rate": 1.1396635578194105e-06,
      "loss": 0.5539,
      "step": 17651
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.5777847621281404,
      "learning_rate": 1.1392045016977582e-06,
      "loss": 0.6413,
      "step": 17652
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.339265038727438,
      "learning_rate": 1.1387455261617587e-06,
      "loss": 0.5195,
      "step": 17653
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.772291480156985,
      "learning_rate": 1.1382866312209884e-06,
      "loss": 0.4871,
      "step": 17654
    },
    {
      "epoch": 1.58,
      "grad_norm": 7.556524185158354,
      "learning_rate": 1.1378278168850281e-06,
      "loss": 0.5538,
      "step": 17655
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.2158788639142015,
      "learning_rate": 1.1373690831634554e-06,
      "loss": 0.5623,
      "step": 17656
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.292218152188463,
      "learning_rate": 1.1369104300658424e-06,
      "loss": 0.5719,
      "step": 17657
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.811378328784184,
      "learning_rate": 1.1364518576017646e-06,
      "loss": 0.5493,
      "step": 17658
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.279648308308718,
      "learning_rate": 1.135993365780793e-06,
      "loss": 0.5831,
      "step": 17659
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.4619990909974785,
      "learning_rate": 1.1355349546124977e-06,
      "loss": 0.5972,
      "step": 17660
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.132199600284785,
      "learning_rate": 1.1350766241064492e-06,
      "loss": 0.5293,
      "step": 17661
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.4525953187982354,
      "learning_rate": 1.13461837427221e-06,
      "loss": 0.5937,
      "step": 17662
    },
    {
      "epoch": 1.58,
      "grad_norm": 8.799719374880995,
      "learning_rate": 1.1341602051193478e-06,
      "loss": 0.6274,
      "step": 17663
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.946026697565971,
      "learning_rate": 1.1337021166574247e-06,
      "loss": 0.5322,
      "step": 17664
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.9911416949183876,
      "learning_rate": 1.1332441088960032e-06,
      "loss": 0.6066,
      "step": 17665
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.402167870042317,
      "learning_rate": 1.1327861818446428e-06,
      "loss": 0.5597,
      "step": 17666
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.016841521583817,
      "learning_rate": 1.1323283355129034e-06,
      "loss": 0.5952,
      "step": 17667
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.357475724652351,
      "learning_rate": 1.1318705699103383e-06,
      "loss": 0.5984,
      "step": 17668
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.2110813929138775,
      "learning_rate": 1.1314128850465056e-06,
      "loss": 0.5688,
      "step": 17669
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.725419305558041,
      "learning_rate": 1.1309552809309553e-06,
      "loss": 0.4859,
      "step": 17670
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.934242950524787,
      "learning_rate": 1.1304977575732406e-06,
      "loss": 0.546,
      "step": 17671
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.091958764988245,
      "learning_rate": 1.130040314982911e-06,
      "loss": 0.5939,
      "step": 17672
    },
    {
      "epoch": 1.58,
      "grad_norm": 7.48677561714069,
      "learning_rate": 1.1295829531695146e-06,
      "loss": 0.5472,
      "step": 17673
    },
    {
      "epoch": 1.58,
      "grad_norm": 7.294167758087319,
      "learning_rate": 1.1291256721425998e-06,
      "loss": 0.6545,
      "step": 17674
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.851636419862357,
      "learning_rate": 1.1286684719117075e-06,
      "loss": 0.6196,
      "step": 17675
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.691897401135834,
      "learning_rate": 1.128211352486383e-06,
      "loss": 0.6005,
      "step": 17676
    },
    {
      "epoch": 1.58,
      "grad_norm": 10.849557928834944,
      "learning_rate": 1.1277543138761677e-06,
      "loss": 0.5473,
      "step": 17677
    },
    {
      "epoch": 1.58,
      "grad_norm": 7.017934869629899,
      "learning_rate": 1.127297356090601e-06,
      "loss": 0.5922,
      "step": 17678
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.5809124151661615,
      "learning_rate": 1.1268404791392212e-06,
      "loss": 0.5994,
      "step": 17679
    },
    {
      "epoch": 1.58,
      "grad_norm": 7.69663292637957,
      "learning_rate": 1.126383683031565e-06,
      "loss": 0.5861,
      "step": 17680
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.710745142889105,
      "learning_rate": 1.1259269677771656e-06,
      "loss": 0.5377,
      "step": 17681
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.922999184560165,
      "learning_rate": 1.1254703333855577e-06,
      "loss": 0.5694,
      "step": 17682
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.667874140926641,
      "learning_rate": 1.1250137798662703e-06,
      "loss": 0.5236,
      "step": 17683
    },
    {
      "epoch": 1.58,
      "grad_norm": 7.88009997981002,
      "learning_rate": 1.124557307228834e-06,
      "loss": 0.6224,
      "step": 17684
    },
    {
      "epoch": 1.58,
      "grad_norm": 8.318224975180792,
      "learning_rate": 1.124100915482777e-06,
      "loss": 0.5704,
      "step": 17685
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.653214432750371,
      "learning_rate": 1.1236446046376247e-06,
      "loss": 0.6105,
      "step": 17686
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.795984918594723,
      "learning_rate": 1.1231883747029042e-06,
      "loss": 0.5865,
      "step": 17687
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.313947423645107,
      "learning_rate": 1.122732225688135e-06,
      "loss": 0.5343,
      "step": 17688
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.593911455145975,
      "learning_rate": 1.1222761576028395e-06,
      "loss": 0.5655,
      "step": 17689
    },
    {
      "epoch": 1.58,
      "grad_norm": 8.452867654013565,
      "learning_rate": 1.1218201704565369e-06,
      "loss": 0.5362,
      "step": 17690
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.6077937599457695,
      "learning_rate": 1.1213642642587452e-06,
      "loss": 0.6037,
      "step": 17691
    },
    {
      "epoch": 1.58,
      "grad_norm": 7.06267875882567,
      "learning_rate": 1.1209084390189807e-06,
      "loss": 0.5483,
      "step": 17692
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.621951030899895,
      "learning_rate": 1.1204526947467587e-06,
      "loss": 0.5802,
      "step": 17693
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.372409619993127,
      "learning_rate": 1.119997031451589e-06,
      "loss": 0.5555,
      "step": 17694
    },
    {
      "epoch": 1.58,
      "grad_norm": 7.956070895443782,
      "learning_rate": 1.1195414491429846e-06,
      "loss": 0.5284,
      "step": 17695
    },
    {
      "epoch": 1.58,
      "grad_norm": 9.874166519113821,
      "learning_rate": 1.1190859478304556e-06,
      "loss": 0.6203,
      "step": 17696
    },
    {
      "epoch": 1.58,
      "grad_norm": 7.956959852078902,
      "learning_rate": 1.1186305275235064e-06,
      "loss": 0.5907,
      "step": 17697
    },
    {
      "epoch": 1.58,
      "grad_norm": 7.920914129514998,
      "learning_rate": 1.118175188231646e-06,
      "loss": 0.6417,
      "step": 17698
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.59495518621229,
      "learning_rate": 1.1177199299643765e-06,
      "loss": 0.5589,
      "step": 17699
    },
    {
      "epoch": 1.58,
      "grad_norm": 9.482105263370439,
      "learning_rate": 1.1172647527312036e-06,
      "loss": 0.6238,
      "step": 17700
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.481394261149703,
      "learning_rate": 1.1168096565416241e-06,
      "loss": 0.5802,
      "step": 17701
    },
    {
      "epoch": 1.58,
      "grad_norm": 7.789736639306126,
      "learning_rate": 1.1163546414051392e-06,
      "loss": 0.6241,
      "step": 17702
    },
    {
      "epoch": 1.58,
      "grad_norm": 7.193404325897423,
      "learning_rate": 1.115899707331246e-06,
      "loss": 0.569,
      "step": 17703
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.100987165619187,
      "learning_rate": 1.11544485432944e-06,
      "loss": 0.5206,
      "step": 17704
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.328007613583612,
      "learning_rate": 1.1149900824092164e-06,
      "loss": 0.547,
      "step": 17705
    },
    {
      "epoch": 1.58,
      "grad_norm": 9.561106162539094,
      "learning_rate": 1.1145353915800677e-06,
      "loss": 0.5553,
      "step": 17706
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.678303277414667,
      "learning_rate": 1.114080781851482e-06,
      "loss": 0.5616,
      "step": 17707
    },
    {
      "epoch": 1.58,
      "grad_norm": 11.551402481705045,
      "learning_rate": 1.1136262532329506e-06,
      "loss": 0.5104,
      "step": 17708
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.060719489729847,
      "learning_rate": 1.1131718057339597e-06,
      "loss": 0.6152,
      "step": 17709
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.293374329352161,
      "learning_rate": 1.1127174393639972e-06,
      "loss": 0.5854,
      "step": 17710
    },
    {
      "epoch": 1.58,
      "grad_norm": 7.487383975367874,
      "learning_rate": 1.1122631541325434e-06,
      "loss": 0.5251,
      "step": 17711
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.42005142352032,
      "learning_rate": 1.1118089500490825e-06,
      "loss": 0.5867,
      "step": 17712
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.9773956462611215,
      "learning_rate": 1.1113548271230961e-06,
      "loss": 0.55,
      "step": 17713
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.7930934978371456,
      "learning_rate": 1.1109007853640603e-06,
      "loss": 0.5151,
      "step": 17714
    },
    {
      "epoch": 1.58,
      "grad_norm": 7.551132356965308,
      "learning_rate": 1.1104468247814531e-06,
      "loss": 0.543,
      "step": 17715
    },
    {
      "epoch": 1.58,
      "grad_norm": 7.270968295741264,
      "learning_rate": 1.1099929453847513e-06,
      "loss": 0.5592,
      "step": 17716
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.536533418728756,
      "learning_rate": 1.109539147183427e-06,
      "loss": 0.6386,
      "step": 17717
    },
    {
      "epoch": 1.58,
      "grad_norm": 8.064194912985668,
      "learning_rate": 1.1090854301869535e-06,
      "loss": 0.5601,
      "step": 17718
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.259881509723264,
      "learning_rate": 1.108631794404802e-06,
      "loss": 0.556,
      "step": 17719
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.887729594770233,
      "learning_rate": 1.1081782398464385e-06,
      "loss": 0.5206,
      "step": 17720
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.841013845276308,
      "learning_rate": 1.1077247665213315e-06,
      "loss": 0.5894,
      "step": 17721
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.412811913463078,
      "learning_rate": 1.1072713744389457e-06,
      "loss": 0.5794,
      "step": 17722
    },
    {
      "epoch": 1.58,
      "grad_norm": 9.952991823220742,
      "learning_rate": 1.106818063608745e-06,
      "loss": 0.5901,
      "step": 17723
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.7409951695840755,
      "learning_rate": 1.1063648340401917e-06,
      "loss": 0.5957,
      "step": 17724
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.801133686735875,
      "learning_rate": 1.1059116857427465e-06,
      "loss": 0.5703,
      "step": 17725
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.708503282693633,
      "learning_rate": 1.1054586187258675e-06,
      "loss": 0.5421,
      "step": 17726
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.624638061538166,
      "learning_rate": 1.1050056329990088e-06,
      "loss": 0.5714,
      "step": 17727
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.523456884517037,
      "learning_rate": 1.1045527285716285e-06,
      "loss": 0.5089,
      "step": 17728
    },
    {
      "epoch": 1.58,
      "grad_norm": 8.34611735204096,
      "learning_rate": 1.1040999054531782e-06,
      "loss": 0.5162,
      "step": 17729
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.539722577345801,
      "learning_rate": 1.1036471636531109e-06,
      "loss": 0.5436,
      "step": 17730
    },
    {
      "epoch": 1.58,
      "grad_norm": 9.439643711616217,
      "learning_rate": 1.1031945031808766e-06,
      "loss": 0.5824,
      "step": 17731
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.490443016983458,
      "learning_rate": 1.102741924045924e-06,
      "loss": 0.5805,
      "step": 17732
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.867945960677296,
      "learning_rate": 1.1022894262576977e-06,
      "loss": 0.5828,
      "step": 17733
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.898182935104365,
      "learning_rate": 1.1018370098256438e-06,
      "loss": 0.5542,
      "step": 17734
    },
    {
      "epoch": 1.58,
      "grad_norm": 7.35400791048111,
      "learning_rate": 1.1013846747592054e-06,
      "loss": 0.6802,
      "step": 17735
    },
    {
      "epoch": 1.58,
      "grad_norm": 7.256036469806392,
      "learning_rate": 1.1009324210678241e-06,
      "loss": 0.5645,
      "step": 17736
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.392053705164908,
      "learning_rate": 1.1004802487609412e-06,
      "loss": 0.5175,
      "step": 17737
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.918740236424694,
      "learning_rate": 1.100028157847992e-06,
      "loss": 0.5794,
      "step": 17738
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.726331262929888,
      "learning_rate": 1.0995761483384154e-06,
      "loss": 0.5734,
      "step": 17739
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.07235344206029,
      "learning_rate": 1.0991242202416436e-06,
      "loss": 0.5432,
      "step": 17740
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.5209443307153565,
      "learning_rate": 1.0986723735671112e-06,
      "loss": 0.5768,
      "step": 17741
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.189145490494776,
      "learning_rate": 1.0982206083242486e-06,
      "loss": 0.5434,
      "step": 17742
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.233959160485008,
      "learning_rate": 1.0977689245224866e-06,
      "loss": 0.5461,
      "step": 17743
    },
    {
      "epoch": 1.58,
      "grad_norm": 8.252156029758718,
      "learning_rate": 1.0973173221712525e-06,
      "loss": 0.5704,
      "step": 17744
    },
    {
      "epoch": 1.58,
      "grad_norm": 7.109437680917873,
      "learning_rate": 1.0968658012799738e-06,
      "loss": 0.5485,
      "step": 17745
    },
    {
      "epoch": 1.58,
      "grad_norm": 8.413060250308108,
      "learning_rate": 1.0964143618580724e-06,
      "loss": 0.5912,
      "step": 17746
    },
    {
      "epoch": 1.58,
      "grad_norm": 8.593120382772229,
      "learning_rate": 1.0959630039149727e-06,
      "loss": 0.5295,
      "step": 17747
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.544339664796516,
      "learning_rate": 1.0955117274600952e-06,
      "loss": 0.5104,
      "step": 17748
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.058131026871833,
      "learning_rate": 1.0950605325028603e-06,
      "loss": 0.5737,
      "step": 17749
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.296662434684804,
      "learning_rate": 1.0946094190526857e-06,
      "loss": 0.5233,
      "step": 17750
    },
    {
      "epoch": 1.58,
      "grad_norm": 7.839372883811948,
      "learning_rate": 1.0941583871189859e-06,
      "loss": 0.5891,
      "step": 17751
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.90961878181536,
      "learning_rate": 1.0937074367111761e-06,
      "loss": 0.5487,
      "step": 17752
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.885026000082986,
      "learning_rate": 1.0932565678386686e-06,
      "loss": 0.6232,
      "step": 17753
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.478300776562269,
      "learning_rate": 1.0928057805108766e-06,
      "loss": 0.5585,
      "step": 17754
    },
    {
      "epoch": 1.58,
      "grad_norm": 7.374362122676819,
      "learning_rate": 1.0923550747372058e-06,
      "loss": 0.6231,
      "step": 17755
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.708365307632975,
      "learning_rate": 1.091904450527065e-06,
      "loss": 0.5851,
      "step": 17756
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.6851067766590315,
      "learning_rate": 1.0914539078898617e-06,
      "loss": 0.5702,
      "step": 17757
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.2750267580916805,
      "learning_rate": 1.0910034468349966e-06,
      "loss": 0.6326,
      "step": 17758
    },
    {
      "epoch": 1.58,
      "grad_norm": 9.650695614065741,
      "learning_rate": 1.090553067371875e-06,
      "loss": 0.5245,
      "step": 17759
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.860630450871063,
      "learning_rate": 1.090102769509896e-06,
      "loss": 0.594,
      "step": 17760
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.574096636205051,
      "learning_rate": 1.089652553258459e-06,
      "loss": 0.5299,
      "step": 17761
    },
    {
      "epoch": 1.58,
      "grad_norm": 8.332461574230216,
      "learning_rate": 1.0892024186269617e-06,
      "loss": 0.5467,
      "step": 17762
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.4375445676229495,
      "learning_rate": 1.0887523656248007e-06,
      "loss": 0.5871,
      "step": 17763
    },
    {
      "epoch": 1.58,
      "grad_norm": 9.261211107713168,
      "learning_rate": 1.0883023942613675e-06,
      "loss": 0.5681,
      "step": 17764
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.196270508884942,
      "learning_rate": 1.0878525045460547e-06,
      "loss": 0.6373,
      "step": 17765
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.064010773737531,
      "learning_rate": 1.087402696488254e-06,
      "loss": 0.5424,
      "step": 17766
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.243937581007242,
      "learning_rate": 1.0869529700973535e-06,
      "loss": 0.6306,
      "step": 17767
    },
    {
      "epoch": 1.59,
      "grad_norm": 5.904216207153296,
      "learning_rate": 1.0865033253827411e-06,
      "loss": 0.5861,
      "step": 17768
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.990579666496277,
      "learning_rate": 1.0860537623538008e-06,
      "loss": 0.6083,
      "step": 17769
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.096826427565851,
      "learning_rate": 1.085604281019918e-06,
      "loss": 0.5337,
      "step": 17770
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.28618132287405,
      "learning_rate": 1.0851548813904717e-06,
      "loss": 0.6189,
      "step": 17771
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.163644500256609,
      "learning_rate": 1.0847055634748443e-06,
      "loss": 0.5671,
      "step": 17772
    },
    {
      "epoch": 1.59,
      "grad_norm": 8.928538379805731,
      "learning_rate": 1.0842563272824135e-06,
      "loss": 0.5324,
      "step": 17773
    },
    {
      "epoch": 1.59,
      "grad_norm": 5.312560945268219,
      "learning_rate": 1.083807172822557e-06,
      "loss": 0.5815,
      "step": 17774
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.872664348330219,
      "learning_rate": 1.083358100104649e-06,
      "loss": 0.5656,
      "step": 17775
    },
    {
      "epoch": 1.59,
      "grad_norm": 5.787312677290027,
      "learning_rate": 1.082909109138065e-06,
      "loss": 0.608,
      "step": 17776
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.635507936883284,
      "learning_rate": 1.0824601999321737e-06,
      "loss": 0.6158,
      "step": 17777
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.263178158344926,
      "learning_rate": 1.0820113724963466e-06,
      "loss": 0.5486,
      "step": 17778
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.521753862970267,
      "learning_rate": 1.0815626268399514e-06,
      "loss": 0.5775,
      "step": 17779
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.8979017475594135,
      "learning_rate": 1.0811139629723554e-06,
      "loss": 0.5326,
      "step": 17780
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.222733420324145,
      "learning_rate": 1.0806653809029232e-06,
      "loss": 0.6071,
      "step": 17781
    },
    {
      "epoch": 1.59,
      "grad_norm": 10.383767741238573,
      "learning_rate": 1.080216880641019e-06,
      "loss": 0.5978,
      "step": 17782
    },
    {
      "epoch": 1.59,
      "grad_norm": 8.308116932836388,
      "learning_rate": 1.0797684621960036e-06,
      "loss": 0.624,
      "step": 17783
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.902020971840928,
      "learning_rate": 1.079320125577235e-06,
      "loss": 0.5757,
      "step": 17784
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.344343498406879,
      "learning_rate": 1.0788718707940733e-06,
      "loss": 0.5659,
      "step": 17785
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.570604203909546,
      "learning_rate": 1.0784236978558732e-06,
      "loss": 0.5649,
      "step": 17786
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.255248243184729,
      "learning_rate": 1.0779756067719904e-06,
      "loss": 0.5186,
      "step": 17787
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.747212977056284,
      "learning_rate": 1.077527597551778e-06,
      "loss": 0.6498,
      "step": 17788
    },
    {
      "epoch": 1.59,
      "grad_norm": 8.002220535377601,
      "learning_rate": 1.0770796702045883e-06,
      "loss": 0.5705,
      "step": 17789
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.135299744019238,
      "learning_rate": 1.076631824739768e-06,
      "loss": 0.5447,
      "step": 17790
    },
    {
      "epoch": 1.59,
      "grad_norm": 8.453854530225575,
      "learning_rate": 1.0761840611666663e-06,
      "loss": 0.5993,
      "step": 17791
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.969544962737254,
      "learning_rate": 1.0757363794946296e-06,
      "loss": 0.5941,
      "step": 17792
    },
    {
      "epoch": 1.59,
      "grad_norm": 8.68358230543144,
      "learning_rate": 1.0752887797330013e-06,
      "loss": 0.5784,
      "step": 17793
    },
    {
      "epoch": 1.59,
      "grad_norm": 9.061466529685534,
      "learning_rate": 1.0748412618911252e-06,
      "loss": 0.5841,
      "step": 17794
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.722506629732372,
      "learning_rate": 1.0743938259783432e-06,
      "loss": 0.5721,
      "step": 17795
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.5981482440850465,
      "learning_rate": 1.0739464720039917e-06,
      "loss": 0.5717,
      "step": 17796
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.801759063294678,
      "learning_rate": 1.0734991999774114e-06,
      "loss": 0.6548,
      "step": 17797
    },
    {
      "epoch": 1.59,
      "grad_norm": 5.806590692207091,
      "learning_rate": 1.0730520099079344e-06,
      "loss": 0.5441,
      "step": 17798
    },
    {
      "epoch": 1.59,
      "grad_norm": 5.635888149628336,
      "learning_rate": 1.0726049018048973e-06,
      "loss": 0.5414,
      "step": 17799
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.098682450096499,
      "learning_rate": 1.072157875677632e-06,
      "loss": 0.4946,
      "step": 17800
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.710716472867745,
      "learning_rate": 1.071710931535469e-06,
      "loss": 0.576,
      "step": 17801
    },
    {
      "epoch": 1.59,
      "grad_norm": 9.193788776004538,
      "learning_rate": 1.071264069387739e-06,
      "loss": 0.5718,
      "step": 17802
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.408977051020424,
      "learning_rate": 1.0708172892437663e-06,
      "loss": 0.565,
      "step": 17803
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.888604011482237,
      "learning_rate": 1.070370591112878e-06,
      "loss": 0.5446,
      "step": 17804
    },
    {
      "epoch": 1.59,
      "grad_norm": 5.917138532387253,
      "learning_rate": 1.0699239750043977e-06,
      "loss": 0.5455,
      "step": 17805
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.499181678780615,
      "learning_rate": 1.0694774409276482e-06,
      "loss": 0.5622,
      "step": 17806
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.001052303073451,
      "learning_rate": 1.0690309888919486e-06,
      "loss": 0.6349,
      "step": 17807
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.8295293123916565,
      "learning_rate": 1.0685846189066208e-06,
      "loss": 0.6075,
      "step": 17808
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.780082007565848,
      "learning_rate": 1.0681383309809773e-06,
      "loss": 0.5788,
      "step": 17809
    },
    {
      "epoch": 1.59,
      "grad_norm": 5.107598562288877,
      "learning_rate": 1.067692125124336e-06,
      "loss": 0.513,
      "step": 17810
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.4331468625276464,
      "learning_rate": 1.0672460013460107e-06,
      "loss": 0.6001,
      "step": 17811
    },
    {
      "epoch": 1.59,
      "grad_norm": 4.754641019969943,
      "learning_rate": 1.066799959655312e-06,
      "loss": 0.5313,
      "step": 17812
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.504273239894298,
      "learning_rate": 1.0663540000615497e-06,
      "loss": 0.6187,
      "step": 17813
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.30204082150503,
      "learning_rate": 1.0659081225740337e-06,
      "loss": 0.5977,
      "step": 17814
    },
    {
      "epoch": 1.59,
      "grad_norm": 5.5580178233010535,
      "learning_rate": 1.0654623272020714e-06,
      "loss": 0.5918,
      "step": 17815
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.705899725047185,
      "learning_rate": 1.065016613954965e-06,
      "loss": 0.5866,
      "step": 17816
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.18351206553548,
      "learning_rate": 1.0645709828420197e-06,
      "loss": 0.5814,
      "step": 17817
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.868224854281957,
      "learning_rate": 1.0641254338725366e-06,
      "loss": 0.5602,
      "step": 17818
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.160273082915747,
      "learning_rate": 1.0636799670558152e-06,
      "loss": 0.5908,
      "step": 17819
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.777197151151375,
      "learning_rate": 1.0632345824011558e-06,
      "loss": 0.6076,
      "step": 17820
    },
    {
      "epoch": 1.59,
      "grad_norm": 8.996773313510069,
      "learning_rate": 1.0627892799178519e-06,
      "loss": 0.5623,
      "step": 17821
    },
    {
      "epoch": 1.59,
      "grad_norm": 5.592931544479642,
      "learning_rate": 1.062344059615199e-06,
      "loss": 0.5519,
      "step": 17822
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.879203915935024,
      "learning_rate": 1.0618989215024906e-06,
      "loss": 0.5959,
      "step": 17823
    },
    {
      "epoch": 1.59,
      "grad_norm": 4.287736841737791,
      "learning_rate": 1.0614538655890184e-06,
      "loss": 0.5449,
      "step": 17824
    },
    {
      "epoch": 1.59,
      "grad_norm": 5.3857028850049105,
      "learning_rate": 1.061008891884071e-06,
      "loss": 0.5308,
      "step": 17825
    },
    {
      "epoch": 1.59,
      "grad_norm": 8.013516535495173,
      "learning_rate": 1.060564000396938e-06,
      "loss": 0.6342,
      "step": 17826
    },
    {
      "epoch": 1.59,
      "grad_norm": 5.570478011760584,
      "learning_rate": 1.0601191911369035e-06,
      "loss": 0.5189,
      "step": 17827
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.387882034458008,
      "learning_rate": 1.059674464113254e-06,
      "loss": 0.6003,
      "step": 17828
    },
    {
      "epoch": 1.59,
      "grad_norm": 8.381526059383503,
      "learning_rate": 1.059229819335269e-06,
      "loss": 0.5862,
      "step": 17829
    },
    {
      "epoch": 1.59,
      "grad_norm": 5.7110127841859075,
      "learning_rate": 1.0587852568122313e-06,
      "loss": 0.5873,
      "step": 17830
    },
    {
      "epoch": 1.59,
      "grad_norm": 8.566206534977958,
      "learning_rate": 1.0583407765534208e-06,
      "loss": 0.5806,
      "step": 17831
    },
    {
      "epoch": 1.59,
      "grad_norm": 8.878977940324667,
      "learning_rate": 1.0578963785681146e-06,
      "loss": 0.5767,
      "step": 17832
    },
    {
      "epoch": 1.59,
      "grad_norm": 8.281668846928035,
      "learning_rate": 1.0574520628655893e-06,
      "loss": 0.6024,
      "step": 17833
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.196724762141909,
      "learning_rate": 1.0570078294551173e-06,
      "loss": 0.6052,
      "step": 17834
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.139092853789234,
      "learning_rate": 1.0565636783459716e-06,
      "loss": 0.5689,
      "step": 17835
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.692549126222678,
      "learning_rate": 1.056119609547423e-06,
      "loss": 0.594,
      "step": 17836
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.882844845287688,
      "learning_rate": 1.0556756230687404e-06,
      "loss": 0.5634,
      "step": 17837
    },
    {
      "epoch": 1.59,
      "grad_norm": 8.633520614034097,
      "learning_rate": 1.0552317189191918e-06,
      "loss": 0.6378,
      "step": 17838
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.547676428937197,
      "learning_rate": 1.0547878971080434e-06,
      "loss": 0.6131,
      "step": 17839
    },
    {
      "epoch": 1.59,
      "grad_norm": 5.868919060663789,
      "learning_rate": 1.0543441576445573e-06,
      "loss": 0.5583,
      "step": 17840
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.9106206197607305,
      "learning_rate": 1.0539005005379948e-06,
      "loss": 0.5847,
      "step": 17841
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.336278129303098,
      "learning_rate": 1.053456925797618e-06,
      "loss": 0.5285,
      "step": 17842
    },
    {
      "epoch": 1.59,
      "grad_norm": 5.9401043502473145,
      "learning_rate": 1.0530134334326852e-06,
      "loss": 0.5644,
      "step": 17843
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.170904236183107,
      "learning_rate": 1.0525700234524527e-06,
      "loss": 0.5996,
      "step": 17844
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.070065680514238,
      "learning_rate": 1.0521266958661763e-06,
      "loss": 0.5936,
      "step": 17845
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.0022738450111195,
      "learning_rate": 1.0516834506831108e-06,
      "loss": 0.5893,
      "step": 17846
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.317311933933317,
      "learning_rate": 1.0512402879125055e-06,
      "loss": 0.5829,
      "step": 17847
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.412783155747906,
      "learning_rate": 1.0507972075636114e-06,
      "loss": 0.5976,
      "step": 17848
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.1555088405656,
      "learning_rate": 1.050354209645677e-06,
      "loss": 0.5349,
      "step": 17849
    },
    {
      "epoch": 1.59,
      "grad_norm": 9.001218295400314,
      "learning_rate": 1.049911294167949e-06,
      "loss": 0.6117,
      "step": 17850
    },
    {
      "epoch": 1.59,
      "grad_norm": 8.197922673965673,
      "learning_rate": 1.0494684611396723e-06,
      "loss": 0.5884,
      "step": 17851
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.447160999464562,
      "learning_rate": 1.0490257105700912e-06,
      "loss": 0.6193,
      "step": 17852
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.750135448385365,
      "learning_rate": 1.0485830424684445e-06,
      "loss": 0.5654,
      "step": 17853
    },
    {
      "epoch": 1.59,
      "grad_norm": 4.9668117626147446,
      "learning_rate": 1.048140456843973e-06,
      "loss": 0.5597,
      "step": 17854
    },
    {
      "epoch": 1.59,
      "grad_norm": 8.24674428608199,
      "learning_rate": 1.0476979537059169e-06,
      "loss": 0.5391,
      "step": 17855
    },
    {
      "epoch": 1.59,
      "grad_norm": 9.302652039303435,
      "learning_rate": 1.0472555330635093e-06,
      "loss": 0.5168,
      "step": 17856
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.269103699811801,
      "learning_rate": 1.0468131949259858e-06,
      "loss": 0.5416,
      "step": 17857
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.543041449120049,
      "learning_rate": 1.0463709393025794e-06,
      "loss": 0.596,
      "step": 17858
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.1948759877539725,
      "learning_rate": 1.045928766202523e-06,
      "loss": 0.5908,
      "step": 17859
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.8516659451070705,
      "learning_rate": 1.0454866756350429e-06,
      "loss": 0.5849,
      "step": 17860
    },
    {
      "epoch": 1.59,
      "grad_norm": 5.786807343905919,
      "learning_rate": 1.0450446676093679e-06,
      "loss": 0.6407,
      "step": 17861
    },
    {
      "epoch": 1.59,
      "grad_norm": 5.64601993650089,
      "learning_rate": 1.0446027421347243e-06,
      "loss": 0.6478,
      "step": 17862
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.6596466746364955,
      "learning_rate": 1.0441608992203366e-06,
      "loss": 0.6206,
      "step": 17863
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.33169474420827,
      "learning_rate": 1.043719138875427e-06,
      "loss": 0.5401,
      "step": 17864
    },
    {
      "epoch": 1.59,
      "grad_norm": 5.8527383994451165,
      "learning_rate": 1.0432774611092178e-06,
      "loss": 0.5835,
      "step": 17865
    },
    {
      "epoch": 1.59,
      "grad_norm": 8.117905511179607,
      "learning_rate": 1.0428358659309245e-06,
      "loss": 0.6231,
      "step": 17866
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.059053953460864,
      "learning_rate": 1.0423943533497671e-06,
      "loss": 0.5188,
      "step": 17867
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.541342704647159,
      "learning_rate": 1.0419529233749603e-06,
      "loss": 0.6044,
      "step": 17868
    },
    {
      "epoch": 1.59,
      "grad_norm": 5.816980523470719,
      "learning_rate": 1.0415115760157196e-06,
      "loss": 0.5509,
      "step": 17869
    },
    {
      "epoch": 1.59,
      "grad_norm": 5.994279221239179,
      "learning_rate": 1.0410703112812542e-06,
      "loss": 0.5982,
      "step": 17870
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.553976484046335,
      "learning_rate": 1.0406291291807769e-06,
      "loss": 0.5404,
      "step": 17871
    },
    {
      "epoch": 1.59,
      "grad_norm": 5.949487129269515,
      "learning_rate": 1.0401880297234963e-06,
      "loss": 0.6214,
      "step": 17872
    },
    {
      "epoch": 1.59,
      "grad_norm": 5.093023167912436,
      "learning_rate": 1.0397470129186177e-06,
      "loss": 0.5863,
      "step": 17873
    },
    {
      "epoch": 1.59,
      "grad_norm": 10.150629838619027,
      "learning_rate": 1.0393060787753472e-06,
      "loss": 0.5641,
      "step": 17874
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.257900662946631,
      "learning_rate": 1.0388652273028887e-06,
      "loss": 0.5688,
      "step": 17875
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.165158257296282,
      "learning_rate": 1.0384244585104435e-06,
      "loss": 0.5484,
      "step": 17876
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.327781142237763,
      "learning_rate": 1.0379837724072122e-06,
      "loss": 0.5791,
      "step": 17877
    },
    {
      "epoch": 1.59,
      "grad_norm": 9.143195088751693,
      "learning_rate": 1.0375431690023946e-06,
      "loss": 0.492,
      "step": 17878
    },
    {
      "epoch": 1.59,
      "grad_norm": 5.691840694717053,
      "learning_rate": 1.0371026483051843e-06,
      "loss": 0.5307,
      "step": 17879
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.395680615334547,
      "learning_rate": 1.0366622103247775e-06,
      "loss": 0.6097,
      "step": 17880
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.331842125765225,
      "learning_rate": 1.0362218550703674e-06,
      "loss": 0.56,
      "step": 17881
    },
    {
      "epoch": 1.6,
      "grad_norm": 7.340207433401166,
      "learning_rate": 1.0357815825511457e-06,
      "loss": 0.5532,
      "step": 17882
    },
    {
      "epoch": 1.6,
      "grad_norm": 8.11571916083722,
      "learning_rate": 1.0353413927763034e-06,
      "loss": 0.5661,
      "step": 17883
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.573559146599973,
      "learning_rate": 1.0349012857550256e-06,
      "loss": 0.5606,
      "step": 17884
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.851948934406516,
      "learning_rate": 1.0344612614965017e-06,
      "loss": 0.5749,
      "step": 17885
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.727476626141845,
      "learning_rate": 1.0340213200099124e-06,
      "loss": 0.5869,
      "step": 17886
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.048040298937964,
      "learning_rate": 1.0335814613044437e-06,
      "loss": 0.5422,
      "step": 17887
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.767757136300132,
      "learning_rate": 1.033141685389275e-06,
      "loss": 0.574,
      "step": 17888
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.986327850397047,
      "learning_rate": 1.0327019922735865e-06,
      "loss": 0.5545,
      "step": 17889
    },
    {
      "epoch": 1.6,
      "grad_norm": 7.171196785220543,
      "learning_rate": 1.0322623819665561e-06,
      "loss": 0.5628,
      "step": 17890
    },
    {
      "epoch": 1.6,
      "grad_norm": 8.154903852106468,
      "learning_rate": 1.0318228544773601e-06,
      "loss": 0.6194,
      "step": 17891
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.905868906887148,
      "learning_rate": 1.031383409815171e-06,
      "loss": 0.6186,
      "step": 17892
    },
    {
      "epoch": 1.6,
      "grad_norm": 8.204149627428391,
      "learning_rate": 1.0309440479891614e-06,
      "loss": 0.5553,
      "step": 17893
    },
    {
      "epoch": 1.6,
      "grad_norm": 8.439976331496068,
      "learning_rate": 1.030504769008503e-06,
      "loss": 0.6057,
      "step": 17894
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.669718611817225,
      "learning_rate": 1.030065572882365e-06,
      "loss": 0.5796,
      "step": 17895
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.687794437756498,
      "learning_rate": 1.029626459619915e-06,
      "loss": 0.6103,
      "step": 17896
    },
    {
      "epoch": 1.6,
      "grad_norm": 10.852417348612583,
      "learning_rate": 1.0291874292303162e-06,
      "loss": 0.5679,
      "step": 17897
    },
    {
      "epoch": 1.6,
      "grad_norm": 4.906574637240102,
      "learning_rate": 1.0287484817227355e-06,
      "loss": 0.5795,
      "step": 17898
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.491049326842565,
      "learning_rate": 1.0283096171063323e-06,
      "loss": 0.5738,
      "step": 17899
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.494878796091423,
      "learning_rate": 1.0278708353902672e-06,
      "loss": 0.5763,
      "step": 17900
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.263773414060625,
      "learning_rate": 1.0274321365837004e-06,
      "loss": 0.5503,
      "step": 17901
    },
    {
      "epoch": 1.6,
      "grad_norm": 8.973000855070655,
      "learning_rate": 1.0269935206957877e-06,
      "loss": 0.5439,
      "step": 17902
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.856203949969184,
      "learning_rate": 1.0265549877356845e-06,
      "loss": 0.5877,
      "step": 17903
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.625037269434562,
      "learning_rate": 1.0261165377125455e-06,
      "loss": 0.5882,
      "step": 17904
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.257673586606543,
      "learning_rate": 1.0256781706355195e-06,
      "loss": 0.5704,
      "step": 17905
    },
    {
      "epoch": 1.6,
      "grad_norm": 4.667115250922777,
      "learning_rate": 1.0252398865137585e-06,
      "loss": 0.5305,
      "step": 17906
    },
    {
      "epoch": 1.6,
      "grad_norm": 7.81372116738008,
      "learning_rate": 1.0248016853564103e-06,
      "loss": 0.5911,
      "step": 17907
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.91788836998909,
      "learning_rate": 1.024363567172621e-06,
      "loss": 0.5817,
      "step": 17908
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.754402470034206,
      "learning_rate": 1.0239255319715375e-06,
      "loss": 0.5946,
      "step": 17909
    },
    {
      "epoch": 1.6,
      "grad_norm": 7.224993216483908,
      "learning_rate": 1.0234875797622996e-06,
      "loss": 0.5629,
      "step": 17910
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.556795619134239,
      "learning_rate": 1.0230497105540499e-06,
      "loss": 0.5456,
      "step": 17911
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.007123396013225,
      "learning_rate": 1.0226119243559296e-06,
      "loss": 0.542,
      "step": 17912
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.6920822989340225,
      "learning_rate": 1.0221742211770736e-06,
      "loss": 0.5731,
      "step": 17913
    },
    {
      "epoch": 1.6,
      "grad_norm": 7.714460393338287,
      "learning_rate": 1.0217366010266195e-06,
      "loss": 0.5324,
      "step": 17914
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.5964535056578875,
      "learning_rate": 1.0212990639137016e-06,
      "loss": 0.5147,
      "step": 17915
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.932281550265375,
      "learning_rate": 1.020861609847454e-06,
      "loss": 0.6094,
      "step": 17916
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.427404379885244,
      "learning_rate": 1.020424238837005e-06,
      "loss": 0.57,
      "step": 17917
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.941842004995757,
      "learning_rate": 1.0199869508914845e-06,
      "loss": 0.5406,
      "step": 17918
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.909774034379287,
      "learning_rate": 1.0195497460200204e-06,
      "loss": 0.5572,
      "step": 17919
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.162914029904658,
      "learning_rate": 1.0191126242317384e-06,
      "loss": 0.5873,
      "step": 17920
    },
    {
      "epoch": 1.6,
      "grad_norm": 7.824884703504038,
      "learning_rate": 1.0186755855357622e-06,
      "loss": 0.5778,
      "step": 17921
    },
    {
      "epoch": 1.6,
      "grad_norm": 8.776991746339222,
      "learning_rate": 1.0182386299412155e-06,
      "loss": 0.5782,
      "step": 17922
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.660437497236797,
      "learning_rate": 1.0178017574572163e-06,
      "loss": 0.5608,
      "step": 17923
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.335076318755636,
      "learning_rate": 1.017364968092885e-06,
      "loss": 0.5019,
      "step": 17924
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.015201904414893,
      "learning_rate": 1.0169282618573372e-06,
      "loss": 0.5379,
      "step": 17925
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.391434608219474,
      "learning_rate": 1.0164916387596895e-06,
      "loss": 0.5552,
      "step": 17926
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.670931095303166,
      "learning_rate": 1.0160550988090567e-06,
      "loss": 0.5532,
      "step": 17927
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.475913534876052,
      "learning_rate": 1.0156186420145481e-06,
      "loss": 0.5439,
      "step": 17928
    },
    {
      "epoch": 1.6,
      "grad_norm": 7.801711744677141,
      "learning_rate": 1.0151822683852751e-06,
      "loss": 0.5882,
      "step": 17929
    },
    {
      "epoch": 1.6,
      "grad_norm": 10.58160274986975,
      "learning_rate": 1.0147459779303448e-06,
      "loss": 0.5873,
      "step": 17930
    },
    {
      "epoch": 1.6,
      "grad_norm": 7.874222126969351,
      "learning_rate": 1.0143097706588644e-06,
      "loss": 0.5666,
      "step": 17931
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.824714289315292,
      "learning_rate": 1.0138736465799392e-06,
      "loss": 0.5587,
      "step": 17932
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.175451958735273,
      "learning_rate": 1.0134376057026723e-06,
      "loss": 0.5381,
      "step": 17933
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.164983455541102,
      "learning_rate": 1.0130016480361648e-06,
      "loss": 0.6112,
      "step": 17934
    },
    {
      "epoch": 1.6,
      "grad_norm": 4.966470028820992,
      "learning_rate": 1.0125657735895183e-06,
      "loss": 0.5614,
      "step": 17935
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.765476108518543,
      "learning_rate": 1.0121299823718273e-06,
      "loss": 0.6289,
      "step": 17936
    },
    {
      "epoch": 1.6,
      "grad_norm": 8.040670272872225,
      "learning_rate": 1.0116942743921894e-06,
      "loss": 0.6106,
      "step": 17937
    },
    {
      "epoch": 1.6,
      "grad_norm": 4.924396206151737,
      "learning_rate": 1.0112586496596993e-06,
      "loss": 0.5606,
      "step": 17938
    },
    {
      "epoch": 1.6,
      "grad_norm": 7.144947512052589,
      "learning_rate": 1.0108231081834496e-06,
      "loss": 0.5773,
      "step": 17939
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.762917369082842,
      "learning_rate": 1.0103876499725318e-06,
      "loss": 0.5784,
      "step": 17940
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.339976656452219,
      "learning_rate": 1.0099522750360358e-06,
      "loss": 0.5999,
      "step": 17941
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.69336274950625,
      "learning_rate": 1.0095169833830481e-06,
      "loss": 0.4993,
      "step": 17942
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.883742156638753,
      "learning_rate": 1.009081775022653e-06,
      "loss": 0.5567,
      "step": 17943
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.108165981320566,
      "learning_rate": 1.0086466499639359e-06,
      "loss": 0.5578,
      "step": 17944
    },
    {
      "epoch": 1.6,
      "grad_norm": 7.898492172591261,
      "learning_rate": 1.0082116082159787e-06,
      "loss": 0.5635,
      "step": 17945
    },
    {
      "epoch": 1.6,
      "grad_norm": 9.116390084168836,
      "learning_rate": 1.0077766497878626e-06,
      "loss": 0.5573,
      "step": 17946
    },
    {
      "epoch": 1.6,
      "grad_norm": 7.773513822313401,
      "learning_rate": 1.007341774688666e-06,
      "loss": 0.5879,
      "step": 17947
    },
    {
      "epoch": 1.6,
      "grad_norm": 8.716741745423962,
      "learning_rate": 1.0069069829274674e-06,
      "loss": 0.5305,
      "step": 17948
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.466805175680484,
      "learning_rate": 1.00647227451334e-06,
      "loss": 0.5644,
      "step": 17949
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.602322894069889,
      "learning_rate": 1.0060376494553576e-06,
      "loss": 0.5627,
      "step": 17950
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.657535894055436,
      "learning_rate": 1.0056031077625928e-06,
      "loss": 0.5719,
      "step": 17951
    },
    {
      "epoch": 1.6,
      "grad_norm": 7.8299182180177045,
      "learning_rate": 1.005168649444116e-06,
      "loss": 0.6256,
      "step": 17952
    },
    {
      "epoch": 1.6,
      "grad_norm": 8.283424314138635,
      "learning_rate": 1.0047342745089945e-06,
      "loss": 0.5263,
      "step": 17953
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.596214865853448,
      "learning_rate": 1.0042999829662969e-06,
      "loss": 0.5497,
      "step": 17954
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.47954095715288,
      "learning_rate": 1.0038657748250857e-06,
      "loss": 0.5773,
      "step": 17955
    },
    {
      "epoch": 1.6,
      "grad_norm": 7.279826020334558,
      "learning_rate": 1.0034316500944264e-06,
      "loss": 0.5422,
      "step": 17956
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.239460555781768,
      "learning_rate": 1.0029976087833775e-06,
      "loss": 0.5782,
      "step": 17957
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.743679715730344,
      "learning_rate": 1.002563650901e-06,
      "loss": 0.5259,
      "step": 17958
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.983877867469228,
      "learning_rate": 1.0021297764563525e-06,
      "loss": 0.5879,
      "step": 17959
    },
    {
      "epoch": 1.6,
      "grad_norm": 8.988793410527816,
      "learning_rate": 1.0016959854584907e-06,
      "loss": 0.6043,
      "step": 17960
    },
    {
      "epoch": 1.6,
      "grad_norm": 8.616293180661032,
      "learning_rate": 1.0012622779164705e-06,
      "loss": 0.5924,
      "step": 17961
    },
    {
      "epoch": 1.6,
      "grad_norm": 7.354137099705865,
      "learning_rate": 1.0008286538393413e-06,
      "loss": 0.5641,
      "step": 17962
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.492538431726884,
      "learning_rate": 1.0003951132361567e-06,
      "loss": 0.549,
      "step": 17963
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.395931494875349,
      "learning_rate": 9.99961656115964e-07,
      "loss": 0.6514,
      "step": 17964
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.2551491776408,
      "learning_rate": 9.995282824878126e-07,
      "loss": 0.547,
      "step": 17965
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.803667568057199,
      "learning_rate": 9.990949923607475e-07,
      "loss": 0.6259,
      "step": 17966
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.99188088864589,
      "learning_rate": 9.986617857438142e-07,
      "loss": 0.5972,
      "step": 17967
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.48495625254666,
      "learning_rate": 9.982286626460513e-07,
      "loss": 0.6347,
      "step": 17968
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.423958116520265,
      "learning_rate": 9.977956230765012e-07,
      "loss": 0.5277,
      "step": 17969
    },
    {
      "epoch": 1.6,
      "grad_norm": 8.649402742421797,
      "learning_rate": 9.97362667044205e-07,
      "loss": 0.5798,
      "step": 17970
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.517187609907048,
      "learning_rate": 9.969297945581952e-07,
      "loss": 0.4766,
      "step": 17971
    },
    {
      "epoch": 1.6,
      "grad_norm": 10.055484090328658,
      "learning_rate": 9.964970056275103e-07,
      "loss": 0.531,
      "step": 17972
    },
    {
      "epoch": 1.6,
      "grad_norm": 7.560072759646931,
      "learning_rate": 9.96064300261182e-07,
      "loss": 0.5643,
      "step": 17973
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.448675832494569,
      "learning_rate": 9.956316784682447e-07,
      "loss": 0.5663,
      "step": 17974
    },
    {
      "epoch": 1.6,
      "grad_norm": 8.01787937007255,
      "learning_rate": 9.951991402577255e-07,
      "loss": 0.5199,
      "step": 17975
    },
    {
      "epoch": 1.6,
      "grad_norm": 7.083516418009018,
      "learning_rate": 9.947666856386539e-07,
      "loss": 0.5869,
      "step": 17976
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.54914215738966,
      "learning_rate": 9.943343146200562e-07,
      "loss": 0.5817,
      "step": 17977
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.444040045450929,
      "learning_rate": 9.939020272109573e-07,
      "loss": 0.5433,
      "step": 17978
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.932660118985743,
      "learning_rate": 9.934698234203806e-07,
      "loss": 0.5879,
      "step": 17979
    },
    {
      "epoch": 1.6,
      "grad_norm": 7.030407642824599,
      "learning_rate": 9.930377032573484e-07,
      "loss": 0.5981,
      "step": 17980
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.192101452125295,
      "learning_rate": 9.92605666730878e-07,
      "loss": 0.5833,
      "step": 17981
    },
    {
      "epoch": 1.6,
      "grad_norm": 8.465799717657243,
      "learning_rate": 9.92173713849988e-07,
      "loss": 0.602,
      "step": 17982
    },
    {
      "epoch": 1.6,
      "grad_norm": 4.497635446413826,
      "learning_rate": 9.917418446236948e-07,
      "loss": 0.5102,
      "step": 17983
    },
    {
      "epoch": 1.6,
      "grad_norm": 7.6673657941356685,
      "learning_rate": 9.913100590610142e-07,
      "loss": 0.5859,
      "step": 17984
    },
    {
      "epoch": 1.6,
      "grad_norm": 7.509090966173564,
      "learning_rate": 9.908783571709557e-07,
      "loss": 0.5493,
      "step": 17985
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.5947902460955,
      "learning_rate": 9.904467389625316e-07,
      "loss": 0.5423,
      "step": 17986
    },
    {
      "epoch": 1.6,
      "grad_norm": 7.855177744175048,
      "learning_rate": 9.900152044447526e-07,
      "loss": 0.5456,
      "step": 17987
    },
    {
      "epoch": 1.6,
      "grad_norm": 4.8348466250041255,
      "learning_rate": 9.895837536266228e-07,
      "loss": 0.5553,
      "step": 17988
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.057372022037554,
      "learning_rate": 9.891523865171499e-07,
      "loss": 0.5495,
      "step": 17989
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.519113164080511,
      "learning_rate": 9.88721103125337e-07,
      "loss": 0.6254,
      "step": 17990
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.723823069202138,
      "learning_rate": 9.882899034601867e-07,
      "loss": 0.5083,
      "step": 17991
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.369328881633095,
      "learning_rate": 9.878587875307004e-07,
      "loss": 0.5325,
      "step": 17992
    },
    {
      "epoch": 1.61,
      "grad_norm": 8.651705855498967,
      "learning_rate": 9.874277553458739e-07,
      "loss": 0.5902,
      "step": 17993
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.267116287098495,
      "learning_rate": 9.869968069147063e-07,
      "loss": 0.5903,
      "step": 17994
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.1115460899195035,
      "learning_rate": 9.865659422461915e-07,
      "loss": 0.5779,
      "step": 17995
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.742900131274435,
      "learning_rate": 9.861351613493237e-07,
      "loss": 0.533,
      "step": 17996
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.939148698926264,
      "learning_rate": 9.857044642330938e-07,
      "loss": 0.5221,
      "step": 17997
    },
    {
      "epoch": 1.61,
      "grad_norm": 8.352423426737401,
      "learning_rate": 9.852738509064935e-07,
      "loss": 0.5826,
      "step": 17998
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.074680716908424,
      "learning_rate": 9.848433213785096e-07,
      "loss": 0.5723,
      "step": 17999
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.011741785447262,
      "learning_rate": 9.844128756581267e-07,
      "loss": 0.5814,
      "step": 18000
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.052003390364944,
      "learning_rate": 9.839825137543313e-07,
      "loss": 0.5808,
      "step": 18001
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.321803259024591,
      "learning_rate": 9.835522356761062e-07,
      "loss": 0.5946,
      "step": 18002
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.463009381255687,
      "learning_rate": 9.83122041432432e-07,
      "loss": 0.5535,
      "step": 18003
    },
    {
      "epoch": 1.61,
      "grad_norm": 8.080380687985928,
      "learning_rate": 9.826919310322885e-07,
      "loss": 0.6401,
      "step": 18004
    },
    {
      "epoch": 1.61,
      "grad_norm": 9.839408560340061,
      "learning_rate": 9.822619044846542e-07,
      "loss": 0.6069,
      "step": 18005
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.6711660311823495,
      "learning_rate": 9.818319617985033e-07,
      "loss": 0.5354,
      "step": 18006
    },
    {
      "epoch": 1.61,
      "grad_norm": 10.270535068295798,
      "learning_rate": 9.814021029828103e-07,
      "loss": 0.5814,
      "step": 18007
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.11920750412422,
      "learning_rate": 9.809723280465472e-07,
      "loss": 0.5658,
      "step": 18008
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.1753964861662745,
      "learning_rate": 9.80542636998686e-07,
      "loss": 0.6164,
      "step": 18009
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.426947887606235,
      "learning_rate": 9.801130298481943e-07,
      "loss": 0.5783,
      "step": 18010
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.121693078933553,
      "learning_rate": 9.79683506604041e-07,
      "loss": 0.5818,
      "step": 18011
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.229544697844796,
      "learning_rate": 9.792540672751894e-07,
      "loss": 0.5771,
      "step": 18012
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.372668040645134,
      "learning_rate": 9.788247118706046e-07,
      "loss": 0.6475,
      "step": 18013
    },
    {
      "epoch": 1.61,
      "grad_norm": 9.209672970319012,
      "learning_rate": 9.783954403992469e-07,
      "loss": 0.6316,
      "step": 18014
    },
    {
      "epoch": 1.61,
      "grad_norm": 4.5970458846866,
      "learning_rate": 9.779662528700767e-07,
      "loss": 0.6519,
      "step": 18015
    },
    {
      "epoch": 1.61,
      "grad_norm": 8.894697246666857,
      "learning_rate": 9.775371492920533e-07,
      "loss": 0.5726,
      "step": 18016
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.130659306484971,
      "learning_rate": 9.771081296741324e-07,
      "loss": 0.5552,
      "step": 18017
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.870486430407558,
      "learning_rate": 9.766791940252713e-07,
      "loss": 0.629,
      "step": 18018
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.8728440336023064,
      "learning_rate": 9.762503423544189e-07,
      "loss": 0.5838,
      "step": 18019
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.884969695351848,
      "learning_rate": 9.758215746705297e-07,
      "loss": 0.5177,
      "step": 18020
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.672025775242217,
      "learning_rate": 9.753928909825517e-07,
      "loss": 0.4861,
      "step": 18021
    },
    {
      "epoch": 1.61,
      "grad_norm": 10.288569374391328,
      "learning_rate": 9.749642912994333e-07,
      "loss": 0.6394,
      "step": 18022
    },
    {
      "epoch": 1.61,
      "grad_norm": 8.740721509826953,
      "learning_rate": 9.74535775630121e-07,
      "loss": 0.5529,
      "step": 18023
    },
    {
      "epoch": 1.61,
      "grad_norm": 8.375824903345414,
      "learning_rate": 9.741073439835602e-07,
      "loss": 0.6015,
      "step": 18024
    },
    {
      "epoch": 1.61,
      "grad_norm": 11.260175340558872,
      "learning_rate": 9.736789963686905e-07,
      "loss": 0.637,
      "step": 18025
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.380123289173198,
      "learning_rate": 9.73250732794454e-07,
      "loss": 0.6042,
      "step": 18026
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.995545044959576,
      "learning_rate": 9.728225532697904e-07,
      "loss": 0.5772,
      "step": 18027
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.93833155522414,
      "learning_rate": 9.723944578036377e-07,
      "loss": 0.5632,
      "step": 18028
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.066695618865758,
      "learning_rate": 9.719664464049288e-07,
      "loss": 0.5477,
      "step": 18029
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.232785468837888,
      "learning_rate": 9.715385190826e-07,
      "loss": 0.5712,
      "step": 18030
    },
    {
      "epoch": 1.61,
      "grad_norm": 8.82081922891611,
      "learning_rate": 9.71110675845583e-07,
      "loss": 0.6058,
      "step": 18031
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.3196415182183285,
      "learning_rate": 9.706829167028065e-07,
      "loss": 0.6185,
      "step": 18032
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.96483592228943,
      "learning_rate": 9.702552416632e-07,
      "loss": 0.5336,
      "step": 18033
    },
    {
      "epoch": 1.61,
      "grad_norm": 8.206738380919841,
      "learning_rate": 9.698276507356902e-07,
      "loss": 0.6332,
      "step": 18034
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.805018256680471,
      "learning_rate": 9.694001439292023e-07,
      "loss": 0.5762,
      "step": 18035
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.172714142509349,
      "learning_rate": 9.689727212526595e-07,
      "loss": 0.5694,
      "step": 18036
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.316395020481393,
      "learning_rate": 9.68545382714985e-07,
      "loss": 0.5591,
      "step": 18037
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.50723767082141,
      "learning_rate": 9.68118128325095e-07,
      "loss": 0.5189,
      "step": 18038
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.4464019216834965,
      "learning_rate": 9.676909580919104e-07,
      "loss": 0.5395,
      "step": 18039
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.456851870142474,
      "learning_rate": 9.672638720243455e-07,
      "loss": 0.5606,
      "step": 18040
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.27135875505036,
      "learning_rate": 9.668368701313163e-07,
      "loss": 0.5923,
      "step": 18041
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.2691206161530575,
      "learning_rate": 9.664099524217357e-07,
      "loss": 0.6302,
      "step": 18042
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.840339803473557,
      "learning_rate": 9.659831189045131e-07,
      "loss": 0.5394,
      "step": 18043
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.773929094086008,
      "learning_rate": 9.655563695885595e-07,
      "loss": 0.5973,
      "step": 18044
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.582149149519686,
      "learning_rate": 9.6512970448278e-07,
      "loss": 0.6163,
      "step": 18045
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.987381089511987,
      "learning_rate": 9.647031235960823e-07,
      "loss": 0.5758,
      "step": 18046
    },
    {
      "epoch": 1.61,
      "grad_norm": 8.91635882671805,
      "learning_rate": 9.642766269373694e-07,
      "loss": 0.5619,
      "step": 18047
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.3212839491229165,
      "learning_rate": 9.638502145155437e-07,
      "loss": 0.5565,
      "step": 18048
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.509118485109363,
      "learning_rate": 9.634238863395063e-07,
      "loss": 0.6023,
      "step": 18049
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.281743591361238,
      "learning_rate": 9.629976424181559e-07,
      "loss": 0.6121,
      "step": 18050
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.140922322496242,
      "learning_rate": 9.625714827603883e-07,
      "loss": 0.5938,
      "step": 18051
    },
    {
      "epoch": 1.61,
      "grad_norm": 9.61034415215165,
      "learning_rate": 9.621454073750985e-07,
      "loss": 0.6023,
      "step": 18052
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.656095541778473,
      "learning_rate": 9.617194162711807e-07,
      "loss": 0.6243,
      "step": 18053
    },
    {
      "epoch": 1.61,
      "grad_norm": 9.367966395941323,
      "learning_rate": 9.612935094575266e-07,
      "loss": 0.5512,
      "step": 18054
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.176830376566611,
      "learning_rate": 9.608676869430256e-07,
      "loss": 0.5224,
      "step": 18055
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.931181996276936,
      "learning_rate": 9.604419487365678e-07,
      "loss": 0.6333,
      "step": 18056
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.405671132052591,
      "learning_rate": 9.600162948470375e-07,
      "loss": 0.6084,
      "step": 18057
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.732338912349537,
      "learning_rate": 9.595907252833186e-07,
      "loss": 0.5166,
      "step": 18058
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.1826508953102906,
      "learning_rate": 9.591652400542944e-07,
      "loss": 0.5534,
      "step": 18059
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.293707759915436,
      "learning_rate": 9.587398391688474e-07,
      "loss": 0.5868,
      "step": 18060
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.513590001232611,
      "learning_rate": 9.583145226358553e-07,
      "loss": 0.5036,
      "step": 18061
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.757857985886947,
      "learning_rate": 9.578892904641968e-07,
      "loss": 0.5454,
      "step": 18062
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.210660811821863,
      "learning_rate": 9.574641426627485e-07,
      "loss": 0.5832,
      "step": 18063
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.941386405073426,
      "learning_rate": 9.570390792403817e-07,
      "loss": 0.5554,
      "step": 18064
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.283088427749256,
      "learning_rate": 9.566141002059704e-07,
      "loss": 0.5792,
      "step": 18065
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.903695155945653,
      "learning_rate": 9.561892055683853e-07,
      "loss": 0.541,
      "step": 18066
    },
    {
      "epoch": 1.61,
      "grad_norm": 8.400972708696255,
      "learning_rate": 9.557643953364941e-07,
      "loss": 0.5204,
      "step": 18067
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.356171884780086,
      "learning_rate": 9.553396695191658e-07,
      "loss": 0.5825,
      "step": 18068
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.623470942125871,
      "learning_rate": 9.549150281252633e-07,
      "loss": 0.6098,
      "step": 18069
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.520050788493921,
      "learning_rate": 9.54490471163651e-07,
      "loss": 0.5763,
      "step": 18070
    },
    {
      "epoch": 1.61,
      "grad_norm": 4.968596769330583,
      "learning_rate": 9.540659986431915e-07,
      "loss": 0.6135,
      "step": 18071
    },
    {
      "epoch": 1.61,
      "grad_norm": 8.469512765493613,
      "learning_rate": 9.536416105727425e-07,
      "loss": 0.5946,
      "step": 18072
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.12278582604844,
      "learning_rate": 9.532173069611639e-07,
      "loss": 0.615,
      "step": 18073
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.958245490925318,
      "learning_rate": 9.527930878173114e-07,
      "loss": 0.5816,
      "step": 18074
    },
    {
      "epoch": 1.61,
      "grad_norm": 8.316766192917656,
      "learning_rate": 9.523689531500413e-07,
      "loss": 0.6127,
      "step": 18075
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.193514553598102,
      "learning_rate": 9.519449029682043e-07,
      "loss": 0.5993,
      "step": 18076
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.963136570665678,
      "learning_rate": 9.515209372806516e-07,
      "loss": 0.6115,
      "step": 18077
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.952080538996575,
      "learning_rate": 9.510970560962341e-07,
      "loss": 0.544,
      "step": 18078
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.212647188095568,
      "learning_rate": 9.506732594237983e-07,
      "loss": 0.619,
      "step": 18079
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.80202706092716,
      "learning_rate": 9.502495472721901e-07,
      "loss": 0.5578,
      "step": 18080
    },
    {
      "epoch": 1.61,
      "grad_norm": 8.666234992907185,
      "learning_rate": 9.498259196502551e-07,
      "loss": 0.558,
      "step": 18081
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.382497703328379,
      "learning_rate": 9.494023765668331e-07,
      "loss": 0.5952,
      "step": 18082
    },
    {
      "epoch": 1.61,
      "grad_norm": 8.032225442150004,
      "learning_rate": 9.489789180307663e-07,
      "loss": 0.6117,
      "step": 18083
    },
    {
      "epoch": 1.61,
      "grad_norm": 9.018436103334455,
      "learning_rate": 9.485555440508926e-07,
      "loss": 0.5011,
      "step": 18084
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.806528618356726,
      "learning_rate": 9.481322546360506e-07,
      "loss": 0.5781,
      "step": 18085
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.460995935337576,
      "learning_rate": 9.47709049795073e-07,
      "loss": 0.5743,
      "step": 18086
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.754529860536604,
      "learning_rate": 9.472859295367948e-07,
      "loss": 0.5801,
      "step": 18087
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.895373505525724,
      "learning_rate": 9.468628938700486e-07,
      "loss": 0.5663,
      "step": 18088
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.586870168147743,
      "learning_rate": 9.464399428036624e-07,
      "loss": 0.5567,
      "step": 18089
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.792852011115068,
      "learning_rate": 9.460170763464649e-07,
      "loss": 0.5966,
      "step": 18090
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.760744210161751,
      "learning_rate": 9.455942945072832e-07,
      "loss": 0.5565,
      "step": 18091
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.396715946723238,
      "learning_rate": 9.451715972949411e-07,
      "loss": 0.5727,
      "step": 18092
    },
    {
      "epoch": 1.61,
      "grad_norm": 8.58112569948942,
      "learning_rate": 9.447489847182623e-07,
      "loss": 0.5698,
      "step": 18093
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.372931367343568,
      "learning_rate": 9.443264567860689e-07,
      "loss": 0.5097,
      "step": 18094
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.172780107970153,
      "learning_rate": 9.439040135071775e-07,
      "loss": 0.5947,
      "step": 18095
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.582566467700685,
      "learning_rate": 9.434816548904075e-07,
      "loss": 0.5538,
      "step": 18096
    },
    {
      "epoch": 1.61,
      "grad_norm": 8.554207934452519,
      "learning_rate": 9.430593809445737e-07,
      "loss": 0.545,
      "step": 18097
    },
    {
      "epoch": 1.61,
      "grad_norm": 8.129507860853522,
      "learning_rate": 9.426371916784915e-07,
      "loss": 0.6629,
      "step": 18098
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.465862128530161,
      "learning_rate": 9.422150871009722e-07,
      "loss": 0.6073,
      "step": 18099
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.1068664945842315,
      "learning_rate": 9.417930672208275e-07,
      "loss": 0.5436,
      "step": 18100
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.111132390917083,
      "learning_rate": 9.413711320468655e-07,
      "loss": 0.5927,
      "step": 18101
    },
    {
      "epoch": 1.61,
      "grad_norm": 8.194102502326075,
      "learning_rate": 9.409492815878913e-07,
      "loss": 0.5917,
      "step": 18102
    },
    {
      "epoch": 1.61,
      "grad_norm": 8.904928591343761,
      "learning_rate": 9.405275158527122e-07,
      "loss": 0.5543,
      "step": 18103
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.794164152941984,
      "learning_rate": 9.401058348501308e-07,
      "loss": 0.5024,
      "step": 18104
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.56897720431886,
      "learning_rate": 9.396842385889488e-07,
      "loss": 0.5491,
      "step": 18105
    },
    {
      "epoch": 1.62,
      "grad_norm": 8.179582842668646,
      "learning_rate": 9.39262727077967e-07,
      "loss": 0.5528,
      "step": 18106
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.793682072915202,
      "learning_rate": 9.388413003259838e-07,
      "loss": 0.6067,
      "step": 18107
    },
    {
      "epoch": 1.62,
      "grad_norm": 9.306617016808637,
      "learning_rate": 9.38419958341793e-07,
      "loss": 0.5805,
      "step": 18108
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.500871052048031,
      "learning_rate": 9.379987011341917e-07,
      "loss": 0.6502,
      "step": 18109
    },
    {
      "epoch": 1.62,
      "grad_norm": 7.760892511905849,
      "learning_rate": 9.375775287119715e-07,
      "loss": 0.5615,
      "step": 18110
    },
    {
      "epoch": 1.62,
      "grad_norm": 8.650118440272678,
      "learning_rate": 9.371564410839235e-07,
      "loss": 0.6076,
      "step": 18111
    },
    {
      "epoch": 1.62,
      "grad_norm": 8.412104921769325,
      "learning_rate": 9.367354382588378e-07,
      "loss": 0.5963,
      "step": 18112
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.0778008201289815,
      "learning_rate": 9.363145202455026e-07,
      "loss": 0.615,
      "step": 18113
    },
    {
      "epoch": 1.62,
      "grad_norm": 8.983094967584496,
      "learning_rate": 9.358936870527025e-07,
      "loss": 0.6083,
      "step": 18114
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.88732145503271,
      "learning_rate": 9.354729386892198e-07,
      "loss": 0.5522,
      "step": 18115
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.546466854323706,
      "learning_rate": 9.350522751638386e-07,
      "loss": 0.5408,
      "step": 18116
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.809403545064443,
      "learning_rate": 9.346316964853391e-07,
      "loss": 0.5425,
      "step": 18117
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.663529293114206,
      "learning_rate": 9.342112026625e-07,
      "loss": 0.5507,
      "step": 18118
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.41937619631052,
      "learning_rate": 9.33790793704098e-07,
      "loss": 0.6171,
      "step": 18119
    },
    {
      "epoch": 1.62,
      "grad_norm": 8.204743906589467,
      "learning_rate": 9.3337046961891e-07,
      "loss": 0.6251,
      "step": 18120
    },
    {
      "epoch": 1.62,
      "grad_norm": 8.289674382428082,
      "learning_rate": 9.329502304157057e-07,
      "loss": 0.592,
      "step": 18121
    },
    {
      "epoch": 1.62,
      "grad_norm": 7.308616186988823,
      "learning_rate": 9.32530076103259e-07,
      "loss": 0.549,
      "step": 18122
    },
    {
      "epoch": 1.62,
      "grad_norm": 7.4421348802028575,
      "learning_rate": 9.321100066903394e-07,
      "loss": 0.5451,
      "step": 18123
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.5302337458914455,
      "learning_rate": 9.316900221857151e-07,
      "loss": 0.6012,
      "step": 18124
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.524583398293805,
      "learning_rate": 9.312701225981518e-07,
      "loss": 0.5795,
      "step": 18125
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.593520559160312,
      "learning_rate": 9.308503079364156e-07,
      "loss": 0.5382,
      "step": 18126
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.445094326605866,
      "learning_rate": 9.304305782092666e-07,
      "loss": 0.5515,
      "step": 18127
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.303265705376453,
      "learning_rate": 9.300109334254676e-07,
      "loss": 0.5719,
      "step": 18128
    },
    {
      "epoch": 1.62,
      "grad_norm": 8.090124058031757,
      "learning_rate": 9.295913735937778e-07,
      "loss": 0.5305,
      "step": 18129
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.682520079975063,
      "learning_rate": 9.291718987229531e-07,
      "loss": 0.5346,
      "step": 18130
    },
    {
      "epoch": 1.62,
      "grad_norm": 7.4746586132778035,
      "learning_rate": 9.287525088217503e-07,
      "loss": 0.5723,
      "step": 18131
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.945535703779306,
      "learning_rate": 9.283332038989223e-07,
      "loss": 0.6007,
      "step": 18132
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.425709410167282,
      "learning_rate": 9.27913983963224e-07,
      "loss": 0.5145,
      "step": 18133
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.893429475420057,
      "learning_rate": 9.27494849023402e-07,
      "loss": 0.6322,
      "step": 18134
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.993502580428424,
      "learning_rate": 9.270757990882062e-07,
      "loss": 0.5916,
      "step": 18135
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.6011847511657775,
      "learning_rate": 9.266568341663834e-07,
      "loss": 0.5758,
      "step": 18136
    },
    {
      "epoch": 1.62,
      "grad_norm": 7.46842099981995,
      "learning_rate": 9.262379542666789e-07,
      "loss": 0.6099,
      "step": 18137
    },
    {
      "epoch": 1.62,
      "grad_norm": 7.7779002170033396,
      "learning_rate": 9.258191593978361e-07,
      "loss": 0.6607,
      "step": 18138
    },
    {
      "epoch": 1.62,
      "grad_norm": 7.280392844724336,
      "learning_rate": 9.25400449568597e-07,
      "loss": 0.5746,
      "step": 18139
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.997898779026832,
      "learning_rate": 9.249818247876985e-07,
      "loss": 0.5594,
      "step": 18140
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.430487917373639,
      "learning_rate": 9.245632850638808e-07,
      "loss": 0.6142,
      "step": 18141
    },
    {
      "epoch": 1.62,
      "grad_norm": 7.775979933853435,
      "learning_rate": 9.241448304058797e-07,
      "loss": 0.6309,
      "step": 18142
    },
    {
      "epoch": 1.62,
      "grad_norm": 7.0164742740793615,
      "learning_rate": 9.2372646082243e-07,
      "loss": 0.5896,
      "step": 18143
    },
    {
      "epoch": 1.62,
      "grad_norm": 8.06522074831444,
      "learning_rate": 9.233081763222629e-07,
      "loss": 0.5374,
      "step": 18144
    },
    {
      "epoch": 1.62,
      "grad_norm": 8.90661086655355,
      "learning_rate": 9.228899769141092e-07,
      "loss": 0.6205,
      "step": 18145
    },
    {
      "epoch": 1.62,
      "grad_norm": 8.862815489606257,
      "learning_rate": 9.224718626067003e-07,
      "loss": 0.595,
      "step": 18146
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.364263405771982,
      "learning_rate": 9.220538334087603e-07,
      "loss": 0.5547,
      "step": 18147
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.647371388757861,
      "learning_rate": 9.216358893290156e-07,
      "loss": 0.5403,
      "step": 18148
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.4180001227946075,
      "learning_rate": 9.212180303761902e-07,
      "loss": 0.5645,
      "step": 18149
    },
    {
      "epoch": 1.62,
      "grad_norm": 7.180771357419384,
      "learning_rate": 9.208002565590063e-07,
      "loss": 0.6358,
      "step": 18150
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.777220386331927,
      "learning_rate": 9.203825678861855e-07,
      "loss": 0.5419,
      "step": 18151
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.947930827615376,
      "learning_rate": 9.199649643664427e-07,
      "loss": 0.5598,
      "step": 18152
    },
    {
      "epoch": 1.62,
      "grad_norm": 7.542144097294394,
      "learning_rate": 9.195474460084957e-07,
      "loss": 0.5338,
      "step": 18153
    },
    {
      "epoch": 1.62,
      "grad_norm": 7.6298807296404645,
      "learning_rate": 9.191300128210601e-07,
      "loss": 0.5691,
      "step": 18154
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.815459108308268,
      "learning_rate": 9.187126648128486e-07,
      "loss": 0.5316,
      "step": 18155
    },
    {
      "epoch": 1.62,
      "grad_norm": 4.93702853052063,
      "learning_rate": 9.182954019925722e-07,
      "loss": 0.5578,
      "step": 18156
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.2538920253433465,
      "learning_rate": 9.178782243689422e-07,
      "loss": 0.5497,
      "step": 18157
    },
    {
      "epoch": 1.62,
      "grad_norm": 7.481762684575561,
      "learning_rate": 9.174611319506627e-07,
      "loss": 0.6307,
      "step": 18158
    },
    {
      "epoch": 1.62,
      "grad_norm": 7.274444107403817,
      "learning_rate": 9.170441247464429e-07,
      "loss": 0.5542,
      "step": 18159
    },
    {
      "epoch": 1.62,
      "grad_norm": 7.146270320364464,
      "learning_rate": 9.166272027649842e-07,
      "loss": 0.5264,
      "step": 18160
    },
    {
      "epoch": 1.62,
      "grad_norm": 7.319799178275437,
      "learning_rate": 9.162103660149907e-07,
      "loss": 0.5936,
      "step": 18161
    },
    {
      "epoch": 1.62,
      "grad_norm": 9.345474281761662,
      "learning_rate": 9.157936145051616e-07,
      "loss": 0.5404,
      "step": 18162
    },
    {
      "epoch": 1.62,
      "grad_norm": 7.591932417497094,
      "learning_rate": 9.153769482441971e-07,
      "loss": 0.5788,
      "step": 18163
    },
    {
      "epoch": 1.62,
      "grad_norm": 7.3519645834185745,
      "learning_rate": 9.149603672407948e-07,
      "loss": 0.5785,
      "step": 18164
    },
    {
      "epoch": 1.62,
      "grad_norm": 8.629589389749933,
      "learning_rate": 9.145438715036476e-07,
      "loss": 0.571,
      "step": 18165
    },
    {
      "epoch": 1.62,
      "grad_norm": 7.90974301994723,
      "learning_rate": 9.141274610414502e-07,
      "loss": 0.6099,
      "step": 18166
    },
    {
      "epoch": 1.62,
      "grad_norm": 8.058936816321117,
      "learning_rate": 9.137111358628936e-07,
      "loss": 0.5566,
      "step": 18167
    },
    {
      "epoch": 1.62,
      "grad_norm": 8.54178005201007,
      "learning_rate": 9.132948959766685e-07,
      "loss": 0.569,
      "step": 18168
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.083389351051305,
      "learning_rate": 9.12878741391463e-07,
      "loss": 0.591,
      "step": 18169
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.937526517419829,
      "learning_rate": 9.124626721159641e-07,
      "loss": 0.5185,
      "step": 18170
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.642668794253535,
      "learning_rate": 9.120466881588541e-07,
      "loss": 0.6023,
      "step": 18171
    },
    {
      "epoch": 1.62,
      "grad_norm": 7.5647297583179665,
      "learning_rate": 9.116307895288184e-07,
      "loss": 0.5742,
      "step": 18172
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.550828858481529,
      "learning_rate": 9.112149762345351e-07,
      "loss": 0.5813,
      "step": 18173
    },
    {
      "epoch": 1.62,
      "grad_norm": 7.251916190084358,
      "learning_rate": 9.107992482846851e-07,
      "loss": 0.6446,
      "step": 18174
    },
    {
      "epoch": 1.62,
      "grad_norm": 4.755232465377591,
      "learning_rate": 9.103836056879456e-07,
      "loss": 0.4867,
      "step": 18175
    },
    {
      "epoch": 1.62,
      "grad_norm": 9.11630851057307,
      "learning_rate": 9.099680484529922e-07,
      "loss": 0.5677,
      "step": 18176
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.4569925344765995,
      "learning_rate": 9.095525765884999e-07,
      "loss": 0.5735,
      "step": 18177
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.0885229444357,
      "learning_rate": 9.091371901031387e-07,
      "loss": 0.5577,
      "step": 18178
    },
    {
      "epoch": 1.62,
      "grad_norm": 7.111844569047185,
      "learning_rate": 9.087218890055794e-07,
      "loss": 0.5698,
      "step": 18179
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.5841871307804745,
      "learning_rate": 9.08306673304491e-07,
      "loss": 0.5376,
      "step": 18180
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.39593420742997,
      "learning_rate": 9.078915430085406e-07,
      "loss": 0.5835,
      "step": 18181
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.65293981552305,
      "learning_rate": 9.074764981263922e-07,
      "loss": 0.5567,
      "step": 18182
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.672798146785269,
      "learning_rate": 9.070615386667108e-07,
      "loss": 0.5872,
      "step": 18183
    },
    {
      "epoch": 1.62,
      "grad_norm": 8.071761118214333,
      "learning_rate": 9.066466646381549e-07,
      "loss": 0.5935,
      "step": 18184
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.96932499253134,
      "learning_rate": 9.062318760493861e-07,
      "loss": 0.5416,
      "step": 18185
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.768353946733788,
      "learning_rate": 9.058171729090626e-07,
      "loss": 0.5886,
      "step": 18186
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.6956942295698685,
      "learning_rate": 9.054025552258384e-07,
      "loss": 0.5377,
      "step": 18187
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.7887822762732215,
      "learning_rate": 9.049880230083686e-07,
      "loss": 0.6197,
      "step": 18188
    },
    {
      "epoch": 1.62,
      "grad_norm": 8.413489361931383,
      "learning_rate": 9.045735762653062e-07,
      "loss": 0.5703,
      "step": 18189
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.861512754231773,
      "learning_rate": 9.041592150053025e-07,
      "loss": 0.6017,
      "step": 18190
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.568912488610636,
      "learning_rate": 9.037449392370046e-07,
      "loss": 0.5553,
      "step": 18191
    },
    {
      "epoch": 1.62,
      "grad_norm": 8.638129074091095,
      "learning_rate": 9.033307489690602e-07,
      "loss": 0.6256,
      "step": 18192
    },
    {
      "epoch": 1.62,
      "grad_norm": 7.07565636234224,
      "learning_rate": 9.029166442101151e-07,
      "loss": 0.5528,
      "step": 18193
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.172719327698289,
      "learning_rate": 9.025026249688124e-07,
      "loss": 0.6111,
      "step": 18194
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.35347233747857,
      "learning_rate": 9.02088691253794e-07,
      "loss": 0.5738,
      "step": 18195
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.545685696964957,
      "learning_rate": 9.01674843073701e-07,
      "loss": 0.5307,
      "step": 18196
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.778536255055442,
      "learning_rate": 9.012610804371691e-07,
      "loss": 0.5791,
      "step": 18197
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.150687859433288,
      "learning_rate": 9.008474033528364e-07,
      "loss": 0.5596,
      "step": 18198
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.665846221871463,
      "learning_rate": 9.004338118293371e-07,
      "loss": 0.5095,
      "step": 18199
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.2974351778042745,
      "learning_rate": 9.000203058753043e-07,
      "loss": 0.5658,
      "step": 18200
    },
    {
      "epoch": 1.62,
      "grad_norm": 4.938608079163612,
      "learning_rate": 8.9960688549937e-07,
      "loss": 0.585,
      "step": 18201
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.878645190647172,
      "learning_rate": 8.991935507101607e-07,
      "loss": 0.4916,
      "step": 18202
    },
    {
      "epoch": 1.62,
      "grad_norm": 7.44400088175061,
      "learning_rate": 8.987803015163071e-07,
      "loss": 0.5172,
      "step": 18203
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.929406495873715,
      "learning_rate": 8.983671379264313e-07,
      "loss": 0.5805,
      "step": 18204
    },
    {
      "epoch": 1.62,
      "grad_norm": 7.483117842506898,
      "learning_rate": 8.979540599491598e-07,
      "loss": 0.5957,
      "step": 18205
    },
    {
      "epoch": 1.62,
      "grad_norm": 4.781265618241893,
      "learning_rate": 8.975410675931134e-07,
      "loss": 0.572,
      "step": 18206
    },
    {
      "epoch": 1.62,
      "grad_norm": 7.4633718299878975,
      "learning_rate": 8.971281608669131e-07,
      "loss": 0.5789,
      "step": 18207
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.806068469258499,
      "learning_rate": 8.967153397791778e-07,
      "loss": 0.5418,
      "step": 18208
    },
    {
      "epoch": 1.62,
      "grad_norm": 7.028874673802735,
      "learning_rate": 8.963026043385243e-07,
      "loss": 0.5481,
      "step": 18209
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.839980543359812,
      "learning_rate": 8.958899545535659e-07,
      "loss": 0.5443,
      "step": 18210
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.907919751102488,
      "learning_rate": 8.954773904329172e-07,
      "loss": 0.5791,
      "step": 18211
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.4091860759876695,
      "learning_rate": 8.95064911985189e-07,
      "loss": 0.555,
      "step": 18212
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.105995657521063,
      "learning_rate": 8.946525192189909e-07,
      "loss": 0.5658,
      "step": 18213
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.083985179423467,
      "learning_rate": 8.942402121429311e-07,
      "loss": 0.5575,
      "step": 18214
    },
    {
      "epoch": 1.62,
      "grad_norm": 8.157613019451317,
      "learning_rate": 8.938279907656172e-07,
      "loss": 0.523,
      "step": 18215
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.404431203815068,
      "learning_rate": 8.934158550956507e-07,
      "loss": 0.5383,
      "step": 18216
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.06904448834443,
      "learning_rate": 8.930038051416345e-07,
      "loss": 0.5917,
      "step": 18217
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.411463415487744,
      "learning_rate": 8.925918409121692e-07,
      "loss": 0.536,
      "step": 18218
    },
    {
      "epoch": 1.63,
      "grad_norm": 8.44280171565088,
      "learning_rate": 8.921799624158545e-07,
      "loss": 0.5371,
      "step": 18219
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.989123998941415,
      "learning_rate": 8.917681696612868e-07,
      "loss": 0.5949,
      "step": 18220
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.411109279427316,
      "learning_rate": 8.913564626570614e-07,
      "loss": 0.5828,
      "step": 18221
    },
    {
      "epoch": 1.63,
      "grad_norm": 11.50710360909609,
      "learning_rate": 8.909448414117739e-07,
      "loss": 0.6403,
      "step": 18222
    },
    {
      "epoch": 1.63,
      "grad_norm": 8.171957785944814,
      "learning_rate": 8.905333059340126e-07,
      "loss": 0.5338,
      "step": 18223
    },
    {
      "epoch": 1.63,
      "grad_norm": 9.368755365464496,
      "learning_rate": 8.901218562323693e-07,
      "loss": 0.6161,
      "step": 18224
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.627411603439027,
      "learning_rate": 8.897104923154314e-07,
      "loss": 0.5799,
      "step": 18225
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.5215441214076915,
      "learning_rate": 8.892992141917855e-07,
      "loss": 0.5806,
      "step": 18226
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.254197858888724,
      "learning_rate": 8.88888021870018e-07,
      "loss": 0.5572,
      "step": 18227
    },
    {
      "epoch": 1.63,
      "grad_norm": 8.07503258611836,
      "learning_rate": 8.88476915358708e-07,
      "loss": 0.5766,
      "step": 18228
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.243080125410569,
      "learning_rate": 8.880658946664389e-07,
      "loss": 0.5344,
      "step": 18229
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.724137989824178,
      "learning_rate": 8.876549598017902e-07,
      "loss": 0.5555,
      "step": 18230
    },
    {
      "epoch": 1.63,
      "grad_norm": 10.768252537941144,
      "learning_rate": 8.87244110773337e-07,
      "loss": 0.6036,
      "step": 18231
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.927711266582771,
      "learning_rate": 8.868333475896562e-07,
      "loss": 0.5998,
      "step": 18232
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.3889397044285,
      "learning_rate": 8.864226702593221e-07,
      "loss": 0.5324,
      "step": 18233
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.516824043163474,
      "learning_rate": 8.860120787909054e-07,
      "loss": 0.5646,
      "step": 18234
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.159444074707801,
      "learning_rate": 8.85601573192979e-07,
      "loss": 0.5946,
      "step": 18235
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.050766597138478,
      "learning_rate": 8.851911534741075e-07,
      "loss": 0.5491,
      "step": 18236
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.330062601257174,
      "learning_rate": 8.847808196428598e-07,
      "loss": 0.6027,
      "step": 18237
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.45215967532486,
      "learning_rate": 8.843705717078005e-07,
      "loss": 0.5824,
      "step": 18238
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.742406344323576,
      "learning_rate": 8.839604096774918e-07,
      "loss": 0.5417,
      "step": 18239
    },
    {
      "epoch": 1.63,
      "grad_norm": 9.617684973221731,
      "learning_rate": 8.835503335604972e-07,
      "loss": 0.6004,
      "step": 18240
    },
    {
      "epoch": 1.63,
      "grad_norm": 9.355380586833276,
      "learning_rate": 8.831403433653735e-07,
      "loss": 0.6032,
      "step": 18241
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.985660832061222,
      "learning_rate": 8.827304391006791e-07,
      "loss": 0.6157,
      "step": 18242
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.875050018215632,
      "learning_rate": 8.823206207749702e-07,
      "loss": 0.5809,
      "step": 18243
    },
    {
      "epoch": 1.63,
      "grad_norm": 9.104459750617547,
      "learning_rate": 8.81910888396802e-07,
      "loss": 0.6527,
      "step": 18244
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.280077704248827,
      "learning_rate": 8.815012419747243e-07,
      "loss": 0.5491,
      "step": 18245
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.965060210899694,
      "learning_rate": 8.810916815172888e-07,
      "loss": 0.6062,
      "step": 18246
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.943564385547581,
      "learning_rate": 8.806822070330456e-07,
      "loss": 0.5667,
      "step": 18247
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.408258668217703,
      "learning_rate": 8.802728185305387e-07,
      "loss": 0.515,
      "step": 18248
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.423691706315809,
      "learning_rate": 8.798635160183145e-07,
      "loss": 0.6134,
      "step": 18249
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.431795502419906,
      "learning_rate": 8.794542995049166e-07,
      "loss": 0.567,
      "step": 18250
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.685827884063977,
      "learning_rate": 8.790451689988866e-07,
      "loss": 0.5943,
      "step": 18251
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.130129509872973,
      "learning_rate": 8.786361245087638e-07,
      "loss": 0.5533,
      "step": 18252
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.670509791079872,
      "learning_rate": 8.782271660430875e-07,
      "loss": 0.5534,
      "step": 18253
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.294371334691596,
      "learning_rate": 8.778182936103913e-07,
      "loss": 0.5361,
      "step": 18254
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.864417934365739,
      "learning_rate": 8.774095072192107e-07,
      "loss": 0.5456,
      "step": 18255
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.787434693839971,
      "learning_rate": 8.770008068780789e-07,
      "loss": 0.5451,
      "step": 18256
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.0494658661463685,
      "learning_rate": 8.765921925955256e-07,
      "loss": 0.5724,
      "step": 18257
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.315313633618835,
      "learning_rate": 8.761836643800814e-07,
      "loss": 0.504,
      "step": 18258
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.466149561554944,
      "learning_rate": 8.75775222240271e-07,
      "loss": 0.5926,
      "step": 18259
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.430287093721745,
      "learning_rate": 8.75366866184622e-07,
      "loss": 0.5838,
      "step": 18260
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.413733393376896,
      "learning_rate": 8.749585962216555e-07,
      "loss": 0.611,
      "step": 18261
    },
    {
      "epoch": 1.63,
      "grad_norm": 8.230063366444163,
      "learning_rate": 8.745504123598953e-07,
      "loss": 0.5847,
      "step": 18262
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.682211424666847,
      "learning_rate": 8.741423146078604e-07,
      "loss": 0.5727,
      "step": 18263
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.275878777534016,
      "learning_rate": 8.737343029740685e-07,
      "loss": 0.56,
      "step": 18264
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.115965077793975,
      "learning_rate": 8.733263774670375e-07,
      "loss": 0.5803,
      "step": 18265
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.358341814071191,
      "learning_rate": 8.72918538095282e-07,
      "loss": 0.5739,
      "step": 18266
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.28456975753307,
      "learning_rate": 8.725107848673126e-07,
      "loss": 0.5502,
      "step": 18267
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.686623893231531,
      "learning_rate": 8.721031177916417e-07,
      "loss": 0.5896,
      "step": 18268
    },
    {
      "epoch": 1.63,
      "grad_norm": 4.973361288364852,
      "learning_rate": 8.716955368767783e-07,
      "loss": 0.5404,
      "step": 18269
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.459976452132031,
      "learning_rate": 8.712880421312297e-07,
      "loss": 0.5348,
      "step": 18270
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.804935385000885,
      "learning_rate": 8.708806335635012e-07,
      "loss": 0.6035,
      "step": 18271
    },
    {
      "epoch": 1.63,
      "grad_norm": 9.144761355081727,
      "learning_rate": 8.704733111820985e-07,
      "loss": 0.5538,
      "step": 18272
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.552658767156932,
      "learning_rate": 8.70066074995522e-07,
      "loss": 0.5763,
      "step": 18273
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.439231166025338,
      "learning_rate": 8.6965892501227e-07,
      "loss": 0.5602,
      "step": 18274
    },
    {
      "epoch": 1.63,
      "grad_norm": 8.355927689980527,
      "learning_rate": 8.69251861240843e-07,
      "loss": 0.61,
      "step": 18275
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.6491556802078415,
      "learning_rate": 8.688448836897373e-07,
      "loss": 0.5976,
      "step": 18276
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.6985071160082565,
      "learning_rate": 8.684379923674474e-07,
      "loss": 0.6608,
      "step": 18277
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.910953575720706,
      "learning_rate": 8.680311872824665e-07,
      "loss": 0.5628,
      "step": 18278
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.26750072139701,
      "learning_rate": 8.676244684432866e-07,
      "loss": 0.5808,
      "step": 18279
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.95859683877297,
      "learning_rate": 8.672178358583955e-07,
      "loss": 0.5811,
      "step": 18280
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.447412100279043,
      "learning_rate": 8.668112895362812e-07,
      "loss": 0.6012,
      "step": 18281
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.9770528790982596,
      "learning_rate": 8.664048294854299e-07,
      "loss": 0.5857,
      "step": 18282
    },
    {
      "epoch": 1.63,
      "grad_norm": 8.00016604063552,
      "learning_rate": 8.659984557143248e-07,
      "loss": 0.6173,
      "step": 18283
    },
    {
      "epoch": 1.63,
      "grad_norm": 9.09395605703525,
      "learning_rate": 8.655921682314489e-07,
      "loss": 0.5814,
      "step": 18284
    },
    {
      "epoch": 1.63,
      "grad_norm": 9.654424856580553,
      "learning_rate": 8.651859670452839e-07,
      "loss": 0.6362,
      "step": 18285
    },
    {
      "epoch": 1.63,
      "grad_norm": 8.086065815868347,
      "learning_rate": 8.647798521643053e-07,
      "loss": 0.6027,
      "step": 18286
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.996278365937551,
      "learning_rate": 8.643738235969923e-07,
      "loss": 0.5732,
      "step": 18287
    },
    {
      "epoch": 1.63,
      "grad_norm": 10.683845737971797,
      "learning_rate": 8.639678813518176e-07,
      "loss": 0.6267,
      "step": 18288
    },
    {
      "epoch": 1.63,
      "grad_norm": 8.447386372645871,
      "learning_rate": 8.635620254372562e-07,
      "loss": 0.5446,
      "step": 18289
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.609530180147943,
      "learning_rate": 8.631562558617784e-07,
      "loss": 0.6245,
      "step": 18290
    },
    {
      "epoch": 1.63,
      "grad_norm": 4.91763858528652,
      "learning_rate": 8.627505726338542e-07,
      "loss": 0.5398,
      "step": 18291
    },
    {
      "epoch": 1.63,
      "grad_norm": 9.5815907877699,
      "learning_rate": 8.623449757619529e-07,
      "loss": 0.6624,
      "step": 18292
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.270017800213312,
      "learning_rate": 8.619394652545376e-07,
      "loss": 0.5495,
      "step": 18293
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.991917250804576,
      "learning_rate": 8.615340411200734e-07,
      "loss": 0.5894,
      "step": 18294
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.465456364022198,
      "learning_rate": 8.611287033670234e-07,
      "loss": 0.5374,
      "step": 18295
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.387797504034625,
      "learning_rate": 8.607234520038477e-07,
      "loss": 0.5616,
      "step": 18296
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.996561667529507,
      "learning_rate": 8.603182870390048e-07,
      "loss": 0.5509,
      "step": 18297
    },
    {
      "epoch": 1.63,
      "grad_norm": 8.091799103368901,
      "learning_rate": 8.599132084809536e-07,
      "loss": 0.6327,
      "step": 18298
    },
    {
      "epoch": 1.63,
      "grad_norm": 11.983132548471609,
      "learning_rate": 8.595082163381458e-07,
      "loss": 0.56,
      "step": 18299
    },
    {
      "epoch": 1.63,
      "grad_norm": 9.926060371433811,
      "learning_rate": 8.591033106190372e-07,
      "loss": 0.5755,
      "step": 18300
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.970776273723415,
      "learning_rate": 8.586984913320778e-07,
      "loss": 0.6005,
      "step": 18301
    },
    {
      "epoch": 1.63,
      "grad_norm": 4.68789766673741,
      "learning_rate": 8.582937584857198e-07,
      "loss": 0.5841,
      "step": 18302
    },
    {
      "epoch": 1.63,
      "grad_norm": 4.882176588253618,
      "learning_rate": 8.578891120884081e-07,
      "loss": 0.5497,
      "step": 18303
    },
    {
      "epoch": 1.63,
      "grad_norm": 8.244517265503795,
      "learning_rate": 8.574845521485897e-07,
      "loss": 0.5553,
      "step": 18304
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.54278847806572,
      "learning_rate": 8.570800786747107e-07,
      "loss": 0.5874,
      "step": 18305
    },
    {
      "epoch": 1.63,
      "grad_norm": 8.557764684020256,
      "learning_rate": 8.566756916752117e-07,
      "loss": 0.6201,
      "step": 18306
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.631090945695476,
      "learning_rate": 8.56271391158533e-07,
      "loss": 0.57,
      "step": 18307
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.631327422169233,
      "learning_rate": 8.55867177133115e-07,
      "loss": 0.5737,
      "step": 18308
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.265102732421731,
      "learning_rate": 8.55463049607394e-07,
      "loss": 0.5393,
      "step": 18309
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.67828985027148,
      "learning_rate": 8.550590085898053e-07,
      "loss": 0.5777,
      "step": 18310
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.504979533999024,
      "learning_rate": 8.546550540887838e-07,
      "loss": 0.5829,
      "step": 18311
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.584739560634968,
      "learning_rate": 8.542511861127589e-07,
      "loss": 0.6253,
      "step": 18312
    },
    {
      "epoch": 1.63,
      "grad_norm": 8.037830336119118,
      "learning_rate": 8.538474046701611e-07,
      "loss": 0.5596,
      "step": 18313
    },
    {
      "epoch": 1.63,
      "grad_norm": 11.11895971433573,
      "learning_rate": 8.534437097694192e-07,
      "loss": 0.5497,
      "step": 18314
    },
    {
      "epoch": 1.63,
      "grad_norm": 4.26677665529349,
      "learning_rate": 8.530401014189588e-07,
      "loss": 0.5064,
      "step": 18315
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.102938683446105,
      "learning_rate": 8.526365796272057e-07,
      "loss": 0.5786,
      "step": 18316
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.323490014530909,
      "learning_rate": 8.522331444025805e-07,
      "loss": 0.5206,
      "step": 18317
    },
    {
      "epoch": 1.63,
      "grad_norm": 8.945711117075765,
      "learning_rate": 8.51829795753506e-07,
      "loss": 0.6376,
      "step": 18318
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.805545505385139,
      "learning_rate": 8.514265336883993e-07,
      "loss": 0.5946,
      "step": 18319
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.115220529627622,
      "learning_rate": 8.510233582156779e-07,
      "loss": 0.5462,
      "step": 18320
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.6669012259309595,
      "learning_rate": 8.506202693437582e-07,
      "loss": 0.5461,
      "step": 18321
    },
    {
      "epoch": 1.63,
      "grad_norm": 8.527211049594577,
      "learning_rate": 8.502172670810532e-07,
      "loss": 0.6009,
      "step": 18322
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.380680942774459,
      "learning_rate": 8.498143514359764e-07,
      "loss": 0.5918,
      "step": 18323
    },
    {
      "epoch": 1.63,
      "grad_norm": 4.948377079877318,
      "learning_rate": 8.494115224169347e-07,
      "loss": 0.6099,
      "step": 18324
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.4111374338024145,
      "learning_rate": 8.490087800323376e-07,
      "loss": 0.6559,
      "step": 18325
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.000457531444563,
      "learning_rate": 8.486061242905924e-07,
      "loss": 0.5705,
      "step": 18326
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.980226250514791,
      "learning_rate": 8.482035552001028e-07,
      "loss": 0.5882,
      "step": 18327
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.99171010435227,
      "learning_rate": 8.478010727692714e-07,
      "loss": 0.5775,
      "step": 18328
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.892311478463849,
      "learning_rate": 8.473986770065007e-07,
      "loss": 0.5754,
      "step": 18329
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.643841575072553,
      "learning_rate": 8.469963679201887e-07,
      "loss": 0.6252,
      "step": 18330
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.424929581527143,
      "learning_rate": 8.465941455187316e-07,
      "loss": 0.5828,
      "step": 18331
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.208605091484272,
      "learning_rate": 8.461920098105253e-07,
      "loss": 0.5443,
      "step": 18332
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.916766019255351,
      "learning_rate": 8.457899608039644e-07,
      "loss": 0.5783,
      "step": 18333
    },
    {
      "epoch": 1.64,
      "grad_norm": 10.508676987467702,
      "learning_rate": 8.453879985074409e-07,
      "loss": 0.5832,
      "step": 18334
    },
    {
      "epoch": 1.64,
      "grad_norm": 8.693719207924325,
      "learning_rate": 8.449861229293438e-07,
      "loss": 0.562,
      "step": 18335
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.797911418691826,
      "learning_rate": 8.445843340780641e-07,
      "loss": 0.5117,
      "step": 18336
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.7738113844680665,
      "learning_rate": 8.441826319619844e-07,
      "loss": 0.4941,
      "step": 18337
    },
    {
      "epoch": 1.64,
      "grad_norm": 8.391707892142685,
      "learning_rate": 8.43781016589491e-07,
      "loss": 0.57,
      "step": 18338
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.301899553051192,
      "learning_rate": 8.433794879689672e-07,
      "loss": 0.5887,
      "step": 18339
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.102860058531695,
      "learning_rate": 8.429780461087939e-07,
      "loss": 0.6253,
      "step": 18340
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.100805384983346,
      "learning_rate": 8.425766910173499e-07,
      "loss": 0.5561,
      "step": 18341
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.9389241001993,
      "learning_rate": 8.421754227030143e-07,
      "loss": 0.5569,
      "step": 18342
    },
    {
      "epoch": 1.64,
      "grad_norm": 8.768141404192999,
      "learning_rate": 8.417742411741603e-07,
      "loss": 0.5864,
      "step": 18343
    },
    {
      "epoch": 1.64,
      "grad_norm": 8.90398598080573,
      "learning_rate": 8.413731464391622e-07,
      "loss": 0.5808,
      "step": 18344
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.083274152815814,
      "learning_rate": 8.409721385063935e-07,
      "loss": 0.5402,
      "step": 18345
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.7148916109339485,
      "learning_rate": 8.405712173842229e-07,
      "loss": 0.5446,
      "step": 18346
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.253897270564336,
      "learning_rate": 8.401703830810182e-07,
      "loss": 0.5212,
      "step": 18347
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.810033767274006,
      "learning_rate": 8.397696356051471e-07,
      "loss": 0.5609,
      "step": 18348
    },
    {
      "epoch": 1.64,
      "grad_norm": 9.323178434173952,
      "learning_rate": 8.393689749649758e-07,
      "loss": 0.611,
      "step": 18349
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.9527752556389775,
      "learning_rate": 8.389684011688642e-07,
      "loss": 0.5718,
      "step": 18350
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.955998385340085,
      "learning_rate": 8.385679142251746e-07,
      "loss": 0.5334,
      "step": 18351
    },
    {
      "epoch": 1.64,
      "grad_norm": 8.080138389775241,
      "learning_rate": 8.381675141422663e-07,
      "loss": 0.5473,
      "step": 18352
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.914079198764076,
      "learning_rate": 8.377672009284965e-07,
      "loss": 0.5867,
      "step": 18353
    },
    {
      "epoch": 1.64,
      "grad_norm": 11.82875431786201,
      "learning_rate": 8.373669745922219e-07,
      "loss": 0.6017,
      "step": 18354
    },
    {
      "epoch": 1.64,
      "grad_norm": 8.648855566181956,
      "learning_rate": 8.369668351417965e-07,
      "loss": 0.5278,
      "step": 18355
    },
    {
      "epoch": 1.64,
      "grad_norm": 8.665460835922133,
      "learning_rate": 8.365667825855706e-07,
      "loss": 0.5637,
      "step": 18356
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.53132647307799,
      "learning_rate": 8.361668169318954e-07,
      "loss": 0.5692,
      "step": 18357
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.70737034368289,
      "learning_rate": 8.357669381891192e-07,
      "loss": 0.5326,
      "step": 18358
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.141488774714178,
      "learning_rate": 8.353671463655894e-07,
      "loss": 0.5387,
      "step": 18359
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.567339501490505,
      "learning_rate": 8.349674414696496e-07,
      "loss": 0.6217,
      "step": 18360
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.757725841017735,
      "learning_rate": 8.345678235096428e-07,
      "loss": 0.5118,
      "step": 18361
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.937835327762397,
      "learning_rate": 8.341682924939115e-07,
      "loss": 0.6571,
      "step": 18362
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.117620642865933,
      "learning_rate": 8.337688484307932e-07,
      "loss": 0.5836,
      "step": 18363
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.398479725230202,
      "learning_rate": 8.333694913286261e-07,
      "loss": 0.4731,
      "step": 18364
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.260003169560137,
      "learning_rate": 8.329702211957463e-07,
      "loss": 0.5406,
      "step": 18365
    },
    {
      "epoch": 1.64,
      "grad_norm": 8.253706875146898,
      "learning_rate": 8.325710380404872e-07,
      "loss": 0.5858,
      "step": 18366
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.797217830115273,
      "learning_rate": 8.321719418711816e-07,
      "loss": 0.6039,
      "step": 18367
    },
    {
      "epoch": 1.64,
      "grad_norm": 8.769105305474817,
      "learning_rate": 8.317729326961598e-07,
      "loss": 0.6089,
      "step": 18368
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.077064818402232,
      "learning_rate": 8.313740105237494e-07,
      "loss": 0.5426,
      "step": 18369
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.480427270604181,
      "learning_rate": 8.309751753622769e-07,
      "loss": 0.5913,
      "step": 18370
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.386080714696423,
      "learning_rate": 8.305764272200678e-07,
      "loss": 0.5179,
      "step": 18371
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.8064521341152195,
      "learning_rate": 8.301777661054449e-07,
      "loss": 0.6527,
      "step": 18372
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.334846924589883,
      "learning_rate": 8.297791920267295e-07,
      "loss": 0.607,
      "step": 18373
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.071571096892972,
      "learning_rate": 8.29380704992242e-07,
      "loss": 0.5353,
      "step": 18374
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.114224187000775,
      "learning_rate": 8.289823050102985e-07,
      "loss": 0.5656,
      "step": 18375
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.945071232802345,
      "learning_rate": 8.285839920892141e-07,
      "loss": 0.5793,
      "step": 18376
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.699113478821209,
      "learning_rate": 8.281857662373033e-07,
      "loss": 0.6275,
      "step": 18377
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.743398708774488,
      "learning_rate": 8.277876274628788e-07,
      "loss": 0.5668,
      "step": 18378
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.375104460711118,
      "learning_rate": 8.273895757742506e-07,
      "loss": 0.5446,
      "step": 18379
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.531842992497854,
      "learning_rate": 8.269916111797266e-07,
      "loss": 0.5505,
      "step": 18380
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.649366647491276,
      "learning_rate": 8.265937336876156e-07,
      "loss": 0.5271,
      "step": 18381
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.109736115688753,
      "learning_rate": 8.2619594330622e-07,
      "loss": 0.5538,
      "step": 18382
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.294945830809467,
      "learning_rate": 8.25798240043843e-07,
      "loss": 0.5528,
      "step": 18383
    },
    {
      "epoch": 1.64,
      "grad_norm": 9.259944562229338,
      "learning_rate": 8.254006239087864e-07,
      "loss": 0.5818,
      "step": 18384
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.394268344918695,
      "learning_rate": 8.250030949093502e-07,
      "loss": 0.5325,
      "step": 18385
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.868022885956684,
      "learning_rate": 8.246056530538304e-07,
      "loss": 0.5926,
      "step": 18386
    },
    {
      "epoch": 1.64,
      "grad_norm": 8.239883290625055,
      "learning_rate": 8.242082983505256e-07,
      "loss": 0.6212,
      "step": 18387
    },
    {
      "epoch": 1.64,
      "grad_norm": 8.042459110384563,
      "learning_rate": 8.238110308077274e-07,
      "loss": 0.6344,
      "step": 18388
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.876540589125333,
      "learning_rate": 8.234138504337274e-07,
      "loss": 0.589,
      "step": 18389
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.044362738141847,
      "learning_rate": 8.230167572368164e-07,
      "loss": 0.5972,
      "step": 18390
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.965982345219756,
      "learning_rate": 8.226197512252832e-07,
      "loss": 0.5222,
      "step": 18391
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.384640688243956,
      "learning_rate": 8.222228324074144e-07,
      "loss": 0.6227,
      "step": 18392
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.78366962350662,
      "learning_rate": 8.218260007914947e-07,
      "loss": 0.5409,
      "step": 18393
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.657027019780211,
      "learning_rate": 8.214292563858084e-07,
      "loss": 0.544,
      "step": 18394
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.180439904615951,
      "learning_rate": 8.21032599198634e-07,
      "loss": 0.5869,
      "step": 18395
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.494839270125743,
      "learning_rate": 8.206360292382525e-07,
      "loss": 0.555,
      "step": 18396
    },
    {
      "epoch": 1.64,
      "grad_norm": 8.421076162304619,
      "learning_rate": 8.202395465129414e-07,
      "loss": 0.5447,
      "step": 18397
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.59237618498641,
      "learning_rate": 8.198431510309762e-07,
      "loss": 0.6146,
      "step": 18398
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.190719618868607,
      "learning_rate": 8.194468428006319e-07,
      "loss": 0.5597,
      "step": 18399
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.509380054737271,
      "learning_rate": 8.190506218301786e-07,
      "loss": 0.5445,
      "step": 18400
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.623266655006669,
      "learning_rate": 8.186544881278874e-07,
      "loss": 0.5398,
      "step": 18401
    },
    {
      "epoch": 1.64,
      "grad_norm": 8.033686402570199,
      "learning_rate": 8.182584417020267e-07,
      "loss": 0.605,
      "step": 18402
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.68802285528086,
      "learning_rate": 8.178624825608645e-07,
      "loss": 0.5394,
      "step": 18403
    },
    {
      "epoch": 1.64,
      "grad_norm": 8.398796979763894,
      "learning_rate": 8.174666107126627e-07,
      "loss": 0.5697,
      "step": 18404
    },
    {
      "epoch": 1.64,
      "grad_norm": 9.312847106730391,
      "learning_rate": 8.170708261656857e-07,
      "loss": 0.6253,
      "step": 18405
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.987843709007214,
      "learning_rate": 8.166751289281965e-07,
      "loss": 0.619,
      "step": 18406
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.57136475148598,
      "learning_rate": 8.16279519008451e-07,
      "loss": 0.5493,
      "step": 18407
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.324105977867103,
      "learning_rate": 8.158839964147086e-07,
      "loss": 0.6139,
      "step": 18408
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.60270649238533,
      "learning_rate": 8.154885611552244e-07,
      "loss": 0.5928,
      "step": 18409
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.22559608870969,
      "learning_rate": 8.150932132382528e-07,
      "loss": 0.583,
      "step": 18410
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.44855956952106,
      "learning_rate": 8.146979526720456e-07,
      "loss": 0.5986,
      "step": 18411
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.9899424013144635,
      "learning_rate": 8.143027794648544e-07,
      "loss": 0.5674,
      "step": 18412
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.154382386812026,
      "learning_rate": 8.139076936249246e-07,
      "loss": 0.5701,
      "step": 18413
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.030091264898395,
      "learning_rate": 8.135126951605044e-07,
      "loss": 0.5131,
      "step": 18414
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.780913175496152,
      "learning_rate": 8.131177840798388e-07,
      "loss": 0.5146,
      "step": 18415
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.413298266505573,
      "learning_rate": 8.127229603911696e-07,
      "loss": 0.5451,
      "step": 18416
    },
    {
      "epoch": 1.64,
      "grad_norm": 8.690138379896133,
      "learning_rate": 8.123282241027403e-07,
      "loss": 0.5163,
      "step": 18417
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.54807653231746,
      "learning_rate": 8.119335752227875e-07,
      "loss": 0.5594,
      "step": 18418
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.1192575435019,
      "learning_rate": 8.115390137595503e-07,
      "loss": 0.5914,
      "step": 18419
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.96023587962027,
      "learning_rate": 8.111445397212625e-07,
      "loss": 0.5595,
      "step": 18420
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.33857444804644,
      "learning_rate": 8.107501531161593e-07,
      "loss": 0.6191,
      "step": 18421
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.4072587021959295,
      "learning_rate": 8.10355853952472e-07,
      "loss": 0.513,
      "step": 18422
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.959006438093087,
      "learning_rate": 8.099616422384316e-07,
      "loss": 0.547,
      "step": 18423
    },
    {
      "epoch": 1.64,
      "grad_norm": 9.481280122107533,
      "learning_rate": 8.095675179822654e-07,
      "loss": 0.543,
      "step": 18424
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.57089957219226,
      "learning_rate": 8.09173481192202e-07,
      "loss": 0.5833,
      "step": 18425
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.719521127469208,
      "learning_rate": 8.08779531876463e-07,
      "loss": 0.6178,
      "step": 18426
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.755967958378412,
      "learning_rate": 8.083856700432724e-07,
      "loss": 0.5795,
      "step": 18427
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.742435810393466,
      "learning_rate": 8.079918957008514e-07,
      "loss": 0.5873,
      "step": 18428
    },
    {
      "epoch": 1.64,
      "grad_norm": 4.810896209349775,
      "learning_rate": 8.075982088574197e-07,
      "loss": 0.6162,
      "step": 18429
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.047397681986977,
      "learning_rate": 8.072046095211938e-07,
      "loss": 0.6128,
      "step": 18430
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.753924884547794,
      "learning_rate": 8.068110977003912e-07,
      "loss": 0.6073,
      "step": 18431
    },
    {
      "epoch": 1.64,
      "grad_norm": 10.163625071799478,
      "learning_rate": 8.064176734032236e-07,
      "loss": 0.6688,
      "step": 18432
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.011963361798186,
      "learning_rate": 8.060243366379028e-07,
      "loss": 0.5849,
      "step": 18433
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.358761392769661,
      "learning_rate": 8.056310874126383e-07,
      "loss": 0.5552,
      "step": 18434
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.193508048280206,
      "learning_rate": 8.052379257356401e-07,
      "loss": 0.5183,
      "step": 18435
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.317180704207342,
      "learning_rate": 8.048448516151136e-07,
      "loss": 0.5916,
      "step": 18436
    },
    {
      "epoch": 1.64,
      "grad_norm": 9.623279080583254,
      "learning_rate": 8.044518650592637e-07,
      "loss": 0.5934,
      "step": 18437
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.5029563716056185,
      "learning_rate": 8.040589660762937e-07,
      "loss": 0.6271,
      "step": 18438
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.120324634867505,
      "learning_rate": 8.036661546744035e-07,
      "loss": 0.6195,
      "step": 18439
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.198340049825372,
      "learning_rate": 8.032734308617917e-07,
      "loss": 0.6135,
      "step": 18440
    },
    {
      "epoch": 1.65,
      "grad_norm": 5.999704104701878,
      "learning_rate": 8.028807946466566e-07,
      "loss": 0.5485,
      "step": 18441
    },
    {
      "epoch": 1.65,
      "grad_norm": 10.640950897463302,
      "learning_rate": 8.024882460371935e-07,
      "loss": 0.6243,
      "step": 18442
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.474663814946242,
      "learning_rate": 8.020957850415962e-07,
      "loss": 0.5873,
      "step": 18443
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.981735709323859,
      "learning_rate": 8.017034116680567e-07,
      "loss": 0.5842,
      "step": 18444
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.0281706129418025,
      "learning_rate": 8.013111259247635e-07,
      "loss": 0.5734,
      "step": 18445
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.315713533005397,
      "learning_rate": 8.00918927819907e-07,
      "loss": 0.6075,
      "step": 18446
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.490775916356115,
      "learning_rate": 8.005268173616703e-07,
      "loss": 0.5974,
      "step": 18447
    },
    {
      "epoch": 1.65,
      "grad_norm": 5.1232602399488405,
      "learning_rate": 8.001347945582399e-07,
      "loss": 0.5679,
      "step": 18448
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.39486736688238,
      "learning_rate": 7.99742859417798e-07,
      "loss": 0.575,
      "step": 18449
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.411909072468878,
      "learning_rate": 7.993510119485254e-07,
      "loss": 0.5741,
      "step": 18450
    },
    {
      "epoch": 1.65,
      "grad_norm": 9.133180080065307,
      "learning_rate": 7.989592521586026e-07,
      "loss": 0.5506,
      "step": 18451
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.603410467080666,
      "learning_rate": 7.985675800562042e-07,
      "loss": 0.5638,
      "step": 18452
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.179075120207025,
      "learning_rate": 7.981759956495067e-07,
      "loss": 0.5614,
      "step": 18453
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.1022844683869915,
      "learning_rate": 7.977844989466827e-07,
      "loss": 0.5352,
      "step": 18454
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.703531469440243,
      "learning_rate": 7.973930899559052e-07,
      "loss": 0.5491,
      "step": 18455
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.452095987625953,
      "learning_rate": 7.970017686853432e-07,
      "loss": 0.5444,
      "step": 18456
    },
    {
      "epoch": 1.65,
      "grad_norm": 11.929247271004897,
      "learning_rate": 7.966105351431664e-07,
      "loss": 0.5453,
      "step": 18457
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.9246923717523075,
      "learning_rate": 7.962193893375375e-07,
      "loss": 0.6027,
      "step": 18458
    },
    {
      "epoch": 1.65,
      "grad_norm": 5.593179876701987,
      "learning_rate": 7.958283312766235e-07,
      "loss": 0.5647,
      "step": 18459
    },
    {
      "epoch": 1.65,
      "grad_norm": 8.234437584685127,
      "learning_rate": 7.954373609685867e-07,
      "loss": 0.6105,
      "step": 18460
    },
    {
      "epoch": 1.65,
      "grad_norm": 5.196807662606069,
      "learning_rate": 7.950464784215861e-07,
      "loss": 0.5822,
      "step": 18461
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.915972218287418,
      "learning_rate": 7.946556836437818e-07,
      "loss": 0.6111,
      "step": 18462
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.461879084297356,
      "learning_rate": 7.942649766433308e-07,
      "loss": 0.608,
      "step": 18463
    },
    {
      "epoch": 1.65,
      "grad_norm": 10.332750622298514,
      "learning_rate": 7.938743574283886e-07,
      "loss": 0.6431,
      "step": 18464
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.184942410920042,
      "learning_rate": 7.93483826007107e-07,
      "loss": 0.5329,
      "step": 18465
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.414006023388483,
      "learning_rate": 7.930933823876386e-07,
      "loss": 0.5826,
      "step": 18466
    },
    {
      "epoch": 1.65,
      "grad_norm": 5.189780697896888,
      "learning_rate": 7.927030265781327e-07,
      "loss": 0.6814,
      "step": 18467
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.006386787402832,
      "learning_rate": 7.923127585867374e-07,
      "loss": 0.5934,
      "step": 18468
    },
    {
      "epoch": 1.65,
      "grad_norm": 9.238356359596235,
      "learning_rate": 7.919225784215984e-07,
      "loss": 0.5701,
      "step": 18469
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.326573433735392,
      "learning_rate": 7.915324860908613e-07,
      "loss": 0.5945,
      "step": 18470
    },
    {
      "epoch": 1.65,
      "grad_norm": 5.872185336640947,
      "learning_rate": 7.911424816026658e-07,
      "loss": 0.5914,
      "step": 18471
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.775453461371875,
      "learning_rate": 7.907525649651537e-07,
      "loss": 0.631,
      "step": 18472
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.267934552548118,
      "learning_rate": 7.90362736186464e-07,
      "loss": 0.5714,
      "step": 18473
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.145931557851759,
      "learning_rate": 7.899729952747332e-07,
      "loss": 0.6003,
      "step": 18474
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.703159097242164,
      "learning_rate": 7.895833422380977e-07,
      "loss": 0.6376,
      "step": 18475
    },
    {
      "epoch": 1.65,
      "grad_norm": 4.8372389857358975,
      "learning_rate": 7.89193777084688e-07,
      "loss": 0.5405,
      "step": 18476
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.081276596007639,
      "learning_rate": 7.888042998226376e-07,
      "loss": 0.5466,
      "step": 18477
    },
    {
      "epoch": 1.65,
      "grad_norm": 9.762203965758758,
      "learning_rate": 7.884149104600741e-07,
      "loss": 0.51,
      "step": 18478
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.113591453398985,
      "learning_rate": 7.880256090051258e-07,
      "loss": 0.5533,
      "step": 18479
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.866316368024405,
      "learning_rate": 7.876363954659194e-07,
      "loss": 0.5866,
      "step": 18480
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.0532276095778625,
      "learning_rate": 7.872472698505779e-07,
      "loss": 0.5627,
      "step": 18481
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.088422148583732,
      "learning_rate": 7.868582321672252e-07,
      "loss": 0.6264,
      "step": 18482
    },
    {
      "epoch": 1.65,
      "grad_norm": 10.004520893253094,
      "learning_rate": 7.864692824239794e-07,
      "loss": 0.621,
      "step": 18483
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.291895097514731,
      "learning_rate": 7.860804206289597e-07,
      "loss": 0.5691,
      "step": 18484
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.59977913718888,
      "learning_rate": 7.856916467902825e-07,
      "loss": 0.5368,
      "step": 18485
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.252789079921171,
      "learning_rate": 7.853029609160634e-07,
      "loss": 0.5964,
      "step": 18486
    },
    {
      "epoch": 1.65,
      "grad_norm": 8.68925027825286,
      "learning_rate": 7.849143630144146e-07,
      "loss": 0.5587,
      "step": 18487
    },
    {
      "epoch": 1.65,
      "grad_norm": 5.432049502858594,
      "learning_rate": 7.845258530934491e-07,
      "loss": 0.5983,
      "step": 18488
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.4365609766453495,
      "learning_rate": 7.841374311612731e-07,
      "loss": 0.5912,
      "step": 18489
    },
    {
      "epoch": 1.65,
      "grad_norm": 5.6075287678113295,
      "learning_rate": 7.83749097225997e-07,
      "loss": 0.5826,
      "step": 18490
    },
    {
      "epoch": 1.65,
      "grad_norm": 8.467804311962306,
      "learning_rate": 7.833608512957241e-07,
      "loss": 0.6157,
      "step": 18491
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.249071809458059,
      "learning_rate": 7.829726933785591e-07,
      "loss": 0.6396,
      "step": 18492
    },
    {
      "epoch": 1.65,
      "grad_norm": 9.02085045649851,
      "learning_rate": 7.82584623482604e-07,
      "loss": 0.6119,
      "step": 18493
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.772931637112214,
      "learning_rate": 7.821966416159588e-07,
      "loss": 0.5112,
      "step": 18494
    },
    {
      "epoch": 1.65,
      "grad_norm": 5.560831521670898,
      "learning_rate": 7.818087477867231e-07,
      "loss": 0.5692,
      "step": 18495
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.90167753048076,
      "learning_rate": 7.814209420029911e-07,
      "loss": 0.521,
      "step": 18496
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.830930956939064,
      "learning_rate": 7.81033224272858e-07,
      "loss": 0.5248,
      "step": 18497
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.896053930836321,
      "learning_rate": 7.806455946044173e-07,
      "loss": 0.5823,
      "step": 18498
    },
    {
      "epoch": 1.65,
      "grad_norm": 4.502397172353447,
      "learning_rate": 7.802580530057597e-07,
      "loss": 0.5971,
      "step": 18499
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.210820175915279,
      "learning_rate": 7.798705994849737e-07,
      "loss": 0.5363,
      "step": 18500
    },
    {
      "epoch": 1.65,
      "grad_norm": 5.413295166889457,
      "learning_rate": 7.794832340501484e-07,
      "loss": 0.5768,
      "step": 18501
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.921624855741802,
      "learning_rate": 7.790959567093664e-07,
      "loss": 0.5316,
      "step": 18502
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.261767629419919,
      "learning_rate": 7.787087674707134e-07,
      "loss": 0.5857,
      "step": 18503
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.702468537812182,
      "learning_rate": 7.78321666342271e-07,
      "loss": 0.638,
      "step": 18504
    },
    {
      "epoch": 1.65,
      "grad_norm": 10.504380768477803,
      "learning_rate": 7.779346533321175e-07,
      "loss": 0.618,
      "step": 18505
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.413499594140099,
      "learning_rate": 7.775477284483318e-07,
      "loss": 0.6134,
      "step": 18506
    },
    {
      "epoch": 1.65,
      "grad_norm": 8.344582191540109,
      "learning_rate": 7.771608916989904e-07,
      "loss": 0.5579,
      "step": 18507
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.726693913995241,
      "learning_rate": 7.767741430921688e-07,
      "loss": 0.5534,
      "step": 18508
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.723186845811823,
      "learning_rate": 7.763874826359369e-07,
      "loss": 0.5765,
      "step": 18509
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.870819049650064,
      "learning_rate": 7.760009103383665e-07,
      "loss": 0.5589,
      "step": 18510
    },
    {
      "epoch": 1.65,
      "grad_norm": 5.467456378735843,
      "learning_rate": 7.756144262075272e-07,
      "loss": 0.5985,
      "step": 18511
    },
    {
      "epoch": 1.65,
      "grad_norm": 9.622277009121818,
      "learning_rate": 7.752280302514853e-07,
      "loss": 0.6046,
      "step": 18512
    },
    {
      "epoch": 1.65,
      "grad_norm": 5.941247032868286,
      "learning_rate": 7.748417224783062e-07,
      "loss": 0.5295,
      "step": 18513
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.132259224279332,
      "learning_rate": 7.744555028960543e-07,
      "loss": 0.5811,
      "step": 18514
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.834414961003755,
      "learning_rate": 7.74069371512789e-07,
      "loss": 0.5906,
      "step": 18515
    },
    {
      "epoch": 1.65,
      "grad_norm": 5.463538861285599,
      "learning_rate": 7.736833283365702e-07,
      "loss": 0.5498,
      "step": 18516
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.902282305239494,
      "learning_rate": 7.732973733754573e-07,
      "loss": 0.5644,
      "step": 18517
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.2951555160481245,
      "learning_rate": 7.729115066375065e-07,
      "loss": 0.6084,
      "step": 18518
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.755316027174562,
      "learning_rate": 7.725257281307691e-07,
      "loss": 0.5538,
      "step": 18519
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.674924485166208,
      "learning_rate": 7.721400378632999e-07,
      "loss": 0.6013,
      "step": 18520
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.69693025873178,
      "learning_rate": 7.717544358431489e-07,
      "loss": 0.5791,
      "step": 18521
    },
    {
      "epoch": 1.65,
      "grad_norm": 8.398408769112057,
      "learning_rate": 7.713689220783637e-07,
      "loss": 0.5703,
      "step": 18522
    },
    {
      "epoch": 1.65,
      "grad_norm": 5.401809854124345,
      "learning_rate": 7.709834965769919e-07,
      "loss": 0.5405,
      "step": 18523
    },
    {
      "epoch": 1.65,
      "grad_norm": 5.861215483641893,
      "learning_rate": 7.70598159347078e-07,
      "loss": 0.5569,
      "step": 18524
    },
    {
      "epoch": 1.65,
      "grad_norm": 8.172857486881899,
      "learning_rate": 7.702129103966649e-07,
      "loss": 0.5503,
      "step": 18525
    },
    {
      "epoch": 1.65,
      "grad_norm": 5.072229724791138,
      "learning_rate": 7.698277497337947e-07,
      "loss": 0.5852,
      "step": 18526
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.615578688952974,
      "learning_rate": 7.694426773665076e-07,
      "loss": 0.4889,
      "step": 18527
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.8136153243673165,
      "learning_rate": 7.690576933028382e-07,
      "loss": 0.5178,
      "step": 18528
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.376655723428289,
      "learning_rate": 7.68672797550824e-07,
      "loss": 0.5437,
      "step": 18529
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.667955408410495,
      "learning_rate": 7.682879901184986e-07,
      "loss": 0.555,
      "step": 18530
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.960045254437673,
      "learning_rate": 7.679032710138945e-07,
      "loss": 0.554,
      "step": 18531
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.4586732033704,
      "learning_rate": 7.675186402450424e-07,
      "loss": 0.5475,
      "step": 18532
    },
    {
      "epoch": 1.65,
      "grad_norm": 8.425712057531369,
      "learning_rate": 7.671340978199682e-07,
      "loss": 0.5436,
      "step": 18533
    },
    {
      "epoch": 1.65,
      "grad_norm": 8.520293919042784,
      "learning_rate": 7.667496437467015e-07,
      "loss": 0.5539,
      "step": 18534
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.271113254641803,
      "learning_rate": 7.663652780332642e-07,
      "loss": 0.5848,
      "step": 18535
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.7972004587419494,
      "learning_rate": 7.6598100068768e-07,
      "loss": 0.5975,
      "step": 18536
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.722343714256844,
      "learning_rate": 7.655968117179702e-07,
      "loss": 0.5459,
      "step": 18537
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.794889242191422,
      "learning_rate": 7.652127111321539e-07,
      "loss": 0.5832,
      "step": 18538
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.309449431028304,
      "learning_rate": 7.648286989382481e-07,
      "loss": 0.5542,
      "step": 18539
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.712729513139768,
      "learning_rate": 7.64444775144269e-07,
      "loss": 0.5926,
      "step": 18540
    },
    {
      "epoch": 1.65,
      "grad_norm": 9.881349408506232,
      "learning_rate": 7.640609397582294e-07,
      "loss": 0.5904,
      "step": 18541
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.353552072120926,
      "learning_rate": 7.636771927881404e-07,
      "loss": 0.5574,
      "step": 18542
    },
    {
      "epoch": 1.65,
      "grad_norm": 5.1315001725763425,
      "learning_rate": 7.632935342420128e-07,
      "loss": 0.5645,
      "step": 18543
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.619942365285354,
      "learning_rate": 7.629099641278542e-07,
      "loss": 0.5992,
      "step": 18544
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.279536057247027,
      "learning_rate": 7.62526482453671e-07,
      "loss": 0.5587,
      "step": 18545
    },
    {
      "epoch": 1.65,
      "grad_norm": 5.677882106829552,
      "learning_rate": 7.621430892274689e-07,
      "loss": 0.5187,
      "step": 18546
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.431041271984883,
      "learning_rate": 7.617597844572494e-07,
      "loss": 0.6119,
      "step": 18547
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.42173809337769,
      "learning_rate": 7.613765681510116e-07,
      "loss": 0.5893,
      "step": 18548
    },
    {
      "epoch": 1.65,
      "grad_norm": 8.215803908928306,
      "learning_rate": 7.609934403167552e-07,
      "loss": 0.5892,
      "step": 18549
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.592797953327762,
      "learning_rate": 7.606104009624776e-07,
      "loss": 0.5404,
      "step": 18550
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.277501009061264,
      "learning_rate": 7.602274500961737e-07,
      "loss": 0.5549,
      "step": 18551
    },
    {
      "epoch": 1.66,
      "grad_norm": 7.343844922343488,
      "learning_rate": 7.598445877258365e-07,
      "loss": 0.5004,
      "step": 18552
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.105406915340262,
      "learning_rate": 7.594618138594595e-07,
      "loss": 0.5705,
      "step": 18553
    },
    {
      "epoch": 1.66,
      "grad_norm": 8.647893967063476,
      "learning_rate": 7.590791285050292e-07,
      "loss": 0.5926,
      "step": 18554
    },
    {
      "epoch": 1.66,
      "grad_norm": 7.6957850584052645,
      "learning_rate": 7.586965316705347e-07,
      "loss": 0.5453,
      "step": 18555
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.636153183401164,
      "learning_rate": 7.583140233639613e-07,
      "loss": 0.6279,
      "step": 18556
    },
    {
      "epoch": 1.66,
      "grad_norm": 8.877912695616361,
      "learning_rate": 7.579316035932943e-07,
      "loss": 0.5933,
      "step": 18557
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.772342885794856,
      "learning_rate": 7.575492723665157e-07,
      "loss": 0.5501,
      "step": 18558
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.434665221273716,
      "learning_rate": 7.57167029691604e-07,
      "loss": 0.5629,
      "step": 18559
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.5926326407758635,
      "learning_rate": 7.567848755765394e-07,
      "loss": 0.5608,
      "step": 18560
    },
    {
      "epoch": 1.66,
      "grad_norm": 7.471635996649574,
      "learning_rate": 7.564028100292987e-07,
      "loss": 0.5194,
      "step": 18561
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.753792910954868,
      "learning_rate": 7.56020833057855e-07,
      "loss": 0.5801,
      "step": 18562
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.604729982440136,
      "learning_rate": 7.556389446701829e-07,
      "loss": 0.5661,
      "step": 18563
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.359980825236722,
      "learning_rate": 7.552571448742518e-07,
      "loss": 0.5824,
      "step": 18564
    },
    {
      "epoch": 1.66,
      "grad_norm": 7.705366826042654,
      "learning_rate": 7.548754336780328e-07,
      "loss": 0.5585,
      "step": 18565
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.511369826601275,
      "learning_rate": 7.544938110894934e-07,
      "loss": 0.5417,
      "step": 18566
    },
    {
      "epoch": 1.66,
      "grad_norm": 9.0263730328385,
      "learning_rate": 7.541122771165965e-07,
      "loss": 0.5613,
      "step": 18567
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.912206738248883,
      "learning_rate": 7.537308317673081e-07,
      "loss": 0.5843,
      "step": 18568
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.9347954118101605,
      "learning_rate": 7.53349475049589e-07,
      "loss": 0.5683,
      "step": 18569
    },
    {
      "epoch": 1.66,
      "grad_norm": 9.534064835232995,
      "learning_rate": 7.529682069714e-07,
      "loss": 0.5509,
      "step": 18570
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.008849574292464,
      "learning_rate": 7.525870275406999e-07,
      "loss": 0.5723,
      "step": 18571
    },
    {
      "epoch": 1.66,
      "grad_norm": 4.422661883713512,
      "learning_rate": 7.522059367654427e-07,
      "loss": 0.6093,
      "step": 18572
    },
    {
      "epoch": 1.66,
      "grad_norm": 7.673561217653939,
      "learning_rate": 7.518249346535838e-07,
      "loss": 0.5752,
      "step": 18573
    },
    {
      "epoch": 1.66,
      "grad_norm": 8.066847278067392,
      "learning_rate": 7.514440212130764e-07,
      "loss": 0.5904,
      "step": 18574
    },
    {
      "epoch": 1.66,
      "grad_norm": 9.237768216850789,
      "learning_rate": 7.510631964518706e-07,
      "loss": 0.5894,
      "step": 18575
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.200674847406414,
      "learning_rate": 7.506824603779173e-07,
      "loss": 0.4793,
      "step": 18576
    },
    {
      "epoch": 1.66,
      "grad_norm": 8.754645973365312,
      "learning_rate": 7.503018129991601e-07,
      "loss": 0.6063,
      "step": 18577
    },
    {
      "epoch": 1.66,
      "grad_norm": 7.615162151614659,
      "learning_rate": 7.499212543235468e-07,
      "loss": 0.6375,
      "step": 18578
    },
    {
      "epoch": 1.66,
      "grad_norm": 7.328288127089694,
      "learning_rate": 7.495407843590192e-07,
      "loss": 0.5686,
      "step": 18579
    },
    {
      "epoch": 1.66,
      "grad_norm": 7.290477639710003,
      "learning_rate": 7.491604031135191e-07,
      "loss": 0.5565,
      "step": 18580
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.515284576952235,
      "learning_rate": 7.487801105949865e-07,
      "loss": 0.5527,
      "step": 18581
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.836041542520796,
      "learning_rate": 7.48399906811359e-07,
      "loss": 0.5229,
      "step": 18582
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.513283737398512,
      "learning_rate": 7.480197917705729e-07,
      "loss": 0.5749,
      "step": 18583
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.015498933113641,
      "learning_rate": 7.476397654805629e-07,
      "loss": 0.6021,
      "step": 18584
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.244230955940849,
      "learning_rate": 7.472598279492588e-07,
      "loss": 0.5485,
      "step": 18585
    },
    {
      "epoch": 1.66,
      "grad_norm": 7.155943675251089,
      "learning_rate": 7.468799791845926e-07,
      "loss": 0.5685,
      "step": 18586
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.822337924308729,
      "learning_rate": 7.465002191944931e-07,
      "loss": 0.5514,
      "step": 18587
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.294443635171104,
      "learning_rate": 7.461205479868861e-07,
      "loss": 0.5667,
      "step": 18588
    },
    {
      "epoch": 1.66,
      "grad_norm": 7.995757602430947,
      "learning_rate": 7.457409655696974e-07,
      "loss": 0.5448,
      "step": 18589
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.2407626508897875,
      "learning_rate": 7.453614719508495e-07,
      "loss": 0.613,
      "step": 18590
    },
    {
      "epoch": 1.66,
      "grad_norm": 7.669333371211093,
      "learning_rate": 7.449820671382641e-07,
      "loss": 0.4954,
      "step": 18591
    },
    {
      "epoch": 1.66,
      "grad_norm": 7.00261761749898,
      "learning_rate": 7.446027511398584e-07,
      "loss": 0.5046,
      "step": 18592
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.54298594135585,
      "learning_rate": 7.442235239635514e-07,
      "loss": 0.6041,
      "step": 18593
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.610275450575934,
      "learning_rate": 7.438443856172578e-07,
      "loss": 0.5612,
      "step": 18594
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.03283612323505,
      "learning_rate": 7.434653361088922e-07,
      "loss": 0.6092,
      "step": 18595
    },
    {
      "epoch": 1.66,
      "grad_norm": 8.837800825025695,
      "learning_rate": 7.430863754463663e-07,
      "loss": 0.5989,
      "step": 18596
    },
    {
      "epoch": 1.66,
      "grad_norm": 8.328324395523541,
      "learning_rate": 7.427075036375903e-07,
      "loss": 0.5673,
      "step": 18597
    },
    {
      "epoch": 1.66,
      "grad_norm": 7.577505155465997,
      "learning_rate": 7.423287206904712e-07,
      "loss": 0.567,
      "step": 18598
    },
    {
      "epoch": 1.66,
      "grad_norm": 7.104704592604466,
      "learning_rate": 7.419500266129159e-07,
      "loss": 0.5739,
      "step": 18599
    },
    {
      "epoch": 1.66,
      "grad_norm": 8.449294448627578,
      "learning_rate": 7.415714214128283e-07,
      "loss": 0.5674,
      "step": 18600
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.762521253783584,
      "learning_rate": 7.41192905098112e-07,
      "loss": 0.5794,
      "step": 18601
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.932503510482977,
      "learning_rate": 7.408144776766674e-07,
      "loss": 0.5584,
      "step": 18602
    },
    {
      "epoch": 1.66,
      "grad_norm": 7.864705191898024,
      "learning_rate": 7.404361391563936e-07,
      "loss": 0.5684,
      "step": 18603
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.675134821013098,
      "learning_rate": 7.400578895451876e-07,
      "loss": 0.532,
      "step": 18604
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.226041500383883,
      "learning_rate": 7.396797288509428e-07,
      "loss": 0.5226,
      "step": 18605
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.906564392208084,
      "learning_rate": 7.393016570815536e-07,
      "loss": 0.5608,
      "step": 18606
    },
    {
      "epoch": 1.66,
      "grad_norm": 8.659609487294826,
      "learning_rate": 7.389236742449113e-07,
      "loss": 0.5328,
      "step": 18607
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.049722274616988,
      "learning_rate": 7.38545780348906e-07,
      "loss": 0.5175,
      "step": 18608
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.7377850034518225,
      "learning_rate": 7.381679754014254e-07,
      "loss": 0.5437,
      "step": 18609
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.733821648543844,
      "learning_rate": 7.377902594103559e-07,
      "loss": 0.589,
      "step": 18610
    },
    {
      "epoch": 1.66,
      "grad_norm": 7.999109437840631,
      "learning_rate": 7.37412632383579e-07,
      "loss": 0.6328,
      "step": 18611
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.623477265812949,
      "learning_rate": 7.370350943289795e-07,
      "loss": 0.5535,
      "step": 18612
    },
    {
      "epoch": 1.66,
      "grad_norm": 8.357716136220102,
      "learning_rate": 7.366576452544361e-07,
      "loss": 0.5365,
      "step": 18613
    },
    {
      "epoch": 1.66,
      "grad_norm": 8.942351944531023,
      "learning_rate": 7.362802851678275e-07,
      "loss": 0.5213,
      "step": 18614
    },
    {
      "epoch": 1.66,
      "grad_norm": 7.177693660022129,
      "learning_rate": 7.359030140770312e-07,
      "loss": 0.5869,
      "step": 18615
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.808089679525784,
      "learning_rate": 7.355258319899222e-07,
      "loss": 0.5726,
      "step": 18616
    },
    {
      "epoch": 1.66,
      "grad_norm": 7.32354715951005,
      "learning_rate": 7.351487389143707e-07,
      "loss": 0.6671,
      "step": 18617
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.101060869939424,
      "learning_rate": 7.3477173485825e-07,
      "loss": 0.5667,
      "step": 18618
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.3293919194574855,
      "learning_rate": 7.343948198294299e-07,
      "loss": 0.5676,
      "step": 18619
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.532579942291563,
      "learning_rate": 7.34017993835775e-07,
      "loss": 0.5663,
      "step": 18620
    },
    {
      "epoch": 1.66,
      "grad_norm": 8.056883143697942,
      "learning_rate": 7.336412568851525e-07,
      "loss": 0.6157,
      "step": 18621
    },
    {
      "epoch": 1.66,
      "grad_norm": 7.9824187760464325,
      "learning_rate": 7.332646089854251e-07,
      "loss": 0.5587,
      "step": 18622
    },
    {
      "epoch": 1.66,
      "grad_norm": 7.1030030105412925,
      "learning_rate": 7.328880501444563e-07,
      "loss": 0.5674,
      "step": 18623
    },
    {
      "epoch": 1.66,
      "grad_norm": 7.257108709007746,
      "learning_rate": 7.325115803701038e-07,
      "loss": 0.5365,
      "step": 18624
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.707902743865074,
      "learning_rate": 7.321351996702259e-07,
      "loss": 0.5337,
      "step": 18625
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.131817029556142,
      "learning_rate": 7.317589080526794e-07,
      "loss": 0.5768,
      "step": 18626
    },
    {
      "epoch": 1.66,
      "grad_norm": 8.904293194958266,
      "learning_rate": 7.313827055253187e-07,
      "loss": 0.5186,
      "step": 18627
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.875943064379474,
      "learning_rate": 7.310065920959958e-07,
      "loss": 0.58,
      "step": 18628
    },
    {
      "epoch": 1.66,
      "grad_norm": 7.173600680305668,
      "learning_rate": 7.306305677725622e-07,
      "loss": 0.5499,
      "step": 18629
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.319265895298604,
      "learning_rate": 7.302546325628651e-07,
      "loss": 0.5554,
      "step": 18630
    },
    {
      "epoch": 1.66,
      "grad_norm": 8.264558975444503,
      "learning_rate": 7.298787864747514e-07,
      "loss": 0.5391,
      "step": 18631
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.897479923246432,
      "learning_rate": 7.295030295160671e-07,
      "loss": 0.6011,
      "step": 18632
    },
    {
      "epoch": 1.66,
      "grad_norm": 8.392347004516212,
      "learning_rate": 7.291273616946554e-07,
      "loss": 0.5826,
      "step": 18633
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.769094701899047,
      "learning_rate": 7.287517830183566e-07,
      "loss": 0.6514,
      "step": 18634
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.123494272997277,
      "learning_rate": 7.2837629349501e-07,
      "loss": 0.6001,
      "step": 18635
    },
    {
      "epoch": 1.66,
      "grad_norm": 10.87676657515052,
      "learning_rate": 7.280008931324545e-07,
      "loss": 0.5966,
      "step": 18636
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.783529813420451,
      "learning_rate": 7.276255819385241e-07,
      "loss": 0.5567,
      "step": 18637
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.27853489836686,
      "learning_rate": 7.272503599210534e-07,
      "loss": 0.5889,
      "step": 18638
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.1056379581585185,
      "learning_rate": 7.268752270878737e-07,
      "loss": 0.577,
      "step": 18639
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.022013039104781,
      "learning_rate": 7.265001834468161e-07,
      "loss": 0.6192,
      "step": 18640
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.304297807417819,
      "learning_rate": 7.261252290057085e-07,
      "loss": 0.5629,
      "step": 18641
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.74885450241976,
      "learning_rate": 7.257503637723778e-07,
      "loss": 0.5454,
      "step": 18642
    },
    {
      "epoch": 1.66,
      "grad_norm": 7.982176648689857,
      "learning_rate": 7.25375587754647e-07,
      "loss": 0.57,
      "step": 18643
    },
    {
      "epoch": 1.66,
      "grad_norm": 8.541529336419861,
      "learning_rate": 7.250009009603398e-07,
      "loss": 0.5751,
      "step": 18644
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.650103547473954,
      "learning_rate": 7.246263033972762e-07,
      "loss": 0.6437,
      "step": 18645
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.748917654076977,
      "learning_rate": 7.242517950732758e-07,
      "loss": 0.5764,
      "step": 18646
    },
    {
      "epoch": 1.66,
      "grad_norm": 4.911560682782207,
      "learning_rate": 7.238773759961564e-07,
      "loss": 0.5681,
      "step": 18647
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.7079078184641645,
      "learning_rate": 7.235030461737313e-07,
      "loss": 0.5691,
      "step": 18648
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.777847785098711,
      "learning_rate": 7.231288056138159e-07,
      "loss": 0.5349,
      "step": 18649
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.76743775229453,
      "learning_rate": 7.227546543242193e-07,
      "loss": 0.5476,
      "step": 18650
    },
    {
      "epoch": 1.66,
      "grad_norm": 8.141345749716066,
      "learning_rate": 7.223805923127525e-07,
      "loss": 0.552,
      "step": 18651
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.175912857858846,
      "learning_rate": 7.220066195872228e-07,
      "loss": 0.5819,
      "step": 18652
    },
    {
      "epoch": 1.66,
      "grad_norm": 4.651008140553142,
      "learning_rate": 7.216327361554359e-07,
      "loss": 0.6192,
      "step": 18653
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.265950931793344,
      "learning_rate": 7.212589420251981e-07,
      "loss": 0.6281,
      "step": 18654
    },
    {
      "epoch": 1.66,
      "grad_norm": 8.377950438085541,
      "learning_rate": 7.208852372043079e-07,
      "loss": 0.5185,
      "step": 18655
    },
    {
      "epoch": 1.66,
      "grad_norm": 8.227941306183167,
      "learning_rate": 7.20511621700567e-07,
      "loss": 0.5885,
      "step": 18656
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.675789424899326,
      "learning_rate": 7.201380955217746e-07,
      "loss": 0.5971,
      "step": 18657
    },
    {
      "epoch": 1.66,
      "grad_norm": 8.12593115679971,
      "learning_rate": 7.197646586757268e-07,
      "loss": 0.5305,
      "step": 18658
    },
    {
      "epoch": 1.66,
      "grad_norm": 7.520701729621175,
      "learning_rate": 7.193913111702178e-07,
      "loss": 0.5425,
      "step": 18659
    },
    {
      "epoch": 1.66,
      "grad_norm": 7.248413749899023,
      "learning_rate": 7.190180530130419e-07,
      "loss": 0.5508,
      "step": 18660
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.158315606482092,
      "learning_rate": 7.186448842119886e-07,
      "loss": 0.5598,
      "step": 18661
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.984874999296665,
      "learning_rate": 7.182718047748477e-07,
      "loss": 0.5444,
      "step": 18662
    },
    {
      "epoch": 1.66,
      "grad_norm": 7.110748210597638,
      "learning_rate": 7.178988147094051e-07,
      "loss": 0.5024,
      "step": 18663
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.586822036717307,
      "learning_rate": 7.175259140234469e-07,
      "loss": 0.5832,
      "step": 18664
    },
    {
      "epoch": 1.67,
      "grad_norm": 8.315036721549612,
      "learning_rate": 7.171531027247575e-07,
      "loss": 0.6356,
      "step": 18665
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.691040891667781,
      "learning_rate": 7.167803808211171e-07,
      "loss": 0.5229,
      "step": 18666
    },
    {
      "epoch": 1.67,
      "grad_norm": 9.681325005006624,
      "learning_rate": 7.164077483203075e-07,
      "loss": 0.5439,
      "step": 18667
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.439042971441842,
      "learning_rate": 7.160352052301039e-07,
      "loss": 0.5764,
      "step": 18668
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.28602361474621,
      "learning_rate": 7.156627515582843e-07,
      "loss": 0.6075,
      "step": 18669
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.568860549564265,
      "learning_rate": 7.152903873126216e-07,
      "loss": 0.5176,
      "step": 18670
    },
    {
      "epoch": 1.67,
      "grad_norm": 9.048156275294534,
      "learning_rate": 7.149181125008892e-07,
      "loss": 0.5836,
      "step": 18671
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.89993615772897,
      "learning_rate": 7.145459271308564e-07,
      "loss": 0.607,
      "step": 18672
    },
    {
      "epoch": 1.67,
      "grad_norm": 8.285752985668125,
      "learning_rate": 7.141738312102942e-07,
      "loss": 0.5917,
      "step": 18673
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.369506918216218,
      "learning_rate": 7.138018247469658e-07,
      "loss": 0.5862,
      "step": 18674
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.710053193021664,
      "learning_rate": 7.134299077486379e-07,
      "loss": 0.5193,
      "step": 18675
    },
    {
      "epoch": 1.67,
      "grad_norm": 8.072010237071831,
      "learning_rate": 7.130580802230735e-07,
      "loss": 0.6417,
      "step": 18676
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.245515967719974,
      "learning_rate": 7.126863421780344e-07,
      "loss": 0.5734,
      "step": 18677
    },
    {
      "epoch": 1.67,
      "grad_norm": 9.232498580156582,
      "learning_rate": 7.123146936212777e-07,
      "loss": 0.5279,
      "step": 18678
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.664449334847816,
      "learning_rate": 7.119431345605615e-07,
      "loss": 0.5666,
      "step": 18679
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.688688098254868,
      "learning_rate": 7.115716650036431e-07,
      "loss": 0.5477,
      "step": 18680
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.075171477177617,
      "learning_rate": 7.112002849582738e-07,
      "loss": 0.5649,
      "step": 18681
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.217644918887227,
      "learning_rate": 7.10828994432206e-07,
      "loss": 0.5912,
      "step": 18682
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.23185791464326,
      "learning_rate": 7.104577934331897e-07,
      "loss": 0.6209,
      "step": 18683
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.117528406015411,
      "learning_rate": 7.100866819689734e-07,
      "loss": 0.537,
      "step": 18684
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.72666004121027,
      "learning_rate": 7.097156600473027e-07,
      "loss": 0.6062,
      "step": 18685
    },
    {
      "epoch": 1.67,
      "grad_norm": 9.56803529877836,
      "learning_rate": 7.09344727675923e-07,
      "loss": 0.6482,
      "step": 18686
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.1979789265901175,
      "learning_rate": 7.089738848625749e-07,
      "loss": 0.5351,
      "step": 18687
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.274290392117034,
      "learning_rate": 7.086031316149994e-07,
      "loss": 0.6011,
      "step": 18688
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.881622047557984,
      "learning_rate": 7.082324679409359e-07,
      "loss": 0.6245,
      "step": 18689
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.417306388040407,
      "learning_rate": 7.078618938481207e-07,
      "loss": 0.5234,
      "step": 18690
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.792101741780786,
      "learning_rate": 7.074914093442897e-07,
      "loss": 0.5923,
      "step": 18691
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.73437880659578,
      "learning_rate": 7.071210144371743e-07,
      "loss": 0.5458,
      "step": 18692
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.6269077018477756,
      "learning_rate": 7.067507091345078e-07,
      "loss": 0.5754,
      "step": 18693
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.471929078348774,
      "learning_rate": 7.063804934440166e-07,
      "loss": 0.5746,
      "step": 18694
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.904466060542601,
      "learning_rate": 7.060103673734297e-07,
      "loss": 0.5795,
      "step": 18695
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.105323647203277,
      "learning_rate": 7.056403309304733e-07,
      "loss": 0.5102,
      "step": 18696
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.414766894151342,
      "learning_rate": 7.052703841228697e-07,
      "loss": 0.5483,
      "step": 18697
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.64479570396298,
      "learning_rate": 7.04900526958342e-07,
      "loss": 0.5845,
      "step": 18698
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.754061116078944,
      "learning_rate": 7.045307594446111e-07,
      "loss": 0.5373,
      "step": 18699
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.350383735338322,
      "learning_rate": 7.041610815893918e-07,
      "loss": 0.5393,
      "step": 18700
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.003381062909269,
      "learning_rate": 7.037914934004025e-07,
      "loss": 0.5338,
      "step": 18701
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.088959912344128,
      "learning_rate": 7.034219948853577e-07,
      "loss": 0.5661,
      "step": 18702
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.733725708744355,
      "learning_rate": 7.030525860519688e-07,
      "loss": 0.5823,
      "step": 18703
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.528017707801505,
      "learning_rate": 7.026832669079476e-07,
      "loss": 0.5615,
      "step": 18704
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.925601395804873,
      "learning_rate": 7.023140374610027e-07,
      "loss": 0.6236,
      "step": 18705
    },
    {
      "epoch": 1.67,
      "grad_norm": 9.542842135152291,
      "learning_rate": 7.019448977188409e-07,
      "loss": 0.595,
      "step": 18706
    },
    {
      "epoch": 1.67,
      "grad_norm": 8.356890442689716,
      "learning_rate": 7.015758476891659e-07,
      "loss": 0.6024,
      "step": 18707
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.997886209615091,
      "learning_rate": 7.012068873796812e-07,
      "loss": 0.6002,
      "step": 18708
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.796145082397513,
      "learning_rate": 7.008380167980888e-07,
      "loss": 0.5537,
      "step": 18709
    },
    {
      "epoch": 1.67,
      "grad_norm": 9.239963522210958,
      "learning_rate": 7.004692359520876e-07,
      "loss": 0.617,
      "step": 18710
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.321059568232503,
      "learning_rate": 7.001005448493752e-07,
      "loss": 0.5837,
      "step": 18711
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.311299161676711,
      "learning_rate": 6.997319434976491e-07,
      "loss": 0.6128,
      "step": 18712
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.215005720842257,
      "learning_rate": 6.993634319045995e-07,
      "loss": 0.6269,
      "step": 18713
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.456462943936085,
      "learning_rate": 6.989950100779203e-07,
      "loss": 0.566,
      "step": 18714
    },
    {
      "epoch": 1.67,
      "grad_norm": 8.834941726039688,
      "learning_rate": 6.98626678025301e-07,
      "loss": 0.5575,
      "step": 18715
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.888359619502168,
      "learning_rate": 6.982584357544297e-07,
      "loss": 0.5285,
      "step": 18716
    },
    {
      "epoch": 1.67,
      "grad_norm": 8.098747007326779,
      "learning_rate": 6.978902832729934e-07,
      "loss": 0.5933,
      "step": 18717
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.544345259790728,
      "learning_rate": 6.975222205886772e-07,
      "loss": 0.5085,
      "step": 18718
    },
    {
      "epoch": 1.67,
      "grad_norm": 8.147526508603773,
      "learning_rate": 6.971542477091609e-07,
      "loss": 0.5928,
      "step": 18719
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.241956454003021,
      "learning_rate": 6.96786364642128e-07,
      "loss": 0.53,
      "step": 18720
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.0727553428761025,
      "learning_rate": 6.964185713952543e-07,
      "loss": 0.6135,
      "step": 18721
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.159982449471985,
      "learning_rate": 6.960508679762185e-07,
      "loss": 0.5235,
      "step": 18722
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.864894478884077,
      "learning_rate": 6.956832543926956e-07,
      "loss": 0.5829,
      "step": 18723
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.642084860783315,
      "learning_rate": 6.953157306523578e-07,
      "loss": 0.5657,
      "step": 18724
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.220734909336617,
      "learning_rate": 6.94948296762879e-07,
      "loss": 0.5868,
      "step": 18725
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.07685488984557,
      "learning_rate": 6.945809527319247e-07,
      "loss": 0.509,
      "step": 18726
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.703640295270385,
      "learning_rate": 6.942136985671649e-07,
      "loss": 0.5856,
      "step": 18727
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.810009436506186,
      "learning_rate": 6.938465342762646e-07,
      "loss": 0.5828,
      "step": 18728
    },
    {
      "epoch": 1.67,
      "grad_norm": 9.965003471760866,
      "learning_rate": 6.93479459866887e-07,
      "loss": 0.6228,
      "step": 18729
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.109170609810283,
      "learning_rate": 6.931124753466961e-07,
      "loss": 0.5992,
      "step": 18730
    },
    {
      "epoch": 1.67,
      "grad_norm": 8.237722885073138,
      "learning_rate": 6.927455807233491e-07,
      "loss": 0.581,
      "step": 18731
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.20712208856185,
      "learning_rate": 6.923787760045059e-07,
      "loss": 0.5876,
      "step": 18732
    },
    {
      "epoch": 1.67,
      "grad_norm": 10.085029075098388,
      "learning_rate": 6.920120611978215e-07,
      "loss": 0.5611,
      "step": 18733
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.6692405277020335,
      "learning_rate": 6.916454363109526e-07,
      "loss": 0.5453,
      "step": 18734
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.6831749969968675,
      "learning_rate": 6.912789013515486e-07,
      "loss": 0.5694,
      "step": 18735
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.3160071959518245,
      "learning_rate": 6.909124563272618e-07,
      "loss": 0.5396,
      "step": 18736
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.853178534807961,
      "learning_rate": 6.905461012457415e-07,
      "loss": 0.5342,
      "step": 18737
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.688804390068174,
      "learning_rate": 6.901798361146333e-07,
      "loss": 0.6195,
      "step": 18738
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.9489721934282755,
      "learning_rate": 6.898136609415823e-07,
      "loss": 0.5903,
      "step": 18739
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.185749223244961,
      "learning_rate": 6.894475757342323e-07,
      "loss": 0.5755,
      "step": 18740
    },
    {
      "epoch": 1.67,
      "grad_norm": 4.504288492776687,
      "learning_rate": 6.890815805002238e-07,
      "loss": 0.55,
      "step": 18741
    },
    {
      "epoch": 1.67,
      "grad_norm": 9.002430118806888,
      "learning_rate": 6.887156752471963e-07,
      "loss": 0.5743,
      "step": 18742
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.154452086167655,
      "learning_rate": 6.883498599827892e-07,
      "loss": 0.5793,
      "step": 18743
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.589666126714272,
      "learning_rate": 6.879841347146354e-07,
      "loss": 0.5672,
      "step": 18744
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.6695765238608224,
      "learning_rate": 6.876184994503693e-07,
      "loss": 0.5278,
      "step": 18745
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.487172838426258,
      "learning_rate": 6.87252954197623e-07,
      "loss": 0.5495,
      "step": 18746
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.585840125163654,
      "learning_rate": 6.868874989640267e-07,
      "loss": 0.5679,
      "step": 18747
    },
    {
      "epoch": 1.67,
      "grad_norm": 8.378562312539302,
      "learning_rate": 6.865221337572081e-07,
      "loss": 0.5725,
      "step": 18748
    },
    {
      "epoch": 1.67,
      "grad_norm": 8.211779636757607,
      "learning_rate": 6.86156858584795e-07,
      "loss": 0.5933,
      "step": 18749
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.989439060179455,
      "learning_rate": 6.857916734544101e-07,
      "loss": 0.5798,
      "step": 18750
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.212689113794405,
      "learning_rate": 6.854265783736752e-07,
      "loss": 0.5645,
      "step": 18751
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.333440539377955,
      "learning_rate": 6.85061573350212e-07,
      "loss": 0.5282,
      "step": 18752
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.607271820728085,
      "learning_rate": 6.846966583916381e-07,
      "loss": 0.5245,
      "step": 18753
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.391278403664465,
      "learning_rate": 6.843318335055715e-07,
      "loss": 0.5442,
      "step": 18754
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.043909403547669,
      "learning_rate": 6.839670986996267e-07,
      "loss": 0.6064,
      "step": 18755
    },
    {
      "epoch": 1.67,
      "grad_norm": 9.378228013436457,
      "learning_rate": 6.836024539814179e-07,
      "loss": 0.5873,
      "step": 18756
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.318488985062379,
      "learning_rate": 6.832378993585543e-07,
      "loss": 0.581,
      "step": 18757
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.756793223529936,
      "learning_rate": 6.828734348386457e-07,
      "loss": 0.4844,
      "step": 18758
    },
    {
      "epoch": 1.67,
      "grad_norm": 11.335201612410094,
      "learning_rate": 6.825090604292995e-07,
      "loss": 0.5738,
      "step": 18759
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.023905326919885,
      "learning_rate": 6.821447761381216e-07,
      "loss": 0.5791,
      "step": 18760
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.078287489474752,
      "learning_rate": 6.817805819727164e-07,
      "loss": 0.5299,
      "step": 18761
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.245017843699664,
      "learning_rate": 6.814164779406851e-07,
      "loss": 0.6011,
      "step": 18762
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.76484488573197,
      "learning_rate": 6.810524640496274e-07,
      "loss": 0.5457,
      "step": 18763
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.087221737572406,
      "learning_rate": 6.8068854030714e-07,
      "loss": 0.596,
      "step": 18764
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.2581433648729226,
      "learning_rate": 6.803247067208207e-07,
      "loss": 0.5334,
      "step": 18765
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.699906844330567,
      "learning_rate": 6.799609632982629e-07,
      "loss": 0.6023,
      "step": 18766
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.9108143228811665,
      "learning_rate": 6.795973100470587e-07,
      "loss": 0.5721,
      "step": 18767
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.211843891150087,
      "learning_rate": 6.792337469748e-07,
      "loss": 0.5872,
      "step": 18768
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.702407954562579,
      "learning_rate": 6.788702740890752e-07,
      "loss": 0.5537,
      "step": 18769
    },
    {
      "epoch": 1.67,
      "grad_norm": 8.372038122474496,
      "learning_rate": 6.785068913974696e-07,
      "loss": 0.5813,
      "step": 18770
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.616550770931809,
      "learning_rate": 6.781435989075679e-07,
      "loss": 0.6036,
      "step": 18771
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.4304274760396405,
      "learning_rate": 6.777803966269548e-07,
      "loss": 0.5797,
      "step": 18772
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.616027884722334,
      "learning_rate": 6.774172845632099e-07,
      "loss": 0.5341,
      "step": 18773
    },
    {
      "epoch": 1.67,
      "grad_norm": 8.009925409586025,
      "learning_rate": 6.770542627239129e-07,
      "loss": 0.5396,
      "step": 18774
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.922225594482779,
      "learning_rate": 6.766913311166423e-07,
      "loss": 0.5921,
      "step": 18775
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.933845732344708,
      "learning_rate": 6.763284897489714e-07,
      "loss": 0.5064,
      "step": 18776
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.530040494106385,
      "learning_rate": 6.759657386284741e-07,
      "loss": 0.5509,
      "step": 18777
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.465859774759495,
      "learning_rate": 6.756030777627242e-07,
      "loss": 0.6278,
      "step": 18778
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.980956829343796,
      "learning_rate": 6.752405071592882e-07,
      "loss": 0.521,
      "step": 18779
    },
    {
      "epoch": 1.68,
      "grad_norm": 8.609240288737123,
      "learning_rate": 6.74878026825736e-07,
      "loss": 0.5745,
      "step": 18780
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.670527244111905,
      "learning_rate": 6.745156367696326e-07,
      "loss": 0.5913,
      "step": 18781
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.780588340578137,
      "learning_rate": 6.741533369985442e-07,
      "loss": 0.632,
      "step": 18782
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.776412430991458,
      "learning_rate": 6.737911275200299e-07,
      "loss": 0.5739,
      "step": 18783
    },
    {
      "epoch": 1.68,
      "grad_norm": 11.70878239371988,
      "learning_rate": 6.734290083416517e-07,
      "loss": 0.5665,
      "step": 18784
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.592203350338689,
      "learning_rate": 6.730669794709682e-07,
      "loss": 0.5828,
      "step": 18785
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.429209256142616,
      "learning_rate": 6.727050409155356e-07,
      "loss": 0.5476,
      "step": 18786
    },
    {
      "epoch": 1.68,
      "grad_norm": 8.42008412669541,
      "learning_rate": 6.723431926829089e-07,
      "loss": 0.5732,
      "step": 18787
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.97629881317841,
      "learning_rate": 6.719814347806414e-07,
      "loss": 0.5316,
      "step": 18788
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.663631241160309,
      "learning_rate": 6.716197672162822e-07,
      "loss": 0.5595,
      "step": 18789
    },
    {
      "epoch": 1.68,
      "grad_norm": 8.116874781957778,
      "learning_rate": 6.712581899973814e-07,
      "loss": 0.5232,
      "step": 18790
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.018151464113047,
      "learning_rate": 6.708967031314856e-07,
      "loss": 0.5506,
      "step": 18791
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.218145677341601,
      "learning_rate": 6.705353066261421e-07,
      "loss": 0.4842,
      "step": 18792
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.675144095481751,
      "learning_rate": 6.701740004888918e-07,
      "loss": 0.5459,
      "step": 18793
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.692457789866794,
      "learning_rate": 6.698127847272767e-07,
      "loss": 0.5617,
      "step": 18794
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.108632728383405,
      "learning_rate": 6.694516593488381e-07,
      "loss": 0.5815,
      "step": 18795
    },
    {
      "epoch": 1.68,
      "grad_norm": 8.053903319691106,
      "learning_rate": 6.690906243611112e-07,
      "loss": 0.567,
      "step": 18796
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.658944717574081,
      "learning_rate": 6.687296797716331e-07,
      "loss": 0.5905,
      "step": 18797
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.39532687998217,
      "learning_rate": 6.683688255879378e-07,
      "loss": 0.5759,
      "step": 18798
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.49582086520836,
      "learning_rate": 6.680080618175566e-07,
      "loss": 0.5526,
      "step": 18799
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.994009290491478,
      "learning_rate": 6.676473884680207e-07,
      "loss": 0.5778,
      "step": 18800
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.7239755606589515,
      "learning_rate": 6.672868055468584e-07,
      "loss": 0.5672,
      "step": 18801
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.349110959927907,
      "learning_rate": 6.669263130615949e-07,
      "loss": 0.5272,
      "step": 18802
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.198362178394492,
      "learning_rate": 6.665659110197554e-07,
      "loss": 0.5086,
      "step": 18803
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.074779064356047,
      "learning_rate": 6.662055994288625e-07,
      "loss": 0.5406,
      "step": 18804
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.459916391306315,
      "learning_rate": 6.658453782964363e-07,
      "loss": 0.4976,
      "step": 18805
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.136413191241806,
      "learning_rate": 6.654852476299978e-07,
      "loss": 0.5703,
      "step": 18806
    },
    {
      "epoch": 1.68,
      "grad_norm": 9.503814022949408,
      "learning_rate": 6.651252074370612e-07,
      "loss": 0.5617,
      "step": 18807
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.7500483386980665,
      "learning_rate": 6.64765257725144e-07,
      "loss": 0.5354,
      "step": 18808
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.814369833312284,
      "learning_rate": 6.644053985017562e-07,
      "loss": 0.541,
      "step": 18809
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.25137842869322,
      "learning_rate": 6.640456297744114e-07,
      "loss": 0.5397,
      "step": 18810
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.3502201467293435,
      "learning_rate": 6.636859515506183e-07,
      "loss": 0.588,
      "step": 18811
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.810472748926223,
      "learning_rate": 6.633263638378851e-07,
      "loss": 0.6125,
      "step": 18812
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.721245419139586,
      "learning_rate": 6.629668666437177e-07,
      "loss": 0.5359,
      "step": 18813
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.87606356444877,
      "learning_rate": 6.626074599756177e-07,
      "loss": 0.5623,
      "step": 18814
    },
    {
      "epoch": 1.68,
      "grad_norm": 8.813522865544947,
      "learning_rate": 6.622481438410883e-07,
      "loss": 0.5718,
      "step": 18815
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.7073305565248384,
      "learning_rate": 6.618889182476296e-07,
      "loss": 0.5058,
      "step": 18816
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.516695352234643,
      "learning_rate": 6.615297832027395e-07,
      "loss": 0.5857,
      "step": 18817
    },
    {
      "epoch": 1.68,
      "grad_norm": 8.018468838088008,
      "learning_rate": 6.611707387139138e-07,
      "loss": 0.613,
      "step": 18818
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.318215491126707,
      "learning_rate": 6.608117847886481e-07,
      "loss": 0.6463,
      "step": 18819
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.55661203817161,
      "learning_rate": 6.60452921434433e-07,
      "loss": 0.5531,
      "step": 18820
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.799424692881998,
      "learning_rate": 6.600941486587603e-07,
      "loss": 0.5896,
      "step": 18821
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.601144408050781,
      "learning_rate": 6.597354664691175e-07,
      "loss": 0.573,
      "step": 18822
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.6409893772529225,
      "learning_rate": 6.593768748729917e-07,
      "loss": 0.5406,
      "step": 18823
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.726992401520035,
      "learning_rate": 6.590183738778677e-07,
      "loss": 0.584,
      "step": 18824
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.192460261554137,
      "learning_rate": 6.586599634912283e-07,
      "loss": 0.5349,
      "step": 18825
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.344657429268577,
      "learning_rate": 6.583016437205564e-07,
      "loss": 0.5964,
      "step": 18826
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.503027925962039,
      "learning_rate": 6.579434145733276e-07,
      "loss": 0.5824,
      "step": 18827
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.911923641265136,
      "learning_rate": 6.575852760570217e-07,
      "loss": 0.5303,
      "step": 18828
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.858446698722333,
      "learning_rate": 6.572272281791137e-07,
      "loss": 0.5656,
      "step": 18829
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.805142288336392,
      "learning_rate": 6.568692709470765e-07,
      "loss": 0.5615,
      "step": 18830
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.3257943037876,
      "learning_rate": 6.565114043683818e-07,
      "loss": 0.4914,
      "step": 18831
    },
    {
      "epoch": 1.68,
      "grad_norm": 8.715461340094754,
      "learning_rate": 6.56153628450501e-07,
      "loss": 0.5991,
      "step": 18832
    },
    {
      "epoch": 1.68,
      "grad_norm": 29.321924962930918,
      "learning_rate": 6.557959432008987e-07,
      "loss": 0.7496,
      "step": 18833
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.786428072304873,
      "learning_rate": 6.554383486270427e-07,
      "loss": 0.5416,
      "step": 18834
    },
    {
      "epoch": 1.68,
      "grad_norm": 8.656064950369164,
      "learning_rate": 6.550808447363977e-07,
      "loss": 0.5306,
      "step": 18835
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.117994606244157,
      "learning_rate": 6.547234315364243e-07,
      "loss": 0.6213,
      "step": 18836
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.445865189010255,
      "learning_rate": 6.543661090345827e-07,
      "loss": 0.6063,
      "step": 18837
    },
    {
      "epoch": 1.68,
      "grad_norm": 8.251337492712066,
      "learning_rate": 6.540088772383318e-07,
      "loss": 0.5543,
      "step": 18838
    },
    {
      "epoch": 1.68,
      "grad_norm": 8.95099479739495,
      "learning_rate": 6.536517361551292e-07,
      "loss": 0.5794,
      "step": 18839
    },
    {
      "epoch": 1.68,
      "grad_norm": 8.177532868716206,
      "learning_rate": 6.532946857924277e-07,
      "loss": 0.5515,
      "step": 18840
    },
    {
      "epoch": 1.68,
      "grad_norm": 11.589111429873162,
      "learning_rate": 6.529377261576797e-07,
      "loss": 0.6229,
      "step": 18841
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.1505555137886425,
      "learning_rate": 6.525808572583376e-07,
      "loss": 0.5879,
      "step": 18842
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.710191984099238,
      "learning_rate": 6.522240791018486e-07,
      "loss": 0.6313,
      "step": 18843
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.335278358144622,
      "learning_rate": 6.518673916956608e-07,
      "loss": 0.5043,
      "step": 18844
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.706837346865468,
      "learning_rate": 6.515107950472199e-07,
      "loss": 0.5547,
      "step": 18845
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.055635324130055,
      "learning_rate": 6.511542891639677e-07,
      "loss": 0.5829,
      "step": 18846
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.482139094299594,
      "learning_rate": 6.50797874053345e-07,
      "loss": 0.5505,
      "step": 18847
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.119250792768566,
      "learning_rate": 6.504415497227928e-07,
      "loss": 0.5558,
      "step": 18848
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.864679887938728,
      "learning_rate": 6.500853161797477e-07,
      "loss": 0.5417,
      "step": 18849
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.606704184365366,
      "learning_rate": 6.497291734316463e-07,
      "loss": 0.5505,
      "step": 18850
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.9258781260246405,
      "learning_rate": 6.493731214859206e-07,
      "loss": 0.619,
      "step": 18851
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.039306097892574,
      "learning_rate": 6.490171603500039e-07,
      "loss": 0.5956,
      "step": 18852
    },
    {
      "epoch": 1.68,
      "grad_norm": 10.132254962253365,
      "learning_rate": 6.48661290031325e-07,
      "loss": 0.5945,
      "step": 18853
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.872016844322838,
      "learning_rate": 6.48305510537312e-07,
      "loss": 0.5515,
      "step": 18854
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.307926348347676,
      "learning_rate": 6.479498218753921e-07,
      "loss": 0.5882,
      "step": 18855
    },
    {
      "epoch": 1.68,
      "grad_norm": 9.33701233142072,
      "learning_rate": 6.475942240529881e-07,
      "loss": 0.5361,
      "step": 18856
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.474261443104198,
      "learning_rate": 6.472387170775235e-07,
      "loss": 0.5223,
      "step": 18857
    },
    {
      "epoch": 1.68,
      "grad_norm": 8.5826781119489,
      "learning_rate": 6.468833009564191e-07,
      "loss": 0.5545,
      "step": 18858
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.753580568110589,
      "learning_rate": 6.465279756970916e-07,
      "loss": 0.5861,
      "step": 18859
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.817911468054885,
      "learning_rate": 6.461727413069586e-07,
      "loss": 0.5665,
      "step": 18860
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.272251931746201,
      "learning_rate": 6.45817597793435e-07,
      "loss": 0.5635,
      "step": 18861
    },
    {
      "epoch": 1.68,
      "grad_norm": 8.669558691619732,
      "learning_rate": 6.45462545163934e-07,
      "loss": 0.6146,
      "step": 18862
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.69145720017826,
      "learning_rate": 6.451075834258652e-07,
      "loss": 0.6444,
      "step": 18863
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.607288333135034,
      "learning_rate": 6.447527125866405e-07,
      "loss": 0.5825,
      "step": 18864
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.365069050573747,
      "learning_rate": 6.443979326536648e-07,
      "loss": 0.5694,
      "step": 18865
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.83092805065577,
      "learning_rate": 6.440432436343425e-07,
      "loss": 0.5441,
      "step": 18866
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.342621958679847,
      "learning_rate": 6.43688645536078e-07,
      "loss": 0.6119,
      "step": 18867
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.865565666350034,
      "learning_rate": 6.433341383662728e-07,
      "loss": 0.5185,
      "step": 18868
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.098905082581208,
      "learning_rate": 6.429797221323264e-07,
      "loss": 0.5526,
      "step": 18869
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.779841551122888,
      "learning_rate": 6.426253968416368e-07,
      "loss": 0.5396,
      "step": 18870
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.984466253006608,
      "learning_rate": 6.422711625016009e-07,
      "loss": 0.5772,
      "step": 18871
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.24311890322206,
      "learning_rate": 6.4191701911961e-07,
      "loss": 0.5197,
      "step": 18872
    },
    {
      "epoch": 1.68,
      "grad_norm": 4.810332912484384,
      "learning_rate": 6.415629667030571e-07,
      "loss": 0.6034,
      "step": 18873
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.222246773034977,
      "learning_rate": 6.412090052593328e-07,
      "loss": 0.5928,
      "step": 18874
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.452383838966274,
      "learning_rate": 6.408551347958247e-07,
      "loss": 0.5246,
      "step": 18875
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.490950268250788,
      "learning_rate": 6.40501355319919e-07,
      "loss": 0.5707,
      "step": 18876
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.400643621422679,
      "learning_rate": 6.401476668390016e-07,
      "loss": 0.5239,
      "step": 18877
    },
    {
      "epoch": 1.68,
      "grad_norm": 9.424243891981137,
      "learning_rate": 6.397940693604543e-07,
      "loss": 0.5428,
      "step": 18878
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.124580133931576,
      "learning_rate": 6.394405628916555e-07,
      "loss": 0.6021,
      "step": 18879
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.406309208254688,
      "learning_rate": 6.390871474399857e-07,
      "loss": 0.5572,
      "step": 18880
    },
    {
      "epoch": 1.68,
      "grad_norm": 4.877279978578899,
      "learning_rate": 6.387338230128209e-07,
      "loss": 0.5857,
      "step": 18881
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.5690298671131115,
      "learning_rate": 6.383805896175371e-07,
      "loss": 0.5783,
      "step": 18882
    },
    {
      "epoch": 1.68,
      "grad_norm": 8.580830253656917,
      "learning_rate": 6.380274472615067e-07,
      "loss": 0.5534,
      "step": 18883
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.3086429205889,
      "learning_rate": 6.37674395952102e-07,
      "loss": 0.5023,
      "step": 18884
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.017225910855992,
      "learning_rate": 6.373214356966894e-07,
      "loss": 0.5878,
      "step": 18885
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.948973237846084,
      "learning_rate": 6.369685665026381e-07,
      "loss": 0.614,
      "step": 18886
    },
    {
      "epoch": 1.68,
      "grad_norm": 8.300911575829993,
      "learning_rate": 6.366157883773128e-07,
      "loss": 0.6188,
      "step": 18887
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.178797171149819,
      "learning_rate": 6.362631013280774e-07,
      "loss": 0.5527,
      "step": 18888
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.0327066672801175,
      "learning_rate": 6.359105053622943e-07,
      "loss": 0.6266,
      "step": 18889
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.762705302196657,
      "learning_rate": 6.355580004873214e-07,
      "loss": 0.5629,
      "step": 18890
    },
    {
      "epoch": 1.69,
      "grad_norm": 7.178532817370461,
      "learning_rate": 6.352055867105173e-07,
      "loss": 0.5019,
      "step": 18891
    },
    {
      "epoch": 1.69,
      "grad_norm": 7.794980904578449,
      "learning_rate": 6.348532640392375e-07,
      "loss": 0.6092,
      "step": 18892
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.676391199337664,
      "learning_rate": 6.345010324808376e-07,
      "loss": 0.6671,
      "step": 18893
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.565822694884817,
      "learning_rate": 6.341488920426675e-07,
      "loss": 0.6108,
      "step": 18894
    },
    {
      "epoch": 1.69,
      "grad_norm": 8.190790186502118,
      "learning_rate": 6.337968427320778e-07,
      "loss": 0.5507,
      "step": 18895
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.376414282470727,
      "learning_rate": 6.334448845564184e-07,
      "loss": 0.5287,
      "step": 18896
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.458842056780733,
      "learning_rate": 6.330930175230337e-07,
      "loss": 0.5662,
      "step": 18897
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.437514913712253,
      "learning_rate": 6.327412416392687e-07,
      "loss": 0.6585,
      "step": 18898
    },
    {
      "epoch": 1.69,
      "grad_norm": 8.029905089667606,
      "learning_rate": 6.32389556912466e-07,
      "loss": 0.6368,
      "step": 18899
    },
    {
      "epoch": 1.69,
      "grad_norm": 7.496775116932812,
      "learning_rate": 6.320379633499663e-07,
      "loss": 0.5654,
      "step": 18900
    },
    {
      "epoch": 1.69,
      "grad_norm": 9.165777463870116,
      "learning_rate": 6.316864609591084e-07,
      "loss": 0.5619,
      "step": 18901
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.8614028401292435,
      "learning_rate": 6.313350497472304e-07,
      "loss": 0.5023,
      "step": 18902
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.105953023442856,
      "learning_rate": 6.309837297216653e-07,
      "loss": 0.5781,
      "step": 18903
    },
    {
      "epoch": 1.69,
      "grad_norm": 8.128712358683185,
      "learning_rate": 6.306325008897463e-07,
      "loss": 0.6142,
      "step": 18904
    },
    {
      "epoch": 1.69,
      "grad_norm": 10.108277967013555,
      "learning_rate": 6.302813632588056e-07,
      "loss": 0.5957,
      "step": 18905
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.799660623405033,
      "learning_rate": 6.299303168361714e-07,
      "loss": 0.5837,
      "step": 18906
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.999237164966168,
      "learning_rate": 6.295793616291729e-07,
      "loss": 0.568,
      "step": 18907
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.422449586789886,
      "learning_rate": 6.292284976451335e-07,
      "loss": 0.5297,
      "step": 18908
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.9693723742317815,
      "learning_rate": 6.28877724891378e-07,
      "loss": 0.6109,
      "step": 18909
    },
    {
      "epoch": 1.69,
      "grad_norm": 7.045375818049848,
      "learning_rate": 6.285270433752261e-07,
      "loss": 0.5692,
      "step": 18910
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.553112804293156,
      "learning_rate": 6.281764531039991e-07,
      "loss": 0.5596,
      "step": 18911
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.210412775616269,
      "learning_rate": 6.278259540850145e-07,
      "loss": 0.5821,
      "step": 18912
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.002990630612759,
      "learning_rate": 6.274755463255883e-07,
      "loss": 0.514,
      "step": 18913
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.613229149851825,
      "learning_rate": 6.271252298330344e-07,
      "loss": 0.5406,
      "step": 18914
    },
    {
      "epoch": 1.69,
      "grad_norm": 12.180184953606888,
      "learning_rate": 6.267750046146659e-07,
      "loss": 0.5771,
      "step": 18915
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.175987278095403,
      "learning_rate": 6.264248706777908e-07,
      "loss": 0.4995,
      "step": 18916
    },
    {
      "epoch": 1.69,
      "grad_norm": 7.285196121682943,
      "learning_rate": 6.260748280297186e-07,
      "loss": 0.6042,
      "step": 18917
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.922147842959956,
      "learning_rate": 6.257248766777558e-07,
      "loss": 0.5817,
      "step": 18918
    },
    {
      "epoch": 1.69,
      "grad_norm": 7.326015210127596,
      "learning_rate": 6.253750166292067e-07,
      "loss": 0.5729,
      "step": 18919
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.579192779897657,
      "learning_rate": 6.250252478913738e-07,
      "loss": 0.5559,
      "step": 18920
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.317479280248291,
      "learning_rate": 6.246755704715585e-07,
      "loss": 0.5606,
      "step": 18921
    },
    {
      "epoch": 1.69,
      "grad_norm": 7.3536689154034285,
      "learning_rate": 6.243259843770594e-07,
      "loss": 0.5459,
      "step": 18922
    },
    {
      "epoch": 1.69,
      "grad_norm": 8.159817965950001,
      "learning_rate": 6.239764896151713e-07,
      "loss": 0.5832,
      "step": 18923
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.824987391693191,
      "learning_rate": 6.236270861931909e-07,
      "loss": 0.5565,
      "step": 18924
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.389013804926853,
      "learning_rate": 6.23277774118411e-07,
      "loss": 0.5639,
      "step": 18925
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.717059472870761,
      "learning_rate": 6.229285533981228e-07,
      "loss": 0.514,
      "step": 18926
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.609567994480676,
      "learning_rate": 6.225794240396155e-07,
      "loss": 0.5433,
      "step": 18927
    },
    {
      "epoch": 1.69,
      "grad_norm": 9.00184275801144,
      "learning_rate": 6.222303860501771e-07,
      "loss": 0.5665,
      "step": 18928
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.702379104834645,
      "learning_rate": 6.218814394370915e-07,
      "loss": 0.5551,
      "step": 18929
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.792444230298871,
      "learning_rate": 6.215325842076425e-07,
      "loss": 0.519,
      "step": 18930
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.317182325276829,
      "learning_rate": 6.211838203691123e-07,
      "loss": 0.5645,
      "step": 18931
    },
    {
      "epoch": 1.69,
      "grad_norm": 7.698935367909271,
      "learning_rate": 6.208351479287805e-07,
      "loss": 0.5637,
      "step": 18932
    },
    {
      "epoch": 1.69,
      "grad_norm": 7.093805217477961,
      "learning_rate": 6.204865668939247e-07,
      "loss": 0.5755,
      "step": 18933
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.894567783421155,
      "learning_rate": 6.201380772718218e-07,
      "loss": 0.5622,
      "step": 18934
    },
    {
      "epoch": 1.69,
      "grad_norm": 8.033144682150617,
      "learning_rate": 6.197896790697439e-07,
      "loss": 0.5912,
      "step": 18935
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.00842784026854,
      "learning_rate": 6.194413722949649e-07,
      "loss": 0.5261,
      "step": 18936
    },
    {
      "epoch": 1.69,
      "grad_norm": 7.688505475149964,
      "learning_rate": 6.190931569547532e-07,
      "loss": 0.5735,
      "step": 18937
    },
    {
      "epoch": 1.69,
      "grad_norm": 8.5165943292433,
      "learning_rate": 6.187450330563777e-07,
      "loss": 0.5977,
      "step": 18938
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.678833741666529,
      "learning_rate": 6.183970006071049e-07,
      "loss": 0.5226,
      "step": 18939
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.357243303401968,
      "learning_rate": 6.180490596141992e-07,
      "loss": 0.5885,
      "step": 18940
    },
    {
      "epoch": 1.69,
      "grad_norm": 8.393318789680654,
      "learning_rate": 6.177012100849239e-07,
      "loss": 0.5508,
      "step": 18941
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.897529237829079,
      "learning_rate": 6.173534520265379e-07,
      "loss": 0.5704,
      "step": 18942
    },
    {
      "epoch": 1.69,
      "grad_norm": 8.069498098497583,
      "learning_rate": 6.170057854463002e-07,
      "loss": 0.5459,
      "step": 18943
    },
    {
      "epoch": 1.69,
      "grad_norm": 8.853617140536203,
      "learning_rate": 6.16658210351469e-07,
      "loss": 0.5531,
      "step": 18944
    },
    {
      "epoch": 1.69,
      "grad_norm": 7.30186853384495,
      "learning_rate": 6.163107267492974e-07,
      "loss": 0.551,
      "step": 18945
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.951803501283672,
      "learning_rate": 6.159633346470401e-07,
      "loss": 0.6103,
      "step": 18946
    },
    {
      "epoch": 1.69,
      "grad_norm": 7.993567274307509,
      "learning_rate": 6.156160340519479e-07,
      "loss": 0.6061,
      "step": 18947
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.750480489614243,
      "learning_rate": 6.152688249712685e-07,
      "loss": 0.5605,
      "step": 18948
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.740640642173404,
      "learning_rate": 6.149217074122499e-07,
      "loss": 0.5319,
      "step": 18949
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.724266978606212,
      "learning_rate": 6.145746813821374e-07,
      "loss": 0.5427,
      "step": 18950
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.413920940265159,
      "learning_rate": 6.142277468881758e-07,
      "loss": 0.5638,
      "step": 18951
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.670657278821217,
      "learning_rate": 6.138809039376042e-07,
      "loss": 0.5087,
      "step": 18952
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.607921597932446,
      "learning_rate": 6.135341525376631e-07,
      "loss": 0.5884,
      "step": 18953
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.566989064524504,
      "learning_rate": 6.131874926955917e-07,
      "loss": 0.5261,
      "step": 18954
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.840820572344754,
      "learning_rate": 6.128409244186228e-07,
      "loss": 0.563,
      "step": 18955
    },
    {
      "epoch": 1.69,
      "grad_norm": 8.83385485003084,
      "learning_rate": 6.124944477139927e-07,
      "loss": 0.5739,
      "step": 18956
    },
    {
      "epoch": 1.69,
      "grad_norm": 7.557325344300222,
      "learning_rate": 6.12148062588932e-07,
      "loss": 0.5326,
      "step": 18957
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.5211304099675855,
      "learning_rate": 6.118017690506717e-07,
      "loss": 0.5695,
      "step": 18958
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.479612808945009,
      "learning_rate": 6.114555671064388e-07,
      "loss": 0.5532,
      "step": 18959
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.606172282171865,
      "learning_rate": 6.111094567634618e-07,
      "loss": 0.5348,
      "step": 18960
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.869737179515736,
      "learning_rate": 6.107634380289624e-07,
      "loss": 0.5046,
      "step": 18961
    },
    {
      "epoch": 1.69,
      "grad_norm": 7.813271500961132,
      "learning_rate": 6.104175109101635e-07,
      "loss": 0.5368,
      "step": 18962
    },
    {
      "epoch": 1.69,
      "grad_norm": 7.09299052541404,
      "learning_rate": 6.10071675414286e-07,
      "loss": 0.5557,
      "step": 18963
    },
    {
      "epoch": 1.69,
      "grad_norm": 9.145060674427496,
      "learning_rate": 6.097259315485487e-07,
      "loss": 0.5439,
      "step": 18964
    },
    {
      "epoch": 1.69,
      "grad_norm": 8.416444108411651,
      "learning_rate": 6.093802793201692e-07,
      "loss": 0.535,
      "step": 18965
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.715219276674035,
      "learning_rate": 6.090347187363599e-07,
      "loss": 0.6482,
      "step": 18966
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.272779095841833,
      "learning_rate": 6.086892498043362e-07,
      "loss": 0.6281,
      "step": 18967
    },
    {
      "epoch": 1.69,
      "grad_norm": 9.729364301083235,
      "learning_rate": 6.083438725313063e-07,
      "loss": 0.5668,
      "step": 18968
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.7402057779917826,
      "learning_rate": 6.079985869244803e-07,
      "loss": 0.6069,
      "step": 18969
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.5990215912099055,
      "learning_rate": 6.076533929910655e-07,
      "loss": 0.5031,
      "step": 18970
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.239932938188158,
      "learning_rate": 6.073082907382677e-07,
      "loss": 0.5054,
      "step": 18971
    },
    {
      "epoch": 1.69,
      "grad_norm": 7.979842494624825,
      "learning_rate": 6.0696328017329e-07,
      "loss": 0.5722,
      "step": 18972
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.860412031989105,
      "learning_rate": 6.066183613033322e-07,
      "loss": 0.5885,
      "step": 18973
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.4122212320912775,
      "learning_rate": 6.062735341355947e-07,
      "loss": 0.6083,
      "step": 18974
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.900810240446636,
      "learning_rate": 6.059287986772749e-07,
      "loss": 0.562,
      "step": 18975
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.88898251353852,
      "learning_rate": 6.055841549355689e-07,
      "loss": 0.5718,
      "step": 18976
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.666640409884414,
      "learning_rate": 6.052396029176704e-07,
      "loss": 0.5684,
      "step": 18977
    },
    {
      "epoch": 1.69,
      "grad_norm": 7.761843215551754,
      "learning_rate": 6.048951426307714e-07,
      "loss": 0.5738,
      "step": 18978
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.4322337950541675,
      "learning_rate": 6.0455077408206e-07,
      "loss": 0.5696,
      "step": 18979
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.974515221183445,
      "learning_rate": 6.04206497278727e-07,
      "loss": 0.627,
      "step": 18980
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.759700016637714,
      "learning_rate": 6.038623122279552e-07,
      "loss": 0.5917,
      "step": 18981
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.275948403257921,
      "learning_rate": 6.035182189369304e-07,
      "loss": 0.601,
      "step": 18982
    },
    {
      "epoch": 1.69,
      "grad_norm": 8.022046387669423,
      "learning_rate": 6.031742174128346e-07,
      "loss": 0.6206,
      "step": 18983
    },
    {
      "epoch": 1.69,
      "grad_norm": 7.6544756360027275,
      "learning_rate": 6.028303076628478e-07,
      "loss": 0.611,
      "step": 18984
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.800875896048132,
      "learning_rate": 6.024864896941501e-07,
      "loss": 0.5519,
      "step": 18985
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.44689105881154,
      "learning_rate": 6.021427635139155e-07,
      "loss": 0.537,
      "step": 18986
    },
    {
      "epoch": 1.69,
      "grad_norm": 7.416079584644411,
      "learning_rate": 6.017991291293196e-07,
      "loss": 0.5582,
      "step": 18987
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.000490980293604,
      "learning_rate": 6.014555865475347e-07,
      "loss": 0.5233,
      "step": 18988
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.42536212664747,
      "learning_rate": 6.011121357757322e-07,
      "loss": 0.5126,
      "step": 18989
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.704908591751175,
      "learning_rate": 6.007687768210801e-07,
      "loss": 0.5824,
      "step": 18990
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.52248744774303,
      "learning_rate": 6.004255096907469e-07,
      "loss": 0.5803,
      "step": 18991
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.46517182669096,
      "learning_rate": 6.000823343918954e-07,
      "loss": 0.522,
      "step": 18992
    },
    {
      "epoch": 1.69,
      "grad_norm": 7.013918618622778,
      "learning_rate": 5.99739250931689e-07,
      "loss": 0.534,
      "step": 18993
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.003693457792655,
      "learning_rate": 5.993962593172908e-07,
      "loss": 0.5428,
      "step": 18994
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.898869674535669,
      "learning_rate": 5.990533595558573e-07,
      "loss": 0.544,
      "step": 18995
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.737435201134384,
      "learning_rate": 5.987105516545472e-07,
      "loss": 0.5581,
      "step": 18996
    },
    {
      "epoch": 1.69,
      "grad_norm": 7.303956481437001,
      "learning_rate": 5.983678356205152e-07,
      "loss": 0.5115,
      "step": 18997
    },
    {
      "epoch": 1.69,
      "grad_norm": 8.070635607130546,
      "learning_rate": 5.98025211460917e-07,
      "loss": 0.5834,
      "step": 18998
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.756638783192319,
      "learning_rate": 5.976826791829004e-07,
      "loss": 0.5955,
      "step": 18999
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.694859608618736,
      "learning_rate": 5.973402387936178e-07,
      "loss": 0.547,
      "step": 19000
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.759035837845706,
      "learning_rate": 5.969978903002155e-07,
      "loss": 0.5788,
      "step": 19001
    },
    {
      "epoch": 1.7,
      "grad_norm": 8.933164747964263,
      "learning_rate": 5.9665563370984e-07,
      "loss": 0.6176,
      "step": 19002
    },
    {
      "epoch": 1.7,
      "grad_norm": 8.284503643220216,
      "learning_rate": 5.963134690296352e-07,
      "loss": 0.5917,
      "step": 19003
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.9370573549582595,
      "learning_rate": 5.959713962667435e-07,
      "loss": 0.5778,
      "step": 19004
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.764708084715923,
      "learning_rate": 5.956294154283037e-07,
      "loss": 0.5429,
      "step": 19005
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.203372011396641,
      "learning_rate": 5.952875265214541e-07,
      "loss": 0.6001,
      "step": 19006
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.313506710314377,
      "learning_rate": 5.949457295533311e-07,
      "loss": 0.5295,
      "step": 19007
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.658479796555576,
      "learning_rate": 5.946040245310703e-07,
      "loss": 0.5692,
      "step": 19008
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.186767672643626,
      "learning_rate": 5.942624114618017e-07,
      "loss": 0.5273,
      "step": 19009
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.80295425633945,
      "learning_rate": 5.939208903526572e-07,
      "loss": 0.5312,
      "step": 19010
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.8452938131159184,
      "learning_rate": 5.935794612107659e-07,
      "loss": 0.5455,
      "step": 19011
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.202115531735215,
      "learning_rate": 5.932381240432522e-07,
      "loss": 0.5305,
      "step": 19012
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.49369238170101,
      "learning_rate": 5.92896878857242e-07,
      "loss": 0.4994,
      "step": 19013
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.47511886426814,
      "learning_rate": 5.92555725659858e-07,
      "loss": 0.5703,
      "step": 19014
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.109749053845435,
      "learning_rate": 5.922146644582216e-07,
      "loss": 0.5957,
      "step": 19015
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.016372156623817,
      "learning_rate": 5.918736952594506e-07,
      "loss": 0.5975,
      "step": 19016
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.685974251928895,
      "learning_rate": 5.915328180706636e-07,
      "loss": 0.5177,
      "step": 19017
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.118161984128971,
      "learning_rate": 5.91192032898974e-07,
      "loss": 0.5599,
      "step": 19018
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.309658708779394,
      "learning_rate": 5.908513397514959e-07,
      "loss": 0.5647,
      "step": 19019
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.8917316825747585,
      "learning_rate": 5.905107386353393e-07,
      "loss": 0.5799,
      "step": 19020
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.180850922694717,
      "learning_rate": 5.90170229557615e-07,
      "loss": 0.5549,
      "step": 19021
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.606988086157251,
      "learning_rate": 5.898298125254298e-07,
      "loss": 0.6065,
      "step": 19022
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.97249869673562,
      "learning_rate": 5.894894875458901e-07,
      "loss": 0.5268,
      "step": 19023
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.087853862983345,
      "learning_rate": 5.891492546260985e-07,
      "loss": 0.6073,
      "step": 19024
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.420801045899042,
      "learning_rate": 5.888091137731555e-07,
      "loss": 0.593,
      "step": 19025
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.227025187496682,
      "learning_rate": 5.884690649941616e-07,
      "loss": 0.5959,
      "step": 19026
    },
    {
      "epoch": 1.7,
      "grad_norm": 9.048602507664945,
      "learning_rate": 5.881291082962154e-07,
      "loss": 0.5659,
      "step": 19027
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.149548644332354,
      "learning_rate": 5.877892436864118e-07,
      "loss": 0.5707,
      "step": 19028
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.578912271562676,
      "learning_rate": 5.874494711718459e-07,
      "loss": 0.5857,
      "step": 19029
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.946419220638033,
      "learning_rate": 5.871097907596096e-07,
      "loss": 0.5555,
      "step": 19030
    },
    {
      "epoch": 1.7,
      "grad_norm": 5.0353785807975315,
      "learning_rate": 5.867702024567912e-07,
      "loss": 0.5684,
      "step": 19031
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.1007264288157215,
      "learning_rate": 5.8643070627048e-07,
      "loss": 0.6035,
      "step": 19032
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.118672684902679,
      "learning_rate": 5.860913022077619e-07,
      "loss": 0.5908,
      "step": 19033
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.655065444722704,
      "learning_rate": 5.857519902757219e-07,
      "loss": 0.5513,
      "step": 19034
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.28166942691206,
      "learning_rate": 5.854127704814422e-07,
      "loss": 0.5296,
      "step": 19035
    },
    {
      "epoch": 1.7,
      "grad_norm": 9.720324929351456,
      "learning_rate": 5.850736428320042e-07,
      "loss": 0.5704,
      "step": 19036
    },
    {
      "epoch": 1.7,
      "grad_norm": 9.213751621430028,
      "learning_rate": 5.847346073344856e-07,
      "loss": 0.5676,
      "step": 19037
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.707403114649427,
      "learning_rate": 5.843956639959614e-07,
      "loss": 0.5694,
      "step": 19038
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.4250369773567355,
      "learning_rate": 5.840568128235074e-07,
      "loss": 0.5885,
      "step": 19039
    },
    {
      "epoch": 1.7,
      "grad_norm": 10.053510846861775,
      "learning_rate": 5.837180538241971e-07,
      "loss": 0.597,
      "step": 19040
    },
    {
      "epoch": 1.7,
      "grad_norm": 8.282699263587892,
      "learning_rate": 5.833793870051013e-07,
      "loss": 0.6402,
      "step": 19041
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.005640149043132,
      "learning_rate": 5.830408123732878e-07,
      "loss": 0.5404,
      "step": 19042
    },
    {
      "epoch": 1.7,
      "grad_norm": 5.278164216070422,
      "learning_rate": 5.827023299358259e-07,
      "loss": 0.6239,
      "step": 19043
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.456627154517253,
      "learning_rate": 5.823639396997783e-07,
      "loss": 0.6,
      "step": 19044
    },
    {
      "epoch": 1.7,
      "grad_norm": 5.628871268303757,
      "learning_rate": 5.820256416722081e-07,
      "loss": 0.5716,
      "step": 19045
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.227678145516003,
      "learning_rate": 5.816874358601782e-07,
      "loss": 0.5937,
      "step": 19046
    },
    {
      "epoch": 1.7,
      "grad_norm": 5.114723011126362,
      "learning_rate": 5.813493222707467e-07,
      "loss": 0.5451,
      "step": 19047
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.284074729684461,
      "learning_rate": 5.810113009109731e-07,
      "loss": 0.5732,
      "step": 19048
    },
    {
      "epoch": 1.7,
      "grad_norm": 5.9428137553330656,
      "learning_rate": 5.806733717879093e-07,
      "loss": 0.5449,
      "step": 19049
    },
    {
      "epoch": 1.7,
      "grad_norm": 8.545216563143326,
      "learning_rate": 5.803355349086115e-07,
      "loss": 0.5696,
      "step": 19050
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.288598720501472,
      "learning_rate": 5.799977902801312e-07,
      "loss": 0.6041,
      "step": 19051
    },
    {
      "epoch": 1.7,
      "grad_norm": 10.0338043242748,
      "learning_rate": 5.796601379095163e-07,
      "loss": 0.6047,
      "step": 19052
    },
    {
      "epoch": 1.7,
      "grad_norm": 5.374045580466187,
      "learning_rate": 5.793225778038153e-07,
      "loss": 0.5888,
      "step": 19053
    },
    {
      "epoch": 1.7,
      "grad_norm": 5.696939849381014,
      "learning_rate": 5.789851099700749e-07,
      "loss": 0.6383,
      "step": 19054
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.774685792968355,
      "learning_rate": 5.786477344153385e-07,
      "loss": 0.5209,
      "step": 19055
    },
    {
      "epoch": 1.7,
      "grad_norm": 8.33623400505331,
      "learning_rate": 5.783104511466487e-07,
      "loss": 0.6197,
      "step": 19056
    },
    {
      "epoch": 1.7,
      "grad_norm": 8.330830016985539,
      "learning_rate": 5.779732601710441e-07,
      "loss": 0.572,
      "step": 19057
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.237326240761746,
      "learning_rate": 5.776361614955634e-07,
      "loss": 0.5838,
      "step": 19058
    },
    {
      "epoch": 1.7,
      "grad_norm": 5.410128964517143,
      "learning_rate": 5.772991551272434e-07,
      "loss": 0.5255,
      "step": 19059
    },
    {
      "epoch": 1.7,
      "grad_norm": 8.526514380422007,
      "learning_rate": 5.769622410731173e-07,
      "loss": 0.4987,
      "step": 19060
    },
    {
      "epoch": 1.7,
      "grad_norm": 8.39842176702046,
      "learning_rate": 5.766254193402199e-07,
      "loss": 0.5939,
      "step": 19061
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.9095607285011615,
      "learning_rate": 5.762886899355785e-07,
      "loss": 0.5903,
      "step": 19062
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.96048259313156,
      "learning_rate": 5.759520528662232e-07,
      "loss": 0.5155,
      "step": 19063
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.1020959017181955,
      "learning_rate": 5.756155081391801e-07,
      "loss": 0.6229,
      "step": 19064
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.335859893232375,
      "learning_rate": 5.75279055761474e-07,
      "loss": 0.5208,
      "step": 19065
    },
    {
      "epoch": 1.7,
      "grad_norm": 8.371622693259138,
      "learning_rate": 5.749426957401289e-07,
      "loss": 0.5767,
      "step": 19066
    },
    {
      "epoch": 1.7,
      "grad_norm": 5.882995764122588,
      "learning_rate": 5.746064280821628e-07,
      "loss": 0.5616,
      "step": 19067
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.0930424802961936,
      "learning_rate": 5.74270252794597e-07,
      "loss": 0.6733,
      "step": 19068
    },
    {
      "epoch": 1.7,
      "grad_norm": 5.829926063995253,
      "learning_rate": 5.739341698844469e-07,
      "loss": 0.5871,
      "step": 19069
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.346167663094094,
      "learning_rate": 5.73598179358728e-07,
      "loss": 0.5286,
      "step": 19070
    },
    {
      "epoch": 1.7,
      "grad_norm": 5.419961296831708,
      "learning_rate": 5.732622812244537e-07,
      "loss": 0.5949,
      "step": 19071
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.558881528836285,
      "learning_rate": 5.729264754886349e-07,
      "loss": 0.5405,
      "step": 19072
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.13152413502468,
      "learning_rate": 5.725907621582805e-07,
      "loss": 0.5652,
      "step": 19073
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.914233366666989,
      "learning_rate": 5.722551412403987e-07,
      "loss": 0.627,
      "step": 19074
    },
    {
      "epoch": 1.7,
      "grad_norm": 9.440617423030988,
      "learning_rate": 5.719196127419941e-07,
      "loss": 0.6581,
      "step": 19075
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.444004010168954,
      "learning_rate": 5.715841766700697e-07,
      "loss": 0.5526,
      "step": 19076
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.679297435636929,
      "learning_rate": 5.712488330316274e-07,
      "loss": 0.5817,
      "step": 19077
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.325270157643071,
      "learning_rate": 5.709135818336675e-07,
      "loss": 0.5483,
      "step": 19078
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.796893557436882,
      "learning_rate": 5.705784230831862e-07,
      "loss": 0.5223,
      "step": 19079
    },
    {
      "epoch": 1.7,
      "grad_norm": 5.249236754851243,
      "learning_rate": 5.702433567871818e-07,
      "loss": 0.5322,
      "step": 19080
    },
    {
      "epoch": 1.7,
      "grad_norm": 5.490670248649101,
      "learning_rate": 5.699083829526464e-07,
      "loss": 0.5513,
      "step": 19081
    },
    {
      "epoch": 1.7,
      "grad_norm": 5.181595580610327,
      "learning_rate": 5.6957350158657e-07,
      "loss": 0.5493,
      "step": 19082
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.223291196004908,
      "learning_rate": 5.692387126959447e-07,
      "loss": 0.5633,
      "step": 19083
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.516392319790798,
      "learning_rate": 5.689040162877585e-07,
      "loss": 0.5593,
      "step": 19084
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.554175770037426,
      "learning_rate": 5.685694123689966e-07,
      "loss": 0.5319,
      "step": 19085
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.004001563774497,
      "learning_rate": 5.682349009466437e-07,
      "loss": 0.52,
      "step": 19086
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.434594190870763,
      "learning_rate": 5.679004820276824e-07,
      "loss": 0.6095,
      "step": 19087
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.508152782857549,
      "learning_rate": 5.67566155619092e-07,
      "loss": 0.5507,
      "step": 19088
    },
    {
      "epoch": 1.7,
      "grad_norm": 9.709586457668753,
      "learning_rate": 5.67231921727851e-07,
      "loss": 0.5345,
      "step": 19089
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.425309453709852,
      "learning_rate": 5.66897780360936e-07,
      "loss": 0.6127,
      "step": 19090
    },
    {
      "epoch": 1.7,
      "grad_norm": 5.876887345335138,
      "learning_rate": 5.665637315253214e-07,
      "loss": 0.5789,
      "step": 19091
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.415390748097458,
      "learning_rate": 5.6622977522798e-07,
      "loss": 0.5207,
      "step": 19092
    },
    {
      "epoch": 1.7,
      "grad_norm": 9.676761435104092,
      "learning_rate": 5.658959114758833e-07,
      "loss": 0.6297,
      "step": 19093
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.99334780251559,
      "learning_rate": 5.655621402759976e-07,
      "loss": 0.6306,
      "step": 19094
    },
    {
      "epoch": 1.7,
      "grad_norm": 10.628010178878265,
      "learning_rate": 5.652284616352921e-07,
      "loss": 0.597,
      "step": 19095
    },
    {
      "epoch": 1.7,
      "grad_norm": 8.446380783433563,
      "learning_rate": 5.648948755607297e-07,
      "loss": 0.5833,
      "step": 19096
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.1836746926029695,
      "learning_rate": 5.645613820592744e-07,
      "loss": 0.582,
      "step": 19097
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.038312623744834,
      "learning_rate": 5.64227981137886e-07,
      "loss": 0.6372,
      "step": 19098
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.468294230075106,
      "learning_rate": 5.63894672803525e-07,
      "loss": 0.5325,
      "step": 19099
    },
    {
      "epoch": 1.7,
      "grad_norm": 8.26228035471566,
      "learning_rate": 5.635614570631481e-07,
      "loss": 0.5398,
      "step": 19100
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.318730178731674,
      "learning_rate": 5.632283339237094e-07,
      "loss": 0.6522,
      "step": 19101
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.574191265879604,
      "learning_rate": 5.628953033921631e-07,
      "loss": 0.5473,
      "step": 19102
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.006764740308048,
      "learning_rate": 5.625623654754603e-07,
      "loss": 0.53,
      "step": 19103
    },
    {
      "epoch": 1.7,
      "grad_norm": 8.843452390600442,
      "learning_rate": 5.6222952018055e-07,
      "loss": 0.6279,
      "step": 19104
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.566153395420582,
      "learning_rate": 5.618967675143799e-07,
      "loss": 0.5443,
      "step": 19105
    },
    {
      "epoch": 1.7,
      "grad_norm": 8.095969023942317,
      "learning_rate": 5.615641074838968e-07,
      "loss": 0.5824,
      "step": 19106
    },
    {
      "epoch": 1.7,
      "grad_norm": 5.789883465374371,
      "learning_rate": 5.612315400960417e-07,
      "loss": 0.591,
      "step": 19107
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.566956511066094,
      "learning_rate": 5.608990653577578e-07,
      "loss": 0.5376,
      "step": 19108
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.1867502166602355,
      "learning_rate": 5.605666832759849e-07,
      "loss": 0.5229,
      "step": 19109
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.72448010142811,
      "learning_rate": 5.602343938576599e-07,
      "loss": 0.5298,
      "step": 19110
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.075153092149934,
      "learning_rate": 5.599021971097185e-07,
      "loss": 0.5406,
      "step": 19111
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.389073607069378,
      "learning_rate": 5.595700930390952e-07,
      "loss": 0.5709,
      "step": 19112
    },
    {
      "epoch": 1.71,
      "grad_norm": 9.733563550270771,
      "learning_rate": 5.592380816527232e-07,
      "loss": 0.5816,
      "step": 19113
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.031636149005233,
      "learning_rate": 5.589061629575298e-07,
      "loss": 0.5061,
      "step": 19114
    },
    {
      "epoch": 1.71,
      "grad_norm": 9.553733823258716,
      "learning_rate": 5.585743369604446e-07,
      "loss": 0.5967,
      "step": 19115
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.904688821346027,
      "learning_rate": 5.58242603668393e-07,
      "loss": 0.5867,
      "step": 19116
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.007371172734647,
      "learning_rate": 5.579109630883006e-07,
      "loss": 0.5959,
      "step": 19117
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.704403188457832,
      "learning_rate": 5.57579415227088e-07,
      "loss": 0.5724,
      "step": 19118
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.736607442586065,
      "learning_rate": 5.57247960091678e-07,
      "loss": 0.5904,
      "step": 19119
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.4250227416406025,
      "learning_rate": 5.569165976889862e-07,
      "loss": 0.5725,
      "step": 19120
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.4791677676016555,
      "learning_rate": 5.565853280259304e-07,
      "loss": 0.5811,
      "step": 19121
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.797238458010775,
      "learning_rate": 5.562541511094249e-07,
      "loss": 0.5966,
      "step": 19122
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.755331007286548,
      "learning_rate": 5.559230669463823e-07,
      "loss": 0.5501,
      "step": 19123
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.97121304661432,
      "learning_rate": 5.555920755437138e-07,
      "loss": 0.6285,
      "step": 19124
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.790935184549811,
      "learning_rate": 5.552611769083271e-07,
      "loss": 0.654,
      "step": 19125
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.046526389961184,
      "learning_rate": 5.54930371047131e-07,
      "loss": 0.548,
      "step": 19126
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.290750568409491,
      "learning_rate": 5.545996579670271e-07,
      "loss": 0.555,
      "step": 19127
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.709688716628476,
      "learning_rate": 5.542690376749205e-07,
      "loss": 0.583,
      "step": 19128
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.762497765200341,
      "learning_rate": 5.539385101777117e-07,
      "loss": 0.5282,
      "step": 19129
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.416600369855901,
      "learning_rate": 5.536080754822992e-07,
      "loss": 0.5762,
      "step": 19130
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.637230739764618,
      "learning_rate": 5.532777335955813e-07,
      "loss": 0.5341,
      "step": 19131
    },
    {
      "epoch": 1.71,
      "grad_norm": 11.934854255215145,
      "learning_rate": 5.529474845244537e-07,
      "loss": 0.6211,
      "step": 19132
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.904980820546287,
      "learning_rate": 5.526173282758074e-07,
      "loss": 0.5581,
      "step": 19133
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.565211822897581,
      "learning_rate": 5.522872648565342e-07,
      "loss": 0.5676,
      "step": 19134
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.388129333356261,
      "learning_rate": 5.519572942735246e-07,
      "loss": 0.5533,
      "step": 19135
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.86501650203383,
      "learning_rate": 5.516274165336655e-07,
      "loss": 0.5577,
      "step": 19136
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.8999527947595425,
      "learning_rate": 5.512976316438429e-07,
      "loss": 0.5973,
      "step": 19137
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.151992448064016,
      "learning_rate": 5.509679396109391e-07,
      "loss": 0.5131,
      "step": 19138
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.418728025390095,
      "learning_rate": 5.50638340441837e-07,
      "loss": 0.5679,
      "step": 19139
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.582562648639339,
      "learning_rate": 5.503088341434149e-07,
      "loss": 0.6134,
      "step": 19140
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.442655526492507,
      "learning_rate": 5.499794207225512e-07,
      "loss": 0.5896,
      "step": 19141
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.140845742317754,
      "learning_rate": 5.496501001861216e-07,
      "loss": 0.5971,
      "step": 19142
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.825858284010325,
      "learning_rate": 5.493208725409998e-07,
      "loss": 0.6031,
      "step": 19143
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.419136127346111,
      "learning_rate": 5.48991737794059e-07,
      "loss": 0.5239,
      "step": 19144
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.009096731450402,
      "learning_rate": 5.486626959521673e-07,
      "loss": 0.584,
      "step": 19145
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.750851383579592,
      "learning_rate": 5.483337470221939e-07,
      "loss": 0.5791,
      "step": 19146
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.462728763440174,
      "learning_rate": 5.480048910110042e-07,
      "loss": 0.5149,
      "step": 19147
    },
    {
      "epoch": 1.71,
      "grad_norm": 8.899297435270581,
      "learning_rate": 5.476761279254628e-07,
      "loss": 0.555,
      "step": 19148
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.2320844764559755,
      "learning_rate": 5.473474577724319e-07,
      "loss": 0.6011,
      "step": 19149
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.955528713722872,
      "learning_rate": 5.470188805587723e-07,
      "loss": 0.5677,
      "step": 19150
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.91801570129733,
      "learning_rate": 5.466903962913411e-07,
      "loss": 0.6123,
      "step": 19151
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.009030119820712,
      "learning_rate": 5.463620049769963e-07,
      "loss": 0.5489,
      "step": 19152
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.290965910772599,
      "learning_rate": 5.460337066225901e-07,
      "loss": 0.5082,
      "step": 19153
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.113003648129236,
      "learning_rate": 5.457055012349765e-07,
      "loss": 0.5617,
      "step": 19154
    },
    {
      "epoch": 1.71,
      "grad_norm": 8.574070506135433,
      "learning_rate": 5.453773888210057e-07,
      "loss": 0.5764,
      "step": 19155
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.451639457273505,
      "learning_rate": 5.450493693875264e-07,
      "loss": 0.5806,
      "step": 19156
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.519702382369728,
      "learning_rate": 5.447214429413861e-07,
      "loss": 0.5856,
      "step": 19157
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.265161181041208,
      "learning_rate": 5.443936094894281e-07,
      "loss": 0.5533,
      "step": 19158
    },
    {
      "epoch": 1.71,
      "grad_norm": 8.33210166084681,
      "learning_rate": 5.440658690384959e-07,
      "loss": 0.6145,
      "step": 19159
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.113219641739811,
      "learning_rate": 5.437382215954306e-07,
      "loss": 0.5508,
      "step": 19160
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.599609456544699,
      "learning_rate": 5.434106671670703e-07,
      "loss": 0.6276,
      "step": 19161
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.173456762507223,
      "learning_rate": 5.43083205760253e-07,
      "loss": 0.6255,
      "step": 19162
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.535875704701766,
      "learning_rate": 5.427558373818143e-07,
      "loss": 0.4989,
      "step": 19163
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.702937746010232,
      "learning_rate": 5.424285620385849e-07,
      "loss": 0.5529,
      "step": 19164
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.425188713686808,
      "learning_rate": 5.421013797373975e-07,
      "loss": 0.5633,
      "step": 19165
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.078964683990989,
      "learning_rate": 5.417742904850814e-07,
      "loss": 0.5862,
      "step": 19166
    },
    {
      "epoch": 1.71,
      "grad_norm": 9.030208770180169,
      "learning_rate": 5.414472942884646e-07,
      "loss": 0.5993,
      "step": 19167
    },
    {
      "epoch": 1.71,
      "grad_norm": 4.879746547300199,
      "learning_rate": 5.4112039115437e-07,
      "loss": 0.5657,
      "step": 19168
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.798628588867052,
      "learning_rate": 5.407935810896231e-07,
      "loss": 0.5903,
      "step": 19169
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.9536106363533685,
      "learning_rate": 5.404668641010452e-07,
      "loss": 0.5284,
      "step": 19170
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.252859025058898,
      "learning_rate": 5.401402401954547e-07,
      "loss": 0.6191,
      "step": 19171
    },
    {
      "epoch": 1.71,
      "grad_norm": 8.175062042890385,
      "learning_rate": 5.398137093796696e-07,
      "loss": 0.6105,
      "step": 19172
    },
    {
      "epoch": 1.71,
      "grad_norm": 8.300608920699718,
      "learning_rate": 5.394872716605054e-07,
      "loss": 0.5959,
      "step": 19173
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.089268615124478,
      "learning_rate": 5.391609270447762e-07,
      "loss": 0.5217,
      "step": 19174
    },
    {
      "epoch": 1.71,
      "grad_norm": 11.894864624684013,
      "learning_rate": 5.388346755392937e-07,
      "loss": 0.6596,
      "step": 19175
    },
    {
      "epoch": 1.71,
      "grad_norm": 8.940059194279009,
      "learning_rate": 5.385085171508686e-07,
      "loss": 0.5618,
      "step": 19176
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.026658441908773,
      "learning_rate": 5.381824518863065e-07,
      "loss": 0.5444,
      "step": 19177
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.892760094991491,
      "learning_rate": 5.37856479752415e-07,
      "loss": 0.5662,
      "step": 19178
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.433885492003733,
      "learning_rate": 5.375306007559972e-07,
      "loss": 0.5552,
      "step": 19179
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.796712401379719,
      "learning_rate": 5.372048149038555e-07,
      "loss": 0.6153,
      "step": 19180
    },
    {
      "epoch": 1.71,
      "grad_norm": 8.834509506144384,
      "learning_rate": 5.368791222027908e-07,
      "loss": 0.5705,
      "step": 19181
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.409687304330135,
      "learning_rate": 5.365535226595997e-07,
      "loss": 0.6201,
      "step": 19182
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.642394786767026,
      "learning_rate": 5.362280162810796e-07,
      "loss": 0.524,
      "step": 19183
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.092196893232274,
      "learning_rate": 5.359026030740233e-07,
      "loss": 0.5633,
      "step": 19184
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.441155030383824,
      "learning_rate": 5.355772830452238e-07,
      "loss": 0.6022,
      "step": 19185
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.331772272716986,
      "learning_rate": 5.352520562014718e-07,
      "loss": 0.5838,
      "step": 19186
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.232479437236316,
      "learning_rate": 5.349269225495557e-07,
      "loss": 0.5508,
      "step": 19187
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.888347364187809,
      "learning_rate": 5.346018820962618e-07,
      "loss": 0.5725,
      "step": 19188
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.751445092278158,
      "learning_rate": 5.342769348483751e-07,
      "loss": 0.5536,
      "step": 19189
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.252832950346197,
      "learning_rate": 5.339520808126769e-07,
      "loss": 0.5771,
      "step": 19190
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.955396944115055,
      "learning_rate": 5.336273199959485e-07,
      "loss": 0.5503,
      "step": 19191
    },
    {
      "epoch": 1.71,
      "grad_norm": 4.223998712183525,
      "learning_rate": 5.333026524049684e-07,
      "loss": 0.5458,
      "step": 19192
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.139431863318768,
      "learning_rate": 5.329780780465138e-07,
      "loss": 0.6049,
      "step": 19193
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.885157642608988,
      "learning_rate": 5.326535969273589e-07,
      "loss": 0.5597,
      "step": 19194
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.823111663416945,
      "learning_rate": 5.323292090542781e-07,
      "loss": 0.53,
      "step": 19195
    },
    {
      "epoch": 1.71,
      "grad_norm": 8.312694344726015,
      "learning_rate": 5.320049144340412e-07,
      "loss": 0.5531,
      "step": 19196
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.656179080445823,
      "learning_rate": 5.316807130734159e-07,
      "loss": 0.5864,
      "step": 19197
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.059279721013413,
      "learning_rate": 5.313566049791696e-07,
      "loss": 0.5776,
      "step": 19198
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.722062718843531,
      "learning_rate": 5.310325901580688e-07,
      "loss": 0.5328,
      "step": 19199
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.168292104870916,
      "learning_rate": 5.307086686168755e-07,
      "loss": 0.4644,
      "step": 19200
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.040838279113908,
      "learning_rate": 5.303848403623513e-07,
      "loss": 0.5143,
      "step": 19201
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.005828551995618,
      "learning_rate": 5.300611054012561e-07,
      "loss": 0.5811,
      "step": 19202
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.812722546965414,
      "learning_rate": 5.297374637403452e-07,
      "loss": 0.5781,
      "step": 19203
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.187542261477316,
      "learning_rate": 5.294139153863753e-07,
      "loss": 0.5623,
      "step": 19204
    },
    {
      "epoch": 1.71,
      "grad_norm": 10.211128732153158,
      "learning_rate": 5.290904603460994e-07,
      "loss": 0.5512,
      "step": 19205
    },
    {
      "epoch": 1.71,
      "grad_norm": 8.185308293820485,
      "learning_rate": 5.287670986262688e-07,
      "loss": 0.5673,
      "step": 19206
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.976479590876485,
      "learning_rate": 5.284438302336337e-07,
      "loss": 0.5615,
      "step": 19207
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.087104575117499,
      "learning_rate": 5.281206551749418e-07,
      "loss": 0.5581,
      "step": 19208
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.046604037626172,
      "learning_rate": 5.277975734569368e-07,
      "loss": 0.5404,
      "step": 19209
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.491873171342029,
      "learning_rate": 5.274745850863649e-07,
      "loss": 0.5453,
      "step": 19210
    },
    {
      "epoch": 1.71,
      "grad_norm": 8.57653264841491,
      "learning_rate": 5.271516900699652e-07,
      "loss": 0.5459,
      "step": 19211
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.18137210299237,
      "learning_rate": 5.268288884144785e-07,
      "loss": 0.5669,
      "step": 19212
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.846701873577875,
      "learning_rate": 5.265061801266425e-07,
      "loss": 0.5903,
      "step": 19213
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.232098063361468,
      "learning_rate": 5.261835652131936e-07,
      "loss": 0.5854,
      "step": 19214
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.070263309547352,
      "learning_rate": 5.258610436808664e-07,
      "loss": 0.5333,
      "step": 19215
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.059854945063307,
      "learning_rate": 5.255386155363901e-07,
      "loss": 0.5016,
      "step": 19216
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.215055142101373,
      "learning_rate": 5.252162807864969e-07,
      "loss": 0.5418,
      "step": 19217
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.61150899569979,
      "learning_rate": 5.248940394379137e-07,
      "loss": 0.633,
      "step": 19218
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.619094490131859,
      "learning_rate": 5.245718914973674e-07,
      "loss": 0.5335,
      "step": 19219
    },
    {
      "epoch": 1.71,
      "grad_norm": 9.826982799291908,
      "learning_rate": 5.242498369715831e-07,
      "loss": 0.5781,
      "step": 19220
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.313756250538101,
      "learning_rate": 5.239278758672805e-07,
      "loss": 0.55,
      "step": 19221
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.1443252251553435,
      "learning_rate": 5.236060081911809e-07,
      "loss": 0.6054,
      "step": 19222
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.491568705036574,
      "learning_rate": 5.232842339500027e-07,
      "loss": 0.5601,
      "step": 19223
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.254470011456196,
      "learning_rate": 5.229625531504623e-07,
      "loss": 0.6025,
      "step": 19224
    },
    {
      "epoch": 1.72,
      "grad_norm": 7.276779902402494,
      "learning_rate": 5.226409657992754e-07,
      "loss": 0.5332,
      "step": 19225
    },
    {
      "epoch": 1.72,
      "grad_norm": 5.451033844015195,
      "learning_rate": 5.223194719031521e-07,
      "loss": 0.6382,
      "step": 19226
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.819546215949375,
      "learning_rate": 5.219980714688044e-07,
      "loss": 0.5366,
      "step": 19227
    },
    {
      "epoch": 1.72,
      "grad_norm": 8.559681168152265,
      "learning_rate": 5.216767645029397e-07,
      "loss": 0.5884,
      "step": 19228
    },
    {
      "epoch": 1.72,
      "grad_norm": 3.9819397891199317,
      "learning_rate": 5.213555510122647e-07,
      "loss": 0.5789,
      "step": 19229
    },
    {
      "epoch": 1.72,
      "grad_norm": 5.7554202294410475,
      "learning_rate": 5.210344310034848e-07,
      "loss": 0.5707,
      "step": 19230
    },
    {
      "epoch": 1.72,
      "grad_norm": 12.005537028557345,
      "learning_rate": 5.207134044833029e-07,
      "loss": 0.557,
      "step": 19231
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.530122962686295,
      "learning_rate": 5.203924714584185e-07,
      "loss": 0.5615,
      "step": 19232
    },
    {
      "epoch": 1.72,
      "grad_norm": 8.10237288132925,
      "learning_rate": 5.200716319355325e-07,
      "loss": 0.5823,
      "step": 19233
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.677227181925658,
      "learning_rate": 5.19750885921339e-07,
      "loss": 0.5635,
      "step": 19234
    },
    {
      "epoch": 1.72,
      "grad_norm": 7.063541718061425,
      "learning_rate": 5.194302334225343e-07,
      "loss": 0.5412,
      "step": 19235
    },
    {
      "epoch": 1.72,
      "grad_norm": 8.001249229832037,
      "learning_rate": 5.191096744458113e-07,
      "loss": 0.5943,
      "step": 19236
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.652211928060776,
      "learning_rate": 5.187892089978608e-07,
      "loss": 0.5794,
      "step": 19237
    },
    {
      "epoch": 1.72,
      "grad_norm": 5.676748666271144,
      "learning_rate": 5.184688370853718e-07,
      "loss": 0.5901,
      "step": 19238
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.9459094465182725,
      "learning_rate": 5.181485587150325e-07,
      "loss": 0.533,
      "step": 19239
    },
    {
      "epoch": 1.72,
      "grad_norm": 7.11192665808139,
      "learning_rate": 5.17828373893527e-07,
      "loss": 0.6778,
      "step": 19240
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.8749805723127135,
      "learning_rate": 5.175082826275374e-07,
      "loss": 0.5864,
      "step": 19241
    },
    {
      "epoch": 1.72,
      "grad_norm": 5.185314423315643,
      "learning_rate": 5.171882849237458e-07,
      "loss": 0.509,
      "step": 19242
    },
    {
      "epoch": 1.72,
      "grad_norm": 7.320260007084134,
      "learning_rate": 5.168683807888313e-07,
      "loss": 0.5879,
      "step": 19243
    },
    {
      "epoch": 1.72,
      "grad_norm": 8.703746557947426,
      "learning_rate": 5.16548570229472e-07,
      "loss": 0.6064,
      "step": 19244
    },
    {
      "epoch": 1.72,
      "grad_norm": 5.479373867619299,
      "learning_rate": 5.162288532523424e-07,
      "loss": 0.528,
      "step": 19245
    },
    {
      "epoch": 1.72,
      "grad_norm": 7.101415704045613,
      "learning_rate": 5.159092298641172e-07,
      "loss": 0.542,
      "step": 19246
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.830933680376042,
      "learning_rate": 5.155897000714661e-07,
      "loss": 0.577,
      "step": 19247
    },
    {
      "epoch": 1.72,
      "grad_norm": 8.47981837693573,
      "learning_rate": 5.152702638810592e-07,
      "loss": 0.5774,
      "step": 19248
    },
    {
      "epoch": 1.72,
      "grad_norm": 8.057985776136146,
      "learning_rate": 5.14950921299564e-07,
      "loss": 0.6631,
      "step": 19249
    },
    {
      "epoch": 1.72,
      "grad_norm": 8.81079428994409,
      "learning_rate": 5.146316723336464e-07,
      "loss": 0.6008,
      "step": 19250
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.8327135673969215,
      "learning_rate": 5.143125169899693e-07,
      "loss": 0.6312,
      "step": 19251
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.757887755134997,
      "learning_rate": 5.139934552751968e-07,
      "loss": 0.5292,
      "step": 19252
    },
    {
      "epoch": 1.72,
      "grad_norm": 7.823177478938973,
      "learning_rate": 5.136744871959859e-07,
      "loss": 0.5955,
      "step": 19253
    },
    {
      "epoch": 1.72,
      "grad_norm": 7.447161056666597,
      "learning_rate": 5.13355612758995e-07,
      "loss": 0.5021,
      "step": 19254
    },
    {
      "epoch": 1.72,
      "grad_norm": 7.0682793689157535,
      "learning_rate": 5.130368319708795e-07,
      "loss": 0.6167,
      "step": 19255
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.053681931780594,
      "learning_rate": 5.127181448382939e-07,
      "loss": 0.5193,
      "step": 19256
    },
    {
      "epoch": 1.72,
      "grad_norm": 7.09226502247413,
      "learning_rate": 5.123995513678904e-07,
      "loss": 0.5567,
      "step": 19257
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.127944996147868,
      "learning_rate": 5.120810515663183e-07,
      "loss": 0.5154,
      "step": 19258
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.101485029334655,
      "learning_rate": 5.117626454402269e-07,
      "loss": 0.5817,
      "step": 19259
    },
    {
      "epoch": 1.72,
      "grad_norm": 8.42937171722228,
      "learning_rate": 5.114443329962604e-07,
      "loss": 0.6305,
      "step": 19260
    },
    {
      "epoch": 1.72,
      "grad_norm": 7.254502111937606,
      "learning_rate": 5.111261142410634e-07,
      "loss": 0.5298,
      "step": 19261
    },
    {
      "epoch": 1.72,
      "grad_norm": 8.340051633725265,
      "learning_rate": 5.108079891812789e-07,
      "loss": 0.6497,
      "step": 19262
    },
    {
      "epoch": 1.72,
      "grad_norm": 4.982442249594672,
      "learning_rate": 5.104899578235456e-07,
      "loss": 0.613,
      "step": 19263
    },
    {
      "epoch": 1.72,
      "grad_norm": 7.09478956960515,
      "learning_rate": 5.10172020174503e-07,
      "loss": 0.5538,
      "step": 19264
    },
    {
      "epoch": 1.72,
      "grad_norm": 9.077518853677551,
      "learning_rate": 5.098541762407883e-07,
      "loss": 0.6039,
      "step": 19265
    },
    {
      "epoch": 1.72,
      "grad_norm": 7.3261375659633385,
      "learning_rate": 5.095364260290331e-07,
      "loss": 0.6406,
      "step": 19266
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.724192105780078,
      "learning_rate": 5.092187695458705e-07,
      "loss": 0.5322,
      "step": 19267
    },
    {
      "epoch": 1.72,
      "grad_norm": 7.471984082158057,
      "learning_rate": 5.089012067979332e-07,
      "loss": 0.6028,
      "step": 19268
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.973116508262568,
      "learning_rate": 5.085837377918468e-07,
      "loss": 0.61,
      "step": 19269
    },
    {
      "epoch": 1.72,
      "grad_norm": 5.76843553253817,
      "learning_rate": 5.082663625342382e-07,
      "loss": 0.5394,
      "step": 19270
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.207002145099065,
      "learning_rate": 5.079490810317333e-07,
      "loss": 0.5868,
      "step": 19271
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.9599250694449,
      "learning_rate": 5.076318932909546e-07,
      "loss": 0.5698,
      "step": 19272
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.713047570783669,
      "learning_rate": 5.073147993185207e-07,
      "loss": 0.6016,
      "step": 19273
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.663640998040794,
      "learning_rate": 5.069977991210523e-07,
      "loss": 0.6446,
      "step": 19274
    },
    {
      "epoch": 1.72,
      "grad_norm": 7.893089910157167,
      "learning_rate": 5.066808927051648e-07,
      "loss": 0.5509,
      "step": 19275
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.651522889505811,
      "learning_rate": 5.063640800774733e-07,
      "loss": 0.5685,
      "step": 19276
    },
    {
      "epoch": 1.72,
      "grad_norm": 8.239088395961359,
      "learning_rate": 5.06047361244591e-07,
      "loss": 0.5212,
      "step": 19277
    },
    {
      "epoch": 1.72,
      "grad_norm": 5.474865591286788,
      "learning_rate": 5.057307362131298e-07,
      "loss": 0.4893,
      "step": 19278
    },
    {
      "epoch": 1.72,
      "grad_norm": 7.838493472446169,
      "learning_rate": 5.054142049896955e-07,
      "loss": 0.487,
      "step": 19279
    },
    {
      "epoch": 1.72,
      "grad_norm": 5.181205621344234,
      "learning_rate": 5.050977675808971e-07,
      "loss": 0.5561,
      "step": 19280
    },
    {
      "epoch": 1.72,
      "grad_norm": 8.283675570722995,
      "learning_rate": 5.047814239933391e-07,
      "loss": 0.6018,
      "step": 19281
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.934600160361875,
      "learning_rate": 5.044651742336254e-07,
      "loss": 0.6249,
      "step": 19282
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.475814835443274,
      "learning_rate": 5.041490183083553e-07,
      "loss": 0.5512,
      "step": 19283
    },
    {
      "epoch": 1.72,
      "grad_norm": 9.062724185903754,
      "learning_rate": 5.038329562241285e-07,
      "loss": 0.6022,
      "step": 19284
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.086858011102138,
      "learning_rate": 5.035169879875435e-07,
      "loss": 0.5679,
      "step": 19285
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.122389201808594,
      "learning_rate": 5.032011136051934e-07,
      "loss": 0.5233,
      "step": 19286
    },
    {
      "epoch": 1.72,
      "grad_norm": 9.805357820931205,
      "learning_rate": 5.02885333083672e-07,
      "loss": 0.5856,
      "step": 19287
    },
    {
      "epoch": 1.72,
      "grad_norm": 5.652777466839532,
      "learning_rate": 5.025696464295704e-07,
      "loss": 0.5289,
      "step": 19288
    },
    {
      "epoch": 1.72,
      "grad_norm": 5.322222051135886,
      "learning_rate": 5.022540536494785e-07,
      "loss": 0.5769,
      "step": 19289
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.522434033769448,
      "learning_rate": 5.019385547499838e-07,
      "loss": 0.5446,
      "step": 19290
    },
    {
      "epoch": 1.72,
      "grad_norm": 7.468669260489667,
      "learning_rate": 5.016231497376717e-07,
      "loss": 0.5778,
      "step": 19291
    },
    {
      "epoch": 1.72,
      "grad_norm": 7.007709250538739,
      "learning_rate": 5.013078386191239e-07,
      "loss": 0.6079,
      "step": 19292
    },
    {
      "epoch": 1.72,
      "grad_norm": 7.208030247197926,
      "learning_rate": 5.009926214009236e-07,
      "loss": 0.5595,
      "step": 19293
    },
    {
      "epoch": 1.72,
      "grad_norm": 5.661476198016453,
      "learning_rate": 5.006774980896495e-07,
      "loss": 0.6075,
      "step": 19294
    },
    {
      "epoch": 1.72,
      "grad_norm": 7.336388046664277,
      "learning_rate": 5.00362468691879e-07,
      "loss": 0.6017,
      "step": 19295
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.982693583438814,
      "learning_rate": 5.00047533214189e-07,
      "loss": 0.6129,
      "step": 19296
    },
    {
      "epoch": 1.72,
      "grad_norm": 7.27886099699583,
      "learning_rate": 4.997326916631512e-07,
      "loss": 0.5441,
      "step": 19297
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.129110353476218,
      "learning_rate": 4.994179440453384e-07,
      "loss": 0.5526,
      "step": 19298
    },
    {
      "epoch": 1.72,
      "grad_norm": 5.856000462473329,
      "learning_rate": 4.991032903673198e-07,
      "loss": 0.5402,
      "step": 19299
    },
    {
      "epoch": 1.72,
      "grad_norm": 8.808762033562825,
      "learning_rate": 4.987887306356625e-07,
      "loss": 0.5456,
      "step": 19300
    },
    {
      "epoch": 1.72,
      "grad_norm": 7.766439359764874,
      "learning_rate": 4.984742648569335e-07,
      "loss": 0.6644,
      "step": 19301
    },
    {
      "epoch": 1.72,
      "grad_norm": 8.557813325310716,
      "learning_rate": 4.981598930376958e-07,
      "loss": 0.6003,
      "step": 19302
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.591130172154082,
      "learning_rate": 4.978456151845124e-07,
      "loss": 0.6241,
      "step": 19303
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.519369507629495,
      "learning_rate": 4.975314313039415e-07,
      "loss": 0.514,
      "step": 19304
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.271098137955858,
      "learning_rate": 4.972173414025416e-07,
      "loss": 0.6202,
      "step": 19305
    },
    {
      "epoch": 1.72,
      "grad_norm": 8.889973568363407,
      "learning_rate": 4.969033454868688e-07,
      "loss": 0.613,
      "step": 19306
    },
    {
      "epoch": 1.72,
      "grad_norm": 4.225439773803859,
      "learning_rate": 4.965894435634772e-07,
      "loss": 0.5878,
      "step": 19307
    },
    {
      "epoch": 1.72,
      "grad_norm": 8.406887708991459,
      "learning_rate": 4.962756356389182e-07,
      "loss": 0.5112,
      "step": 19308
    },
    {
      "epoch": 1.72,
      "grad_norm": 11.288106316938237,
      "learning_rate": 4.959619217197436e-07,
      "loss": 0.5601,
      "step": 19309
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.446747229524267,
      "learning_rate": 4.956483018124991e-07,
      "loss": 0.5036,
      "step": 19310
    },
    {
      "epoch": 1.72,
      "grad_norm": 7.448196832400418,
      "learning_rate": 4.953347759237331e-07,
      "loss": 0.5444,
      "step": 19311
    },
    {
      "epoch": 1.72,
      "grad_norm": 5.854335005664905,
      "learning_rate": 4.950213440599877e-07,
      "loss": 0.5294,
      "step": 19312
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.956740392848371,
      "learning_rate": 4.947080062278059e-07,
      "loss": 0.6221,
      "step": 19313
    },
    {
      "epoch": 1.72,
      "grad_norm": 7.737058727487721,
      "learning_rate": 4.943947624337281e-07,
      "loss": 0.5862,
      "step": 19314
    },
    {
      "epoch": 1.72,
      "grad_norm": 5.6738145587614515,
      "learning_rate": 4.940816126842924e-07,
      "loss": 0.4948,
      "step": 19315
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.497897982584543,
      "learning_rate": 4.937685569860363e-07,
      "loss": 0.5769,
      "step": 19316
    },
    {
      "epoch": 1.72,
      "grad_norm": 5.015728622718626,
      "learning_rate": 4.934555953454917e-07,
      "loss": 0.5065,
      "step": 19317
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.864760898155862,
      "learning_rate": 4.93142727769193e-07,
      "loss": 0.5881,
      "step": 19318
    },
    {
      "epoch": 1.72,
      "grad_norm": 5.5218097842853044,
      "learning_rate": 4.928299542636694e-07,
      "loss": 0.5714,
      "step": 19319
    },
    {
      "epoch": 1.72,
      "grad_norm": 8.083160156770147,
      "learning_rate": 4.925172748354506e-07,
      "loss": 0.6348,
      "step": 19320
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.372166054790135,
      "learning_rate": 4.92204689491062e-07,
      "loss": 0.5414,
      "step": 19321
    },
    {
      "epoch": 1.72,
      "grad_norm": 7.725967718140857,
      "learning_rate": 4.918921982370301e-07,
      "loss": 0.5795,
      "step": 19322
    },
    {
      "epoch": 1.72,
      "grad_norm": 5.9056104550134565,
      "learning_rate": 4.915798010798744e-07,
      "loss": 0.5861,
      "step": 19323
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.881010994061635,
      "learning_rate": 4.912674980261173e-07,
      "loss": 0.5187,
      "step": 19324
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.345883903798714,
      "learning_rate": 4.909552890822783e-07,
      "loss": 0.5332,
      "step": 19325
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.9379887869435235,
      "learning_rate": 4.906431742548717e-07,
      "loss": 0.5362,
      "step": 19326
    },
    {
      "epoch": 1.72,
      "grad_norm": 5.236948955043289,
      "learning_rate": 4.903311535504141e-07,
      "loss": 0.6127,
      "step": 19327
    },
    {
      "epoch": 1.72,
      "grad_norm": 7.486709686912351,
      "learning_rate": 4.900192269754172e-07,
      "loss": 0.5359,
      "step": 19328
    },
    {
      "epoch": 1.72,
      "grad_norm": 4.9003106758823325,
      "learning_rate": 4.897073945363933e-07,
      "loss": 0.6314,
      "step": 19329
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.216385538033852,
      "learning_rate": 4.893956562398494e-07,
      "loss": 0.5444,
      "step": 19330
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.705827142799635,
      "learning_rate": 4.89084012092293e-07,
      "loss": 0.5844,
      "step": 19331
    },
    {
      "epoch": 1.72,
      "grad_norm": 8.232366511184482,
      "learning_rate": 4.887724621002293e-07,
      "loss": 0.5572,
      "step": 19332
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.49548111364951,
      "learning_rate": 4.884610062701606e-07,
      "loss": 0.5663,
      "step": 19333
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.1898167178981085,
      "learning_rate": 4.881496446085887e-07,
      "loss": 0.5579,
      "step": 19334
    },
    {
      "epoch": 1.72,
      "grad_norm": 8.068500430312332,
      "learning_rate": 4.87838377122013e-07,
      "loss": 0.572,
      "step": 19335
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.457498589952259,
      "learning_rate": 4.875272038169288e-07,
      "loss": 0.5507,
      "step": 19336
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.023509125613162,
      "learning_rate": 4.87216124699832e-07,
      "loss": 0.5569,
      "step": 19337
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.968688537484046,
      "learning_rate": 4.869051397772157e-07,
      "loss": 0.5662,
      "step": 19338
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.706046470144435,
      "learning_rate": 4.865942490555709e-07,
      "loss": 0.5329,
      "step": 19339
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.2492840798989295,
      "learning_rate": 4.862834525413884e-07,
      "loss": 0.5478,
      "step": 19340
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.269368514134456,
      "learning_rate": 4.859727502411526e-07,
      "loss": 0.613,
      "step": 19341
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.6593597824437545,
      "learning_rate": 4.85662142161351e-07,
      "loss": 0.5452,
      "step": 19342
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.906582912420032,
      "learning_rate": 4.853516283084658e-07,
      "loss": 0.5737,
      "step": 19343
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.927776107359576,
      "learning_rate": 4.850412086889778e-07,
      "loss": 0.5561,
      "step": 19344
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.954169828803071,
      "learning_rate": 4.847308833093667e-07,
      "loss": 0.5339,
      "step": 19345
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.8864968213449735,
      "learning_rate": 4.844206521761108e-07,
      "loss": 0.5779,
      "step": 19346
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.852329187928231,
      "learning_rate": 4.841105152956844e-07,
      "loss": 0.5467,
      "step": 19347
    },
    {
      "epoch": 1.73,
      "grad_norm": 8.629021958197946,
      "learning_rate": 4.838004726745627e-07,
      "loss": 0.5726,
      "step": 19348
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.904183231014354,
      "learning_rate": 4.834905243192145e-07,
      "loss": 0.6291,
      "step": 19349
    },
    {
      "epoch": 1.73,
      "grad_norm": 9.508433958400593,
      "learning_rate": 4.831806702361108e-07,
      "loss": 0.6327,
      "step": 19350
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.600746207743731,
      "learning_rate": 4.828709104317186e-07,
      "loss": 0.6202,
      "step": 19351
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.551423717402952,
      "learning_rate": 4.825612449125039e-07,
      "loss": 0.5421,
      "step": 19352
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.6849786606963795,
      "learning_rate": 4.822516736849309e-07,
      "loss": 0.5849,
      "step": 19353
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.991540953393871,
      "learning_rate": 4.819421967554605e-07,
      "loss": 0.5412,
      "step": 19354
    },
    {
      "epoch": 1.73,
      "grad_norm": 8.250703871026147,
      "learning_rate": 4.816328141305532e-07,
      "loss": 0.5874,
      "step": 19355
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.73175601384945,
      "learning_rate": 4.813235258166643e-07,
      "loss": 0.6141,
      "step": 19356
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.566532670668538,
      "learning_rate": 4.81014331820251e-07,
      "loss": 0.6336,
      "step": 19357
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.625856191781129,
      "learning_rate": 4.807052321477674e-07,
      "loss": 0.5941,
      "step": 19358
    },
    {
      "epoch": 1.73,
      "grad_norm": 9.666870318758392,
      "learning_rate": 4.80396226805665e-07,
      "loss": 0.5297,
      "step": 19359
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.519065582590679,
      "learning_rate": 4.800873158003938e-07,
      "loss": 0.5747,
      "step": 19360
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.469255760232207,
      "learning_rate": 4.797784991384019e-07,
      "loss": 0.5158,
      "step": 19361
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.871573956959095,
      "learning_rate": 4.794697768261342e-07,
      "loss": 0.6057,
      "step": 19362
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.6511755027024995,
      "learning_rate": 4.791611488700349e-07,
      "loss": 0.5306,
      "step": 19363
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.835016144307784,
      "learning_rate": 4.78852615276546e-07,
      "loss": 0.5912,
      "step": 19364
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.224706729196271,
      "learning_rate": 4.785441760521081e-07,
      "loss": 0.534,
      "step": 19365
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.986751141083325,
      "learning_rate": 4.782358312031582e-07,
      "loss": 0.5887,
      "step": 19366
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.903006714816481,
      "learning_rate": 4.779275807361339e-07,
      "loss": 0.5313,
      "step": 19367
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.279385980730084,
      "learning_rate": 4.776194246574672e-07,
      "loss": 0.5882,
      "step": 19368
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.173874118693422,
      "learning_rate": 4.773113629735926e-07,
      "loss": 0.5801,
      "step": 19369
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.173140860923199,
      "learning_rate": 4.770033956909376e-07,
      "loss": 0.598,
      "step": 19370
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.3641734062848085,
      "learning_rate": 4.766955228159309e-07,
      "loss": 0.5069,
      "step": 19371
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.459408658738114,
      "learning_rate": 4.7638774435500026e-07,
      "loss": 0.5673,
      "step": 19372
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.595546409051166,
      "learning_rate": 4.760800603145682e-07,
      "loss": 0.5688,
      "step": 19373
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.47088545769917,
      "learning_rate": 4.7577247070105856e-07,
      "loss": 0.5754,
      "step": 19374
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.22385880409795,
      "learning_rate": 4.754649755208901e-07,
      "loss": 0.6025,
      "step": 19375
    },
    {
      "epoch": 1.73,
      "grad_norm": 9.136320891195847,
      "learning_rate": 4.7515757478048206e-07,
      "loss": 0.6178,
      "step": 19376
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.9568872313968075,
      "learning_rate": 4.748502684862499e-07,
      "loss": 0.6257,
      "step": 19377
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.303508009911228,
      "learning_rate": 4.745430566446085e-07,
      "loss": 0.5526,
      "step": 19378
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.954122962816615,
      "learning_rate": 4.7423593926197107e-07,
      "loss": 0.5937,
      "step": 19379
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.10097464707848,
      "learning_rate": 4.739289163447464e-07,
      "loss": 0.5678,
      "step": 19380
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.854825842588739,
      "learning_rate": 4.736219878993437e-07,
      "loss": 0.5284,
      "step": 19381
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.057609236861676,
      "learning_rate": 4.7331515393216955e-07,
      "loss": 0.5842,
      "step": 19382
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.116272413425819,
      "learning_rate": 4.7300841444962885e-07,
      "loss": 0.6466,
      "step": 19383
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.048051285019041,
      "learning_rate": 4.7270176945812264e-07,
      "loss": 0.5892,
      "step": 19384
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.7126592333813715,
      "learning_rate": 4.723952189640529e-07,
      "loss": 0.5489,
      "step": 19385
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.652933413170544,
      "learning_rate": 4.720887629738169e-07,
      "loss": 0.5761,
      "step": 19386
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.870849655755925,
      "learning_rate": 4.7178240149381384e-07,
      "loss": 0.6134,
      "step": 19387
    },
    {
      "epoch": 1.73,
      "grad_norm": 9.047434013408829,
      "learning_rate": 4.7147613453043473e-07,
      "loss": 0.5379,
      "step": 19388
    },
    {
      "epoch": 1.73,
      "grad_norm": 8.175338454233335,
      "learning_rate": 4.7116996209007446e-07,
      "loss": 0.5769,
      "step": 19389
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.723106258740303,
      "learning_rate": 4.7086388417912357e-07,
      "loss": 0.5946,
      "step": 19390
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.810193569171156,
      "learning_rate": 4.705579008039696e-07,
      "loss": 0.5567,
      "step": 19391
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.247260855360766,
      "learning_rate": 4.702520119710019e-07,
      "loss": 0.6463,
      "step": 19392
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.1309969964739235,
      "learning_rate": 4.699462176866021e-07,
      "loss": 0.5101,
      "step": 19393
    },
    {
      "epoch": 1.73,
      "grad_norm": 8.485486137340368,
      "learning_rate": 4.6964051795715447e-07,
      "loss": 0.5776,
      "step": 19394
    },
    {
      "epoch": 1.73,
      "grad_norm": 8.353063305764115,
      "learning_rate": 4.693349127890395e-07,
      "loss": 0.5897,
      "step": 19395
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.779164791915477,
      "learning_rate": 4.690294021886366e-07,
      "loss": 0.5606,
      "step": 19396
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.130215520557752,
      "learning_rate": 4.6872398616232215e-07,
      "loss": 0.5555,
      "step": 19397
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.808278991469077,
      "learning_rate": 4.684186647164718e-07,
      "loss": 0.5762,
      "step": 19398
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.282882831319591,
      "learning_rate": 4.681134378574581e-07,
      "loss": 0.5883,
      "step": 19399
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.572389867608683,
      "learning_rate": 4.6780830559165103e-07,
      "loss": 0.5835,
      "step": 19400
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.298789100247537,
      "learning_rate": 4.675032679254199e-07,
      "loss": 0.6434,
      "step": 19401
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.8678607751302865,
      "learning_rate": 4.671983248651324e-07,
      "loss": 0.5881,
      "step": 19402
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.7400369363938015,
      "learning_rate": 4.668934764171529e-07,
      "loss": 0.5245,
      "step": 19403
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.787827952950555,
      "learning_rate": 4.665887225878446e-07,
      "loss": 0.5634,
      "step": 19404
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.643549889337505,
      "learning_rate": 4.662840633835697e-07,
      "loss": 0.6457,
      "step": 19405
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.839191533841817,
      "learning_rate": 4.6597949881068584e-07,
      "loss": 0.5467,
      "step": 19406
    },
    {
      "epoch": 1.73,
      "grad_norm": 10.48631251531014,
      "learning_rate": 4.6567502887555015e-07,
      "loss": 0.5612,
      "step": 19407
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.3906054383085795,
      "learning_rate": 4.6537065358451814e-07,
      "loss": 0.5324,
      "step": 19408
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.263617958800574,
      "learning_rate": 4.650663729439436e-07,
      "loss": 0.5711,
      "step": 19409
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.2754894472017435,
      "learning_rate": 4.6476218696017704e-07,
      "loss": 0.5062,
      "step": 19410
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.95918658256001,
      "learning_rate": 4.6445809563956836e-07,
      "loss": 0.5051,
      "step": 19411
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.6293551593567575,
      "learning_rate": 4.6415409898846466e-07,
      "loss": 0.5644,
      "step": 19412
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.734682719267298,
      "learning_rate": 4.638501970132098e-07,
      "loss": 0.5468,
      "step": 19413
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.15537191739193,
      "learning_rate": 4.6354638972014807e-07,
      "loss": 0.5984,
      "step": 19414
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.869161750853449,
      "learning_rate": 4.6324267711562067e-07,
      "loss": 0.5576,
      "step": 19415
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.41963550394519,
      "learning_rate": 4.629390592059668e-07,
      "loss": 0.5739,
      "step": 19416
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.651653717417522,
      "learning_rate": 4.626355359975243e-07,
      "loss": 0.6304,
      "step": 19417
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.887474044365535,
      "learning_rate": 4.623321074966297e-07,
      "loss": 0.5951,
      "step": 19418
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.717921006027463,
      "learning_rate": 4.620287737096135e-07,
      "loss": 0.4947,
      "step": 19419
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.916811116708596,
      "learning_rate": 4.61725534642809e-07,
      "loss": 0.5177,
      "step": 19420
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.085053110013839,
      "learning_rate": 4.61422390302545e-07,
      "loss": 0.5873,
      "step": 19421
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.868226757163164,
      "learning_rate": 4.6111934069514975e-07,
      "loss": 0.5201,
      "step": 19422
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.412623879367188,
      "learning_rate": 4.608163858269482e-07,
      "loss": 0.5286,
      "step": 19423
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.3677036229341715,
      "learning_rate": 4.6051352570426476e-07,
      "loss": 0.6241,
      "step": 19424
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.221287755135121,
      "learning_rate": 4.602107603334188e-07,
      "loss": 0.5265,
      "step": 19425
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.052411893368286,
      "learning_rate": 4.5990808972073255e-07,
      "loss": 0.5632,
      "step": 19426
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.55027300075432,
      "learning_rate": 4.596055138725214e-07,
      "loss": 0.5647,
      "step": 19427
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.668110890199223,
      "learning_rate": 4.59303032795102e-07,
      "loss": 0.6055,
      "step": 19428
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.312832277198063,
      "learning_rate": 4.590006464947877e-07,
      "loss": 0.564,
      "step": 19429
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.921875474207813,
      "learning_rate": 4.5869835497789007e-07,
      "loss": 0.5284,
      "step": 19430
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.580446871976063,
      "learning_rate": 4.583961582507201e-07,
      "loss": 0.5742,
      "step": 19431
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.127740466810843,
      "learning_rate": 4.5809405631958404e-07,
      "loss": 0.6233,
      "step": 19432
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.509361638584741,
      "learning_rate": 4.5779204919078723e-07,
      "loss": 0.5259,
      "step": 19433
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.910155228137168,
      "learning_rate": 4.5749013687063416e-07,
      "loss": 0.5382,
      "step": 19434
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.084255844239266,
      "learning_rate": 4.57188319365427e-07,
      "loss": 0.6455,
      "step": 19435
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.778452294988229,
      "learning_rate": 4.568865966814645e-07,
      "loss": 0.5586,
      "step": 19436
    },
    {
      "epoch": 1.73,
      "grad_norm": 4.963984387176218,
      "learning_rate": 4.565849688250462e-07,
      "loss": 0.5277,
      "step": 19437
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.9347737759326,
      "learning_rate": 4.562834358024659e-07,
      "loss": 0.5785,
      "step": 19438
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.552812746162098,
      "learning_rate": 4.5598199762001803e-07,
      "loss": 0.5813,
      "step": 19439
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.39216738885419,
      "learning_rate": 4.5568065428399533e-07,
      "loss": 0.6358,
      "step": 19440
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.912036422875484,
      "learning_rate": 4.5537940580068776e-07,
      "loss": 0.5354,
      "step": 19441
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.480778336736666,
      "learning_rate": 4.5507825217638136e-07,
      "loss": 0.556,
      "step": 19442
    },
    {
      "epoch": 1.73,
      "grad_norm": 4.980796344159234,
      "learning_rate": 4.547771934173634e-07,
      "loss": 0.5523,
      "step": 19443
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.442605138064705,
      "learning_rate": 4.5447622952991875e-07,
      "loss": 0.5259,
      "step": 19444
    },
    {
      "epoch": 1.73,
      "grad_norm": 4.688802096088219,
      "learning_rate": 4.5417536052032695e-07,
      "loss": 0.5023,
      "step": 19445
    },
    {
      "epoch": 1.73,
      "grad_norm": 9.871890058454602,
      "learning_rate": 4.538745863948696e-07,
      "loss": 0.5648,
      "step": 19446
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.566036689774609,
      "learning_rate": 4.535739071598244e-07,
      "loss": 0.495,
      "step": 19447
    },
    {
      "epoch": 1.73,
      "grad_norm": 8.05316981616333,
      "learning_rate": 4.532733228214675e-07,
      "loss": 0.6281,
      "step": 19448
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.206722343812226,
      "learning_rate": 4.5297283338607277e-07,
      "loss": 0.6174,
      "step": 19449
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.429438622005737,
      "learning_rate": 4.52672438859913e-07,
      "loss": 0.5687,
      "step": 19450
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.722909491269947,
      "learning_rate": 4.523721392492569e-07,
      "loss": 0.5135,
      "step": 19451
    },
    {
      "epoch": 1.74,
      "grad_norm": 7.75820394411651,
      "learning_rate": 4.5207193456037304e-07,
      "loss": 0.5637,
      "step": 19452
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.6505179748808985,
      "learning_rate": 4.5177182479952787e-07,
      "loss": 0.5885,
      "step": 19453
    },
    {
      "epoch": 1.74,
      "grad_norm": 4.508279026270385,
      "learning_rate": 4.514718099729853e-07,
      "loss": 0.5476,
      "step": 19454
    },
    {
      "epoch": 1.74,
      "grad_norm": 11.11160781272378,
      "learning_rate": 4.5117189008700877e-07,
      "loss": 0.5905,
      "step": 19455
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.644063104018046,
      "learning_rate": 4.50872065147856e-07,
      "loss": 0.5581,
      "step": 19456
    },
    {
      "epoch": 1.74,
      "grad_norm": 7.83409033269218,
      "learning_rate": 4.50572335161788e-07,
      "loss": 0.5476,
      "step": 19457
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.8516078765871535,
      "learning_rate": 4.502727001350582e-07,
      "loss": 0.544,
      "step": 19458
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.929133282275333,
      "learning_rate": 4.4997316007392154e-07,
      "loss": 0.5757,
      "step": 19459
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.2370200660742645,
      "learning_rate": 4.4967371498463144e-07,
      "loss": 0.5548,
      "step": 19460
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.672082077769837,
      "learning_rate": 4.493743648734372e-07,
      "loss": 0.6031,
      "step": 19461
    },
    {
      "epoch": 1.74,
      "grad_norm": 8.71773433719934,
      "learning_rate": 4.490751097465873e-07,
      "loss": 0.6109,
      "step": 19462
    },
    {
      "epoch": 1.74,
      "grad_norm": 7.963620716966938,
      "learning_rate": 4.487759496103294e-07,
      "loss": 0.6297,
      "step": 19463
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.4599310772610075,
      "learning_rate": 4.484768844709053e-07,
      "loss": 0.5753,
      "step": 19464
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.617599894248542,
      "learning_rate": 4.4817791433455883e-07,
      "loss": 0.5972,
      "step": 19465
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.686150666328848,
      "learning_rate": 4.4787903920752997e-07,
      "loss": 0.5704,
      "step": 19466
    },
    {
      "epoch": 1.74,
      "grad_norm": 4.985462343250049,
      "learning_rate": 4.4758025909605706e-07,
      "loss": 0.5933,
      "step": 19467
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.976573467713243,
      "learning_rate": 4.472815740063774e-07,
      "loss": 0.6288,
      "step": 19468
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.804012158339068,
      "learning_rate": 4.469829839447237e-07,
      "loss": 0.5102,
      "step": 19469
    },
    {
      "epoch": 1.74,
      "grad_norm": 8.37295431432846,
      "learning_rate": 4.4668448891733054e-07,
      "loss": 0.5826,
      "step": 19470
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.4797327617345575,
      "learning_rate": 4.463860889304261e-07,
      "loss": 0.5026,
      "step": 19471
    },
    {
      "epoch": 1.74,
      "grad_norm": 7.410786566920826,
      "learning_rate": 4.4608778399023943e-07,
      "loss": 0.581,
      "step": 19472
    },
    {
      "epoch": 1.74,
      "grad_norm": 9.104747470895896,
      "learning_rate": 4.4578957410299774e-07,
      "loss": 0.6567,
      "step": 19473
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.57326787309342,
      "learning_rate": 4.4549145927492545e-07,
      "loss": 0.5768,
      "step": 19474
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.624955460644096,
      "learning_rate": 4.4519343951224536e-07,
      "loss": 0.5421,
      "step": 19475
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.517261562359927,
      "learning_rate": 4.448955148211759e-07,
      "loss": 0.5719,
      "step": 19476
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.696974359447795,
      "learning_rate": 4.4459768520793755e-07,
      "loss": 0.6165,
      "step": 19477
    },
    {
      "epoch": 1.74,
      "grad_norm": 7.416344132161564,
      "learning_rate": 4.4429995067874656e-07,
      "loss": 0.5866,
      "step": 19478
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.113297676799036,
      "learning_rate": 4.4400231123981676e-07,
      "loss": 0.5709,
      "step": 19479
    },
    {
      "epoch": 1.74,
      "grad_norm": 7.833172164564583,
      "learning_rate": 4.4370476689736154e-07,
      "loss": 0.5482,
      "step": 19480
    },
    {
      "epoch": 1.74,
      "grad_norm": 7.69284265061819,
      "learning_rate": 4.434073176575926e-07,
      "loss": 0.5913,
      "step": 19481
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.9865239367218885,
      "learning_rate": 4.4310996352671555e-07,
      "loss": 0.5446,
      "step": 19482
    },
    {
      "epoch": 1.74,
      "grad_norm": 7.663676139550128,
      "learning_rate": 4.4281270451093874e-07,
      "loss": 0.5512,
      "step": 19483
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.247997725914049,
      "learning_rate": 4.425155406164677e-07,
      "loss": 0.6155,
      "step": 19484
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.462384546782858,
      "learning_rate": 4.4221847184950264e-07,
      "loss": 0.5153,
      "step": 19485
    },
    {
      "epoch": 1.74,
      "grad_norm": 9.150770400193105,
      "learning_rate": 4.419214982162462e-07,
      "loss": 0.5833,
      "step": 19486
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.427268855909817,
      "learning_rate": 4.416246197228963e-07,
      "loss": 0.5872,
      "step": 19487
    },
    {
      "epoch": 1.74,
      "grad_norm": 7.055163627749746,
      "learning_rate": 4.413278363756507e-07,
      "loss": 0.5841,
      "step": 19488
    },
    {
      "epoch": 1.74,
      "grad_norm": 11.356795729150889,
      "learning_rate": 4.4103114818070226e-07,
      "loss": 0.5212,
      "step": 19489
    },
    {
      "epoch": 1.74,
      "grad_norm": 9.19756954222699,
      "learning_rate": 4.4073455514424437e-07,
      "loss": 0.6042,
      "step": 19490
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.77521256839744,
      "learning_rate": 4.4043805727246867e-07,
      "loss": 0.5715,
      "step": 19491
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.2174625244851125,
      "learning_rate": 4.4014165457156257e-07,
      "loss": 0.5895,
      "step": 19492
    },
    {
      "epoch": 1.74,
      "grad_norm": 8.210066784928845,
      "learning_rate": 4.3984534704771376e-07,
      "loss": 0.4879,
      "step": 19493
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.685446147847926,
      "learning_rate": 4.395491347071079e-07,
      "loss": 0.5202,
      "step": 19494
    },
    {
      "epoch": 1.74,
      "grad_norm": 7.083276183627929,
      "learning_rate": 4.392530175559251e-07,
      "loss": 0.5038,
      "step": 19495
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.2682418428907,
      "learning_rate": 4.389569956003481e-07,
      "loss": 0.5507,
      "step": 19496
    },
    {
      "epoch": 1.74,
      "grad_norm": 8.017207083682553,
      "learning_rate": 4.386610688465554e-07,
      "loss": 0.5647,
      "step": 19497
    },
    {
      "epoch": 1.74,
      "grad_norm": 9.186013430066737,
      "learning_rate": 4.3836523730072466e-07,
      "loss": 0.5868,
      "step": 19498
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.765438289298087,
      "learning_rate": 4.3806950096902887e-07,
      "loss": 0.5796,
      "step": 19499
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.489491365182356,
      "learning_rate": 4.377738598576414e-07,
      "loss": 0.5306,
      "step": 19500
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.209091495303553,
      "learning_rate": 4.374783139727351e-07,
      "loss": 0.5434,
      "step": 19501
    },
    {
      "epoch": 1.74,
      "grad_norm": 7.171287453550054,
      "learning_rate": 4.3718286332047556e-07,
      "loss": 0.596,
      "step": 19502
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.584728193089351,
      "learning_rate": 4.3688750790703173e-07,
      "loss": 0.6529,
      "step": 19503
    },
    {
      "epoch": 1.74,
      "grad_norm": 7.841150305205505,
      "learning_rate": 4.365922477385681e-07,
      "loss": 0.5513,
      "step": 19504
    },
    {
      "epoch": 1.74,
      "grad_norm": 7.229058464113996,
      "learning_rate": 4.362970828212476e-07,
      "loss": 0.5594,
      "step": 19505
    },
    {
      "epoch": 1.74,
      "grad_norm": 9.524199450805392,
      "learning_rate": 4.360020131612314e-07,
      "loss": 0.5706,
      "step": 19506
    },
    {
      "epoch": 1.74,
      "grad_norm": 7.414270151324955,
      "learning_rate": 4.357070387646789e-07,
      "loss": 0.6619,
      "step": 19507
    },
    {
      "epoch": 1.74,
      "grad_norm": 8.141966767160117,
      "learning_rate": 4.354121596377453e-07,
      "loss": 0.5791,
      "step": 19508
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.721795442160722,
      "learning_rate": 4.3511737578658677e-07,
      "loss": 0.5991,
      "step": 19509
    },
    {
      "epoch": 1.74,
      "grad_norm": 8.272871543053228,
      "learning_rate": 4.348226872173561e-07,
      "loss": 0.566,
      "step": 19510
    },
    {
      "epoch": 1.74,
      "grad_norm": 9.065393872017193,
      "learning_rate": 4.34528093936204e-07,
      "loss": 0.5894,
      "step": 19511
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.9424534078563465,
      "learning_rate": 4.3423359594927994e-07,
      "loss": 0.5841,
      "step": 19512
    },
    {
      "epoch": 1.74,
      "grad_norm": 7.025154010311308,
      "learning_rate": 4.339391932627318e-07,
      "loss": 0.6101,
      "step": 19513
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.714642236206939,
      "learning_rate": 4.3364488588270304e-07,
      "loss": 0.5641,
      "step": 19514
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.209323935102551,
      "learning_rate": 4.3335067381533645e-07,
      "loss": 0.6219,
      "step": 19515
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.502953986171062,
      "learning_rate": 4.330565570667739e-07,
      "loss": 0.5999,
      "step": 19516
    },
    {
      "epoch": 1.74,
      "grad_norm": 7.7691372672810415,
      "learning_rate": 4.327625356431542e-07,
      "loss": 0.5874,
      "step": 19517
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.7821185312419505,
      "learning_rate": 4.324686095506142e-07,
      "loss": 0.5011,
      "step": 19518
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.744629692843127,
      "learning_rate": 4.3217477879528956e-07,
      "loss": 0.5585,
      "step": 19519
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.662715987931336,
      "learning_rate": 4.318810433833143e-07,
      "loss": 0.5716,
      "step": 19520
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.930405151187031,
      "learning_rate": 4.315874033208167e-07,
      "loss": 0.6317,
      "step": 19521
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.908598566831788,
      "learning_rate": 4.312938586139276e-07,
      "loss": 0.5721,
      "step": 19522
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.04129035261346,
      "learning_rate": 4.310004092687742e-07,
      "loss": 0.5262,
      "step": 19523
    },
    {
      "epoch": 1.74,
      "grad_norm": 7.389922616938146,
      "learning_rate": 4.3070705529148106e-07,
      "loss": 0.6032,
      "step": 19524
    },
    {
      "epoch": 1.74,
      "grad_norm": 8.116850144131325,
      "learning_rate": 4.304137966881716e-07,
      "loss": 0.5776,
      "step": 19525
    },
    {
      "epoch": 1.74,
      "grad_norm": 7.2664356210579335,
      "learning_rate": 4.301206334649677e-07,
      "loss": 0.5636,
      "step": 19526
    },
    {
      "epoch": 1.74,
      "grad_norm": 4.714019699834582,
      "learning_rate": 4.2982756562798767e-07,
      "loss": 0.5647,
      "step": 19527
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.3409487131263615,
      "learning_rate": 4.295345931833478e-07,
      "loss": 0.5175,
      "step": 19528
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.718982887953429,
      "learning_rate": 4.2924171613716426e-07,
      "loss": 0.5529,
      "step": 19529
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.743555745622507,
      "learning_rate": 4.289489344955505e-07,
      "loss": 0.5281,
      "step": 19530
    },
    {
      "epoch": 1.74,
      "grad_norm": 9.057076315677909,
      "learning_rate": 4.286562482646167e-07,
      "loss": 0.6085,
      "step": 19531
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.485153116295635,
      "learning_rate": 4.283636574504729e-07,
      "loss": 0.536,
      "step": 19532
    },
    {
      "epoch": 1.74,
      "grad_norm": 11.125507510358856,
      "learning_rate": 4.2807116205922706e-07,
      "loss": 0.6311,
      "step": 19533
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.471770671664696,
      "learning_rate": 4.27778762096982e-07,
      "loss": 0.5579,
      "step": 19534
    },
    {
      "epoch": 1.74,
      "grad_norm": 7.295582893472179,
      "learning_rate": 4.274864575698429e-07,
      "loss": 0.5303,
      "step": 19535
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.362482514196293,
      "learning_rate": 4.2719424848390977e-07,
      "loss": 0.5984,
      "step": 19536
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.849008163059676,
      "learning_rate": 4.269021348452829e-07,
      "loss": 0.5443,
      "step": 19537
    },
    {
      "epoch": 1.74,
      "grad_norm": 7.605104750299689,
      "learning_rate": 4.266101166600589e-07,
      "loss": 0.5992,
      "step": 19538
    },
    {
      "epoch": 1.74,
      "grad_norm": 8.204995873121996,
      "learning_rate": 4.263181939343336e-07,
      "loss": 0.5926,
      "step": 19539
    },
    {
      "epoch": 1.74,
      "grad_norm": 7.3905347631205,
      "learning_rate": 4.2602636667419985e-07,
      "loss": 0.586,
      "step": 19540
    },
    {
      "epoch": 1.74,
      "grad_norm": 9.354597506155264,
      "learning_rate": 4.257346348857483e-07,
      "loss": 0.6128,
      "step": 19541
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.56697419814046,
      "learning_rate": 4.2544299857506967e-07,
      "loss": 0.555,
      "step": 19542
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.244794268370045,
      "learning_rate": 4.2515145774824906e-07,
      "loss": 0.5267,
      "step": 19543
    },
    {
      "epoch": 1.74,
      "grad_norm": 7.021835240042137,
      "learning_rate": 4.248600124113739e-07,
      "loss": 0.6466,
      "step": 19544
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.902044062648817,
      "learning_rate": 4.245686625705259e-07,
      "loss": 0.518,
      "step": 19545
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.779476581420664,
      "learning_rate": 4.24277408231788e-07,
      "loss": 0.5485,
      "step": 19546
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.785942055004502,
      "learning_rate": 4.239862494012381e-07,
      "loss": 0.5707,
      "step": 19547
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.975753411899509,
      "learning_rate": 4.236951860849536e-07,
      "loss": 0.6084,
      "step": 19548
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.732381193617649,
      "learning_rate": 4.2340421828901023e-07,
      "loss": 0.5761,
      "step": 19549
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.436980845332754,
      "learning_rate": 4.231133460194814e-07,
      "loss": 0.5762,
      "step": 19550
    },
    {
      "epoch": 1.74,
      "grad_norm": 7.5183641256989615,
      "learning_rate": 4.228225692824389e-07,
      "loss": 0.595,
      "step": 19551
    },
    {
      "epoch": 1.74,
      "grad_norm": 7.489418278430452,
      "learning_rate": 4.2253188808395074e-07,
      "loss": 0.5498,
      "step": 19552
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.4939550188975215,
      "learning_rate": 4.2224130243008475e-07,
      "loss": 0.4962,
      "step": 19553
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.262098686909548,
      "learning_rate": 4.2195081232690614e-07,
      "loss": 0.6027,
      "step": 19554
    },
    {
      "epoch": 1.74,
      "grad_norm": 10.22436940681599,
      "learning_rate": 4.2166041778047893e-07,
      "loss": 0.6078,
      "step": 19555
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.520070894521177,
      "learning_rate": 4.213701187968655e-07,
      "loss": 0.5413,
      "step": 19556
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.0252144158591365,
      "learning_rate": 4.210799153821221e-07,
      "loss": 0.5761,
      "step": 19557
    },
    {
      "epoch": 1.74,
      "grad_norm": 8.728993546951923,
      "learning_rate": 4.207898075423089e-07,
      "loss": 0.5773,
      "step": 19558
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.197874096537514,
      "learning_rate": 4.204997952834794e-07,
      "loss": 0.5847,
      "step": 19559
    },
    {
      "epoch": 1.74,
      "grad_norm": 7.503700224415165,
      "learning_rate": 4.2020987861168813e-07,
      "loss": 0.5614,
      "step": 19560
    },
    {
      "epoch": 1.75,
      "grad_norm": 7.028679120702142,
      "learning_rate": 4.199200575329854e-07,
      "loss": 0.5806,
      "step": 19561
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.966521346020008,
      "learning_rate": 4.196303320534217e-07,
      "loss": 0.5606,
      "step": 19562
    },
    {
      "epoch": 1.75,
      "grad_norm": 7.974386403767797,
      "learning_rate": 4.1934070217904355e-07,
      "loss": 0.5169,
      "step": 19563
    },
    {
      "epoch": 1.75,
      "grad_norm": 8.32441157132391,
      "learning_rate": 4.190511679158982e-07,
      "loss": 0.6075,
      "step": 19564
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.3441187321156445,
      "learning_rate": 4.187617292700263e-07,
      "loss": 0.5657,
      "step": 19565
    },
    {
      "epoch": 1.75,
      "grad_norm": 7.307415532048551,
      "learning_rate": 4.18472386247471e-07,
      "loss": 0.6333,
      "step": 19566
    },
    {
      "epoch": 1.75,
      "grad_norm": 7.984642000228335,
      "learning_rate": 4.1818313885427064e-07,
      "loss": 0.5422,
      "step": 19567
    },
    {
      "epoch": 1.75,
      "grad_norm": 8.93175364337211,
      "learning_rate": 4.1789398709646377e-07,
      "loss": 0.6203,
      "step": 19568
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.111005548912059,
      "learning_rate": 4.17604930980085e-07,
      "loss": 0.5713,
      "step": 19569
    },
    {
      "epoch": 1.75,
      "grad_norm": 7.901959834357352,
      "learning_rate": 4.17315970511169e-07,
      "loss": 0.5153,
      "step": 19570
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.549466272608496,
      "learning_rate": 4.1702710569574646e-07,
      "loss": 0.6212,
      "step": 19571
    },
    {
      "epoch": 1.75,
      "grad_norm": 4.998539854877805,
      "learning_rate": 4.167383365398453e-07,
      "loss": 0.5805,
      "step": 19572
    },
    {
      "epoch": 1.75,
      "grad_norm": 9.061823567642627,
      "learning_rate": 4.164496630494941e-07,
      "loss": 0.6782,
      "step": 19573
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.175329237579687,
      "learning_rate": 4.161610852307185e-07,
      "loss": 0.6098,
      "step": 19574
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.5759171947605815,
      "learning_rate": 4.158726030895421e-07,
      "loss": 0.4765,
      "step": 19575
    },
    {
      "epoch": 1.75,
      "grad_norm": 8.293575669205936,
      "learning_rate": 4.155842166319857e-07,
      "loss": 0.5614,
      "step": 19576
    },
    {
      "epoch": 1.75,
      "grad_norm": 9.240226721822177,
      "learning_rate": 4.1529592586406986e-07,
      "loss": 0.5638,
      "step": 19577
    },
    {
      "epoch": 1.75,
      "grad_norm": 8.129264392722758,
      "learning_rate": 4.15007730791811e-07,
      "loss": 0.6329,
      "step": 19578
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.5864988472486745,
      "learning_rate": 4.147196314212243e-07,
      "loss": 0.5589,
      "step": 19579
    },
    {
      "epoch": 1.75,
      "grad_norm": 8.088119764851541,
      "learning_rate": 4.144316277583238e-07,
      "loss": 0.5748,
      "step": 19580
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.30953818424619,
      "learning_rate": 4.1414371980912083e-07,
      "loss": 0.5491,
      "step": 19581
    },
    {
      "epoch": 1.75,
      "grad_norm": 9.380457303441355,
      "learning_rate": 4.13855907579625e-07,
      "loss": 0.5846,
      "step": 19582
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.874659520965099,
      "learning_rate": 4.1356819107584436e-07,
      "loss": 0.5474,
      "step": 19583
    },
    {
      "epoch": 1.75,
      "grad_norm": 7.9037714880721675,
      "learning_rate": 4.1328057030378234e-07,
      "loss": 0.5465,
      "step": 19584
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.043225529610049,
      "learning_rate": 4.129930452694453e-07,
      "loss": 0.6186,
      "step": 19585
    },
    {
      "epoch": 1.75,
      "grad_norm": 7.104755531465783,
      "learning_rate": 4.127056159788317e-07,
      "loss": 0.5415,
      "step": 19586
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.495774771769534,
      "learning_rate": 4.1241828243794245e-07,
      "loss": 0.5683,
      "step": 19587
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.853326112838729,
      "learning_rate": 4.121310446527749e-07,
      "loss": 0.5553,
      "step": 19588
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.896284228505057,
      "learning_rate": 4.118439026293247e-07,
      "loss": 0.5775,
      "step": 19589
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.043600940017664,
      "learning_rate": 4.115568563735861e-07,
      "loss": 0.536,
      "step": 19590
    },
    {
      "epoch": 1.75,
      "grad_norm": 4.494709104299569,
      "learning_rate": 4.1126990589154816e-07,
      "loss": 0.5587,
      "step": 19591
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.939396251131095,
      "learning_rate": 4.109830511892021e-07,
      "loss": 0.625,
      "step": 19592
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.909909478371609,
      "learning_rate": 4.106962922725355e-07,
      "loss": 0.5488,
      "step": 19593
    },
    {
      "epoch": 1.75,
      "grad_norm": 7.370413518248144,
      "learning_rate": 4.1040962914753293e-07,
      "loss": 0.592,
      "step": 19594
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.997244266385302,
      "learning_rate": 4.1012306182017846e-07,
      "loss": 0.5191,
      "step": 19595
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.983914973386036,
      "learning_rate": 4.0983659029645405e-07,
      "loss": 0.5806,
      "step": 19596
    },
    {
      "epoch": 1.75,
      "grad_norm": 8.07000655730643,
      "learning_rate": 4.095502145823382e-07,
      "loss": 0.583,
      "step": 19597
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.493505353975712,
      "learning_rate": 4.092639346838084e-07,
      "loss": 0.5598,
      "step": 19598
    },
    {
      "epoch": 1.75,
      "grad_norm": 9.06574225483365,
      "learning_rate": 4.0897775060684097e-07,
      "loss": 0.5379,
      "step": 19599
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.990198161129171,
      "learning_rate": 4.0869166235740834e-07,
      "loss": 0.5242,
      "step": 19600
    },
    {
      "epoch": 1.75,
      "grad_norm": 9.211851700845598,
      "learning_rate": 4.0840566994148236e-07,
      "loss": 0.5492,
      "step": 19601
    },
    {
      "epoch": 1.75,
      "grad_norm": 8.944803565214043,
      "learning_rate": 4.081197733650327e-07,
      "loss": 0.561,
      "step": 19602
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.749754974491633,
      "learning_rate": 4.078339726340275e-07,
      "loss": 0.6271,
      "step": 19603
    },
    {
      "epoch": 1.75,
      "grad_norm": 8.482761969646706,
      "learning_rate": 4.075482677544307e-07,
      "loss": 0.5221,
      "step": 19604
    },
    {
      "epoch": 1.75,
      "grad_norm": 7.033914045969014,
      "learning_rate": 4.072626587322065e-07,
      "loss": 0.533,
      "step": 19605
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.521944947729008,
      "learning_rate": 4.0697714557331624e-07,
      "loss": 0.4996,
      "step": 19606
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.032867736125572,
      "learning_rate": 4.0669172828371905e-07,
      "loss": 0.551,
      "step": 19607
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.71965334245738,
      "learning_rate": 4.064064068693735e-07,
      "loss": 0.5636,
      "step": 19608
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.250768628587512,
      "learning_rate": 4.061211813362353e-07,
      "loss": 0.5883,
      "step": 19609
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.347316546245109,
      "learning_rate": 4.05836051690256e-07,
      "loss": 0.5704,
      "step": 19610
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.488333944777599,
      "learning_rate": 4.055510179373878e-07,
      "loss": 0.5414,
      "step": 19611
    },
    {
      "epoch": 1.75,
      "grad_norm": 7.4774745812018795,
      "learning_rate": 4.0526608008358113e-07,
      "loss": 0.5589,
      "step": 19612
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.333774055737967,
      "learning_rate": 4.0498123813478227e-07,
      "loss": 0.5753,
      "step": 19613
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.740864381298494,
      "learning_rate": 4.046964920969376e-07,
      "loss": 0.5938,
      "step": 19614
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.7873548648798865,
      "learning_rate": 4.0441184197599013e-07,
      "loss": 0.5302,
      "step": 19615
    },
    {
      "epoch": 1.75,
      "grad_norm": 7.828962397026977,
      "learning_rate": 4.0412728777788177e-07,
      "loss": 0.487,
      "step": 19616
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.747157508571282,
      "learning_rate": 4.038428295085506e-07,
      "loss": 0.5104,
      "step": 19617
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.043432526283541,
      "learning_rate": 4.0355846717393456e-07,
      "loss": 0.579,
      "step": 19618
    },
    {
      "epoch": 1.75,
      "grad_norm": 7.474772445135849,
      "learning_rate": 4.0327420077997004e-07,
      "loss": 0.5859,
      "step": 19619
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.510875892457881,
      "learning_rate": 4.0299003033258954e-07,
      "loss": 0.5995,
      "step": 19620
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.638153204280999,
      "learning_rate": 4.02705955837725e-07,
      "loss": 0.6042,
      "step": 19621
    },
    {
      "epoch": 1.75,
      "grad_norm": 7.828864635901741,
      "learning_rate": 4.024219773013066e-07,
      "loss": 0.558,
      "step": 19622
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.804225598889114,
      "learning_rate": 4.0213809472926026e-07,
      "loss": 0.5217,
      "step": 19623
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.529206105044793,
      "learning_rate": 4.018543081275117e-07,
      "loss": 0.632,
      "step": 19624
    },
    {
      "epoch": 1.75,
      "grad_norm": 7.145676611096472,
      "learning_rate": 4.015706175019851e-07,
      "loss": 0.5485,
      "step": 19625
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.035548121425731,
      "learning_rate": 4.012870228586013e-07,
      "loss": 0.6035,
      "step": 19626
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.897499591592463,
      "learning_rate": 4.010035242032806e-07,
      "loss": 0.5592,
      "step": 19627
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.706325618160136,
      "learning_rate": 4.0072012154193874e-07,
      "loss": 0.6221,
      "step": 19628
    },
    {
      "epoch": 1.75,
      "grad_norm": 7.957055338696454,
      "learning_rate": 4.004368148804932e-07,
      "loss": 0.5563,
      "step": 19629
    },
    {
      "epoch": 1.75,
      "grad_norm": 8.083637779593275,
      "learning_rate": 4.001536042248555e-07,
      "loss": 0.6039,
      "step": 19630
    },
    {
      "epoch": 1.75,
      "grad_norm": 7.977073656828253,
      "learning_rate": 3.99870489580938e-07,
      "loss": 0.6233,
      "step": 19631
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.783718591834109,
      "learning_rate": 3.9958747095464936e-07,
      "loss": 0.5667,
      "step": 19632
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.973502012047984,
      "learning_rate": 3.993045483518976e-07,
      "loss": 0.5902,
      "step": 19633
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.656787522549099,
      "learning_rate": 3.990217217785891e-07,
      "loss": 0.5359,
      "step": 19634
    },
    {
      "epoch": 1.75,
      "grad_norm": 7.285131820228348,
      "learning_rate": 3.987389912406253e-07,
      "loss": 0.6403,
      "step": 19635
    },
    {
      "epoch": 1.75,
      "grad_norm": 8.129978007358527,
      "learning_rate": 3.984563567439087e-07,
      "loss": 0.5756,
      "step": 19636
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.873697903969904,
      "learning_rate": 3.9817381829433787e-07,
      "loss": 0.636,
      "step": 19637
    },
    {
      "epoch": 1.75,
      "grad_norm": 8.087186722190522,
      "learning_rate": 3.9789137589781144e-07,
      "loss": 0.5165,
      "step": 19638
    },
    {
      "epoch": 1.75,
      "grad_norm": 7.930119010868773,
      "learning_rate": 3.976090295602236e-07,
      "loss": 0.5646,
      "step": 19639
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.785396388527818,
      "learning_rate": 3.973267792874691e-07,
      "loss": 0.4622,
      "step": 19640
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.681868248652717,
      "learning_rate": 3.9704462508543763e-07,
      "loss": 0.6074,
      "step": 19641
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.320174629464637,
      "learning_rate": 3.9676256696001947e-07,
      "loss": 0.5904,
      "step": 19642
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.149751911381122,
      "learning_rate": 3.964806049171027e-07,
      "loss": 0.5406,
      "step": 19643
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.909769895364347,
      "learning_rate": 3.9619873896257045e-07,
      "loss": 0.4871,
      "step": 19644
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.624790997273116,
      "learning_rate": 3.959169691023079e-07,
      "loss": 0.571,
      "step": 19645
    },
    {
      "epoch": 1.75,
      "grad_norm": 11.803390920989672,
      "learning_rate": 3.956352953421955e-07,
      "loss": 0.6361,
      "step": 19646
    },
    {
      "epoch": 1.75,
      "grad_norm": 9.017606468806923,
      "learning_rate": 3.9535371768811394e-07,
      "loss": 0.6447,
      "step": 19647
    },
    {
      "epoch": 1.75,
      "grad_norm": 7.0913966995425515,
      "learning_rate": 3.950722361459386e-07,
      "loss": 0.5459,
      "step": 19648
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.284812986705919,
      "learning_rate": 3.947908507215459e-07,
      "loss": 0.5229,
      "step": 19649
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.244625809701947,
      "learning_rate": 3.945095614208089e-07,
      "loss": 0.5272,
      "step": 19650
    },
    {
      "epoch": 1.75,
      "grad_norm": 7.467635244493185,
      "learning_rate": 3.9422836824959844e-07,
      "loss": 0.5399,
      "step": 19651
    },
    {
      "epoch": 1.75,
      "grad_norm": 8.853271392184466,
      "learning_rate": 3.939472712137848e-07,
      "loss": 0.5826,
      "step": 19652
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.9165074121467685,
      "learning_rate": 3.936662703192357e-07,
      "loss": 0.6326,
      "step": 19653
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.35021154117626,
      "learning_rate": 3.933853655718145e-07,
      "loss": 0.5486,
      "step": 19654
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.8136929381373585,
      "learning_rate": 3.931045569773856e-07,
      "loss": 0.5998,
      "step": 19655
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.5924693334908655,
      "learning_rate": 3.928238445418098e-07,
      "loss": 0.5502,
      "step": 19656
    },
    {
      "epoch": 1.75,
      "grad_norm": 7.447327998709474,
      "learning_rate": 3.9254322827094795e-07,
      "loss": 0.5442,
      "step": 19657
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.300775980520395,
      "learning_rate": 3.922627081706548e-07,
      "loss": 0.5236,
      "step": 19658
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.699089403460558,
      "learning_rate": 3.919822842467874e-07,
      "loss": 0.553,
      "step": 19659
    },
    {
      "epoch": 1.75,
      "grad_norm": 8.170776528048355,
      "learning_rate": 3.917019565051988e-07,
      "loss": 0.5772,
      "step": 19660
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.9955361881803855,
      "learning_rate": 3.9142172495173935e-07,
      "loss": 0.5276,
      "step": 19661
    },
    {
      "epoch": 1.75,
      "grad_norm": 8.103680470225793,
      "learning_rate": 3.9114158959225824e-07,
      "loss": 0.5698,
      "step": 19662
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.0866402050950965,
      "learning_rate": 3.9086155043260355e-07,
      "loss": 0.6669,
      "step": 19663
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.722969461717848,
      "learning_rate": 3.905816074786206e-07,
      "loss": 0.5767,
      "step": 19664
    },
    {
      "epoch": 1.75,
      "grad_norm": 8.087856638722865,
      "learning_rate": 3.9030176073615144e-07,
      "loss": 0.5561,
      "step": 19665
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.308731137366994,
      "learning_rate": 3.9002201021103914e-07,
      "loss": 0.5659,
      "step": 19666
    },
    {
      "epoch": 1.75,
      "grad_norm": 7.993088919015334,
      "learning_rate": 3.897423559091207e-07,
      "loss": 0.5559,
      "step": 19667
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.734790214046318,
      "learning_rate": 3.8946279783623474e-07,
      "loss": 0.5341,
      "step": 19668
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.904448352937969,
      "learning_rate": 3.8918333599821554e-07,
      "loss": 0.5867,
      "step": 19669
    },
    {
      "epoch": 1.75,
      "grad_norm": 7.344615521789546,
      "learning_rate": 3.8890397040089677e-07,
      "loss": 0.5566,
      "step": 19670
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.072070371082955,
      "learning_rate": 3.886247010501093e-07,
      "loss": 0.5497,
      "step": 19671
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.311034080029039,
      "learning_rate": 3.883455279516835e-07,
      "loss": 0.6418,
      "step": 19672
    },
    {
      "epoch": 1.76,
      "grad_norm": 8.06748364551857,
      "learning_rate": 3.880664511114457e-07,
      "loss": 0.5327,
      "step": 19673
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.038241050006686,
      "learning_rate": 3.8778747053521914e-07,
      "loss": 0.561,
      "step": 19674
    },
    {
      "epoch": 1.76,
      "grad_norm": 11.924382084564126,
      "learning_rate": 3.8750858622882914e-07,
      "loss": 0.6141,
      "step": 19675
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.580861376687991,
      "learning_rate": 3.8722979819809656e-07,
      "loss": 0.6454,
      "step": 19676
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.81660072710745,
      "learning_rate": 3.869511064488396e-07,
      "loss": 0.5433,
      "step": 19677
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.410725997619927,
      "learning_rate": 3.8667251098687576e-07,
      "loss": 0.5452,
      "step": 19678
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.451134818922314,
      "learning_rate": 3.863940118180215e-07,
      "loss": 0.5197,
      "step": 19679
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.812272061215325,
      "learning_rate": 3.8611560894808776e-07,
      "loss": 0.6051,
      "step": 19680
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.35103294818015,
      "learning_rate": 3.858373023828865e-07,
      "loss": 0.5789,
      "step": 19681
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.834090487047694,
      "learning_rate": 3.855590921282265e-07,
      "loss": 0.6213,
      "step": 19682
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.923483746237946,
      "learning_rate": 3.852809781899147e-07,
      "loss": 0.5946,
      "step": 19683
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.6640073399027315,
      "learning_rate": 3.850029605737571e-07,
      "loss": 0.5899,
      "step": 19684
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.63864027281919,
      "learning_rate": 3.847250392855562e-07,
      "loss": 0.5325,
      "step": 19685
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.344464757765747,
      "learning_rate": 3.844472143311134e-07,
      "loss": 0.5803,
      "step": 19686
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.865971007462977,
      "learning_rate": 3.8416948571622593e-07,
      "loss": 0.6159,
      "step": 19687
    },
    {
      "epoch": 1.76,
      "grad_norm": 8.058557885172691,
      "learning_rate": 3.838918534466918e-07,
      "loss": 0.5729,
      "step": 19688
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.442225475485808,
      "learning_rate": 3.836143175283064e-07,
      "loss": 0.5163,
      "step": 19689
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.273984291209826,
      "learning_rate": 3.8333687796686227e-07,
      "loss": 0.5195,
      "step": 19690
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.417060199214073,
      "learning_rate": 3.8305953476815093e-07,
      "loss": 0.5476,
      "step": 19691
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.761896577743787,
      "learning_rate": 3.827822879379611e-07,
      "loss": 0.5601,
      "step": 19692
    },
    {
      "epoch": 1.76,
      "grad_norm": 8.603299699092851,
      "learning_rate": 3.825051374820793e-07,
      "loss": 0.6041,
      "step": 19693
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.128584504828116,
      "learning_rate": 3.822280834062897e-07,
      "loss": 0.5263,
      "step": 19694
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.438728178288485,
      "learning_rate": 3.8195112571637717e-07,
      "loss": 0.5325,
      "step": 19695
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.466573022249899,
      "learning_rate": 3.8167426441812094e-07,
      "loss": 0.601,
      "step": 19696
    },
    {
      "epoch": 1.76,
      "grad_norm": 9.034090999895605,
      "learning_rate": 3.813974995173003e-07,
      "loss": 0.5683,
      "step": 19697
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.95688227074515,
      "learning_rate": 3.811208310196934e-07,
      "loss": 0.5495,
      "step": 19698
    },
    {
      "epoch": 1.76,
      "grad_norm": 4.697796935557323,
      "learning_rate": 3.808442589310729e-07,
      "loss": 0.5166,
      "step": 19699
    },
    {
      "epoch": 1.76,
      "grad_norm": 8.044370306877486,
      "learning_rate": 3.8056778325721413e-07,
      "loss": 0.56,
      "step": 19700
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.400519954001416,
      "learning_rate": 3.802914040038852e-07,
      "loss": 0.5208,
      "step": 19701
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.327879128051387,
      "learning_rate": 3.80015121176856e-07,
      "loss": 0.5337,
      "step": 19702
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.7260738906728506,
      "learning_rate": 3.797389347818942e-07,
      "loss": 0.5595,
      "step": 19703
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.361503936910675,
      "learning_rate": 3.7946284482476337e-07,
      "loss": 0.6005,
      "step": 19704
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.9419428618247165,
      "learning_rate": 3.791868513112279e-07,
      "loss": 0.6043,
      "step": 19705
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.976117698044675,
      "learning_rate": 3.78910954247047e-07,
      "loss": 0.5631,
      "step": 19706
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.097718921842231,
      "learning_rate": 3.7863515363797955e-07,
      "loss": 0.5758,
      "step": 19707
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.828543222660143,
      "learning_rate": 3.783594494897824e-07,
      "loss": 0.5585,
      "step": 19708
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.847761748363026,
      "learning_rate": 3.780838418082111e-07,
      "loss": 0.579,
      "step": 19709
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.141851893266935,
      "learning_rate": 3.778083305990182e-07,
      "loss": 0.5307,
      "step": 19710
    },
    {
      "epoch": 1.76,
      "grad_norm": 10.464513958055207,
      "learning_rate": 3.775329158679536e-07,
      "loss": 0.5844,
      "step": 19711
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.803489293195507,
      "learning_rate": 3.77257597620766e-07,
      "loss": 0.5404,
      "step": 19712
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.249506640416286,
      "learning_rate": 3.769823758632024e-07,
      "loss": 0.5818,
      "step": 19713
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.2149037699577985,
      "learning_rate": 3.767072506010072e-07,
      "loss": 0.5565,
      "step": 19714
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.9714467332937575,
      "learning_rate": 3.7643222183992465e-07,
      "loss": 0.5618,
      "step": 19715
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.076929775087157,
      "learning_rate": 3.7615728958569296e-07,
      "loss": 0.5475,
      "step": 19716
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.796781702767706,
      "learning_rate": 3.75882453844052e-07,
      "loss": 0.5355,
      "step": 19717
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.586681731787227,
      "learning_rate": 3.7560771462073876e-07,
      "loss": 0.6308,
      "step": 19718
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.332125875279872,
      "learning_rate": 3.753330719214865e-07,
      "loss": 0.5621,
      "step": 19719
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.860786109682944,
      "learning_rate": 3.750585257520284e-07,
      "loss": 0.5771,
      "step": 19720
    },
    {
      "epoch": 1.76,
      "grad_norm": 9.947297923046218,
      "learning_rate": 3.7478407611809544e-07,
      "loss": 0.5491,
      "step": 19721
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.345106314638253,
      "learning_rate": 3.745097230254152e-07,
      "loss": 0.64,
      "step": 19722
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.481295585429623,
      "learning_rate": 3.7423546647971654e-07,
      "loss": 0.5628,
      "step": 19723
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.134732626018723,
      "learning_rate": 3.7396130648672093e-07,
      "loss": 0.542,
      "step": 19724
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.0671710304577395,
      "learning_rate": 3.736872430521521e-07,
      "loss": 0.5529,
      "step": 19725
    },
    {
      "epoch": 1.76,
      "grad_norm": 8.304724173653264,
      "learning_rate": 3.7341327618173053e-07,
      "loss": 0.5916,
      "step": 19726
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.030357661682252,
      "learning_rate": 3.731394058811755e-07,
      "loss": 0.5302,
      "step": 19727
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.320311793260315,
      "learning_rate": 3.7286563215620184e-07,
      "loss": 0.5766,
      "step": 19728
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.893324175858843,
      "learning_rate": 3.725919550125262e-07,
      "loss": 0.5099,
      "step": 19729
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.458103224275066,
      "learning_rate": 3.723183744558595e-07,
      "loss": 0.5881,
      "step": 19730
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.858791109486914,
      "learning_rate": 3.720448904919116e-07,
      "loss": 0.6084,
      "step": 19731
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.4961283967112955,
      "learning_rate": 3.7177150312639134e-07,
      "loss": 0.5737,
      "step": 19732
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.450248048136858,
      "learning_rate": 3.714982123650057e-07,
      "loss": 0.6134,
      "step": 19733
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.048636809310132,
      "learning_rate": 3.71225018213458e-07,
      "loss": 0.5653,
      "step": 19734
    },
    {
      "epoch": 1.76,
      "grad_norm": 4.867077477204794,
      "learning_rate": 3.7095192067745145e-07,
      "loss": 0.5277,
      "step": 19735
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.943465998775884,
      "learning_rate": 3.7067891976268697e-07,
      "loss": 0.5379,
      "step": 19736
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.585122244417388,
      "learning_rate": 3.7040601547486175e-07,
      "loss": 0.6059,
      "step": 19737
    },
    {
      "epoch": 1.76,
      "grad_norm": 8.6994514473219,
      "learning_rate": 3.7013320781967175e-07,
      "loss": 0.5257,
      "step": 19738
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.184764067835705,
      "learning_rate": 3.698604968028119e-07,
      "loss": 0.5752,
      "step": 19739
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.6572518244666465,
      "learning_rate": 3.695878824299742e-07,
      "loss": 0.4934,
      "step": 19740
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.806123344110607,
      "learning_rate": 3.693153647068487e-07,
      "loss": 0.533,
      "step": 19741
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.146215916881925,
      "learning_rate": 3.6904294363912576e-07,
      "loss": 0.5193,
      "step": 19742
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.652121123894609,
      "learning_rate": 3.68770619232488e-07,
      "loss": 0.5629,
      "step": 19743
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.285310359550078,
      "learning_rate": 3.684983914926227e-07,
      "loss": 0.5133,
      "step": 19744
    },
    {
      "epoch": 1.76,
      "grad_norm": 9.350167239186117,
      "learning_rate": 3.682262604252096e-07,
      "loss": 0.5507,
      "step": 19745
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.274868050331939,
      "learning_rate": 3.679542260359298e-07,
      "loss": 0.6073,
      "step": 19746
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.936635482477075,
      "learning_rate": 3.67682288330462e-07,
      "loss": 0.5798,
      "step": 19747
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.5961920605956905,
      "learning_rate": 3.6741044731448127e-07,
      "loss": 0.5442,
      "step": 19748
    },
    {
      "epoch": 1.76,
      "grad_norm": 4.8173863664966206,
      "learning_rate": 3.671387029936635e-07,
      "loss": 0.5364,
      "step": 19749
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.304272652554368,
      "learning_rate": 3.668670553736786e-07,
      "loss": 0.5379,
      "step": 19750
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.74849161684889,
      "learning_rate": 3.665955044601971e-07,
      "loss": 0.5789,
      "step": 19751
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.634909395963369,
      "learning_rate": 3.663240502588877e-07,
      "loss": 0.5602,
      "step": 19752
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.403485718847094,
      "learning_rate": 3.6605269277541657e-07,
      "loss": 0.5026,
      "step": 19753
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.050318399358755,
      "learning_rate": 3.657814320154468e-07,
      "loss": 0.5482,
      "step": 19754
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.405296612528212,
      "learning_rate": 3.655102679846423e-07,
      "loss": 0.6187,
      "step": 19755
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.025888001570944,
      "learning_rate": 3.6523920068866014e-07,
      "loss": 0.6519,
      "step": 19756
    },
    {
      "epoch": 1.76,
      "grad_norm": 8.588271850408525,
      "learning_rate": 3.649682301331597e-07,
      "loss": 0.5306,
      "step": 19757
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.668522562020544,
      "learning_rate": 3.646973563237982e-07,
      "loss": 0.5286,
      "step": 19758
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.799509987159986,
      "learning_rate": 3.6442657926622713e-07,
      "loss": 0.5474,
      "step": 19759
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.581517093933606,
      "learning_rate": 3.6415589896609984e-07,
      "loss": 0.5749,
      "step": 19760
    },
    {
      "epoch": 1.76,
      "grad_norm": 8.787511155511702,
      "learning_rate": 3.6388531542906556e-07,
      "loss": 0.5561,
      "step": 19761
    },
    {
      "epoch": 1.76,
      "grad_norm": 10.273321978579942,
      "learning_rate": 3.6361482866077326e-07,
      "loss": 0.6136,
      "step": 19762
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.526908755554605,
      "learning_rate": 3.6334443866686663e-07,
      "loss": 0.6197,
      "step": 19763
    },
    {
      "epoch": 1.76,
      "grad_norm": 8.26608377834526,
      "learning_rate": 3.6307414545299124e-07,
      "loss": 0.5708,
      "step": 19764
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.801813393890665,
      "learning_rate": 3.628039490247881e-07,
      "loss": 0.5803,
      "step": 19765
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.790158674123941,
      "learning_rate": 3.625338493878977e-07,
      "loss": 0.6113,
      "step": 19766
    },
    {
      "epoch": 1.76,
      "grad_norm": 8.263280593630943,
      "learning_rate": 3.622638465479572e-07,
      "loss": 0.634,
      "step": 19767
    },
    {
      "epoch": 1.76,
      "grad_norm": 4.74747306367983,
      "learning_rate": 3.6199394051060266e-07,
      "loss": 0.5422,
      "step": 19768
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.885327655526138,
      "learning_rate": 3.6172413128146734e-07,
      "loss": 0.5585,
      "step": 19769
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.927287605789044,
      "learning_rate": 3.614544188661828e-07,
      "loss": 0.6487,
      "step": 19770
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.734015023086613,
      "learning_rate": 3.6118480327037965e-07,
      "loss": 0.5412,
      "step": 19771
    },
    {
      "epoch": 1.76,
      "grad_norm": 8.311331912711692,
      "learning_rate": 3.6091528449968493e-07,
      "loss": 0.5211,
      "step": 19772
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.212081011957654,
      "learning_rate": 3.606458625597237e-07,
      "loss": 0.5607,
      "step": 19773
    },
    {
      "epoch": 1.76,
      "grad_norm": 4.974769716714912,
      "learning_rate": 3.6037653745612024e-07,
      "loss": 0.5809,
      "step": 19774
    },
    {
      "epoch": 1.76,
      "grad_norm": 8.546083207619608,
      "learning_rate": 3.6010730919449676e-07,
      "loss": 0.5804,
      "step": 19775
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.117149025502638,
      "learning_rate": 3.59838177780471e-07,
      "loss": 0.6099,
      "step": 19776
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.0548294562165665,
      "learning_rate": 3.595691432196613e-07,
      "loss": 0.5846,
      "step": 19777
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.532277937696835,
      "learning_rate": 3.593002055176836e-07,
      "loss": 0.5448,
      "step": 19778
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.9295487164270915,
      "learning_rate": 3.5903136468015134e-07,
      "loss": 0.6589,
      "step": 19779
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.6568565573973695,
      "learning_rate": 3.587626207126754e-07,
      "loss": 0.5938,
      "step": 19780
    },
    {
      "epoch": 1.76,
      "grad_norm": 8.411649343190227,
      "learning_rate": 3.5849397362086704e-07,
      "loss": 0.5294,
      "step": 19781
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.243812073088526,
      "learning_rate": 3.5822542341033105e-07,
      "loss": 0.6329,
      "step": 19782
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.840356616702833,
      "learning_rate": 3.5795697008667363e-07,
      "loss": 0.6179,
      "step": 19783
    },
    {
      "epoch": 1.76,
      "grad_norm": 11.0107135816396,
      "learning_rate": 3.576886136554991e-07,
      "loss": 0.6116,
      "step": 19784
    },
    {
      "epoch": 1.77,
      "grad_norm": 8.242020509575376,
      "learning_rate": 3.5742035412240806e-07,
      "loss": 0.6004,
      "step": 19785
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.579636082252926,
      "learning_rate": 3.5715219149300097e-07,
      "loss": 0.52,
      "step": 19786
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.526261763109346,
      "learning_rate": 3.5688412577287334e-07,
      "loss": 0.4934,
      "step": 19787
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.872883082228418,
      "learning_rate": 3.5661615696762244e-07,
      "loss": 0.5101,
      "step": 19788
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.388711770039403,
      "learning_rate": 3.5634828508283925e-07,
      "loss": 0.572,
      "step": 19789
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.7266480388264505,
      "learning_rate": 3.560805101241166e-07,
      "loss": 0.5758,
      "step": 19790
    },
    {
      "epoch": 1.77,
      "grad_norm": 10.266569134607847,
      "learning_rate": 3.558128320970428e-07,
      "loss": 0.5476,
      "step": 19791
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.9299683055178445,
      "learning_rate": 3.555452510072055e-07,
      "loss": 0.5567,
      "step": 19792
    },
    {
      "epoch": 1.77,
      "grad_norm": 9.048253508202187,
      "learning_rate": 3.552777668601903e-07,
      "loss": 0.5712,
      "step": 19793
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.271724802297707,
      "learning_rate": 3.550103796615806e-07,
      "loss": 0.567,
      "step": 19794
    },
    {
      "epoch": 1.77,
      "grad_norm": 9.02523986090543,
      "learning_rate": 3.5474308941695625e-07,
      "loss": 0.5673,
      "step": 19795
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.922454154372157,
      "learning_rate": 3.544758961318967e-07,
      "loss": 0.5283,
      "step": 19796
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.850571909802351,
      "learning_rate": 3.542087998119792e-07,
      "loss": 0.5819,
      "step": 19797
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.415124762273409,
      "learning_rate": 3.539418004627793e-07,
      "loss": 0.4967,
      "step": 19798
    },
    {
      "epoch": 1.77,
      "grad_norm": 9.01876678519449,
      "learning_rate": 3.536748980898702e-07,
      "loss": 0.5514,
      "step": 19799
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.816093656609749,
      "learning_rate": 3.534080926988215e-07,
      "loss": 0.5981,
      "step": 19800
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.019152331932099,
      "learning_rate": 3.531413842952036e-07,
      "loss": 0.5694,
      "step": 19801
    },
    {
      "epoch": 1.77,
      "grad_norm": 8.310338777569969,
      "learning_rate": 3.5287477288458214e-07,
      "loss": 0.5202,
      "step": 19802
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.998713206641457,
      "learning_rate": 3.526082584725232e-07,
      "loss": 0.5964,
      "step": 19803
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.790716865541022,
      "learning_rate": 3.523418410645885e-07,
      "loss": 0.595,
      "step": 19804
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.140708123764078,
      "learning_rate": 3.520755206663401e-07,
      "loss": 0.589,
      "step": 19805
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.8760759167707155,
      "learning_rate": 3.5180929728333757e-07,
      "loss": 0.6384,
      "step": 19806
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.382948107444717,
      "learning_rate": 3.515431709211353e-07,
      "loss": 0.5672,
      "step": 19807
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.198849219329035,
      "learning_rate": 3.5127714158528945e-07,
      "loss": 0.5979,
      "step": 19808
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.105843434513304,
      "learning_rate": 3.510112092813528e-07,
      "loss": 0.5652,
      "step": 19809
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.4604144045706375,
      "learning_rate": 3.5074537401487643e-07,
      "loss": 0.5487,
      "step": 19810
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.585257963615427,
      "learning_rate": 3.504796357914081e-07,
      "loss": 0.5278,
      "step": 19811
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.85687431488126,
      "learning_rate": 3.502139946164962e-07,
      "loss": 0.6007,
      "step": 19812
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.204431428670946,
      "learning_rate": 3.499484504956835e-07,
      "loss": 0.5582,
      "step": 19813
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.930613096449291,
      "learning_rate": 3.496830034345133e-07,
      "loss": 0.573,
      "step": 19814
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.825722637993685,
      "learning_rate": 3.494176534385263e-07,
      "loss": 0.6175,
      "step": 19815
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.248278537176491,
      "learning_rate": 3.4915240051326236e-07,
      "loss": 0.6196,
      "step": 19816
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.35738898678409,
      "learning_rate": 3.4888724466425605e-07,
      "loss": 0.5608,
      "step": 19817
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.526357044254078,
      "learning_rate": 3.4862218589704235e-07,
      "loss": 0.5674,
      "step": 19818
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.052975924643498,
      "learning_rate": 3.483572242171551e-07,
      "loss": 0.5131,
      "step": 19819
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.815312778465246,
      "learning_rate": 3.4809235963012334e-07,
      "loss": 0.5455,
      "step": 19820
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.709245496910512,
      "learning_rate": 3.478275921414759e-07,
      "loss": 0.5722,
      "step": 19821
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.120071872697752,
      "learning_rate": 3.475629217567394e-07,
      "loss": 0.5583,
      "step": 19822
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.776473695584824,
      "learning_rate": 3.4729834848143795e-07,
      "loss": 0.5678,
      "step": 19823
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.567680665028,
      "learning_rate": 3.4703387232109475e-07,
      "loss": 0.5727,
      "step": 19824
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.764916745623874,
      "learning_rate": 3.467694932812299e-07,
      "loss": 0.5915,
      "step": 19825
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.460790126270922,
      "learning_rate": 3.4650521136736114e-07,
      "loss": 0.5938,
      "step": 19826
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.131463882155137,
      "learning_rate": 3.4624102658500524e-07,
      "loss": 0.5127,
      "step": 19827
    },
    {
      "epoch": 1.77,
      "grad_norm": 8.949487518865457,
      "learning_rate": 3.459769389396761e-07,
      "loss": 0.5432,
      "step": 19828
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.177134577126785,
      "learning_rate": 3.45712948436886e-07,
      "loss": 0.4936,
      "step": 19829
    },
    {
      "epoch": 1.77,
      "grad_norm": 8.065609896269185,
      "learning_rate": 3.4544905508214655e-07,
      "loss": 0.5712,
      "step": 19830
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.648440366391174,
      "learning_rate": 3.4518525888096343e-07,
      "loss": 0.5951,
      "step": 19831
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.300283756571457,
      "learning_rate": 3.4492155983884556e-07,
      "loss": 0.6456,
      "step": 19832
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.828136074145293,
      "learning_rate": 3.4465795796129464e-07,
      "loss": 0.5434,
      "step": 19833
    },
    {
      "epoch": 1.77,
      "grad_norm": 8.303864997169875,
      "learning_rate": 3.443944532538135e-07,
      "loss": 0.562,
      "step": 19834
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.939664514824431,
      "learning_rate": 3.4413104572190324e-07,
      "loss": 0.532,
      "step": 19835
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.598829608815354,
      "learning_rate": 3.438677353710607e-07,
      "loss": 0.5154,
      "step": 19836
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.942962859242887,
      "learning_rate": 3.436045222067824e-07,
      "loss": 0.5506,
      "step": 19837
    },
    {
      "epoch": 1.77,
      "grad_norm": 8.225452730414359,
      "learning_rate": 3.4334140623456356e-07,
      "loss": 0.6093,
      "step": 19838
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.473021262714535,
      "learning_rate": 3.430783874598942e-07,
      "loss": 0.5466,
      "step": 19839
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.3416327601327245,
      "learning_rate": 3.428154658882649e-07,
      "loss": 0.5535,
      "step": 19840
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.236169734630806,
      "learning_rate": 3.4255264152516345e-07,
      "loss": 0.5628,
      "step": 19841
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.260951355414242,
      "learning_rate": 3.4228991437607605e-07,
      "loss": 0.6115,
      "step": 19842
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.292846752515909,
      "learning_rate": 3.4202728444648615e-07,
      "loss": 0.5502,
      "step": 19843
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.559231224773722,
      "learning_rate": 3.417647517418771e-07,
      "loss": 0.5444,
      "step": 19844
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.999625901639536,
      "learning_rate": 3.415023162677278e-07,
      "loss": 0.5805,
      "step": 19845
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.530595238096674,
      "learning_rate": 3.4123997802951447e-07,
      "loss": 0.534,
      "step": 19846
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.052958632332279,
      "learning_rate": 3.4097773703271386e-07,
      "loss": 0.5417,
      "step": 19847
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.833686918324576,
      "learning_rate": 3.407155932828005e-07,
      "loss": 0.5903,
      "step": 19848
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.491370053090296,
      "learning_rate": 3.4045354678524545e-07,
      "loss": 0.5532,
      "step": 19849
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.564490451322855,
      "learning_rate": 3.401915975455178e-07,
      "loss": 0.5619,
      "step": 19850
    },
    {
      "epoch": 1.77,
      "grad_norm": 8.996210561883526,
      "learning_rate": 3.39929745569087e-07,
      "loss": 0.5359,
      "step": 19851
    },
    {
      "epoch": 1.77,
      "grad_norm": 8.824081712006613,
      "learning_rate": 3.3966799086141645e-07,
      "loss": 0.552,
      "step": 19852
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.931618013253737,
      "learning_rate": 3.394063334279707e-07,
      "loss": 0.6182,
      "step": 19853
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.894340030052003,
      "learning_rate": 3.3914477327421094e-07,
      "loss": 0.5615,
      "step": 19854
    },
    {
      "epoch": 1.77,
      "grad_norm": 8.863943483280599,
      "learning_rate": 3.388833104055972e-07,
      "loss": 0.558,
      "step": 19855
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.204441547676108,
      "learning_rate": 3.386219448275868e-07,
      "loss": 0.536,
      "step": 19856
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.454503651815953,
      "learning_rate": 3.383606765456354e-07,
      "loss": 0.5627,
      "step": 19857
    },
    {
      "epoch": 1.77,
      "grad_norm": 8.258598421760949,
      "learning_rate": 3.380995055651959e-07,
      "loss": 0.6474,
      "step": 19858
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.882433445377432,
      "learning_rate": 3.3783843189171996e-07,
      "loss": 0.5382,
      "step": 19859
    },
    {
      "epoch": 1.77,
      "grad_norm": 8.444347492728252,
      "learning_rate": 3.375774555306566e-07,
      "loss": 0.5777,
      "step": 19860
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.597255840515881,
      "learning_rate": 3.373165764874531e-07,
      "loss": 0.6223,
      "step": 19861
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.943766263272918,
      "learning_rate": 3.370557947675546e-07,
      "loss": 0.504,
      "step": 19862
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.1621031127805495,
      "learning_rate": 3.36795110376405e-07,
      "loss": 0.5519,
      "step": 19863
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.044063591679987,
      "learning_rate": 3.3653452331944615e-07,
      "loss": 0.5793,
      "step": 19864
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.2166516687121725,
      "learning_rate": 3.3627403360211586e-07,
      "loss": 0.6351,
      "step": 19865
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.244270833836846,
      "learning_rate": 3.3601364122985094e-07,
      "loss": 0.5477,
      "step": 19866
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.3067780375560565,
      "learning_rate": 3.3575334620808755e-07,
      "loss": 0.5119,
      "step": 19867
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.531854930729359,
      "learning_rate": 3.3549314854225856e-07,
      "loss": 0.5325,
      "step": 19868
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.075554432814713,
      "learning_rate": 3.3523304823779636e-07,
      "loss": 0.5928,
      "step": 19869
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.577545386442936,
      "learning_rate": 3.349730453001271e-07,
      "loss": 0.5395,
      "step": 19870
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.080993615097719,
      "learning_rate": 3.3471313973467923e-07,
      "loss": 0.5691,
      "step": 19871
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.262079866286539,
      "learning_rate": 3.344533315468784e-07,
      "loss": 0.5027,
      "step": 19872
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.376951835057217,
      "learning_rate": 3.34193620742147e-07,
      "loss": 0.5458,
      "step": 19873
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.627236103788549,
      "learning_rate": 3.339340073259056e-07,
      "loss": 0.5524,
      "step": 19874
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.4779485235456615,
      "learning_rate": 3.336744913035733e-07,
      "loss": 0.5455,
      "step": 19875
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.369958093907991,
      "learning_rate": 3.334150726805663e-07,
      "loss": 0.6611,
      "step": 19876
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.294401342969553,
      "learning_rate": 3.3315575146230126e-07,
      "loss": 0.5769,
      "step": 19877
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.014655902332615,
      "learning_rate": 3.32896527654189e-07,
      "loss": 0.609,
      "step": 19878
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.3182136684944705,
      "learning_rate": 3.326374012616407e-07,
      "loss": 0.5464,
      "step": 19879
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.159900933542396,
      "learning_rate": 3.323783722900653e-07,
      "loss": 0.6302,
      "step": 19880
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.707812345858786,
      "learning_rate": 3.321194407448697e-07,
      "loss": 0.5835,
      "step": 19881
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.855744651937401,
      "learning_rate": 3.318606066314589e-07,
      "loss": 0.5771,
      "step": 19882
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.816772915787004,
      "learning_rate": 3.316018699552337e-07,
      "loss": 0.6057,
      "step": 19883
    },
    {
      "epoch": 1.77,
      "grad_norm": 8.477727885597911,
      "learning_rate": 3.3134323072159633e-07,
      "loss": 0.5244,
      "step": 19884
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.934849044716972,
      "learning_rate": 3.310846889359448e-07,
      "loss": 0.5992,
      "step": 19885
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.708116320934139,
      "learning_rate": 3.3082624460367587e-07,
      "loss": 0.5422,
      "step": 19886
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.106721428417182,
      "learning_rate": 3.305678977301835e-07,
      "loss": 0.5262,
      "step": 19887
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.824795206033562,
      "learning_rate": 3.303096483208607e-07,
      "loss": 0.5524,
      "step": 19888
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.961877435484607,
      "learning_rate": 3.30051496381098e-07,
      "loss": 0.5791,
      "step": 19889
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.848628772443872,
      "learning_rate": 3.2979344191628295e-07,
      "loss": 0.5303,
      "step": 19890
    },
    {
      "epoch": 1.77,
      "grad_norm": 9.129635218576166,
      "learning_rate": 3.295354849318017e-07,
      "loss": 0.5339,
      "step": 19891
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.499808763268714,
      "learning_rate": 3.2927762543303877e-07,
      "loss": 0.5304,
      "step": 19892
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.252816620916252,
      "learning_rate": 3.290198634253772e-07,
      "loss": 0.5544,
      "step": 19893
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.652323560548498,
      "learning_rate": 3.287621989141965e-07,
      "loss": 0.5564,
      "step": 19894
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.124341573778996,
      "learning_rate": 3.2850463190487627e-07,
      "loss": 0.532,
      "step": 19895
    },
    {
      "epoch": 1.77,
      "grad_norm": 12.040230421749376,
      "learning_rate": 3.2824716240278997e-07,
      "loss": 0.5727,
      "step": 19896
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.679275173992141,
      "learning_rate": 3.279897904133139e-07,
      "loss": 0.5457,
      "step": 19897
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.3535600027842625,
      "learning_rate": 3.2773251594181874e-07,
      "loss": 0.507,
      "step": 19898
    },
    {
      "epoch": 1.78,
      "grad_norm": 8.012256757711764,
      "learning_rate": 3.274753389936758e-07,
      "loss": 0.5887,
      "step": 19899
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.586176287636549,
      "learning_rate": 3.2721825957425236e-07,
      "loss": 0.5903,
      "step": 19900
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.112065606871026,
      "learning_rate": 3.269612776889153e-07,
      "loss": 0.5484,
      "step": 19901
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.3339476509372625,
      "learning_rate": 3.267043933430275e-07,
      "loss": 0.5646,
      "step": 19902
    },
    {
      "epoch": 1.78,
      "grad_norm": 8.681134005179397,
      "learning_rate": 3.264476065419503e-07,
      "loss": 0.5492,
      "step": 19903
    },
    {
      "epoch": 1.78,
      "grad_norm": 8.318287383606528,
      "learning_rate": 3.261909172910449e-07,
      "loss": 0.5883,
      "step": 19904
    },
    {
      "epoch": 1.78,
      "grad_norm": 11.353144395074082,
      "learning_rate": 3.2593432559566816e-07,
      "loss": 0.6049,
      "step": 19905
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.856366768934138,
      "learning_rate": 3.2567783146117694e-07,
      "loss": 0.5644,
      "step": 19906
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.517478836855554,
      "learning_rate": 3.254214348929241e-07,
      "loss": 0.6249,
      "step": 19907
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.8722163204096,
      "learning_rate": 3.251651358962621e-07,
      "loss": 0.6208,
      "step": 19908
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.606752278814589,
      "learning_rate": 3.249089344765399e-07,
      "loss": 0.5371,
      "step": 19909
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.370040541005374,
      "learning_rate": 3.2465283063910503e-07,
      "loss": 0.5353,
      "step": 19910
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.267514551466646,
      "learning_rate": 3.2439682438930363e-07,
      "loss": 0.6554,
      "step": 19911
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.762833467586259,
      "learning_rate": 3.241409157324793e-07,
      "loss": 0.5756,
      "step": 19912
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.5093480017842245,
      "learning_rate": 3.2388510467397326e-07,
      "loss": 0.5616,
      "step": 19913
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.9079793787060515,
      "learning_rate": 3.2362939121912574e-07,
      "loss": 0.5724,
      "step": 19914
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.4384529111924795,
      "learning_rate": 3.2337377537327297e-07,
      "loss": 0.5957,
      "step": 19915
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.581381527023659,
      "learning_rate": 3.2311825714175126e-07,
      "loss": 0.5328,
      "step": 19916
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.11689208054004,
      "learning_rate": 3.2286283652989416e-07,
      "loss": 0.5658,
      "step": 19917
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.1125420684208,
      "learning_rate": 3.2260751354303174e-07,
      "loss": 0.5828,
      "step": 19918
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.190081231509544,
      "learning_rate": 3.2235228818649433e-07,
      "loss": 0.6032,
      "step": 19919
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.641057504384174,
      "learning_rate": 3.220971604656092e-07,
      "loss": 0.5526,
      "step": 19920
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.609030226601544,
      "learning_rate": 3.2184213038570167e-07,
      "loss": 0.6041,
      "step": 19921
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.8615369959995185,
      "learning_rate": 3.215871979520946e-07,
      "loss": 0.5481,
      "step": 19922
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.491288470638519,
      "learning_rate": 3.2133236317010875e-07,
      "loss": 0.5825,
      "step": 19923
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.7488551780205155,
      "learning_rate": 3.2107762604506377e-07,
      "loss": 0.5388,
      "step": 19924
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.941077614973662,
      "learning_rate": 3.2082298658227704e-07,
      "loss": 0.6139,
      "step": 19925
    },
    {
      "epoch": 1.78,
      "grad_norm": 8.440466377744492,
      "learning_rate": 3.2056844478706273e-07,
      "loss": 0.5848,
      "step": 19926
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.880109375237851,
      "learning_rate": 3.203140006647354e-07,
      "loss": 0.5202,
      "step": 19927
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.728455151003148,
      "learning_rate": 3.2005965422060416e-07,
      "loss": 0.5413,
      "step": 19928
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.85808800180692,
      "learning_rate": 3.198054054599786e-07,
      "loss": 0.5935,
      "step": 19929
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.2079445216345235,
      "learning_rate": 3.1955125438816626e-07,
      "loss": 0.4988,
      "step": 19930
    },
    {
      "epoch": 1.78,
      "grad_norm": 9.390979116043923,
      "learning_rate": 3.192972010104717e-07,
      "loss": 0.541,
      "step": 19931
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.204364360990222,
      "learning_rate": 3.190432453321968e-07,
      "loss": 0.5915,
      "step": 19932
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.6490733845241925,
      "learning_rate": 3.187893873586434e-07,
      "loss": 0.5655,
      "step": 19933
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.902796862462204,
      "learning_rate": 3.185356270951101e-07,
      "loss": 0.5611,
      "step": 19934
    },
    {
      "epoch": 1.78,
      "grad_norm": 8.871835270853392,
      "learning_rate": 3.182819645468932e-07,
      "loss": 0.5676,
      "step": 19935
    },
    {
      "epoch": 1.78,
      "grad_norm": 8.061395383849746,
      "learning_rate": 3.180283997192868e-07,
      "loss": 0.607,
      "step": 19936
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.867495299422822,
      "learning_rate": 3.1777493261758497e-07,
      "loss": 0.5616,
      "step": 19937
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.593568104160547,
      "learning_rate": 3.175215632470768e-07,
      "loss": 0.5132,
      "step": 19938
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.18067524867868,
      "learning_rate": 3.1726829161305204e-07,
      "loss": 0.5689,
      "step": 19939
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.8338065624457505,
      "learning_rate": 3.1701511772079696e-07,
      "loss": 0.6029,
      "step": 19940
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.856658728274167,
      "learning_rate": 3.1676204157559566e-07,
      "loss": 0.5926,
      "step": 19941
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.5134771501440385,
      "learning_rate": 3.1650906318273e-07,
      "loss": 0.5303,
      "step": 19942
    },
    {
      "epoch": 1.78,
      "grad_norm": 4.889318488219453,
      "learning_rate": 3.1625618254748136e-07,
      "loss": 0.6141,
      "step": 19943
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.902986609068479,
      "learning_rate": 3.1600339967512774e-07,
      "loss": 0.5249,
      "step": 19944
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.736455874819436,
      "learning_rate": 3.1575071457094595e-07,
      "loss": 0.5951,
      "step": 19945
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.03421371727678,
      "learning_rate": 3.1549812724020856e-07,
      "loss": 0.5702,
      "step": 19946
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.277024105709893,
      "learning_rate": 3.1524563768819015e-07,
      "loss": 0.5371,
      "step": 19947
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.670900985220695,
      "learning_rate": 3.149932459201588e-07,
      "loss": 0.578,
      "step": 19948
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.461478097027307,
      "learning_rate": 3.14740951941383e-07,
      "loss": 0.5873,
      "step": 19949
    },
    {
      "epoch": 1.78,
      "grad_norm": 8.376348893147746,
      "learning_rate": 3.1448875575712965e-07,
      "loss": 0.5645,
      "step": 19950
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.020813598735923,
      "learning_rate": 3.142366573726624e-07,
      "loss": 0.564,
      "step": 19951
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.508072728340795,
      "learning_rate": 3.139846567932436e-07,
      "loss": 0.6183,
      "step": 19952
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.444977334774294,
      "learning_rate": 3.1373275402413306e-07,
      "loss": 0.5702,
      "step": 19953
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.088837154159336,
      "learning_rate": 3.134809490705881e-07,
      "loss": 0.6158,
      "step": 19954
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.664560501031596,
      "learning_rate": 3.1322924193786517e-07,
      "loss": 0.5365,
      "step": 19955
    },
    {
      "epoch": 1.78,
      "grad_norm": 4.5033430031775445,
      "learning_rate": 3.129776326312178e-07,
      "loss": 0.5174,
      "step": 19956
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.477097912531358,
      "learning_rate": 3.1272612115589793e-07,
      "loss": 0.5167,
      "step": 19957
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.16485114274206,
      "learning_rate": 3.1247470751715583e-07,
      "loss": 0.5916,
      "step": 19958
    },
    {
      "epoch": 1.78,
      "grad_norm": 9.037509202936494,
      "learning_rate": 3.122233917202383e-07,
      "loss": 0.6143,
      "step": 19959
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.163394329996369,
      "learning_rate": 3.119721737703918e-07,
      "loss": 0.544,
      "step": 19960
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.33878713165649,
      "learning_rate": 3.117210536728593e-07,
      "loss": 0.5623,
      "step": 19961
    },
    {
      "epoch": 1.78,
      "grad_norm": 4.9105274370656495,
      "learning_rate": 3.1147003143288225e-07,
      "loss": 0.5111,
      "step": 19962
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.237143804635172,
      "learning_rate": 3.112191070557008e-07,
      "loss": 0.6147,
      "step": 19963
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.369140642497617,
      "learning_rate": 3.109682805465525e-07,
      "loss": 0.5318,
      "step": 19964
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.360607519036494,
      "learning_rate": 3.107175519106725e-07,
      "loss": 0.5274,
      "step": 19965
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.48550638956406,
      "learning_rate": 3.10466921153294e-07,
      "loss": 0.5713,
      "step": 19966
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.779915923415577,
      "learning_rate": 3.1021638827964884e-07,
      "loss": 0.5421,
      "step": 19967
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.09051195617959,
      "learning_rate": 3.099659532949656e-07,
      "loss": 0.6235,
      "step": 19968
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.552184461517378,
      "learning_rate": 3.097156162044723e-07,
      "loss": 0.5354,
      "step": 19969
    },
    {
      "epoch": 1.78,
      "grad_norm": 10.149378475491043,
      "learning_rate": 3.094653770133943e-07,
      "loss": 0.5878,
      "step": 19970
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.613567513605767,
      "learning_rate": 3.092152357269551e-07,
      "loss": 0.5218,
      "step": 19971
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.598520161054192,
      "learning_rate": 3.0896519235037393e-07,
      "loss": 0.6006,
      "step": 19972
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.489200072187707,
      "learning_rate": 3.0871524688887155e-07,
      "loss": 0.5665,
      "step": 19973
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.590907813021528,
      "learning_rate": 3.084653993476655e-07,
      "loss": 0.5378,
      "step": 19974
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.598848156363632,
      "learning_rate": 3.082156497319694e-07,
      "loss": 0.6342,
      "step": 19975
    },
    {
      "epoch": 1.78,
      "grad_norm": 8.148583599775362,
      "learning_rate": 3.079659980469962e-07,
      "loss": 0.569,
      "step": 19976
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.127137398707646,
      "learning_rate": 3.07716444297958e-07,
      "loss": 0.5638,
      "step": 19977
    },
    {
      "epoch": 1.78,
      "grad_norm": 8.175072029865344,
      "learning_rate": 3.0746698849006327e-07,
      "loss": 0.616,
      "step": 19978
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.43241974538059,
      "learning_rate": 3.072176306285185e-07,
      "loss": 0.554,
      "step": 19979
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.63019666956596,
      "learning_rate": 3.069683707185284e-07,
      "loss": 0.5456,
      "step": 19980
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.436098471871223,
      "learning_rate": 3.0671920876529593e-07,
      "loss": 0.5816,
      "step": 19981
    },
    {
      "epoch": 1.78,
      "grad_norm": 8.65944392488762,
      "learning_rate": 3.0647014477402147e-07,
      "loss": 0.6021,
      "step": 19982
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.721634623171956,
      "learning_rate": 3.062211787499048e-07,
      "loss": 0.568,
      "step": 19983
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.408777260532234,
      "learning_rate": 3.059723106981421e-07,
      "loss": 0.5464,
      "step": 19984
    },
    {
      "epoch": 1.78,
      "grad_norm": 8.420219221266803,
      "learning_rate": 3.0572354062392727e-07,
      "loss": 0.5582,
      "step": 19985
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.806484928670248,
      "learning_rate": 3.0547486853245325e-07,
      "loss": 0.5477,
      "step": 19986
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.65389241864085,
      "learning_rate": 3.0522629442891026e-07,
      "loss": 0.5494,
      "step": 19987
    },
    {
      "epoch": 1.78,
      "grad_norm": 8.355409377885232,
      "learning_rate": 3.0497781831848703e-07,
      "loss": 0.5632,
      "step": 19988
    },
    {
      "epoch": 1.78,
      "grad_norm": 8.445521977960574,
      "learning_rate": 3.047294402063705e-07,
      "loss": 0.5303,
      "step": 19989
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.560659638175641,
      "learning_rate": 3.0448116009774433e-07,
      "loss": 0.4908,
      "step": 19990
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.606933352353607,
      "learning_rate": 3.0423297799779096e-07,
      "loss": 0.5919,
      "step": 19991
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.139233276894587,
      "learning_rate": 3.0398489391169016e-07,
      "loss": 0.5856,
      "step": 19992
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.795825923382389,
      "learning_rate": 3.0373690784462065e-07,
      "loss": 0.5371,
      "step": 19993
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.777719644848041,
      "learning_rate": 3.0348901980175873e-07,
      "loss": 0.6136,
      "step": 19994
    },
    {
      "epoch": 1.78,
      "grad_norm": 8.626564971563297,
      "learning_rate": 3.032412297882781e-07,
      "loss": 0.5849,
      "step": 19995
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.725997650649992,
      "learning_rate": 3.029935378093507e-07,
      "loss": 0.6495,
      "step": 19996
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.2333274046757,
      "learning_rate": 3.0274594387014856e-07,
      "loss": 0.5761,
      "step": 19997
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.374000655647602,
      "learning_rate": 3.0249844797583637e-07,
      "loss": 0.5634,
      "step": 19998
    },
    {
      "epoch": 1.78,
      "grad_norm": 4.852321610005478,
      "learning_rate": 3.022510501315823e-07,
      "loss": 0.5247,
      "step": 19999
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.9111073618822605,
      "learning_rate": 3.0200375034254936e-07,
      "loss": 0.6089,
      "step": 20000
    },
    {
      "epoch": 1.78,
      "grad_norm": 8.4327216108502,
      "learning_rate": 3.0175654861390015e-07,
      "loss": 0.5755,
      "step": 20001
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.5583233957231295,
      "learning_rate": 3.015094449507938e-07,
      "loss": 0.5844,
      "step": 20002
    },
    {
      "epoch": 1.78,
      "grad_norm": 4.8598549055235996,
      "learning_rate": 3.012624393583885e-07,
      "loss": 0.5733,
      "step": 20003
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.649337722780526,
      "learning_rate": 3.010155318418406e-07,
      "loss": 0.5675,
      "step": 20004
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.976002327871897,
      "learning_rate": 3.007687224063016e-07,
      "loss": 0.6447,
      "step": 20005
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.61105855163293,
      "learning_rate": 3.0052201105692404e-07,
      "loss": 0.5379,
      "step": 20006
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.930824452874397,
      "learning_rate": 3.002753977988582e-07,
      "loss": 0.5548,
      "step": 20007
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.2894448531210925,
      "learning_rate": 3.000288826372516e-07,
      "loss": 0.5651,
      "step": 20008
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.368295200929125,
      "learning_rate": 2.9978246557724856e-07,
      "loss": 0.544,
      "step": 20009
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.1909488984673535,
      "learning_rate": 2.9953614662399487e-07,
      "loss": 0.5263,
      "step": 20010
    },
    {
      "epoch": 1.79,
      "grad_norm": 4.574841672318607,
      "learning_rate": 2.992899257826287e-07,
      "loss": 0.5005,
      "step": 20011
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.6369663076523295,
      "learning_rate": 2.990438030582915e-07,
      "loss": 0.5912,
      "step": 20012
    },
    {
      "epoch": 1.79,
      "grad_norm": 9.027818270058656,
      "learning_rate": 2.9879777845611966e-07,
      "loss": 0.5875,
      "step": 20013
    },
    {
      "epoch": 1.79,
      "grad_norm": 8.07636476034707,
      "learning_rate": 2.9855185198124916e-07,
      "loss": 0.5281,
      "step": 20014
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.022251470348601,
      "learning_rate": 2.9830602363881245e-07,
      "loss": 0.5855,
      "step": 20015
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.216429239939149,
      "learning_rate": 2.980602934339416e-07,
      "loss": 0.5948,
      "step": 20016
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.2770665488088975,
      "learning_rate": 2.978146613717642e-07,
      "loss": 0.5382,
      "step": 20017
    },
    {
      "epoch": 1.79,
      "grad_norm": 5.883293298891001,
      "learning_rate": 2.975691274574094e-07,
      "loss": 0.5701,
      "step": 20018
    },
    {
      "epoch": 1.79,
      "grad_norm": 5.468611407099075,
      "learning_rate": 2.9732369169599984e-07,
      "loss": 0.59,
      "step": 20019
    },
    {
      "epoch": 1.79,
      "grad_norm": 9.420396199017945,
      "learning_rate": 2.970783540926592e-07,
      "loss": 0.5871,
      "step": 20020
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.077486990142298,
      "learning_rate": 2.968331146525094e-07,
      "loss": 0.5283,
      "step": 20021
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.794369306752069,
      "learning_rate": 2.9658797338066814e-07,
      "loss": 0.5757,
      "step": 20022
    },
    {
      "epoch": 1.79,
      "grad_norm": 4.752776892192753,
      "learning_rate": 2.9634293028225346e-07,
      "loss": 0.5395,
      "step": 20023
    },
    {
      "epoch": 1.79,
      "grad_norm": 5.79792030062455,
      "learning_rate": 2.9609798536237854e-07,
      "loss": 0.575,
      "step": 20024
    },
    {
      "epoch": 1.79,
      "grad_norm": 11.017351965576053,
      "learning_rate": 2.958531386261565e-07,
      "loss": 0.5872,
      "step": 20025
    },
    {
      "epoch": 1.79,
      "grad_norm": 5.90667213472733,
      "learning_rate": 2.956083900786988e-07,
      "loss": 0.5241,
      "step": 20026
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.645793675341324,
      "learning_rate": 2.9536373972511354e-07,
      "loss": 0.5911,
      "step": 20027
    },
    {
      "epoch": 1.79,
      "grad_norm": 5.935833361214213,
      "learning_rate": 2.9511918757050673e-07,
      "loss": 0.5566,
      "step": 20028
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.083138449082243,
      "learning_rate": 2.9487473361998477e-07,
      "loss": 0.5534,
      "step": 20029
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.531448200287632,
      "learning_rate": 2.946303778786475e-07,
      "loss": 0.6642,
      "step": 20030
    },
    {
      "epoch": 1.79,
      "grad_norm": 9.741820981070681,
      "learning_rate": 2.943861203515963e-07,
      "loss": 0.5382,
      "step": 20031
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.287714841712965,
      "learning_rate": 2.9414196104393055e-07,
      "loss": 0.5545,
      "step": 20032
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.15074527272271,
      "learning_rate": 2.938978999607456e-07,
      "loss": 0.5641,
      "step": 20033
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.307327988638788,
      "learning_rate": 2.936539371071351e-07,
      "loss": 0.5998,
      "step": 20034
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.759251423109145,
      "learning_rate": 2.934100724881922e-07,
      "loss": 0.6089,
      "step": 20035
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.610813508605387,
      "learning_rate": 2.9316630610900786e-07,
      "loss": 0.5549,
      "step": 20036
    },
    {
      "epoch": 1.79,
      "grad_norm": 5.02581836063025,
      "learning_rate": 2.9292263797466804e-07,
      "loss": 0.5156,
      "step": 20037
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.171809603361232,
      "learning_rate": 2.926790680902597e-07,
      "loss": 0.5229,
      "step": 20038
    },
    {
      "epoch": 1.79,
      "grad_norm": 8.066923120965326,
      "learning_rate": 2.924355964608666e-07,
      "loss": 0.529,
      "step": 20039
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.08464746908951,
      "learning_rate": 2.921922230915719e-07,
      "loss": 0.5894,
      "step": 20040
    },
    {
      "epoch": 1.79,
      "grad_norm": 4.971039554479666,
      "learning_rate": 2.9194894798745487e-07,
      "loss": 0.6143,
      "step": 20041
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.175729814466849,
      "learning_rate": 2.9170577115359256e-07,
      "loss": 0.5427,
      "step": 20042
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.080903762714071,
      "learning_rate": 2.914626925950609e-07,
      "loss": 0.5623,
      "step": 20043
    },
    {
      "epoch": 1.79,
      "grad_norm": 5.2685744109991015,
      "learning_rate": 2.9121971231693416e-07,
      "loss": 0.5779,
      "step": 20044
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.626158456181663,
      "learning_rate": 2.9097683032428436e-07,
      "loss": 0.5886,
      "step": 20045
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.633412617223977,
      "learning_rate": 2.907340466221809e-07,
      "loss": 0.5824,
      "step": 20046
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.577616841077556,
      "learning_rate": 2.9049136121569066e-07,
      "loss": 0.5658,
      "step": 20047
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.56266342898239,
      "learning_rate": 2.902487741098797e-07,
      "loss": 0.5685,
      "step": 20048
    },
    {
      "epoch": 1.79,
      "grad_norm": 5.919900203647548,
      "learning_rate": 2.900062853098112e-07,
      "loss": 0.5509,
      "step": 20049
    },
    {
      "epoch": 1.79,
      "grad_norm": 8.76893650320256,
      "learning_rate": 2.8976389482054666e-07,
      "loss": 0.5857,
      "step": 20050
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.386722961385307,
      "learning_rate": 2.895216026471459e-07,
      "loss": 0.5011,
      "step": 20051
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.47798903737093,
      "learning_rate": 2.8927940879466543e-07,
      "loss": 0.569,
      "step": 20052
    },
    {
      "epoch": 1.79,
      "grad_norm": 8.630274324488346,
      "learning_rate": 2.890373132681612e-07,
      "loss": 0.5251,
      "step": 20053
    },
    {
      "epoch": 1.79,
      "grad_norm": 8.501999471921446,
      "learning_rate": 2.8879531607268696e-07,
      "loss": 0.6033,
      "step": 20054
    },
    {
      "epoch": 1.79,
      "grad_norm": 5.656707458593224,
      "learning_rate": 2.8855341721329256e-07,
      "loss": 0.5285,
      "step": 20055
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.618707329758404,
      "learning_rate": 2.883116166950273e-07,
      "loss": 0.5577,
      "step": 20056
    },
    {
      "epoch": 1.79,
      "grad_norm": 9.935551070342399,
      "learning_rate": 2.8806991452293873e-07,
      "loss": 0.5234,
      "step": 20057
    },
    {
      "epoch": 1.79,
      "grad_norm": 8.475685462178188,
      "learning_rate": 2.8782831070207185e-07,
      "loss": 0.4757,
      "step": 20058
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.575138999245838,
      "learning_rate": 2.875868052374697e-07,
      "loss": 0.6101,
      "step": 20059
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.743419866100189,
      "learning_rate": 2.873453981341734e-07,
      "loss": 0.5723,
      "step": 20060
    },
    {
      "epoch": 1.79,
      "grad_norm": 5.971624256789703,
      "learning_rate": 2.871040893972216e-07,
      "loss": 0.5488,
      "step": 20061
    },
    {
      "epoch": 1.79,
      "grad_norm": 5.428105035918969,
      "learning_rate": 2.868628790316502e-07,
      "loss": 0.6055,
      "step": 20062
    },
    {
      "epoch": 1.79,
      "grad_norm": 5.930397584315674,
      "learning_rate": 2.866217670424942e-07,
      "loss": 0.5573,
      "step": 20063
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.057558700354545,
      "learning_rate": 2.8638075343478676e-07,
      "loss": 0.52,
      "step": 20064
    },
    {
      "epoch": 1.79,
      "grad_norm": 4.383391848950644,
      "learning_rate": 2.8613983821355886e-07,
      "loss": 0.5659,
      "step": 20065
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.571498570221796,
      "learning_rate": 2.858990213838386e-07,
      "loss": 0.475,
      "step": 20066
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.54414147152359,
      "learning_rate": 2.8565830295065267e-07,
      "loss": 0.5577,
      "step": 20067
    },
    {
      "epoch": 1.79,
      "grad_norm": 8.306921497907714,
      "learning_rate": 2.854176829190253e-07,
      "loss": 0.5579,
      "step": 20068
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.371600005457422,
      "learning_rate": 2.851771612939791e-07,
      "loss": 0.5818,
      "step": 20069
    },
    {
      "epoch": 1.79,
      "grad_norm": 8.29877764730848,
      "learning_rate": 2.84936738080534e-07,
      "loss": 0.5357,
      "step": 20070
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.396033657904993,
      "learning_rate": 2.846964132837088e-07,
      "loss": 0.6002,
      "step": 20071
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.142398131015172,
      "learning_rate": 2.8445618690851995e-07,
      "loss": 0.4959,
      "step": 20072
    },
    {
      "epoch": 1.79,
      "grad_norm": 8.281011624109869,
      "learning_rate": 2.842160589599813e-07,
      "loss": 0.5536,
      "step": 20073
    },
    {
      "epoch": 1.79,
      "grad_norm": 8.658376277847436,
      "learning_rate": 2.839760294431049e-07,
      "loss": 0.6011,
      "step": 20074
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.597513769399369,
      "learning_rate": 2.8373609836290174e-07,
      "loss": 0.642,
      "step": 20075
    },
    {
      "epoch": 1.79,
      "grad_norm": 8.767142909881834,
      "learning_rate": 2.8349626572437836e-07,
      "loss": 0.6042,
      "step": 20076
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.039028439613305,
      "learning_rate": 2.8325653153254086e-07,
      "loss": 0.5743,
      "step": 20077
    },
    {
      "epoch": 1.79,
      "grad_norm": 8.469337560411864,
      "learning_rate": 2.8301689579239457e-07,
      "loss": 0.5987,
      "step": 20078
    },
    {
      "epoch": 1.79,
      "grad_norm": 11.433543756344712,
      "learning_rate": 2.8277735850893995e-07,
      "loss": 0.6014,
      "step": 20079
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.273955263599302,
      "learning_rate": 2.8253791968717857e-07,
      "loss": 0.5375,
      "step": 20080
    },
    {
      "epoch": 1.79,
      "grad_norm": 5.917419973075672,
      "learning_rate": 2.822985793321059e-07,
      "loss": 0.5912,
      "step": 20081
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.278213429846615,
      "learning_rate": 2.8205933744871904e-07,
      "loss": 0.5604,
      "step": 20082
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.307008760094153,
      "learning_rate": 2.8182019404201176e-07,
      "loss": 0.5227,
      "step": 20083
    },
    {
      "epoch": 1.79,
      "grad_norm": 8.844916833471185,
      "learning_rate": 2.8158114911697456e-07,
      "loss": 0.5908,
      "step": 20084
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.012835286282313,
      "learning_rate": 2.813422026785984e-07,
      "loss": 0.5559,
      "step": 20085
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.037807048226236,
      "learning_rate": 2.8110335473187033e-07,
      "loss": 0.5188,
      "step": 20086
    },
    {
      "epoch": 1.79,
      "grad_norm": 5.677385792994709,
      "learning_rate": 2.8086460528177484e-07,
      "loss": 0.5502,
      "step": 20087
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.816195303426281,
      "learning_rate": 2.806259543332962e-07,
      "loss": 0.5844,
      "step": 20088
    },
    {
      "epoch": 1.79,
      "grad_norm": 5.319970857596945,
      "learning_rate": 2.8038740189141544e-07,
      "loss": 0.5195,
      "step": 20089
    },
    {
      "epoch": 1.79,
      "grad_norm": 8.62959581043223,
      "learning_rate": 2.801489479611125e-07,
      "loss": 0.6171,
      "step": 20090
    },
    {
      "epoch": 1.79,
      "grad_norm": 4.823164471673383,
      "learning_rate": 2.799105925473633e-07,
      "loss": 0.5438,
      "step": 20091
    },
    {
      "epoch": 1.79,
      "grad_norm": 8.055849503894976,
      "learning_rate": 2.7967233565514395e-07,
      "loss": 0.5271,
      "step": 20092
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.171608570997626,
      "learning_rate": 2.794341772894277e-07,
      "loss": 0.5432,
      "step": 20093
    },
    {
      "epoch": 1.79,
      "grad_norm": 8.434908647170557,
      "learning_rate": 2.791961174551844e-07,
      "loss": 0.542,
      "step": 20094
    },
    {
      "epoch": 1.79,
      "grad_norm": 5.893723459877089,
      "learning_rate": 2.7895815615738395e-07,
      "loss": 0.5197,
      "step": 20095
    },
    {
      "epoch": 1.79,
      "grad_norm": 4.68024606193015,
      "learning_rate": 2.78720293400993e-07,
      "loss": 0.5191,
      "step": 20096
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.417760912338335,
      "learning_rate": 2.784825291909765e-07,
      "loss": 0.5594,
      "step": 20097
    },
    {
      "epoch": 1.79,
      "grad_norm": 9.024787299493008,
      "learning_rate": 2.782448635322971e-07,
      "loss": 0.5439,
      "step": 20098
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.040007624121557,
      "learning_rate": 2.780072964299163e-07,
      "loss": 0.5906,
      "step": 20099
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.761824104452046,
      "learning_rate": 2.777698278887919e-07,
      "loss": 0.6003,
      "step": 20100
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.183857096685306,
      "learning_rate": 2.775324579138805e-07,
      "loss": 0.6285,
      "step": 20101
    },
    {
      "epoch": 1.79,
      "grad_norm": 9.608729120117474,
      "learning_rate": 2.7729518651013696e-07,
      "loss": 0.5871,
      "step": 20102
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.718157675119265,
      "learning_rate": 2.7705801368251407e-07,
      "loss": 0.5476,
      "step": 20103
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.607760833026992,
      "learning_rate": 2.7682093943596275e-07,
      "loss": 0.5542,
      "step": 20104
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.612835912144923,
      "learning_rate": 2.7658396377543027e-07,
      "loss": 0.5605,
      "step": 20105
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.665896476257027,
      "learning_rate": 2.7634708670586376e-07,
      "loss": 0.5668,
      "step": 20106
    },
    {
      "epoch": 1.79,
      "grad_norm": 8.196592338875362,
      "learning_rate": 2.7611030823220694e-07,
      "loss": 0.5852,
      "step": 20107
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.688859439047316,
      "learning_rate": 2.7587362835940214e-07,
      "loss": 0.5902,
      "step": 20108
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.917909872703726,
      "learning_rate": 2.7563704709238915e-07,
      "loss": 0.5674,
      "step": 20109
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.110566334008197,
      "learning_rate": 2.7540056443610743e-07,
      "loss": 0.6073,
      "step": 20110
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.679001371213017,
      "learning_rate": 2.7516418039549134e-07,
      "loss": 0.5868,
      "step": 20111
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.369106403109633,
      "learning_rate": 2.749278949754769e-07,
      "loss": 0.5003,
      "step": 20112
    },
    {
      "epoch": 1.79,
      "grad_norm": 5.919410758715684,
      "learning_rate": 2.7469170818099467e-07,
      "loss": 0.5187,
      "step": 20113
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.631282369987722,
      "learning_rate": 2.74455620016974e-07,
      "loss": 0.5695,
      "step": 20114
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.980788960100082,
      "learning_rate": 2.742196304883443e-07,
      "loss": 0.5769,
      "step": 20115
    },
    {
      "epoch": 1.79,
      "grad_norm": 5.246336638819135,
      "learning_rate": 2.7398373960003e-07,
      "loss": 0.553,
      "step": 20116
    },
    {
      "epoch": 1.79,
      "grad_norm": 5.200200675840701,
      "learning_rate": 2.73747947356956e-07,
      "loss": 0.5499,
      "step": 20117
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.374147287579982,
      "learning_rate": 2.7351225376404276e-07,
      "loss": 0.486,
      "step": 20118
    },
    {
      "epoch": 1.79,
      "grad_norm": 5.953679671078105,
      "learning_rate": 2.732766588262109e-07,
      "loss": 0.5706,
      "step": 20119
    },
    {
      "epoch": 1.79,
      "grad_norm": 8.072275544364853,
      "learning_rate": 2.7304116254837696e-07,
      "loss": 0.6194,
      "step": 20120
    },
    {
      "epoch": 1.79,
      "grad_norm": 5.949555330510494,
      "learning_rate": 2.7280576493545705e-07,
      "loss": 0.5475,
      "step": 20121
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.462675483886381,
      "learning_rate": 2.725704659923639e-07,
      "loss": 0.5424,
      "step": 20122
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.5105995948920805,
      "learning_rate": 2.723352657240097e-07,
      "loss": 0.5179,
      "step": 20123
    },
    {
      "epoch": 1.8,
      "grad_norm": 9.986808834047178,
      "learning_rate": 2.721001641353044e-07,
      "loss": 0.5718,
      "step": 20124
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.2272071566708656,
      "learning_rate": 2.7186516123115294e-07,
      "loss": 0.5719,
      "step": 20125
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.384344030928999,
      "learning_rate": 2.716302570164625e-07,
      "loss": 0.6195,
      "step": 20126
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.897320981823818,
      "learning_rate": 2.713954514961348e-07,
      "loss": 0.5489,
      "step": 20127
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.019304680050157,
      "learning_rate": 2.711607446750719e-07,
      "loss": 0.55,
      "step": 20128
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.689313764272906,
      "learning_rate": 2.7092613655817223e-07,
      "loss": 0.574,
      "step": 20129
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.784032683477858,
      "learning_rate": 2.70691627150334e-07,
      "loss": 0.5173,
      "step": 20130
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.985281964924715,
      "learning_rate": 2.704572164564506e-07,
      "loss": 0.5751,
      "step": 20131
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.56397800230942,
      "learning_rate": 2.7022290448141477e-07,
      "loss": 0.554,
      "step": 20132
    },
    {
      "epoch": 1.8,
      "grad_norm": 8.661211026814701,
      "learning_rate": 2.699886912301186e-07,
      "loss": 0.6258,
      "step": 20133
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.660108703984498,
      "learning_rate": 2.697545767074494e-07,
      "loss": 0.5797,
      "step": 20134
    },
    {
      "epoch": 1.8,
      "grad_norm": 8.948699010113168,
      "learning_rate": 2.6952056091829384e-07,
      "loss": 0.6447,
      "step": 20135
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.422904983296428,
      "learning_rate": 2.6928664386753734e-07,
      "loss": 0.5593,
      "step": 20136
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.407476945382753,
      "learning_rate": 2.690528255600633e-07,
      "loss": 0.6067,
      "step": 20137
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.167292018908641,
      "learning_rate": 2.6881910600074945e-07,
      "loss": 0.5571,
      "step": 20138
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.984016320395842,
      "learning_rate": 2.685854851944758e-07,
      "loss": 0.5658,
      "step": 20139
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.262631120747608,
      "learning_rate": 2.68351963146119e-07,
      "loss": 0.5426,
      "step": 20140
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.850754359742553,
      "learning_rate": 2.681185398605524e-07,
      "loss": 0.5607,
      "step": 20141
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.3884075369153255,
      "learning_rate": 2.6788521534264866e-07,
      "loss": 0.5297,
      "step": 20142
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.731237016526866,
      "learning_rate": 2.6765198959727846e-07,
      "loss": 0.5714,
      "step": 20143
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.4810396948687865,
      "learning_rate": 2.6741886262930895e-07,
      "loss": 0.5405,
      "step": 20144
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.198120580196714,
      "learning_rate": 2.671858344436062e-07,
      "loss": 0.546,
      "step": 20145
    },
    {
      "epoch": 1.8,
      "grad_norm": 8.608420625930954,
      "learning_rate": 2.669529050450348e-07,
      "loss": 0.567,
      "step": 20146
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.108039014843654,
      "learning_rate": 2.667200744384568e-07,
      "loss": 0.5863,
      "step": 20147
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.695144933653571,
      "learning_rate": 2.664873426287312e-07,
      "loss": 0.6712,
      "step": 20148
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.252375085094501,
      "learning_rate": 2.662547096207158e-07,
      "loss": 0.5698,
      "step": 20149
    },
    {
      "epoch": 1.8,
      "grad_norm": 8.930649568365828,
      "learning_rate": 2.660221754192671e-07,
      "loss": 0.5541,
      "step": 20150
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.688669633855491,
      "learning_rate": 2.6578974002923805e-07,
      "loss": 0.5335,
      "step": 20151
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.18244435583854,
      "learning_rate": 2.6555740345547974e-07,
      "loss": 0.5315,
      "step": 20152
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.229205834832341,
      "learning_rate": 2.6532516570284273e-07,
      "loss": 0.5196,
      "step": 20153
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.160749000997784,
      "learning_rate": 2.650930267761742e-07,
      "loss": 0.55,
      "step": 20154
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.866709075995652,
      "learning_rate": 2.648609866803198e-07,
      "loss": 0.578,
      "step": 20155
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.339963671152519,
      "learning_rate": 2.646290454201228e-07,
      "loss": 0.545,
      "step": 20156
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.256990399892547,
      "learning_rate": 2.643972030004233e-07,
      "loss": 0.5233,
      "step": 20157
    },
    {
      "epoch": 1.8,
      "grad_norm": 9.22568983669438,
      "learning_rate": 2.6416545942606176e-07,
      "loss": 0.5678,
      "step": 20158
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.186034803637399,
      "learning_rate": 2.63933814701875e-07,
      "loss": 0.57,
      "step": 20159
    },
    {
      "epoch": 1.8,
      "grad_norm": 8.515086336854688,
      "learning_rate": 2.637022688326979e-07,
      "loss": 0.6178,
      "step": 20160
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.018057135058336,
      "learning_rate": 2.634708218233634e-07,
      "loss": 0.5687,
      "step": 20161
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.675948241946451,
      "learning_rate": 2.6323947367870364e-07,
      "loss": 0.6308,
      "step": 20162
    },
    {
      "epoch": 1.8,
      "grad_norm": 8.805872540355585,
      "learning_rate": 2.6300822440354644e-07,
      "loss": 0.6143,
      "step": 20163
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.770733787243477,
      "learning_rate": 2.6277707400271855e-07,
      "loss": 0.5573,
      "step": 20164
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.359593674477983,
      "learning_rate": 2.6254602248104446e-07,
      "loss": 0.5852,
      "step": 20165
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.433254958115622,
      "learning_rate": 2.6231506984334686e-07,
      "loss": 0.5307,
      "step": 20166
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.39839573302087,
      "learning_rate": 2.620842160944476e-07,
      "loss": 0.5182,
      "step": 20167
    },
    {
      "epoch": 1.8,
      "grad_norm": 8.666709855565117,
      "learning_rate": 2.6185346123916376e-07,
      "loss": 0.5462,
      "step": 20168
    },
    {
      "epoch": 1.8,
      "grad_norm": 9.514473207529655,
      "learning_rate": 2.6162280528231385e-07,
      "loss": 0.5772,
      "step": 20169
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.258844379299862,
      "learning_rate": 2.613922482287101e-07,
      "loss": 0.5361,
      "step": 20170
    },
    {
      "epoch": 1.8,
      "grad_norm": 8.698236717602715,
      "learning_rate": 2.611617900831653e-07,
      "loss": 0.5526,
      "step": 20171
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.778092805130735,
      "learning_rate": 2.609314308504912e-07,
      "loss": 0.661,
      "step": 20172
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.933060416071443,
      "learning_rate": 2.607011705354945e-07,
      "loss": 0.5092,
      "step": 20173
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.096568076146124,
      "learning_rate": 2.6047100914298187e-07,
      "loss": 0.5249,
      "step": 20174
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.871722454429138,
      "learning_rate": 2.6024094667775833e-07,
      "loss": 0.5452,
      "step": 20175
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.487861840642195,
      "learning_rate": 2.600109831446257e-07,
      "loss": 0.5312,
      "step": 20176
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.34964581755061,
      "learning_rate": 2.597811185483823e-07,
      "loss": 0.5704,
      "step": 20177
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.295483374946834,
      "learning_rate": 2.5955135289382707e-07,
      "loss": 0.5899,
      "step": 20178
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.16422746457735,
      "learning_rate": 2.593216861857556e-07,
      "loss": 0.5234,
      "step": 20179
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.587631067781655,
      "learning_rate": 2.590921184289624e-07,
      "loss": 0.5666,
      "step": 20180
    },
    {
      "epoch": 1.8,
      "grad_norm": 8.352644872255844,
      "learning_rate": 2.5886264962823873e-07,
      "loss": 0.6265,
      "step": 20181
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.523037238592465,
      "learning_rate": 2.5863327978837503e-07,
      "loss": 0.5481,
      "step": 20182
    },
    {
      "epoch": 1.8,
      "grad_norm": 9.092102603594158,
      "learning_rate": 2.584040089141576e-07,
      "loss": 0.5649,
      "step": 20183
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.491995993082319,
      "learning_rate": 2.581748370103726e-07,
      "loss": 0.5705,
      "step": 20184
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.821583778949467,
      "learning_rate": 2.5794576408180336e-07,
      "loss": 0.5242,
      "step": 20185
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.617725783770447,
      "learning_rate": 2.577167901332317e-07,
      "loss": 0.5566,
      "step": 20186
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.418464877121288,
      "learning_rate": 2.574879151694365e-07,
      "loss": 0.5672,
      "step": 20187
    },
    {
      "epoch": 1.8,
      "grad_norm": 9.137378178050847,
      "learning_rate": 2.5725913919519563e-07,
      "loss": 0.5827,
      "step": 20188
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.5065553676603045,
      "learning_rate": 2.5703046221528306e-07,
      "loss": 0.6023,
      "step": 20189
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.3121738196496855,
      "learning_rate": 2.568018842344727e-07,
      "loss": 0.6181,
      "step": 20190
    },
    {
      "epoch": 1.8,
      "grad_norm": 8.595294062484992,
      "learning_rate": 2.565734052575364e-07,
      "loss": 0.5329,
      "step": 20191
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.272949341921487,
      "learning_rate": 2.563450252892413e-07,
      "loss": 0.5955,
      "step": 20192
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.459732987583873,
      "learning_rate": 2.5611674433435597e-07,
      "loss": 0.5526,
      "step": 20193
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.848475052652493,
      "learning_rate": 2.5588856239764424e-07,
      "loss": 0.5465,
      "step": 20194
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.997668478173891,
      "learning_rate": 2.5566047948387016e-07,
      "loss": 0.5143,
      "step": 20195
    },
    {
      "epoch": 1.8,
      "grad_norm": 8.810408219611556,
      "learning_rate": 2.554324955977927e-07,
      "loss": 0.5023,
      "step": 20196
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.661909265502535,
      "learning_rate": 2.552046107441719e-07,
      "loss": 0.5549,
      "step": 20197
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.224249175598741,
      "learning_rate": 2.5497682492776397e-07,
      "loss": 0.5587,
      "step": 20198
    },
    {
      "epoch": 1.8,
      "grad_norm": 8.038842813400061,
      "learning_rate": 2.5474913815332294e-07,
      "loss": 0.6019,
      "step": 20199
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.209182126591057,
      "learning_rate": 2.5452155042560265e-07,
      "loss": 0.5777,
      "step": 20200
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.660550765357079,
      "learning_rate": 2.542940617493517e-07,
      "loss": 0.5454,
      "step": 20201
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.971933064081867,
      "learning_rate": 2.540666721293195e-07,
      "loss": 0.5236,
      "step": 20202
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.3567883050424046,
      "learning_rate": 2.538393815702522e-07,
      "loss": 0.5965,
      "step": 20203
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.427784124290656,
      "learning_rate": 2.53612190076894e-07,
      "loss": 0.5048,
      "step": 20204
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.712114182341302,
      "learning_rate": 2.53385097653987e-07,
      "loss": 0.5217,
      "step": 20205
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.386064799644375,
      "learning_rate": 2.5315810430627084e-07,
      "loss": 0.5181,
      "step": 20206
    },
    {
      "epoch": 1.8,
      "grad_norm": 8.36655898038565,
      "learning_rate": 2.5293121003848396e-07,
      "loss": 0.598,
      "step": 20207
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.361123196420978,
      "learning_rate": 2.527044148553631e-07,
      "loss": 0.5368,
      "step": 20208
    },
    {
      "epoch": 1.8,
      "grad_norm": 8.475683347322963,
      "learning_rate": 2.524777187616401e-07,
      "loss": 0.5551,
      "step": 20209
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.328834547261899,
      "learning_rate": 2.522511217620477e-07,
      "loss": 0.5589,
      "step": 20210
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.281499826677118,
      "learning_rate": 2.520246238613161e-07,
      "loss": 0.5942,
      "step": 20211
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.2196985975073735,
      "learning_rate": 2.5179822506417205e-07,
      "loss": 0.6575,
      "step": 20212
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.254794794472285,
      "learning_rate": 2.515719253753429e-07,
      "loss": 0.5362,
      "step": 20213
    },
    {
      "epoch": 1.8,
      "grad_norm": 8.10013068512519,
      "learning_rate": 2.513457247995499e-07,
      "loss": 0.6038,
      "step": 20214
    },
    {
      "epoch": 1.8,
      "grad_norm": 8.14510569168985,
      "learning_rate": 2.5111962334151585e-07,
      "loss": 0.5891,
      "step": 20215
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.123661629703746,
      "learning_rate": 2.5089362100595983e-07,
      "loss": 0.6303,
      "step": 20216
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.58658268470892,
      "learning_rate": 2.5066771779759855e-07,
      "loss": 0.562,
      "step": 20217
    },
    {
      "epoch": 1.8,
      "grad_norm": 8.368661770724051,
      "learning_rate": 2.5044191372114836e-07,
      "loss": 0.591,
      "step": 20218
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.4507698823086415,
      "learning_rate": 2.50216208781322e-07,
      "loss": 0.5436,
      "step": 20219
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.112014714534617,
      "learning_rate": 2.4999060298283085e-07,
      "loss": 0.5592,
      "step": 20220
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.789597482733471,
      "learning_rate": 2.497650963303827e-07,
      "loss": 0.5485,
      "step": 20221
    },
    {
      "epoch": 1.8,
      "grad_norm": 8.821937137363907,
      "learning_rate": 2.495396888286855e-07,
      "loss": 0.5878,
      "step": 20222
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.962520806929792,
      "learning_rate": 2.493143804824438e-07,
      "loss": 0.5223,
      "step": 20223
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.078670628325172,
      "learning_rate": 2.4908917129636055e-07,
      "loss": 0.6204,
      "step": 20224
    },
    {
      "epoch": 1.8,
      "grad_norm": 10.806981495697736,
      "learning_rate": 2.48864061275137e-07,
      "loss": 0.558,
      "step": 20225
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.248956448085863,
      "learning_rate": 2.4863905042347145e-07,
      "loss": 0.5859,
      "step": 20226
    },
    {
      "epoch": 1.8,
      "grad_norm": 8.002940868574399,
      "learning_rate": 2.484141387460598e-07,
      "loss": 0.5836,
      "step": 20227
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.708152259768651,
      "learning_rate": 2.481893262475976e-07,
      "loss": 0.5205,
      "step": 20228
    },
    {
      "epoch": 1.8,
      "grad_norm": 8.006763682248472,
      "learning_rate": 2.479646129327767e-07,
      "loss": 0.6026,
      "step": 20229
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.037161750533865,
      "learning_rate": 2.4773999880628787e-07,
      "loss": 0.5617,
      "step": 20230
    },
    {
      "epoch": 1.8,
      "grad_norm": 9.497997463141866,
      "learning_rate": 2.4751548387281886e-07,
      "loss": 0.5581,
      "step": 20231
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.78184234266397,
      "learning_rate": 2.472910681370572e-07,
      "loss": 0.5204,
      "step": 20232
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.139118241036792,
      "learning_rate": 2.4706675160368564e-07,
      "loss": 0.5963,
      "step": 20233
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.926055683870987,
      "learning_rate": 2.468425342773878e-07,
      "loss": 0.5462,
      "step": 20234
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.5045596124541225,
      "learning_rate": 2.466184161628421e-07,
      "loss": 0.5637,
      "step": 20235
    },
    {
      "epoch": 1.81,
      "grad_norm": 8.538217192743138,
      "learning_rate": 2.46394397264727e-07,
      "loss": 0.6163,
      "step": 20236
    },
    {
      "epoch": 1.81,
      "grad_norm": 8.13355933098946,
      "learning_rate": 2.4617047758771886e-07,
      "loss": 0.5522,
      "step": 20237
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.939928613629592,
      "learning_rate": 2.45946657136491e-07,
      "loss": 0.5702,
      "step": 20238
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.3412676032403885,
      "learning_rate": 2.4572293591571595e-07,
      "loss": 0.5327,
      "step": 20239
    },
    {
      "epoch": 1.81,
      "grad_norm": 8.91457334607669,
      "learning_rate": 2.454993139300627e-07,
      "loss": 0.6077,
      "step": 20240
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.605575552697134,
      "learning_rate": 2.452757911841991e-07,
      "loss": 0.6251,
      "step": 20241
    },
    {
      "epoch": 1.81,
      "grad_norm": 4.548840595188242,
      "learning_rate": 2.450523676827904e-07,
      "loss": 0.5479,
      "step": 20242
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.658580595516935,
      "learning_rate": 2.448290434305006e-07,
      "loss": 0.5561,
      "step": 20243
    },
    {
      "epoch": 1.81,
      "grad_norm": 9.257071663618545,
      "learning_rate": 2.4460581843199103e-07,
      "loss": 0.648,
      "step": 20244
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.462969411566627,
      "learning_rate": 2.4438269269192125e-07,
      "loss": 0.5774,
      "step": 20245
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.835006273534454,
      "learning_rate": 2.4415966621494757e-07,
      "loss": 0.5829,
      "step": 20246
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.6425944640351196,
      "learning_rate": 2.4393673900572565e-07,
      "loss": 0.5863,
      "step": 20247
    },
    {
      "epoch": 1.81,
      "grad_norm": 8.427490403234392,
      "learning_rate": 2.4371391106890904e-07,
      "loss": 0.5843,
      "step": 20248
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.805059156213748,
      "learning_rate": 2.434911824091485e-07,
      "loss": 0.6033,
      "step": 20249
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.7804428134438455,
      "learning_rate": 2.4326855303109243e-07,
      "loss": 0.5301,
      "step": 20250
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.279891578865532,
      "learning_rate": 2.4304602293938883e-07,
      "loss": 0.5779,
      "step": 20251
    },
    {
      "epoch": 1.81,
      "grad_norm": 8.01653329615071,
      "learning_rate": 2.428235921386818e-07,
      "loss": 0.5792,
      "step": 20252
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.693282072244704,
      "learning_rate": 2.4260126063361424e-07,
      "loss": 0.552,
      "step": 20253
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.409428415424622,
      "learning_rate": 2.423790284288263e-07,
      "loss": 0.5432,
      "step": 20254
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.868955644553105,
      "learning_rate": 2.4215689552895773e-07,
      "loss": 0.5725,
      "step": 20255
    },
    {
      "epoch": 1.81,
      "grad_norm": 4.922575558955085,
      "learning_rate": 2.419348619386441e-07,
      "loss": 0.6063,
      "step": 20256
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.9727995920039625,
      "learning_rate": 2.417129276625202e-07,
      "loss": 0.5953,
      "step": 20257
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.675560956718508,
      "learning_rate": 2.4149109270521943e-07,
      "loss": 0.5438,
      "step": 20258
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.2145684580957665,
      "learning_rate": 2.4126935707137035e-07,
      "loss": 0.5508,
      "step": 20259
    },
    {
      "epoch": 1.81,
      "grad_norm": 8.423860542698833,
      "learning_rate": 2.41047720765602e-07,
      "loss": 0.5719,
      "step": 20260
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.966114480203932,
      "learning_rate": 2.408261837925402e-07,
      "loss": 0.5997,
      "step": 20261
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.495993202105955,
      "learning_rate": 2.406047461568095e-07,
      "loss": 0.6325,
      "step": 20262
    },
    {
      "epoch": 1.81,
      "grad_norm": 9.221114166692875,
      "learning_rate": 2.4038340786303293e-07,
      "loss": 0.596,
      "step": 20263
    },
    {
      "epoch": 1.81,
      "grad_norm": 9.145792665420482,
      "learning_rate": 2.4016216891582846e-07,
      "loss": 0.6185,
      "step": 20264
    },
    {
      "epoch": 1.81,
      "grad_norm": 9.06869178338252,
      "learning_rate": 2.3994102931981565e-07,
      "loss": 0.6484,
      "step": 20265
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.962909196054485,
      "learning_rate": 2.397199890796087e-07,
      "loss": 0.5368,
      "step": 20266
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.862883749916136,
      "learning_rate": 2.394990481998216e-07,
      "loss": 0.554,
      "step": 20267
    },
    {
      "epoch": 1.81,
      "grad_norm": 8.030391744636105,
      "learning_rate": 2.392782066850674e-07,
      "loss": 0.5446,
      "step": 20268
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.086423353242284,
      "learning_rate": 2.390574645399546e-07,
      "loss": 0.5412,
      "step": 20269
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.139246017518816,
      "learning_rate": 2.388368217690906e-07,
      "loss": 0.5763,
      "step": 20270
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.942525645131948,
      "learning_rate": 2.386162783770818e-07,
      "loss": 0.5864,
      "step": 20271
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.362242995371688,
      "learning_rate": 2.3839583436853053e-07,
      "loss": 0.5967,
      "step": 20272
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.369010466697994,
      "learning_rate": 2.3817548974803873e-07,
      "loss": 0.5328,
      "step": 20273
    },
    {
      "epoch": 1.81,
      "grad_norm": 8.515571463138,
      "learning_rate": 2.3795524452020491e-07,
      "loss": 0.5719,
      "step": 20274
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.726760212563206,
      "learning_rate": 2.3773509868962708e-07,
      "loss": 0.5471,
      "step": 20275
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.421711806862612,
      "learning_rate": 2.3751505226090044e-07,
      "loss": 0.5811,
      "step": 20276
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.054321393280794,
      "learning_rate": 2.3729510523861632e-07,
      "loss": 0.5798,
      "step": 20277
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.8725345648529155,
      "learning_rate": 2.3707525762736771e-07,
      "loss": 0.5734,
      "step": 20278
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.477584645140424,
      "learning_rate": 2.3685550943174207e-07,
      "loss": 0.5117,
      "step": 20279
    },
    {
      "epoch": 1.81,
      "grad_norm": 4.822659256609229,
      "learning_rate": 2.3663586065632626e-07,
      "loss": 0.5982,
      "step": 20280
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.470572118417263,
      "learning_rate": 2.3641631130570553e-07,
      "loss": 0.6265,
      "step": 20281
    },
    {
      "epoch": 1.81,
      "grad_norm": 4.698795042756981,
      "learning_rate": 2.3619686138446175e-07,
      "loss": 0.4987,
      "step": 20282
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.527633305099976,
      "learning_rate": 2.3597751089717623e-07,
      "loss": 0.5295,
      "step": 20283
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.45247531869541,
      "learning_rate": 2.3575825984842814e-07,
      "loss": 0.5226,
      "step": 20284
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.2717056198569745,
      "learning_rate": 2.355391082427916e-07,
      "loss": 0.5261,
      "step": 20285
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.025750631293702,
      "learning_rate": 2.3532005608484232e-07,
      "loss": 0.5597,
      "step": 20286
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.754779751304873,
      "learning_rate": 2.3510110337915282e-07,
      "loss": 0.6012,
      "step": 20287
    },
    {
      "epoch": 1.81,
      "grad_norm": 8.282886537852962,
      "learning_rate": 2.348822501302922e-07,
      "loss": 0.6136,
      "step": 20288
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.4521609585655835,
      "learning_rate": 2.3466349634283014e-07,
      "loss": 0.6117,
      "step": 20289
    },
    {
      "epoch": 1.81,
      "grad_norm": 8.037015599327136,
      "learning_rate": 2.3444484202133132e-07,
      "loss": 0.5338,
      "step": 20290
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.888350929403232,
      "learning_rate": 2.3422628717035988e-07,
      "loss": 0.5756,
      "step": 20291
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.97800575102655,
      "learning_rate": 2.3400783179447827e-07,
      "loss": 0.5804,
      "step": 20292
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.013635008901206,
      "learning_rate": 2.3378947589824507e-07,
      "loss": 0.5601,
      "step": 20293
    },
    {
      "epoch": 1.81,
      "grad_norm": 8.258622299466971,
      "learning_rate": 2.3357121948621885e-07,
      "loss": 0.5408,
      "step": 20294
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.817804621970263,
      "learning_rate": 2.3335306256295543e-07,
      "loss": 0.5166,
      "step": 20295
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.912078959927987,
      "learning_rate": 2.3313500513300836e-07,
      "loss": 0.6116,
      "step": 20296
    },
    {
      "epoch": 1.81,
      "grad_norm": 4.768575105397824,
      "learning_rate": 2.3291704720092845e-07,
      "loss": 0.5132,
      "step": 20297
    },
    {
      "epoch": 1.81,
      "grad_norm": 8.953804543936467,
      "learning_rate": 2.326991887712654e-07,
      "loss": 0.5926,
      "step": 20298
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.755694945178848,
      "learning_rate": 2.3248142984856615e-07,
      "loss": 0.5903,
      "step": 20299
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.970329605926503,
      "learning_rate": 2.32263770437377e-07,
      "loss": 0.5757,
      "step": 20300
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.766425488596052,
      "learning_rate": 2.3204621054223998e-07,
      "loss": 0.5234,
      "step": 20301
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.555744698525791,
      "learning_rate": 2.3182875016769746e-07,
      "loss": 0.534,
      "step": 20302
    },
    {
      "epoch": 1.81,
      "grad_norm": 8.0790004250328,
      "learning_rate": 2.3161138931828698e-07,
      "loss": 0.6034,
      "step": 20303
    },
    {
      "epoch": 1.81,
      "grad_norm": 9.068260988058736,
      "learning_rate": 2.31394127998546e-07,
      "loss": 0.6468,
      "step": 20304
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.2744644569363635,
      "learning_rate": 2.3117696621300977e-07,
      "loss": 0.5185,
      "step": 20305
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.652952767291488,
      "learning_rate": 2.3095990396621137e-07,
      "loss": 0.623,
      "step": 20306
    },
    {
      "epoch": 1.81,
      "grad_norm": 10.3061385646532,
      "learning_rate": 2.307429412626805e-07,
      "loss": 0.5231,
      "step": 20307
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.0456883788843925,
      "learning_rate": 2.3052607810694627e-07,
      "loss": 0.6147,
      "step": 20308
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.48923810934394,
      "learning_rate": 2.3030931450353565e-07,
      "loss": 0.5635,
      "step": 20309
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.8509502556002655,
      "learning_rate": 2.3009265045697226e-07,
      "loss": 0.576,
      "step": 20310
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.159652899628439,
      "learning_rate": 2.2987608597177856e-07,
      "loss": 0.5843,
      "step": 20311
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.569583914283195,
      "learning_rate": 2.2965962105247486e-07,
      "loss": 0.5574,
      "step": 20312
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.379800991621863,
      "learning_rate": 2.294432557035803e-07,
      "loss": 0.5765,
      "step": 20313
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.005612108773395,
      "learning_rate": 2.2922698992961011e-07,
      "loss": 0.5852,
      "step": 20314
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.802530881160066,
      "learning_rate": 2.2901082373507965e-07,
      "loss": 0.5243,
      "step": 20315
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.518139807223092,
      "learning_rate": 2.2879475712449918e-07,
      "loss": 0.4985,
      "step": 20316
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.564910749318091,
      "learning_rate": 2.285787901023795e-07,
      "loss": 0.5628,
      "step": 20317
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.436699581239236,
      "learning_rate": 2.2836292267322814e-07,
      "loss": 0.5573,
      "step": 20318
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.387014539461933,
      "learning_rate": 2.281471548415509e-07,
      "loss": 0.5511,
      "step": 20319
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.687883795236095,
      "learning_rate": 2.279314866118526e-07,
      "loss": 0.5797,
      "step": 20320
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.282020325526409,
      "learning_rate": 2.2771591798863292e-07,
      "loss": 0.5292,
      "step": 20321
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.466570479630273,
      "learning_rate": 2.2750044897639322e-07,
      "loss": 0.4911,
      "step": 20322
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.1673479147168235,
      "learning_rate": 2.2728507957962942e-07,
      "loss": 0.5686,
      "step": 20323
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.885223293914555,
      "learning_rate": 2.2706980980283734e-07,
      "loss": 0.4986,
      "step": 20324
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.4644937804045615,
      "learning_rate": 2.2685463965051057e-07,
      "loss": 0.5569,
      "step": 20325
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.720995733154218,
      "learning_rate": 2.2663956912714004e-07,
      "loss": 0.5614,
      "step": 20326
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.281391106730937,
      "learning_rate": 2.2642459823721485e-07,
      "loss": 0.5958,
      "step": 20327
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.185822149745963,
      "learning_rate": 2.2620972698522314e-07,
      "loss": 0.5975,
      "step": 20328
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.602707156663584,
      "learning_rate": 2.2599495537564853e-07,
      "loss": 0.5763,
      "step": 20329
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.248202989030173,
      "learning_rate": 2.2578028341297408e-07,
      "loss": 0.562,
      "step": 20330
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.9621458971478045,
      "learning_rate": 2.2556571110168123e-07,
      "loss": 0.4817,
      "step": 20331
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.3292638724103565,
      "learning_rate": 2.253512384462475e-07,
      "loss": 0.592,
      "step": 20332
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.732958776985698,
      "learning_rate": 2.2513686545115098e-07,
      "loss": 0.5825,
      "step": 20333
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.439820134610554,
      "learning_rate": 2.2492259212086642e-07,
      "loss": 0.5692,
      "step": 20334
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.328872209916557,
      "learning_rate": 2.2470841845986524e-07,
      "loss": 0.617,
      "step": 20335
    },
    {
      "epoch": 1.81,
      "grad_norm": 8.346839092780693,
      "learning_rate": 2.244943444726172e-07,
      "loss": 0.5514,
      "step": 20336
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.164455384443192,
      "learning_rate": 2.2428037016359207e-07,
      "loss": 0.5066,
      "step": 20337
    },
    {
      "epoch": 1.81,
      "grad_norm": 10.162037166219404,
      "learning_rate": 2.2406649553725512e-07,
      "loss": 0.5792,
      "step": 20338
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.332782643582894,
      "learning_rate": 2.2385272059807117e-07,
      "loss": 0.5366,
      "step": 20339
    },
    {
      "epoch": 1.81,
      "grad_norm": 9.91812411166732,
      "learning_rate": 2.236390453505022e-07,
      "loss": 0.6217,
      "step": 20340
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.226934221983952,
      "learning_rate": 2.2342546979900847e-07,
      "loss": 0.5992,
      "step": 20341
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.304173516408155,
      "learning_rate": 2.2321199394804705e-07,
      "loss": 0.5989,
      "step": 20342
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.776586743412852,
      "learning_rate": 2.2299861780207433e-07,
      "loss": 0.6497,
      "step": 20343
    },
    {
      "epoch": 1.81,
      "grad_norm": 4.981550200929377,
      "learning_rate": 2.227853413655434e-07,
      "loss": 0.5079,
      "step": 20344
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.373752041155732,
      "learning_rate": 2.2257216464290743e-07,
      "loss": 0.5626,
      "step": 20345
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.711095571130825,
      "learning_rate": 2.2235908763861447e-07,
      "loss": 0.5841,
      "step": 20346
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.27261517869933,
      "learning_rate": 2.2214611035711376e-07,
      "loss": 0.6024,
      "step": 20347
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.999472446176799,
      "learning_rate": 2.2193323280284896e-07,
      "loss": 0.5451,
      "step": 20348
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.920481885578876,
      "learning_rate": 2.2172045498026483e-07,
      "loss": 0.5568,
      "step": 20349
    },
    {
      "epoch": 1.82,
      "grad_norm": 5.2537681777874115,
      "learning_rate": 2.2150777689380122e-07,
      "loss": 0.5309,
      "step": 20350
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.727950782253402,
      "learning_rate": 2.2129519854789784e-07,
      "loss": 0.5635,
      "step": 20351
    },
    {
      "epoch": 1.82,
      "grad_norm": 5.890640442800136,
      "learning_rate": 2.2108271994699227e-07,
      "loss": 0.5647,
      "step": 20352
    },
    {
      "epoch": 1.82,
      "grad_norm": 7.32152987026481,
      "learning_rate": 2.208703410955193e-07,
      "loss": 0.5503,
      "step": 20353
    },
    {
      "epoch": 1.82,
      "grad_norm": 7.648278347492774,
      "learning_rate": 2.2065806199791262e-07,
      "loss": 0.6538,
      "step": 20354
    },
    {
      "epoch": 1.82,
      "grad_norm": 7.115367528182663,
      "learning_rate": 2.2044588265860145e-07,
      "loss": 0.5933,
      "step": 20355
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.534842227471008,
      "learning_rate": 2.2023380308201504e-07,
      "loss": 0.5525,
      "step": 20356
    },
    {
      "epoch": 1.82,
      "grad_norm": 7.13181886579683,
      "learning_rate": 2.2002182327258092e-07,
      "loss": 0.6053,
      "step": 20357
    },
    {
      "epoch": 1.82,
      "grad_norm": 5.450643871223588,
      "learning_rate": 2.1980994323472338e-07,
      "loss": 0.5309,
      "step": 20358
    },
    {
      "epoch": 1.82,
      "grad_norm": 7.214602193393517,
      "learning_rate": 2.1959816297286497e-07,
      "loss": 0.6458,
      "step": 20359
    },
    {
      "epoch": 1.82,
      "grad_norm": 7.524801737445293,
      "learning_rate": 2.1938648249142602e-07,
      "loss": 0.5763,
      "step": 20360
    },
    {
      "epoch": 1.82,
      "grad_norm": 8.051750166319282,
      "learning_rate": 2.191749017948247e-07,
      "loss": 0.599,
      "step": 20361
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.653680850772671,
      "learning_rate": 2.1896342088747746e-07,
      "loss": 0.5354,
      "step": 20362
    },
    {
      "epoch": 1.82,
      "grad_norm": 4.673092269826447,
      "learning_rate": 2.18752039773798e-07,
      "loss": 0.5441,
      "step": 20363
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.538872380699893,
      "learning_rate": 2.185407584582e-07,
      "loss": 0.5881,
      "step": 20364
    },
    {
      "epoch": 1.82,
      "grad_norm": 5.64364718863824,
      "learning_rate": 2.1832957694509215e-07,
      "loss": 0.5804,
      "step": 20365
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.668728031753405,
      "learning_rate": 2.1811849523888208e-07,
      "loss": 0.5657,
      "step": 20366
    },
    {
      "epoch": 1.82,
      "grad_norm": 8.541118594081484,
      "learning_rate": 2.179075133439773e-07,
      "loss": 0.5883,
      "step": 20367
    },
    {
      "epoch": 1.82,
      "grad_norm": 4.696639277199813,
      "learning_rate": 2.176966312647799e-07,
      "loss": 0.5629,
      "step": 20368
    },
    {
      "epoch": 1.82,
      "grad_norm": 8.22416829046417,
      "learning_rate": 2.1748584900569249e-07,
      "loss": 0.5691,
      "step": 20369
    },
    {
      "epoch": 1.82,
      "grad_norm": 8.768375380788974,
      "learning_rate": 2.172751665711137e-07,
      "loss": 0.6473,
      "step": 20370
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.632056137911872,
      "learning_rate": 2.1706458396544226e-07,
      "loss": 0.5746,
      "step": 20371
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.248121823525085,
      "learning_rate": 2.1685410119307416e-07,
      "loss": 0.6362,
      "step": 20372
    },
    {
      "epoch": 1.82,
      "grad_norm": 9.254585851863926,
      "learning_rate": 2.166437182584008e-07,
      "loss": 0.5677,
      "step": 20373
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.263297139127617,
      "learning_rate": 2.1643343516581428e-07,
      "loss": 0.5926,
      "step": 20374
    },
    {
      "epoch": 1.82,
      "grad_norm": 7.563720904712251,
      "learning_rate": 2.162232519197044e-07,
      "loss": 0.5665,
      "step": 20375
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.9926385921479985,
      "learning_rate": 2.1601316852445765e-07,
      "loss": 0.5229,
      "step": 20376
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.330287588076863,
      "learning_rate": 2.1580318498445885e-07,
      "loss": 0.5752,
      "step": 20377
    },
    {
      "epoch": 1.82,
      "grad_norm": 9.670324581710089,
      "learning_rate": 2.1559330130409285e-07,
      "loss": 0.5971,
      "step": 20378
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.330010512539535,
      "learning_rate": 2.1538351748773834e-07,
      "loss": 0.5424,
      "step": 20379
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.373724027223369,
      "learning_rate": 2.1517383353977407e-07,
      "loss": 0.5695,
      "step": 20380
    },
    {
      "epoch": 1.82,
      "grad_norm": 5.866746359797262,
      "learning_rate": 2.149642494645776e-07,
      "loss": 0.5455,
      "step": 20381
    },
    {
      "epoch": 1.82,
      "grad_norm": 9.02980418262046,
      "learning_rate": 2.1475476526652383e-07,
      "loss": 0.5529,
      "step": 20382
    },
    {
      "epoch": 1.82,
      "grad_norm": 8.03487652768313,
      "learning_rate": 2.1454538094998424e-07,
      "loss": 0.5389,
      "step": 20383
    },
    {
      "epoch": 1.82,
      "grad_norm": 8.09687546602805,
      "learning_rate": 2.143360965193303e-07,
      "loss": 0.6027,
      "step": 20384
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.699081777004381,
      "learning_rate": 2.141269119789302e-07,
      "loss": 0.5738,
      "step": 20385
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.078788152964231,
      "learning_rate": 2.1391782733314937e-07,
      "loss": 0.6008,
      "step": 20386
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.784109823546863,
      "learning_rate": 2.137088425863526e-07,
      "loss": 0.5546,
      "step": 20387
    },
    {
      "epoch": 1.82,
      "grad_norm": 7.743314726582956,
      "learning_rate": 2.13499957742902e-07,
      "loss": 0.5479,
      "step": 20388
    },
    {
      "epoch": 1.82,
      "grad_norm": 7.043306769680538,
      "learning_rate": 2.1329117280715738e-07,
      "loss": 0.5635,
      "step": 20389
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.622676714881441,
      "learning_rate": 2.1308248778347696e-07,
      "loss": 0.5287,
      "step": 20390
    },
    {
      "epoch": 1.82,
      "grad_norm": 7.279210963857248,
      "learning_rate": 2.128739026762172e-07,
      "loss": 0.5399,
      "step": 20391
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.083016477051384,
      "learning_rate": 2.1266541748973024e-07,
      "loss": 0.5516,
      "step": 20392
    },
    {
      "epoch": 1.82,
      "grad_norm": 8.298826184499651,
      "learning_rate": 2.1245703222836923e-07,
      "loss": 0.5709,
      "step": 20393
    },
    {
      "epoch": 1.82,
      "grad_norm": 8.30460091480953,
      "learning_rate": 2.1224874689648235e-07,
      "loss": 0.4972,
      "step": 20394
    },
    {
      "epoch": 1.82,
      "grad_norm": 7.29398130575607,
      "learning_rate": 2.1204056149841833e-07,
      "loss": 0.5604,
      "step": 20395
    },
    {
      "epoch": 1.82,
      "grad_norm": 5.331060689004122,
      "learning_rate": 2.1183247603852152e-07,
      "loss": 0.5269,
      "step": 20396
    },
    {
      "epoch": 1.82,
      "grad_norm": 7.6150439010416,
      "learning_rate": 2.116244905211362e-07,
      "loss": 0.5505,
      "step": 20397
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.350466777641433,
      "learning_rate": 2.114166049506039e-07,
      "loss": 0.5473,
      "step": 20398
    },
    {
      "epoch": 1.82,
      "grad_norm": 9.008374196485153,
      "learning_rate": 2.112088193312628e-07,
      "loss": 0.5543,
      "step": 20399
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.637621747455729,
      "learning_rate": 2.1100113366745e-07,
      "loss": 0.4869,
      "step": 20400
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.352906939206318,
      "learning_rate": 2.1079354796350093e-07,
      "loss": 0.5321,
      "step": 20401
    },
    {
      "epoch": 1.82,
      "grad_norm": 5.628200477132338,
      "learning_rate": 2.1058606222374822e-07,
      "loss": 0.5705,
      "step": 20402
    },
    {
      "epoch": 1.82,
      "grad_norm": 7.1779917967788185,
      "learning_rate": 2.1037867645252285e-07,
      "loss": 0.6331,
      "step": 20403
    },
    {
      "epoch": 1.82,
      "grad_norm": 8.023588815436085,
      "learning_rate": 2.1017139065415414e-07,
      "loss": 0.5904,
      "step": 20404
    },
    {
      "epoch": 1.82,
      "grad_norm": 7.245603987394275,
      "learning_rate": 2.0996420483296753e-07,
      "loss": 0.6412,
      "step": 20405
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.666378280626616,
      "learning_rate": 2.0975711899328843e-07,
      "loss": 0.6069,
      "step": 20406
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.67954607803919,
      "learning_rate": 2.0955013313943952e-07,
      "loss": 0.5165,
      "step": 20407
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.776180826879517,
      "learning_rate": 2.0934324727573953e-07,
      "loss": 0.6086,
      "step": 20408
    },
    {
      "epoch": 1.82,
      "grad_norm": 8.752649705282286,
      "learning_rate": 2.0913646140650835e-07,
      "loss": 0.5747,
      "step": 20409
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.689917978072819,
      "learning_rate": 2.089297755360614e-07,
      "loss": 0.5863,
      "step": 20410
    },
    {
      "epoch": 1.82,
      "grad_norm": 9.650003916958465,
      "learning_rate": 2.087231896687142e-07,
      "loss": 0.5626,
      "step": 20411
    },
    {
      "epoch": 1.82,
      "grad_norm": 8.111181806243508,
      "learning_rate": 2.085167038087771e-07,
      "loss": 0.5216,
      "step": 20412
    },
    {
      "epoch": 1.82,
      "grad_norm": 9.060854453590139,
      "learning_rate": 2.0831031796056e-07,
      "loss": 0.5259,
      "step": 20413
    },
    {
      "epoch": 1.82,
      "grad_norm": 5.74420943501903,
      "learning_rate": 2.081040321283717e-07,
      "loss": 0.5636,
      "step": 20414
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.895155157666161,
      "learning_rate": 2.078978463165182e-07,
      "loss": 0.5941,
      "step": 20415
    },
    {
      "epoch": 1.82,
      "grad_norm": 5.161517244977868,
      "learning_rate": 2.076917605293022e-07,
      "loss": 0.5567,
      "step": 20416
    },
    {
      "epoch": 1.82,
      "grad_norm": 5.07699876221817,
      "learning_rate": 2.0748577477102628e-07,
      "loss": 0.5462,
      "step": 20417
    },
    {
      "epoch": 1.82,
      "grad_norm": 5.237011103189711,
      "learning_rate": 2.0727988904598874e-07,
      "loss": 0.5657,
      "step": 20418
    },
    {
      "epoch": 1.82,
      "grad_norm": 7.552052709643952,
      "learning_rate": 2.0707410335848776e-07,
      "loss": 0.6128,
      "step": 20419
    },
    {
      "epoch": 1.82,
      "grad_norm": 8.60987458526532,
      "learning_rate": 2.068684177128183e-07,
      "loss": 0.5235,
      "step": 20420
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.263312045753581,
      "learning_rate": 2.0666283211327464e-07,
      "loss": 0.577,
      "step": 20421
    },
    {
      "epoch": 1.82,
      "grad_norm": 7.669672053924077,
      "learning_rate": 2.0645734656414672e-07,
      "loss": 0.5412,
      "step": 20422
    },
    {
      "epoch": 1.82,
      "grad_norm": 9.828775305809325,
      "learning_rate": 2.062519610697239e-07,
      "loss": 0.487,
      "step": 20423
    },
    {
      "epoch": 1.82,
      "grad_norm": 8.332992734467085,
      "learning_rate": 2.0604667563429382e-07,
      "loss": 0.5058,
      "step": 20424
    },
    {
      "epoch": 1.82,
      "grad_norm": 7.96663326901817,
      "learning_rate": 2.0584149026213973e-07,
      "loss": 0.605,
      "step": 20425
    },
    {
      "epoch": 1.82,
      "grad_norm": 8.521144792061442,
      "learning_rate": 2.05636404957546e-07,
      "loss": 0.5555,
      "step": 20426
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.03475592442481,
      "learning_rate": 2.054314197247931e-07,
      "loss": 0.5203,
      "step": 20427
    },
    {
      "epoch": 1.82,
      "grad_norm": 5.333731284143635,
      "learning_rate": 2.0522653456815867e-07,
      "loss": 0.5184,
      "step": 20428
    },
    {
      "epoch": 1.82,
      "grad_norm": 4.957549715389349,
      "learning_rate": 2.0502174949192043e-07,
      "loss": 0.5797,
      "step": 20429
    },
    {
      "epoch": 1.82,
      "grad_norm": 9.18609628662794,
      "learning_rate": 2.0481706450035276e-07,
      "loss": 0.6062,
      "step": 20430
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.60962596064513,
      "learning_rate": 2.0461247959772722e-07,
      "loss": 0.5519,
      "step": 20431
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.1850230947131015,
      "learning_rate": 2.044079947883143e-07,
      "loss": 0.6035,
      "step": 20432
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.469050471292772,
      "learning_rate": 2.0420361007638224e-07,
      "loss": 0.6026,
      "step": 20433
    },
    {
      "epoch": 1.82,
      "grad_norm": 5.82918424130985,
      "learning_rate": 2.039993254661976e-07,
      "loss": 0.5146,
      "step": 20434
    },
    {
      "epoch": 1.82,
      "grad_norm": 8.283950625548787,
      "learning_rate": 2.037951409620237e-07,
      "loss": 0.6239,
      "step": 20435
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.15172142841754,
      "learning_rate": 2.0359105656812316e-07,
      "loss": 0.5507,
      "step": 20436
    },
    {
      "epoch": 1.82,
      "grad_norm": 8.62070697102389,
      "learning_rate": 2.0338707228875543e-07,
      "loss": 0.587,
      "step": 20437
    },
    {
      "epoch": 1.82,
      "grad_norm": 10.97755321489954,
      "learning_rate": 2.031831881281776e-07,
      "loss": 0.6402,
      "step": 20438
    },
    {
      "epoch": 1.82,
      "grad_norm": 87.1265144626567,
      "learning_rate": 2.029794040906463e-07,
      "loss": 0.7129,
      "step": 20439
    },
    {
      "epoch": 1.82,
      "grad_norm": 11.6718028617347,
      "learning_rate": 2.0277572018041424e-07,
      "loss": 0.5319,
      "step": 20440
    },
    {
      "epoch": 1.82,
      "grad_norm": 7.770461968969511,
      "learning_rate": 2.0257213640173302e-07,
      "loss": 0.6158,
      "step": 20441
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.781326265654025,
      "learning_rate": 2.0236865275885254e-07,
      "loss": 0.5361,
      "step": 20442
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.624956735971415,
      "learning_rate": 2.0216526925602053e-07,
      "loss": 0.4998,
      "step": 20443
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.480797294743205,
      "learning_rate": 2.0196198589748084e-07,
      "loss": 0.5972,
      "step": 20444
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.901760078470299,
      "learning_rate": 2.017588026874767e-07,
      "loss": 0.6088,
      "step": 20445
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.785967621461534,
      "learning_rate": 2.0155571963025034e-07,
      "loss": 0.6178,
      "step": 20446
    },
    {
      "epoch": 1.82,
      "grad_norm": 11.4221677062742,
      "learning_rate": 2.0135273673003942e-07,
      "loss": 0.5629,
      "step": 20447
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.492879925208589,
      "learning_rate": 2.0114985399108222e-07,
      "loss": 0.515,
      "step": 20448
    },
    {
      "epoch": 1.82,
      "grad_norm": 7.055463088821366,
      "learning_rate": 2.009470714176115e-07,
      "loss": 0.5262,
      "step": 20449
    },
    {
      "epoch": 1.82,
      "grad_norm": 7.339667480675694,
      "learning_rate": 2.007443890138616e-07,
      "loss": 0.5455,
      "step": 20450
    },
    {
      "epoch": 1.82,
      "grad_norm": 7.515248328669999,
      "learning_rate": 2.0054180678406198e-07,
      "loss": 0.5939,
      "step": 20451
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.25963111267426,
      "learning_rate": 2.0033932473244145e-07,
      "loss": 0.5512,
      "step": 20452
    },
    {
      "epoch": 1.82,
      "grad_norm": 7.648022089915887,
      "learning_rate": 2.0013694286322604e-07,
      "loss": 0.5635,
      "step": 20453
    },
    {
      "epoch": 1.82,
      "grad_norm": 7.483444867460799,
      "learning_rate": 1.9993466118064075e-07,
      "loss": 0.5764,
      "step": 20454
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.116851861642821,
      "learning_rate": 1.997324796889083e-07,
      "loss": 0.5632,
      "step": 20455
    },
    {
      "epoch": 1.82,
      "grad_norm": 8.780173175199488,
      "learning_rate": 1.9953039839224698e-07,
      "loss": 0.5372,
      "step": 20456
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.2090648385741325,
      "learning_rate": 1.9932841729487562e-07,
      "loss": 0.5622,
      "step": 20457
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.909724162124632,
      "learning_rate": 1.9912653640101088e-07,
      "loss": 0.5931,
      "step": 20458
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.548760202888542,
      "learning_rate": 1.9892475571486547e-07,
      "loss": 0.5893,
      "step": 20459
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.01112858806052,
      "learning_rate": 1.987230752406516e-07,
      "loss": 0.5782,
      "step": 20460
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.884939750989552,
      "learning_rate": 1.9852149498257977e-07,
      "loss": 0.5956,
      "step": 20461
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.79764087783772,
      "learning_rate": 1.9832001494485553e-07,
      "loss": 0.6102,
      "step": 20462
    },
    {
      "epoch": 1.83,
      "grad_norm": 8.624676222568793,
      "learning_rate": 1.9811863513168604e-07,
      "loss": 0.5921,
      "step": 20463
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.3734844495624285,
      "learning_rate": 1.979173555472741e-07,
      "loss": 0.559,
      "step": 20464
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.320615064425367,
      "learning_rate": 1.9771617619582128e-07,
      "loss": 0.5696,
      "step": 20465
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.298984837552271,
      "learning_rate": 1.9751509708152593e-07,
      "loss": 0.6083,
      "step": 20466
    },
    {
      "epoch": 1.83,
      "grad_norm": 8.676029817425633,
      "learning_rate": 1.973141182085858e-07,
      "loss": 0.5644,
      "step": 20467
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.40319304041225,
      "learning_rate": 1.9711323958119644e-07,
      "loss": 0.5086,
      "step": 20468
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.227176543209952,
      "learning_rate": 1.9691246120354944e-07,
      "loss": 0.5921,
      "step": 20469
    },
    {
      "epoch": 1.83,
      "grad_norm": 8.746354427927626,
      "learning_rate": 1.9671178307983595e-07,
      "loss": 0.5186,
      "step": 20470
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.360235988423362,
      "learning_rate": 1.9651120521424482e-07,
      "loss": 0.5587,
      "step": 20471
    },
    {
      "epoch": 1.83,
      "grad_norm": 8.825659187244895,
      "learning_rate": 1.9631072761096326e-07,
      "loss": 0.5666,
      "step": 20472
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.956631391073639,
      "learning_rate": 1.9611035027417514e-07,
      "loss": 0.5666,
      "step": 20473
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.534469255229354,
      "learning_rate": 1.9591007320806376e-07,
      "loss": 0.5593,
      "step": 20474
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.691856937612313,
      "learning_rate": 1.957098964168086e-07,
      "loss": 0.5315,
      "step": 20475
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.214973825952561,
      "learning_rate": 1.955098199045874e-07,
      "loss": 0.5409,
      "step": 20476
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.5839598885858175,
      "learning_rate": 1.9530984367557792e-07,
      "loss": 0.6,
      "step": 20477
    },
    {
      "epoch": 1.83,
      "grad_norm": 8.849183831160639,
      "learning_rate": 1.951099677339524e-07,
      "loss": 0.618,
      "step": 20478
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.010060078654609,
      "learning_rate": 1.949101920838853e-07,
      "loss": 0.5413,
      "step": 20479
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.858561614479666,
      "learning_rate": 1.9471051672954378e-07,
      "loss": 0.5549,
      "step": 20480
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.985118660322044,
      "learning_rate": 1.9451094167509731e-07,
      "loss": 0.5969,
      "step": 20481
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.750083546594355,
      "learning_rate": 1.9431146692471092e-07,
      "loss": 0.5589,
      "step": 20482
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.8235582399638455,
      "learning_rate": 1.9411209248254848e-07,
      "loss": 0.5898,
      "step": 20483
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.181679531428085,
      "learning_rate": 1.939128183527711e-07,
      "loss": 0.6147,
      "step": 20484
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.0117752321798825,
      "learning_rate": 1.9371364453953822e-07,
      "loss": 0.567,
      "step": 20485
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.440936937138381,
      "learning_rate": 1.9351457104700766e-07,
      "loss": 0.634,
      "step": 20486
    },
    {
      "epoch": 1.83,
      "grad_norm": 8.008518239983848,
      "learning_rate": 1.9331559787933495e-07,
      "loss": 0.6317,
      "step": 20487
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.850205980235074,
      "learning_rate": 1.9311672504067235e-07,
      "loss": 0.5379,
      "step": 20488
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.41262437523909,
      "learning_rate": 1.9291795253517097e-07,
      "loss": 0.5,
      "step": 20489
    },
    {
      "epoch": 1.83,
      "grad_norm": 10.400692954633453,
      "learning_rate": 1.9271928036698028e-07,
      "loss": 0.5931,
      "step": 20490
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.45540512859701,
      "learning_rate": 1.925207085402464e-07,
      "loss": 0.5477,
      "step": 20491
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.6690567359634185,
      "learning_rate": 1.923222370591149e-07,
      "loss": 0.5613,
      "step": 20492
    },
    {
      "epoch": 1.83,
      "grad_norm": 9.801136489273054,
      "learning_rate": 1.9212386592772858e-07,
      "loss": 0.5561,
      "step": 20493
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.5102574063153416,
      "learning_rate": 1.9192559515022747e-07,
      "loss": 0.5462,
      "step": 20494
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.914886929942341,
      "learning_rate": 1.9172742473074934e-07,
      "loss": 0.5857,
      "step": 20495
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.526457650109771,
      "learning_rate": 1.915293546734315e-07,
      "loss": 0.6289,
      "step": 20496
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.766455141920986,
      "learning_rate": 1.9133138498240776e-07,
      "loss": 0.5401,
      "step": 20497
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.493991312445989,
      "learning_rate": 1.9113351566181105e-07,
      "loss": 0.6428,
      "step": 20498
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.3785527867819605,
      "learning_rate": 1.909357467157702e-07,
      "loss": 0.5616,
      "step": 20499
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.531002773947975,
      "learning_rate": 1.9073807814841528e-07,
      "loss": 0.5734,
      "step": 20500
    },
    {
      "epoch": 1.83,
      "grad_norm": 9.863778177938391,
      "learning_rate": 1.905405099638702e-07,
      "loss": 0.6263,
      "step": 20501
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.774935404995329,
      "learning_rate": 1.9034304216625888e-07,
      "loss": 0.5295,
      "step": 20502
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.594889837993192,
      "learning_rate": 1.9014567475970414e-07,
      "loss": 0.5955,
      "step": 20503
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.51213376563046,
      "learning_rate": 1.8994840774832436e-07,
      "loss": 0.5461,
      "step": 20504
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.44399336858761,
      "learning_rate": 1.8975124113623844e-07,
      "loss": 0.5255,
      "step": 20505
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.540364689901551,
      "learning_rate": 1.8955417492756146e-07,
      "loss": 0.5801,
      "step": 20506
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.808116696624899,
      "learning_rate": 1.8935720912640565e-07,
      "loss": 0.5442,
      "step": 20507
    },
    {
      "epoch": 1.83,
      "grad_norm": 8.512301712187089,
      "learning_rate": 1.8916034373688386e-07,
      "loss": 0.5589,
      "step": 20508
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.552659179667349,
      "learning_rate": 1.8896357876310334e-07,
      "loss": 0.4933,
      "step": 20509
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.822747546019246,
      "learning_rate": 1.8876691420917247e-07,
      "loss": 0.6186,
      "step": 20510
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.668168126507062,
      "learning_rate": 1.8857035007919578e-07,
      "loss": 0.5726,
      "step": 20511
    },
    {
      "epoch": 1.83,
      "grad_norm": 9.25252382168987,
      "learning_rate": 1.8837388637727604e-07,
      "loss": 0.5497,
      "step": 20512
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.394982308962592,
      "learning_rate": 1.8817752310751503e-07,
      "loss": 0.5146,
      "step": 20513
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.023439932792848,
      "learning_rate": 1.8798126027400943e-07,
      "loss": 0.5808,
      "step": 20514
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.411326408644978,
      "learning_rate": 1.8778509788085708e-07,
      "loss": 0.5764,
      "step": 20515
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.000731823699352,
      "learning_rate": 1.875890359321525e-07,
      "loss": 0.5576,
      "step": 20516
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.235548447960247,
      "learning_rate": 1.873930744319874e-07,
      "loss": 0.5319,
      "step": 20517
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.201313025011415,
      "learning_rate": 1.8719721338445297e-07,
      "loss": 0.585,
      "step": 20518
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.909144468141908,
      "learning_rate": 1.8700145279363702e-07,
      "loss": 0.5728,
      "step": 20519
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.8916160277305645,
      "learning_rate": 1.8680579266362464e-07,
      "loss": 0.6033,
      "step": 20520
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.056578034269325,
      "learning_rate": 1.8661023299850088e-07,
      "loss": 0.5331,
      "step": 20521
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.81521244705786,
      "learning_rate": 1.8641477380234806e-07,
      "loss": 0.5393,
      "step": 20522
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.9427681369927425,
      "learning_rate": 1.8621941507924512e-07,
      "loss": 0.6002,
      "step": 20523
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.243086083353086,
      "learning_rate": 1.8602415683326936e-07,
      "loss": 0.5631,
      "step": 20524
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.139142010332657,
      "learning_rate": 1.8582899906849695e-07,
      "loss": 0.5531,
      "step": 20525
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.034200215991555,
      "learning_rate": 1.8563394178900185e-07,
      "loss": 0.6206,
      "step": 20526
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.512701892531282,
      "learning_rate": 1.8543898499885415e-07,
      "loss": 0.5549,
      "step": 20527
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.069490271297436,
      "learning_rate": 1.8524412870212448e-07,
      "loss": 0.5648,
      "step": 20528
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.404130774484126,
      "learning_rate": 1.8504937290287905e-07,
      "loss": 0.5587,
      "step": 20529
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.160351574777903,
      "learning_rate": 1.8485471760518347e-07,
      "loss": 0.5191,
      "step": 20530
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.012094233701174,
      "learning_rate": 1.8466016281310117e-07,
      "loss": 0.5937,
      "step": 20531
    },
    {
      "epoch": 1.83,
      "grad_norm": 8.31209278941808,
      "learning_rate": 1.8446570853069225e-07,
      "loss": 0.5911,
      "step": 20532
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.712534670907481,
      "learning_rate": 1.8427135476201562e-07,
      "loss": 0.5809,
      "step": 20533
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.105480755103412,
      "learning_rate": 1.8407710151112867e-07,
      "loss": 0.5462,
      "step": 20534
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.455549501758667,
      "learning_rate": 1.838829487820848e-07,
      "loss": 0.5728,
      "step": 20535
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.007491025938872,
      "learning_rate": 1.8368889657893796e-07,
      "loss": 0.6133,
      "step": 20536
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.569733734313795,
      "learning_rate": 1.834949449057377e-07,
      "loss": 0.6627,
      "step": 20537
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.377258838073588,
      "learning_rate": 1.833010937665325e-07,
      "loss": 0.5487,
      "step": 20538
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.733676934603241,
      "learning_rate": 1.8310734316536905e-07,
      "loss": 0.619,
      "step": 20539
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.781531552477668,
      "learning_rate": 1.8291369310629026e-07,
      "loss": 0.5519,
      "step": 20540
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.021595324472358,
      "learning_rate": 1.82720143593339e-07,
      "loss": 0.5241,
      "step": 20541
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.567755600548578,
      "learning_rate": 1.8252669463055484e-07,
      "loss": 0.5037,
      "step": 20542
    },
    {
      "epoch": 1.83,
      "grad_norm": 8.23301331942959,
      "learning_rate": 1.823333462219762e-07,
      "loss": 0.597,
      "step": 20543
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.344234960928545,
      "learning_rate": 1.821400983716387e-07,
      "loss": 0.5471,
      "step": 20544
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.650330377656289,
      "learning_rate": 1.819469510835753e-07,
      "loss": 0.5905,
      "step": 20545
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.524706681850245,
      "learning_rate": 1.8175390436181827e-07,
      "loss": 0.5502,
      "step": 20546
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.925619267853445,
      "learning_rate": 1.815609582103961e-07,
      "loss": 0.6254,
      "step": 20547
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.4783942201599904,
      "learning_rate": 1.813681126333372e-07,
      "loss": 0.6185,
      "step": 20548
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.023649098377665,
      "learning_rate": 1.8117536763466613e-07,
      "loss": 0.485,
      "step": 20549
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.641043358880881,
      "learning_rate": 1.809827232184064e-07,
      "loss": 0.5897,
      "step": 20550
    },
    {
      "epoch": 1.83,
      "grad_norm": 8.736048692597262,
      "learning_rate": 1.8079017938857968e-07,
      "loss": 0.5537,
      "step": 20551
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.057018733646689,
      "learning_rate": 1.8059773614920285e-07,
      "loss": 0.5242,
      "step": 20552
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.985378865240564,
      "learning_rate": 1.804053935042943e-07,
      "loss": 0.5388,
      "step": 20553
    },
    {
      "epoch": 1.83,
      "grad_norm": 8.499315849638082,
      "learning_rate": 1.802131514578681e-07,
      "loss": 0.5432,
      "step": 20554
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.175874916886208,
      "learning_rate": 1.800210100139377e-07,
      "loss": 0.5524,
      "step": 20555
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.75116381359166,
      "learning_rate": 1.798289691765126e-07,
      "loss": 0.5424,
      "step": 20556
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.666910065694054,
      "learning_rate": 1.7963702894960245e-07,
      "loss": 0.5929,
      "step": 20557
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.17378371706061,
      "learning_rate": 1.794451893372129e-07,
      "loss": 0.5085,
      "step": 20558
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.778893209826911,
      "learning_rate": 1.7925345034334741e-07,
      "loss": 0.5547,
      "step": 20559
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.618248351620059,
      "learning_rate": 1.7906181197200945e-07,
      "loss": 0.5759,
      "step": 20560
    },
    {
      "epoch": 1.83,
      "grad_norm": 8.215854309210181,
      "learning_rate": 1.7887027422719806e-07,
      "loss": 0.5545,
      "step": 20561
    },
    {
      "epoch": 1.83,
      "grad_norm": 8.648398383380762,
      "learning_rate": 1.7867883711291222e-07,
      "loss": 0.568,
      "step": 20562
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.973951410737096,
      "learning_rate": 1.7848750063314714e-07,
      "loss": 0.571,
      "step": 20563
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.827115326116493,
      "learning_rate": 1.782962647918962e-07,
      "loss": 0.5615,
      "step": 20564
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.5611942164720745,
      "learning_rate": 1.781051295931513e-07,
      "loss": 0.525,
      "step": 20565
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.331898476982622,
      "learning_rate": 1.7791409504090252e-07,
      "loss": 0.5641,
      "step": 20566
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.343842515876296,
      "learning_rate": 1.777231611391367e-07,
      "loss": 0.5353,
      "step": 20567
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.256052043421368,
      "learning_rate": 1.7753232789183895e-07,
      "loss": 0.6052,
      "step": 20568
    },
    {
      "epoch": 1.83,
      "grad_norm": 10.150019865955187,
      "learning_rate": 1.7734159530299278e-07,
      "loss": 0.5993,
      "step": 20569
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.257002136699931,
      "learning_rate": 1.7715096337658e-07,
      "loss": 0.58,
      "step": 20570
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.8176732898595755,
      "learning_rate": 1.7696043211657854e-07,
      "loss": 0.5953,
      "step": 20571
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.181080241560242,
      "learning_rate": 1.7677000152696576e-07,
      "loss": 0.5559,
      "step": 20572
    },
    {
      "epoch": 1.84,
      "grad_norm": 9.301970072252466,
      "learning_rate": 1.765796716117163e-07,
      "loss": 0.5428,
      "step": 20573
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.604052057216837,
      "learning_rate": 1.7638944237480305e-07,
      "loss": 0.5683,
      "step": 20574
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.2795176523219745,
      "learning_rate": 1.7619931382019727e-07,
      "loss": 0.5616,
      "step": 20575
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.753000941821374,
      "learning_rate": 1.7600928595186696e-07,
      "loss": 0.585,
      "step": 20576
    },
    {
      "epoch": 1.84,
      "grad_norm": 9.195842247806896,
      "learning_rate": 1.7581935877377775e-07,
      "loss": 0.5479,
      "step": 20577
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.576741740200964,
      "learning_rate": 1.7562953228989487e-07,
      "loss": 0.5742,
      "step": 20578
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.4949020152651045,
      "learning_rate": 1.7543980650418013e-07,
      "loss": 0.5808,
      "step": 20579
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.884247373400638,
      "learning_rate": 1.752501814205948e-07,
      "loss": 0.5317,
      "step": 20580
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.279222604874481,
      "learning_rate": 1.7506065704309517e-07,
      "loss": 0.5885,
      "step": 20581
    },
    {
      "epoch": 1.84,
      "grad_norm": 8.616160351957546,
      "learning_rate": 1.748712333756375e-07,
      "loss": 0.5853,
      "step": 20582
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.179798290897012,
      "learning_rate": 1.7468191042217696e-07,
      "loss": 0.5749,
      "step": 20583
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.5286318134227,
      "learning_rate": 1.744926881866632e-07,
      "loss": 0.5298,
      "step": 20584
    },
    {
      "epoch": 1.84,
      "grad_norm": 7.982310228780247,
      "learning_rate": 1.743035666730475e-07,
      "loss": 0.5605,
      "step": 20585
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.654867160169335,
      "learning_rate": 1.7411454588527666e-07,
      "loss": 0.5707,
      "step": 20586
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.199575587706497,
      "learning_rate": 1.739256258272959e-07,
      "loss": 0.5284,
      "step": 20587
    },
    {
      "epoch": 1.84,
      "grad_norm": 8.504493593250443,
      "learning_rate": 1.7373680650304869e-07,
      "loss": 0.6131,
      "step": 20588
    },
    {
      "epoch": 1.84,
      "grad_norm": 7.007058234551393,
      "learning_rate": 1.7354808791647748e-07,
      "loss": 0.5685,
      "step": 20589
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.215801908787185,
      "learning_rate": 1.733594700715191e-07,
      "loss": 0.5271,
      "step": 20590
    },
    {
      "epoch": 1.84,
      "grad_norm": 9.187340979066992,
      "learning_rate": 1.7317095297211152e-07,
      "loss": 0.5429,
      "step": 20591
    },
    {
      "epoch": 1.84,
      "grad_norm": 11.423582294095121,
      "learning_rate": 1.7298253662219045e-07,
      "loss": 0.6098,
      "step": 20592
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.447013232987154,
      "learning_rate": 1.7279422102568721e-07,
      "loss": 0.5292,
      "step": 20593
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.1436312442008445,
      "learning_rate": 1.7260600618653423e-07,
      "loss": 0.5149,
      "step": 20594
    },
    {
      "epoch": 1.84,
      "grad_norm": 7.412966700829481,
      "learning_rate": 1.724178921086578e-07,
      "loss": 0.5885,
      "step": 20595
    },
    {
      "epoch": 1.84,
      "grad_norm": 7.368196018116045,
      "learning_rate": 1.7222987879598697e-07,
      "loss": 0.545,
      "step": 20596
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.204424713814753,
      "learning_rate": 1.7204196625244362e-07,
      "loss": 0.5535,
      "step": 20597
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.847989482722372,
      "learning_rate": 1.7185415448195187e-07,
      "loss": 0.5556,
      "step": 20598
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.0190238746350415,
      "learning_rate": 1.7166644348843076e-07,
      "loss": 0.5279,
      "step": 20599
    },
    {
      "epoch": 1.84,
      "grad_norm": 7.218219763586711,
      "learning_rate": 1.7147883327579885e-07,
      "loss": 0.6,
      "step": 20600
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.865882020959749,
      "learning_rate": 1.7129132384797186e-07,
      "loss": 0.577,
      "step": 20601
    },
    {
      "epoch": 1.84,
      "grad_norm": 8.767851021780876,
      "learning_rate": 1.7110391520886449e-07,
      "loss": 0.5916,
      "step": 20602
    },
    {
      "epoch": 1.84,
      "grad_norm": 7.7311602834285305,
      "learning_rate": 1.7091660736238745e-07,
      "loss": 0.5576,
      "step": 20603
    },
    {
      "epoch": 1.84,
      "grad_norm": 7.110694541604403,
      "learning_rate": 1.7072940031245043e-07,
      "loss": 0.5825,
      "step": 20604
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.681090915181193,
      "learning_rate": 1.7054229406296086e-07,
      "loss": 0.5971,
      "step": 20605
    },
    {
      "epoch": 1.84,
      "grad_norm": 8.44866720228712,
      "learning_rate": 1.7035528861782502e-07,
      "loss": 0.6166,
      "step": 20606
    },
    {
      "epoch": 1.84,
      "grad_norm": 8.739578940587425,
      "learning_rate": 1.7016838398094648e-07,
      "loss": 0.5325,
      "step": 20607
    },
    {
      "epoch": 1.84,
      "grad_norm": 7.118913271491661,
      "learning_rate": 1.6998158015622546e-07,
      "loss": 0.5272,
      "step": 20608
    },
    {
      "epoch": 1.84,
      "grad_norm": 8.095108419045275,
      "learning_rate": 1.697948771475616e-07,
      "loss": 0.5672,
      "step": 20609
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.460942594749149,
      "learning_rate": 1.6960827495885067e-07,
      "loss": 0.5358,
      "step": 20610
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.82603937270263,
      "learning_rate": 1.6942177359398904e-07,
      "loss": 0.5682,
      "step": 20611
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.805303963182185,
      "learning_rate": 1.692353730568691e-07,
      "loss": 0.59,
      "step": 20612
    },
    {
      "epoch": 1.84,
      "grad_norm": 10.674598605006066,
      "learning_rate": 1.6904907335138166e-07,
      "loss": 0.5725,
      "step": 20613
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.171454943630189,
      "learning_rate": 1.6886287448141524e-07,
      "loss": 0.5612,
      "step": 20614
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.052066638022161,
      "learning_rate": 1.686767764508568e-07,
      "loss": 0.5165,
      "step": 20615
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.5919631351502135,
      "learning_rate": 1.684907792635898e-07,
      "loss": 0.5054,
      "step": 20616
    },
    {
      "epoch": 1.84,
      "grad_norm": 4.7353476754433865,
      "learning_rate": 1.6830488292349677e-07,
      "loss": 0.5124,
      "step": 20617
    },
    {
      "epoch": 1.84,
      "grad_norm": 7.204958966073327,
      "learning_rate": 1.6811908743445794e-07,
      "loss": 0.6068,
      "step": 20618
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.141231863710975,
      "learning_rate": 1.6793339280035181e-07,
      "loss": 0.5395,
      "step": 20619
    },
    {
      "epoch": 1.84,
      "grad_norm": 7.166457008307585,
      "learning_rate": 1.6774779902505422e-07,
      "loss": 0.5799,
      "step": 20620
    },
    {
      "epoch": 1.84,
      "grad_norm": 7.1074055015778494,
      "learning_rate": 1.6756230611243873e-07,
      "loss": 0.514,
      "step": 20621
    },
    {
      "epoch": 1.84,
      "grad_norm": 9.25815280742036,
      "learning_rate": 1.6737691406637724e-07,
      "loss": 0.5643,
      "step": 20622
    },
    {
      "epoch": 1.84,
      "grad_norm": 7.6254913932324095,
      "learning_rate": 1.6719162289074e-07,
      "loss": 0.5724,
      "step": 20623
    },
    {
      "epoch": 1.84,
      "grad_norm": 9.085864675061949,
      "learning_rate": 1.6700643258939275e-07,
      "loss": 0.5756,
      "step": 20624
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.041295468637495,
      "learning_rate": 1.6682134316620301e-07,
      "loss": 0.5489,
      "step": 20625
    },
    {
      "epoch": 1.84,
      "grad_norm": 8.12739041768074,
      "learning_rate": 1.666363546250327e-07,
      "loss": 0.51,
      "step": 20626
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.985130977525909,
      "learning_rate": 1.6645146696974423e-07,
      "loss": 0.5693,
      "step": 20627
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.060312315167556,
      "learning_rate": 1.6626668020419568e-07,
      "loss": 0.5088,
      "step": 20628
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.550445470383628,
      "learning_rate": 1.660819943322445e-07,
      "loss": 0.5307,
      "step": 20629
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.17270312755417,
      "learning_rate": 1.6589740935774534e-07,
      "loss": 0.538,
      "step": 20630
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.098320704753383,
      "learning_rate": 1.6571292528455184e-07,
      "loss": 0.5299,
      "step": 20631
    },
    {
      "epoch": 1.84,
      "grad_norm": 9.412574609084226,
      "learning_rate": 1.655285421165137e-07,
      "loss": 0.5262,
      "step": 20632
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.495406545634089,
      "learning_rate": 1.6534425985748058e-07,
      "loss": 0.6123,
      "step": 20633
    },
    {
      "epoch": 1.84,
      "grad_norm": 7.119548540429959,
      "learning_rate": 1.6516007851129722e-07,
      "loss": 0.6113,
      "step": 20634
    },
    {
      "epoch": 1.84,
      "grad_norm": 7.015780757802399,
      "learning_rate": 1.6497599808180996e-07,
      "loss": 0.5299,
      "step": 20635
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.91290315546167,
      "learning_rate": 1.6479201857285966e-07,
      "loss": 0.5914,
      "step": 20636
    },
    {
      "epoch": 1.84,
      "grad_norm": 8.182518205296983,
      "learning_rate": 1.6460813998828706e-07,
      "loss": 0.5972,
      "step": 20637
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.321433817476633,
      "learning_rate": 1.6442436233193083e-07,
      "loss": 0.5803,
      "step": 20638
    },
    {
      "epoch": 1.84,
      "grad_norm": 7.553572613996043,
      "learning_rate": 1.642406856076262e-07,
      "loss": 0.502,
      "step": 20639
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.965231264948768,
      "learning_rate": 1.640571098192073e-07,
      "loss": 0.5064,
      "step": 20640
    },
    {
      "epoch": 1.84,
      "grad_norm": 8.852506111449145,
      "learning_rate": 1.6387363497050502e-07,
      "loss": 0.5843,
      "step": 20641
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.417552391475473,
      "learning_rate": 1.6369026106535013e-07,
      "loss": 0.5277,
      "step": 20642
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.352046883286437,
      "learning_rate": 1.6350698810756905e-07,
      "loss": 0.6596,
      "step": 20643
    },
    {
      "epoch": 1.84,
      "grad_norm": 9.15741264665353,
      "learning_rate": 1.6332381610098812e-07,
      "loss": 0.6115,
      "step": 20644
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.734780462390892,
      "learning_rate": 1.6314074504943045e-07,
      "loss": 0.5393,
      "step": 20645
    },
    {
      "epoch": 1.84,
      "grad_norm": 8.174027053604659,
      "learning_rate": 1.6295777495671794e-07,
      "loss": 0.5653,
      "step": 20646
    },
    {
      "epoch": 1.84,
      "grad_norm": 7.526712950684415,
      "learning_rate": 1.627749058266681e-07,
      "loss": 0.5756,
      "step": 20647
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.806740319255307,
      "learning_rate": 1.62592137663099e-07,
      "loss": 0.5499,
      "step": 20648
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.805197189264749,
      "learning_rate": 1.6240947046982482e-07,
      "loss": 0.5787,
      "step": 20649
    },
    {
      "epoch": 1.84,
      "grad_norm": 4.879679664692137,
      "learning_rate": 1.6222690425065913e-07,
      "loss": 0.5088,
      "step": 20650
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.66902208922615,
      "learning_rate": 1.6204443900941225e-07,
      "loss": 0.5851,
      "step": 20651
    },
    {
      "epoch": 1.84,
      "grad_norm": 8.610626200010827,
      "learning_rate": 1.6186207474989336e-07,
      "loss": 0.5686,
      "step": 20652
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.879336768763866,
      "learning_rate": 1.6167981147590827e-07,
      "loss": 0.5695,
      "step": 20653
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.444710868017044,
      "learning_rate": 1.614976491912612e-07,
      "loss": 0.5149,
      "step": 20654
    },
    {
      "epoch": 1.84,
      "grad_norm": 9.702470203042676,
      "learning_rate": 1.613155878997541e-07,
      "loss": 0.576,
      "step": 20655
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.521313268884404,
      "learning_rate": 1.611336276051878e-07,
      "loss": 0.6021,
      "step": 20656
    },
    {
      "epoch": 1.84,
      "grad_norm": 7.2175942695285435,
      "learning_rate": 1.6095176831135984e-07,
      "loss": 0.5669,
      "step": 20657
    },
    {
      "epoch": 1.84,
      "grad_norm": 7.7327185475437865,
      "learning_rate": 1.607700100220666e-07,
      "loss": 0.569,
      "step": 20658
    },
    {
      "epoch": 1.84,
      "grad_norm": 7.925217472774974,
      "learning_rate": 1.6058835274110175e-07,
      "loss": 0.5659,
      "step": 20659
    },
    {
      "epoch": 1.84,
      "grad_norm": 8.245009724545737,
      "learning_rate": 1.604067964722572e-07,
      "loss": 0.5274,
      "step": 20660
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.193378223488177,
      "learning_rate": 1.6022534121932165e-07,
      "loss": 0.5313,
      "step": 20661
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.363165464399734,
      "learning_rate": 1.6004398698608315e-07,
      "loss": 0.5404,
      "step": 20662
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.699463848639689,
      "learning_rate": 1.59862733776327e-07,
      "loss": 0.5442,
      "step": 20663
    },
    {
      "epoch": 1.84,
      "grad_norm": 7.225396966498009,
      "learning_rate": 1.5968158159383686e-07,
      "loss": 0.6238,
      "step": 20664
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.779454844668766,
      "learning_rate": 1.5950053044239412e-07,
      "loss": 0.5854,
      "step": 20665
    },
    {
      "epoch": 1.84,
      "grad_norm": 7.0156551760393455,
      "learning_rate": 1.5931958032577743e-07,
      "loss": 0.6239,
      "step": 20666
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.373027958224076,
      "learning_rate": 1.5913873124776268e-07,
      "loss": 0.5647,
      "step": 20667
    },
    {
      "epoch": 1.84,
      "grad_norm": 9.580236403917825,
      "learning_rate": 1.5895798321212575e-07,
      "loss": 0.5732,
      "step": 20668
    },
    {
      "epoch": 1.84,
      "grad_norm": 7.34703536297252,
      "learning_rate": 1.587773362226397e-07,
      "loss": 0.5578,
      "step": 20669
    },
    {
      "epoch": 1.84,
      "grad_norm": 7.206467444491746,
      "learning_rate": 1.5859679028307373e-07,
      "loss": 0.6023,
      "step": 20670
    },
    {
      "epoch": 1.84,
      "grad_norm": 7.936489137472094,
      "learning_rate": 1.584163453971982e-07,
      "loss": 0.5782,
      "step": 20671
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.984013154560975,
      "learning_rate": 1.5823600156877838e-07,
      "loss": 0.5931,
      "step": 20672
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.626837171686694,
      "learning_rate": 1.5805575880157853e-07,
      "loss": 0.5274,
      "step": 20673
    },
    {
      "epoch": 1.84,
      "grad_norm": 9.033289739035647,
      "learning_rate": 1.5787561709936116e-07,
      "loss": 0.5758,
      "step": 20674
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.368076671729043,
      "learning_rate": 1.5769557646588608e-07,
      "loss": 0.6066,
      "step": 20675
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.7553218412895815,
      "learning_rate": 1.5751563690491134e-07,
      "loss": 0.579,
      "step": 20676
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.911690875194166,
      "learning_rate": 1.5733579842019286e-07,
      "loss": 0.5366,
      "step": 20677
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.823752747637525,
      "learning_rate": 1.5715606101548486e-07,
      "loss": 0.5844,
      "step": 20678
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.670256139896281,
      "learning_rate": 1.5697642469453765e-07,
      "loss": 0.5191,
      "step": 20679
    },
    {
      "epoch": 1.84,
      "grad_norm": 7.058900612450303,
      "learning_rate": 1.5679688946110218e-07,
      "loss": 0.6104,
      "step": 20680
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.717185135150641,
      "learning_rate": 1.566174553189248e-07,
      "loss": 0.5512,
      "step": 20681
    },
    {
      "epoch": 1.85,
      "grad_norm": 7.94179318805623,
      "learning_rate": 1.5643812227175147e-07,
      "loss": 0.5279,
      "step": 20682
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.825478602575518,
      "learning_rate": 1.5625889032332475e-07,
      "loss": 0.4881,
      "step": 20683
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.454847107451756,
      "learning_rate": 1.5607975947738608e-07,
      "loss": 0.6306,
      "step": 20684
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.749403955468157,
      "learning_rate": 1.5590072973767523e-07,
      "loss": 0.5705,
      "step": 20685
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.243477390781048,
      "learning_rate": 1.5572180110792755e-07,
      "loss": 0.5854,
      "step": 20686
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.181351908085328,
      "learning_rate": 1.5554297359187842e-07,
      "loss": 0.5901,
      "step": 20687
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.4255511317015905,
      "learning_rate": 1.5536424719326093e-07,
      "loss": 0.6489,
      "step": 20688
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.2962015572329735,
      "learning_rate": 1.5518562191580487e-07,
      "loss": 0.5604,
      "step": 20689
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.107226514259649,
      "learning_rate": 1.5500709776323896e-07,
      "loss": 0.5724,
      "step": 20690
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.9905807097479835,
      "learning_rate": 1.548286747392902e-07,
      "loss": 0.5349,
      "step": 20691
    },
    {
      "epoch": 1.85,
      "grad_norm": 7.513159803246822,
      "learning_rate": 1.5465035284768114e-07,
      "loss": 0.5339,
      "step": 20692
    },
    {
      "epoch": 1.85,
      "grad_norm": 8.619768601955238,
      "learning_rate": 1.5447213209213552e-07,
      "loss": 0.5616,
      "step": 20693
    },
    {
      "epoch": 1.85,
      "grad_norm": 7.951223224382077,
      "learning_rate": 1.54294012476372e-07,
      "loss": 0.5639,
      "step": 20694
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.804612928825943,
      "learning_rate": 1.541159940041098e-07,
      "loss": 0.6178,
      "step": 20695
    },
    {
      "epoch": 1.85,
      "grad_norm": 7.9672974410657575,
      "learning_rate": 1.539380766790638e-07,
      "loss": 0.6019,
      "step": 20696
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.563748236630898,
      "learning_rate": 1.5376026050494708e-07,
      "loss": 0.6108,
      "step": 20697
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.59648961699994,
      "learning_rate": 1.535825454854728e-07,
      "loss": 0.5885,
      "step": 20698
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.375746423288026,
      "learning_rate": 1.534049316243491e-07,
      "loss": 0.5861,
      "step": 20699
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.919163397385305,
      "learning_rate": 1.5322741892528302e-07,
      "loss": 0.528,
      "step": 20700
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.902082114687394,
      "learning_rate": 1.530500073919805e-07,
      "loss": 0.6444,
      "step": 20701
    },
    {
      "epoch": 1.85,
      "grad_norm": 8.140864365959969,
      "learning_rate": 1.5287269702814466e-07,
      "loss": 0.6305,
      "step": 20702
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.367806913185126,
      "learning_rate": 1.5269548783747646e-07,
      "loss": 0.5972,
      "step": 20703
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.135615776841737,
      "learning_rate": 1.5251837982367402e-07,
      "loss": 0.5133,
      "step": 20704
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.295566547168751,
      "learning_rate": 1.5234137299043494e-07,
      "loss": 0.562,
      "step": 20705
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.001047020620564,
      "learning_rate": 1.5216446734145352e-07,
      "loss": 0.6087,
      "step": 20706
    },
    {
      "epoch": 1.85,
      "grad_norm": 10.608212647899032,
      "learning_rate": 1.5198766288042178e-07,
      "loss": 0.5224,
      "step": 20707
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.660837851279341,
      "learning_rate": 1.5181095961103122e-07,
      "loss": 0.5568,
      "step": 20708
    },
    {
      "epoch": 1.85,
      "grad_norm": 8.741720853501345,
      "learning_rate": 1.5163435753696943e-07,
      "loss": 0.574,
      "step": 20709
    },
    {
      "epoch": 1.85,
      "grad_norm": 7.816033902830379,
      "learning_rate": 1.514578566619229e-07,
      "loss": 0.5683,
      "step": 20710
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.306716640962868,
      "learning_rate": 1.5128145698957486e-07,
      "loss": 0.6095,
      "step": 20711
    },
    {
      "epoch": 1.85,
      "grad_norm": 7.668564815262459,
      "learning_rate": 1.5110515852360842e-07,
      "loss": 0.5726,
      "step": 20712
    },
    {
      "epoch": 1.85,
      "grad_norm": 7.812263292813327,
      "learning_rate": 1.5092896126770228e-07,
      "loss": 0.5464,
      "step": 20713
    },
    {
      "epoch": 1.85,
      "grad_norm": 11.67851967749272,
      "learning_rate": 1.5075286522553521e-07,
      "loss": 0.5075,
      "step": 20714
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.508848499774697,
      "learning_rate": 1.5057687040078207e-07,
      "loss": 0.5227,
      "step": 20715
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.52348829954265,
      "learning_rate": 1.5040097679711763e-07,
      "loss": 0.5905,
      "step": 20716
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.361670595288296,
      "learning_rate": 1.5022518441821176e-07,
      "loss": 0.5709,
      "step": 20717
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.6156427270427125,
      "learning_rate": 1.5004949326773376e-07,
      "loss": 0.5605,
      "step": 20718
    },
    {
      "epoch": 1.85,
      "grad_norm": 8.059451107160532,
      "learning_rate": 1.498739033493518e-07,
      "loss": 0.5048,
      "step": 20719
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.962871173530319,
      "learning_rate": 1.4969841466673075e-07,
      "loss": 0.5331,
      "step": 20720
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.3847774780850886,
      "learning_rate": 1.4952302722353319e-07,
      "loss": 0.5216,
      "step": 20721
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.130320238681146,
      "learning_rate": 1.4934774102342065e-07,
      "loss": 0.6532,
      "step": 20722
    },
    {
      "epoch": 1.85,
      "grad_norm": 7.561056384119012,
      "learning_rate": 1.4917255607005022e-07,
      "loss": 0.5603,
      "step": 20723
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.344088377169165,
      "learning_rate": 1.489974723670806e-07,
      "loss": 0.5112,
      "step": 20724
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.254261475571364,
      "learning_rate": 1.488224899181645e-07,
      "loss": 0.5632,
      "step": 20725
    },
    {
      "epoch": 1.85,
      "grad_norm": 7.190624237221434,
      "learning_rate": 1.4864760872695504e-07,
      "loss": 0.5705,
      "step": 20726
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.5855103262337185,
      "learning_rate": 1.4847282879710268e-07,
      "loss": 0.5835,
      "step": 20727
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.447979700376296,
      "learning_rate": 1.4829815013225502e-07,
      "loss": 0.5879,
      "step": 20728
    },
    {
      "epoch": 1.85,
      "grad_norm": 7.006537134390363,
      "learning_rate": 1.4812357273605916e-07,
      "loss": 0.6094,
      "step": 20729
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.1509200504931885,
      "learning_rate": 1.4794909661215773e-07,
      "loss": 0.5813,
      "step": 20730
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.840434545625151,
      "learning_rate": 1.4777472176419284e-07,
      "loss": 0.6159,
      "step": 20731
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.008186713956841,
      "learning_rate": 1.4760044819580488e-07,
      "loss": 0.5735,
      "step": 20732
    },
    {
      "epoch": 1.85,
      "grad_norm": 8.551331632345947,
      "learning_rate": 1.4742627591063096e-07,
      "loss": 0.5805,
      "step": 20733
    },
    {
      "epoch": 1.85,
      "grad_norm": 12.372222775031402,
      "learning_rate": 1.4725220491230597e-07,
      "loss": 0.5552,
      "step": 20734
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.881821916252152,
      "learning_rate": 1.470782352044653e-07,
      "loss": 0.5411,
      "step": 20735
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.493200857403462,
      "learning_rate": 1.4690436679073772e-07,
      "loss": 0.5579,
      "step": 20736
    },
    {
      "epoch": 1.85,
      "grad_norm": 8.082053577151761,
      "learning_rate": 1.467305996747531e-07,
      "loss": 0.5706,
      "step": 20737
    },
    {
      "epoch": 1.85,
      "grad_norm": 8.135973036058362,
      "learning_rate": 1.4655693386013914e-07,
      "loss": 0.6296,
      "step": 20738
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.657370109019666,
      "learning_rate": 1.4638336935052067e-07,
      "loss": 0.5509,
      "step": 20739
    },
    {
      "epoch": 1.85,
      "grad_norm": 8.457991224327344,
      "learning_rate": 1.462099061495198e-07,
      "loss": 0.5412,
      "step": 20740
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.806944106259335,
      "learning_rate": 1.4603654426075752e-07,
      "loss": 0.5778,
      "step": 20741
    },
    {
      "epoch": 1.85,
      "grad_norm": 8.699963751859817,
      "learning_rate": 1.4586328368785262e-07,
      "loss": 0.595,
      "step": 20742
    },
    {
      "epoch": 1.85,
      "grad_norm": 9.774215037343122,
      "learning_rate": 1.456901244344211e-07,
      "loss": 0.6045,
      "step": 20743
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.616621245571267,
      "learning_rate": 1.4551706650407727e-07,
      "loss": 0.6048,
      "step": 20744
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.719868377370761,
      "learning_rate": 1.453441099004338e-07,
      "loss": 0.5295,
      "step": 20745
    },
    {
      "epoch": 1.85,
      "grad_norm": 8.960038654519149,
      "learning_rate": 1.4517125462710001e-07,
      "loss": 0.6466,
      "step": 20746
    },
    {
      "epoch": 1.85,
      "grad_norm": 7.082738278824946,
      "learning_rate": 1.4499850068768474e-07,
      "loss": 0.5853,
      "step": 20747
    },
    {
      "epoch": 1.85,
      "grad_norm": 7.370544068563108,
      "learning_rate": 1.448258480857939e-07,
      "loss": 0.5584,
      "step": 20748
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.2634582266000844,
      "learning_rate": 1.4465329682503082e-07,
      "loss": 0.5014,
      "step": 20749
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.714959066875765,
      "learning_rate": 1.4448084690899643e-07,
      "loss": 0.5435,
      "step": 20750
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.6697051482836525,
      "learning_rate": 1.4430849834129123e-07,
      "loss": 0.5463,
      "step": 20751
    },
    {
      "epoch": 1.85,
      "grad_norm": 7.160913512351617,
      "learning_rate": 1.4413625112551233e-07,
      "loss": 0.5489,
      "step": 20752
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.534741951836044,
      "learning_rate": 1.4396410526525571e-07,
      "loss": 0.6571,
      "step": 20753
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.796412790025985,
      "learning_rate": 1.43792060764113e-07,
      "loss": 0.5198,
      "step": 20754
    },
    {
      "epoch": 1.85,
      "grad_norm": 7.113735277983158,
      "learning_rate": 1.4362011762567686e-07,
      "loss": 0.6048,
      "step": 20755
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.993419552464397,
      "learning_rate": 1.4344827585353492e-07,
      "loss": 0.5444,
      "step": 20756
    },
    {
      "epoch": 1.85,
      "grad_norm": 7.6414674370545566,
      "learning_rate": 1.432765354512744e-07,
      "loss": 0.5535,
      "step": 20757
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.266687094261093,
      "learning_rate": 1.4310489642248071e-07,
      "loss": 0.5557,
      "step": 20758
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.820637049068365,
      "learning_rate": 1.4293335877073545e-07,
      "loss": 0.5465,
      "step": 20759
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.898243643478518,
      "learning_rate": 1.4276192249961907e-07,
      "loss": 0.4958,
      "step": 20760
    },
    {
      "epoch": 1.85,
      "grad_norm": 10.9913655452052,
      "learning_rate": 1.425905876127115e-07,
      "loss": 0.5509,
      "step": 20761
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.436011563774006,
      "learning_rate": 1.4241935411358766e-07,
      "loss": 0.5622,
      "step": 20762
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.500006090618431,
      "learning_rate": 1.4224822200582134e-07,
      "loss": 0.5736,
      "step": 20763
    },
    {
      "epoch": 1.85,
      "grad_norm": 8.136794612910185,
      "learning_rate": 1.4207719129298524e-07,
      "loss": 0.5803,
      "step": 20764
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.353492221688485,
      "learning_rate": 1.4190626197864876e-07,
      "loss": 0.5763,
      "step": 20765
    },
    {
      "epoch": 1.85,
      "grad_norm": 9.132683403428883,
      "learning_rate": 1.4173543406638068e-07,
      "loss": 0.6115,
      "step": 20766
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.5486957638558865,
      "learning_rate": 1.4156470755974593e-07,
      "loss": 0.557,
      "step": 20767
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.8945024644753214,
      "learning_rate": 1.4139408246230834e-07,
      "loss": 0.5156,
      "step": 20768
    },
    {
      "epoch": 1.85,
      "grad_norm": 9.965774831561184,
      "learning_rate": 1.4122355877762895e-07,
      "loss": 0.5065,
      "step": 20769
    },
    {
      "epoch": 1.85,
      "grad_norm": 7.2362957058002095,
      "learning_rate": 1.4105313650926655e-07,
      "loss": 0.5478,
      "step": 20770
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.647771875611597,
      "learning_rate": 1.4088281566077945e-07,
      "loss": 0.5278,
      "step": 20771
    },
    {
      "epoch": 1.85,
      "grad_norm": 9.333874294630814,
      "learning_rate": 1.4071259623572198e-07,
      "loss": 0.5102,
      "step": 20772
    },
    {
      "epoch": 1.85,
      "grad_norm": 10.199259542717519,
      "learning_rate": 1.4054247823764799e-07,
      "loss": 0.6378,
      "step": 20773
    },
    {
      "epoch": 1.85,
      "grad_norm": 8.585911573462356,
      "learning_rate": 1.4037246167010797e-07,
      "loss": 0.5811,
      "step": 20774
    },
    {
      "epoch": 1.85,
      "grad_norm": 9.315257002957978,
      "learning_rate": 1.4020254653664966e-07,
      "loss": 0.5754,
      "step": 20775
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.8464622808205915,
      "learning_rate": 1.4003273284082076e-07,
      "loss": 0.5394,
      "step": 20776
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.974948297992769,
      "learning_rate": 1.3986302058616507e-07,
      "loss": 0.6491,
      "step": 20777
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.977434017019341,
      "learning_rate": 1.3969340977622537e-07,
      "loss": 0.5342,
      "step": 20778
    },
    {
      "epoch": 1.85,
      "grad_norm": 7.021562676372832,
      "learning_rate": 1.3952390041454267e-07,
      "loss": 0.5317,
      "step": 20779
    },
    {
      "epoch": 1.85,
      "grad_norm": 7.535531062988102,
      "learning_rate": 1.393544925046536e-07,
      "loss": 0.5866,
      "step": 20780
    },
    {
      "epoch": 1.85,
      "grad_norm": 8.902467961139712,
      "learning_rate": 1.391851860500948e-07,
      "loss": 0.5785,
      "step": 20781
    },
    {
      "epoch": 1.85,
      "grad_norm": 9.505384638575148,
      "learning_rate": 1.3901598105440063e-07,
      "loss": 0.5561,
      "step": 20782
    },
    {
      "epoch": 1.85,
      "grad_norm": 7.662422340849546,
      "learning_rate": 1.3884687752110216e-07,
      "loss": 0.5131,
      "step": 20783
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.929889555351067,
      "learning_rate": 1.3867787545372936e-07,
      "loss": 0.6142,
      "step": 20784
    },
    {
      "epoch": 1.85,
      "grad_norm": 7.177602044673022,
      "learning_rate": 1.385089748558094e-07,
      "loss": 0.5389,
      "step": 20785
    },
    {
      "epoch": 1.85,
      "grad_norm": 7.81832985850233,
      "learning_rate": 1.3834017573086945e-07,
      "loss": 0.6021,
      "step": 20786
    },
    {
      "epoch": 1.85,
      "grad_norm": 7.309583836331467,
      "learning_rate": 1.3817147808243003e-07,
      "loss": 0.5492,
      "step": 20787
    },
    {
      "epoch": 1.85,
      "grad_norm": 7.624355433945249,
      "learning_rate": 1.3800288191401445e-07,
      "loss": 0.5648,
      "step": 20788
    },
    {
      "epoch": 1.85,
      "grad_norm": 8.52686104361023,
      "learning_rate": 1.37834387229141e-07,
      "loss": 0.5744,
      "step": 20789
    },
    {
      "epoch": 1.85,
      "grad_norm": 7.4883410230955985,
      "learning_rate": 1.3766599403132687e-07,
      "loss": 0.5737,
      "step": 20790
    },
    {
      "epoch": 1.85,
      "grad_norm": 7.104573207856971,
      "learning_rate": 1.3749770232408643e-07,
      "loss": 0.5482,
      "step": 20791
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.1341930326638865,
      "learning_rate": 1.373295121109336e-07,
      "loss": 0.6058,
      "step": 20792
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.347110008424269,
      "learning_rate": 1.3716142339537718e-07,
      "loss": 0.5119,
      "step": 20793
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.447875234027292,
      "learning_rate": 1.3699343618092664e-07,
      "loss": 0.5528,
      "step": 20794
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.699203313825528,
      "learning_rate": 1.368255504710886e-07,
      "loss": 0.5578,
      "step": 20795
    },
    {
      "epoch": 1.86,
      "grad_norm": 8.579204373057108,
      "learning_rate": 1.36657766269368e-07,
      "loss": 0.6023,
      "step": 20796
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.431015194307729,
      "learning_rate": 1.3649008357926485e-07,
      "loss": 0.6472,
      "step": 20797
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.566677492814019,
      "learning_rate": 1.3632250240428024e-07,
      "loss": 0.6206,
      "step": 20798
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.658093124515533,
      "learning_rate": 1.3615502274791304e-07,
      "loss": 0.5456,
      "step": 20799
    },
    {
      "epoch": 1.86,
      "grad_norm": 6.119947799283983,
      "learning_rate": 1.359876446136571e-07,
      "loss": 0.6193,
      "step": 20800
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.601273303765995,
      "learning_rate": 1.3582036800500741e-07,
      "loss": 0.5532,
      "step": 20801
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.919294445629335,
      "learning_rate": 1.3565319292545508e-07,
      "loss": 0.6154,
      "step": 20802
    },
    {
      "epoch": 1.86,
      "grad_norm": 6.956389198203883,
      "learning_rate": 1.3548611937848953e-07,
      "loss": 0.6708,
      "step": 20803
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.660266923665702,
      "learning_rate": 1.3531914736759853e-07,
      "loss": 0.5521,
      "step": 20804
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.09289370177187,
      "learning_rate": 1.351522768962671e-07,
      "loss": 0.5698,
      "step": 20805
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.808385857494279,
      "learning_rate": 1.3498550796797737e-07,
      "loss": 0.557,
      "step": 20806
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.926376441595453,
      "learning_rate": 1.3481884058621108e-07,
      "loss": 0.5682,
      "step": 20807
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.96441644043271,
      "learning_rate": 1.3465227475444708e-07,
      "loss": 0.4399,
      "step": 20808
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.1681812128371565,
      "learning_rate": 1.344858104761615e-07,
      "loss": 0.5539,
      "step": 20809
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.781405172357765,
      "learning_rate": 1.3431944775482984e-07,
      "loss": 0.5094,
      "step": 20810
    },
    {
      "epoch": 1.86,
      "grad_norm": 6.1363584471298145,
      "learning_rate": 1.341531865939244e-07,
      "loss": 0.5563,
      "step": 20811
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.476718555566055,
      "learning_rate": 1.3398702699691512e-07,
      "loss": 0.6181,
      "step": 20812
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.76993116161206,
      "learning_rate": 1.338209689672698e-07,
      "loss": 0.534,
      "step": 20813
    },
    {
      "epoch": 1.86,
      "grad_norm": 4.605057315364204,
      "learning_rate": 1.336550125084546e-07,
      "loss": 0.5288,
      "step": 20814
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.196881878408631,
      "learning_rate": 1.3348915762393445e-07,
      "loss": 0.5766,
      "step": 20815
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.4163874436204145,
      "learning_rate": 1.333234043171705e-07,
      "loss": 0.6161,
      "step": 20816
    },
    {
      "epoch": 1.86,
      "grad_norm": 6.419143592644683,
      "learning_rate": 1.3315775259162277e-07,
      "loss": 0.5454,
      "step": 20817
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.045115092518291,
      "learning_rate": 1.3299220245074905e-07,
      "loss": 0.5819,
      "step": 20818
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.637496955232273,
      "learning_rate": 1.3282675389800382e-07,
      "loss": 0.5535,
      "step": 20819
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.353833345294148,
      "learning_rate": 1.3266140693684148e-07,
      "loss": 0.5358,
      "step": 20820
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.091492953576116,
      "learning_rate": 1.3249616157071322e-07,
      "loss": 0.5758,
      "step": 20821
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.654040464489238,
      "learning_rate": 1.3233101780306734e-07,
      "loss": 0.6387,
      "step": 20822
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.171394925134168,
      "learning_rate": 1.3216597563735167e-07,
      "loss": 0.4972,
      "step": 20823
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.989950266775283,
      "learning_rate": 1.320010350770118e-07,
      "loss": 0.5936,
      "step": 20824
    },
    {
      "epoch": 1.86,
      "grad_norm": 6.40998223035195,
      "learning_rate": 1.3183619612548882e-07,
      "loss": 0.5748,
      "step": 20825
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.960343529002361,
      "learning_rate": 1.3167145878622446e-07,
      "loss": 0.5498,
      "step": 20826
    },
    {
      "epoch": 1.86,
      "grad_norm": 8.074714944985482,
      "learning_rate": 1.315068230626565e-07,
      "loss": 0.587,
      "step": 20827
    },
    {
      "epoch": 1.86,
      "grad_norm": 10.930983615210451,
      "learning_rate": 1.3134228895822166e-07,
      "loss": 0.5997,
      "step": 20828
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.833404069106863,
      "learning_rate": 1.3117785647635438e-07,
      "loss": 0.5204,
      "step": 20829
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.2435678506377705,
      "learning_rate": 1.3101352562048698e-07,
      "loss": 0.5362,
      "step": 20830
    },
    {
      "epoch": 1.86,
      "grad_norm": 4.9831901449143645,
      "learning_rate": 1.3084929639404997e-07,
      "loss": 0.5828,
      "step": 20831
    },
    {
      "epoch": 1.86,
      "grad_norm": 8.808364014579539,
      "learning_rate": 1.306851688004701e-07,
      "loss": 0.6058,
      "step": 20832
    },
    {
      "epoch": 1.86,
      "grad_norm": 6.259270901014186,
      "learning_rate": 1.3052114284317351e-07,
      "loss": 0.5461,
      "step": 20833
    },
    {
      "epoch": 1.86,
      "grad_norm": 9.385996076363854,
      "learning_rate": 1.3035721852558413e-07,
      "loss": 0.5949,
      "step": 20834
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.876072029494765,
      "learning_rate": 1.3019339585112368e-07,
      "loss": 0.5218,
      "step": 20835
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.732081845177594,
      "learning_rate": 1.3002967482321106e-07,
      "loss": 0.5774,
      "step": 20836
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.692333821457527,
      "learning_rate": 1.2986605544526465e-07,
      "loss": 0.6332,
      "step": 20837
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.651818823129153,
      "learning_rate": 1.297025377206984e-07,
      "loss": 0.5356,
      "step": 20838
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.153943524509841,
      "learning_rate": 1.2953912165292626e-07,
      "loss": 0.586,
      "step": 20839
    },
    {
      "epoch": 1.86,
      "grad_norm": 6.24557584481248,
      "learning_rate": 1.293758072453588e-07,
      "loss": 0.5762,
      "step": 20840
    },
    {
      "epoch": 1.86,
      "grad_norm": 6.415173725690853,
      "learning_rate": 1.2921259450140445e-07,
      "loss": 0.5409,
      "step": 20841
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.337528505154716,
      "learning_rate": 1.2904948342447043e-07,
      "loss": 0.5569,
      "step": 20842
    },
    {
      "epoch": 1.86,
      "grad_norm": 8.087247129376046,
      "learning_rate": 1.2888647401796129e-07,
      "loss": 0.5116,
      "step": 20843
    },
    {
      "epoch": 1.86,
      "grad_norm": 6.288121198820275,
      "learning_rate": 1.2872356628527982e-07,
      "loss": 0.5713,
      "step": 20844
    },
    {
      "epoch": 1.86,
      "grad_norm": 9.361416278766429,
      "learning_rate": 1.2856076022982557e-07,
      "loss": 0.6492,
      "step": 20845
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.226769858454966,
      "learning_rate": 1.2839805585499744e-07,
      "loss": 0.596,
      "step": 20846
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.005302319448655,
      "learning_rate": 1.2823545316419105e-07,
      "loss": 0.5067,
      "step": 20847
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.01187269728679,
      "learning_rate": 1.280729521608004e-07,
      "loss": 0.5797,
      "step": 20848
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.620131355174027,
      "learning_rate": 1.2791055284821774e-07,
      "loss": 0.5408,
      "step": 20849
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.62215942999358,
      "learning_rate": 1.2774825522983314e-07,
      "loss": 0.5317,
      "step": 20850
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.6831874515070515,
      "learning_rate": 1.275860593090328e-07,
      "loss": 0.5427,
      "step": 20851
    },
    {
      "epoch": 1.86,
      "grad_norm": 6.295267420827264,
      "learning_rate": 1.2742396508920342e-07,
      "loss": 0.495,
      "step": 20852
    },
    {
      "epoch": 1.86,
      "grad_norm": 8.049834909850775,
      "learning_rate": 1.2726197257372786e-07,
      "loss": 0.5081,
      "step": 20853
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.0816144956393074,
      "learning_rate": 1.2710008176598843e-07,
      "loss": 0.5619,
      "step": 20854
    },
    {
      "epoch": 1.86,
      "grad_norm": 6.191829277850562,
      "learning_rate": 1.269382926693624e-07,
      "loss": 0.5489,
      "step": 20855
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.426504896641189,
      "learning_rate": 1.2677660528722768e-07,
      "loss": 0.532,
      "step": 20856
    },
    {
      "epoch": 1.86,
      "grad_norm": 6.530443049887126,
      "learning_rate": 1.2661501962295985e-07,
      "loss": 0.5781,
      "step": 20857
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.371552076556492,
      "learning_rate": 1.2645353567993014e-07,
      "loss": 0.5566,
      "step": 20858
    },
    {
      "epoch": 1.86,
      "grad_norm": 6.945652668476601,
      "learning_rate": 1.2629215346151025e-07,
      "loss": 0.5395,
      "step": 20859
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.150754472671299,
      "learning_rate": 1.2613087297106863e-07,
      "loss": 0.4997,
      "step": 20860
    },
    {
      "epoch": 1.86,
      "grad_norm": 6.2640518658192885,
      "learning_rate": 1.259696942119709e-07,
      "loss": 0.5972,
      "step": 20861
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.17709732377082,
      "learning_rate": 1.2580861718758275e-07,
      "loss": 0.5554,
      "step": 20862
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.668104139429877,
      "learning_rate": 1.2564764190126477e-07,
      "loss": 0.5183,
      "step": 20863
    },
    {
      "epoch": 1.86,
      "grad_norm": 6.8538791527927065,
      "learning_rate": 1.2548676835637762e-07,
      "loss": 0.5451,
      "step": 20864
    },
    {
      "epoch": 1.86,
      "grad_norm": 6.031819516555116,
      "learning_rate": 1.253259965562792e-07,
      "loss": 0.5951,
      "step": 20865
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.5835770701555685,
      "learning_rate": 1.2516532650432512e-07,
      "loss": 0.5433,
      "step": 20866
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.904840384640677,
      "learning_rate": 1.2500475820386936e-07,
      "loss": 0.5453,
      "step": 20867
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.49736475495323,
      "learning_rate": 1.2484429165826374e-07,
      "loss": 0.5838,
      "step": 20868
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.474244509951982,
      "learning_rate": 1.2468392687085663e-07,
      "loss": 0.5884,
      "step": 20869
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.347121883761366,
      "learning_rate": 1.2452366384499592e-07,
      "loss": 0.5726,
      "step": 20870
    },
    {
      "epoch": 1.86,
      "grad_norm": 8.336977061124907,
      "learning_rate": 1.2436350258402618e-07,
      "loss": 0.6086,
      "step": 20871
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.483167921849278,
      "learning_rate": 1.242034430912914e-07,
      "loss": 0.561,
      "step": 20872
    },
    {
      "epoch": 1.86,
      "grad_norm": 6.553893645184795,
      "learning_rate": 1.2404348537013167e-07,
      "loss": 0.5229,
      "step": 20873
    },
    {
      "epoch": 1.86,
      "grad_norm": 9.888150082564447,
      "learning_rate": 1.2388362942388544e-07,
      "loss": 0.5729,
      "step": 20874
    },
    {
      "epoch": 1.86,
      "grad_norm": 6.166723804715248,
      "learning_rate": 1.2372387525589114e-07,
      "loss": 0.5586,
      "step": 20875
    },
    {
      "epoch": 1.86,
      "grad_norm": 6.031752396610235,
      "learning_rate": 1.235642228694811e-07,
      "loss": 0.5922,
      "step": 20876
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.636527516106462,
      "learning_rate": 1.2340467226798935e-07,
      "loss": 0.6135,
      "step": 20877
    },
    {
      "epoch": 1.86,
      "grad_norm": 8.491756950911448,
      "learning_rate": 1.232452234547449e-07,
      "loss": 0.5919,
      "step": 20878
    },
    {
      "epoch": 1.86,
      "grad_norm": 9.593700584333767,
      "learning_rate": 1.230858764330767e-07,
      "loss": 0.5372,
      "step": 20879
    },
    {
      "epoch": 1.86,
      "grad_norm": 6.620912018552872,
      "learning_rate": 1.2292663120631042e-07,
      "loss": 0.5331,
      "step": 20880
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.113053622045773,
      "learning_rate": 1.2276748777777124e-07,
      "loss": 0.6008,
      "step": 20881
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.0007817788367195,
      "learning_rate": 1.226084461507787e-07,
      "loss": 0.5586,
      "step": 20882
    },
    {
      "epoch": 1.86,
      "grad_norm": 8.566830754484338,
      "learning_rate": 1.2244950632865405e-07,
      "loss": 0.5883,
      "step": 20883
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.055544912452524,
      "learning_rate": 1.2229066831471404e-07,
      "loss": 0.5592,
      "step": 20884
    },
    {
      "epoch": 1.86,
      "grad_norm": 6.506648661483889,
      "learning_rate": 1.221319321122738e-07,
      "loss": 0.6495,
      "step": 20885
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.552023336318937,
      "learning_rate": 1.2197329772464793e-07,
      "loss": 0.5753,
      "step": 20886
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.438698453303239,
      "learning_rate": 1.21814765155146e-07,
      "loss": 0.5488,
      "step": 20887
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.954586186042278,
      "learning_rate": 1.216563344070787e-07,
      "loss": 0.5422,
      "step": 20888
    },
    {
      "epoch": 1.86,
      "grad_norm": 8.804441995457722,
      "learning_rate": 1.2149800548375167e-07,
      "loss": 0.5825,
      "step": 20889
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.234809180184427,
      "learning_rate": 1.213397783884701e-07,
      "loss": 0.496,
      "step": 20890
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.548833529805375,
      "learning_rate": 1.2118165312453633e-07,
      "loss": 0.5632,
      "step": 20891
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.539988231818698,
      "learning_rate": 1.2102362969525107e-07,
      "loss": 0.5504,
      "step": 20892
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.211054118868753,
      "learning_rate": 1.2086570810391275e-07,
      "loss": 0.5877,
      "step": 20893
    },
    {
      "epoch": 1.86,
      "grad_norm": 8.198400075389241,
      "learning_rate": 1.207078883538182e-07,
      "loss": 0.6082,
      "step": 20894
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.831563359671443,
      "learning_rate": 1.2055017044826033e-07,
      "loss": 0.5316,
      "step": 20895
    },
    {
      "epoch": 1.86,
      "grad_norm": 6.960349363198147,
      "learning_rate": 1.203925543905321e-07,
      "loss": 0.4851,
      "step": 20896
    },
    {
      "epoch": 1.86,
      "grad_norm": 6.347508470135518,
      "learning_rate": 1.2023504018392363e-07,
      "loss": 0.542,
      "step": 20897
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.668066741262089,
      "learning_rate": 1.200776278317217e-07,
      "loss": 0.5246,
      "step": 20898
    },
    {
      "epoch": 1.86,
      "grad_norm": 8.318392218451947,
      "learning_rate": 1.1992031733721265e-07,
      "loss": 0.5722,
      "step": 20899
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.3442136685444215,
      "learning_rate": 1.1976310870367935e-07,
      "loss": 0.4921,
      "step": 20900
    },
    {
      "epoch": 1.86,
      "grad_norm": 6.285326748487118,
      "learning_rate": 1.1960600193440418e-07,
      "loss": 0.6312,
      "step": 20901
    },
    {
      "epoch": 1.86,
      "grad_norm": 6.6632030455943045,
      "learning_rate": 1.1944899703266566e-07,
      "loss": 0.5718,
      "step": 20902
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.882800978957217,
      "learning_rate": 1.1929209400174058e-07,
      "loss": 0.5611,
      "step": 20903
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.551947262804913,
      "learning_rate": 1.1913529284490522e-07,
      "loss": 0.5343,
      "step": 20904
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.936624604155431,
      "learning_rate": 1.1897859356543085e-07,
      "loss": 0.5736,
      "step": 20905
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.756978794402099,
      "learning_rate": 1.1882199616658985e-07,
      "loss": 0.512,
      "step": 20906
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.1468495364979425,
      "learning_rate": 1.1866550065165017e-07,
      "loss": 0.5894,
      "step": 20907
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.674355866854298,
      "learning_rate": 1.1850910702387752e-07,
      "loss": 0.5549,
      "step": 20908
    },
    {
      "epoch": 1.87,
      "grad_norm": 8.934674538597452,
      "learning_rate": 1.1835281528653763e-07,
      "loss": 0.5953,
      "step": 20909
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.311335061490136,
      "learning_rate": 1.1819662544289178e-07,
      "loss": 0.5661,
      "step": 20910
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.720342663478497,
      "learning_rate": 1.1804053749620015e-07,
      "loss": 0.4986,
      "step": 20911
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.933016112160173,
      "learning_rate": 1.1788455144972178e-07,
      "loss": 0.5975,
      "step": 20912
    },
    {
      "epoch": 1.87,
      "grad_norm": 5.853590284522304,
      "learning_rate": 1.1772866730671128e-07,
      "loss": 0.5763,
      "step": 20913
    },
    {
      "epoch": 1.87,
      "grad_norm": 5.609953813246108,
      "learning_rate": 1.1757288507042331e-07,
      "loss": 0.5236,
      "step": 20914
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.1080928441080165,
      "learning_rate": 1.17417204744108e-07,
      "loss": 0.5645,
      "step": 20915
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.319052581198775,
      "learning_rate": 1.1726162633101668e-07,
      "loss": 0.5532,
      "step": 20916
    },
    {
      "epoch": 1.87,
      "grad_norm": 8.300198881567914,
      "learning_rate": 1.1710614983439561e-07,
      "loss": 0.6368,
      "step": 20917
    },
    {
      "epoch": 1.87,
      "grad_norm": 8.485380255574585,
      "learning_rate": 1.1695077525748999e-07,
      "loss": 0.6009,
      "step": 20918
    },
    {
      "epoch": 1.87,
      "grad_norm": 8.273778558564048,
      "learning_rate": 1.1679550260354334e-07,
      "loss": 0.5272,
      "step": 20919
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.059699153916687,
      "learning_rate": 1.1664033187579749e-07,
      "loss": 0.61,
      "step": 20920
    },
    {
      "epoch": 1.87,
      "grad_norm": 8.274940176486949,
      "learning_rate": 1.164852630774893e-07,
      "loss": 0.5857,
      "step": 20921
    },
    {
      "epoch": 1.87,
      "grad_norm": 5.672126319038237,
      "learning_rate": 1.1633029621185677e-07,
      "loss": 0.6091,
      "step": 20922
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.810306771626415,
      "learning_rate": 1.161754312821345e-07,
      "loss": 0.5741,
      "step": 20923
    },
    {
      "epoch": 1.87,
      "grad_norm": 8.37046194831419,
      "learning_rate": 1.1602066829155434e-07,
      "loss": 0.5647,
      "step": 20924
    },
    {
      "epoch": 1.87,
      "grad_norm": 5.998727723491224,
      "learning_rate": 1.1586600724334707e-07,
      "loss": 0.5206,
      "step": 20925
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.77527443332122,
      "learning_rate": 1.157114481407412e-07,
      "loss": 0.583,
      "step": 20926
    },
    {
      "epoch": 1.87,
      "grad_norm": 9.44946736760188,
      "learning_rate": 1.155569909869625e-07,
      "loss": 0.5324,
      "step": 20927
    },
    {
      "epoch": 1.87,
      "grad_norm": 10.616989072778397,
      "learning_rate": 1.1540263578523447e-07,
      "loss": 0.5438,
      "step": 20928
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.702694502700782,
      "learning_rate": 1.1524838253877956e-07,
      "loss": 0.5663,
      "step": 20929
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.639624171003314,
      "learning_rate": 1.1509423125081687e-07,
      "loss": 0.5548,
      "step": 20930
    },
    {
      "epoch": 1.87,
      "grad_norm": 5.129699777622099,
      "learning_rate": 1.1494018192456491e-07,
      "loss": 0.5153,
      "step": 20931
    },
    {
      "epoch": 1.87,
      "grad_norm": 5.229384835533354,
      "learning_rate": 1.1478623456323778e-07,
      "loss": 0.5887,
      "step": 20932
    },
    {
      "epoch": 1.87,
      "grad_norm": 8.334385949405162,
      "learning_rate": 1.146323891700507e-07,
      "loss": 0.5369,
      "step": 20933
    },
    {
      "epoch": 1.87,
      "grad_norm": 10.752366608220989,
      "learning_rate": 1.1447864574821277e-07,
      "loss": 0.6436,
      "step": 20934
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.002575225995843,
      "learning_rate": 1.1432500430093418e-07,
      "loss": 0.5511,
      "step": 20935
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.088028013609253,
      "learning_rate": 1.1417146483142183e-07,
      "loss": 0.5521,
      "step": 20936
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.6453532235425055,
      "learning_rate": 1.1401802734287981e-07,
      "loss": 0.5975,
      "step": 20937
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.321766881540197,
      "learning_rate": 1.1386469183851223e-07,
      "loss": 0.6044,
      "step": 20938
    },
    {
      "epoch": 1.87,
      "grad_norm": 8.173029638014793,
      "learning_rate": 1.1371145832151875e-07,
      "loss": 0.6119,
      "step": 20939
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.131477204313819,
      "learning_rate": 1.1355832679509737e-07,
      "loss": 0.5596,
      "step": 20940
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.580047662161449,
      "learning_rate": 1.1340529726244497e-07,
      "loss": 0.6023,
      "step": 20941
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.619435017507248,
      "learning_rate": 1.1325236972675513e-07,
      "loss": 0.5588,
      "step": 20942
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.683804453365449,
      "learning_rate": 1.1309954419122083e-07,
      "loss": 0.5869,
      "step": 20943
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.09944224744641,
      "learning_rate": 1.1294682065903062e-07,
      "loss": 0.6749,
      "step": 20944
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.5487041492200975,
      "learning_rate": 1.127941991333742e-07,
      "loss": 0.5304,
      "step": 20945
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.446764296626316,
      "learning_rate": 1.1264167961743511e-07,
      "loss": 0.547,
      "step": 20946
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.2545087510964334,
      "learning_rate": 1.1248926211439804e-07,
      "loss": 0.5487,
      "step": 20947
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.510454450931272,
      "learning_rate": 1.1233694662744433e-07,
      "loss": 0.604,
      "step": 20948
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.054648532294999,
      "learning_rate": 1.1218473315975309e-07,
      "loss": 0.5751,
      "step": 20949
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.07277547793017,
      "learning_rate": 1.1203262171450125e-07,
      "loss": 0.5575,
      "step": 20950
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.336523506779946,
      "learning_rate": 1.1188061229486457e-07,
      "loss": 0.5812,
      "step": 20951
    },
    {
      "epoch": 1.87,
      "grad_norm": 8.672214679922554,
      "learning_rate": 1.1172870490401443e-07,
      "loss": 0.5661,
      "step": 20952
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.9845773082643925,
      "learning_rate": 1.1157689954512274e-07,
      "loss": 0.6505,
      "step": 20953
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.350269679368775,
      "learning_rate": 1.114251962213575e-07,
      "loss": 0.6356,
      "step": 20954
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.744722879055875,
      "learning_rate": 1.1127359493588675e-07,
      "loss": 0.5528,
      "step": 20955
    },
    {
      "epoch": 1.87,
      "grad_norm": 5.24211728915567,
      "learning_rate": 1.1112209569187237e-07,
      "loss": 0.6083,
      "step": 20956
    },
    {
      "epoch": 1.87,
      "grad_norm": 5.003613297908868,
      "learning_rate": 1.1097069849247799e-07,
      "loss": 0.5467,
      "step": 20957
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.354998931589032,
      "learning_rate": 1.1081940334086383e-07,
      "loss": 0.6045,
      "step": 20958
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.334818328275256,
      "learning_rate": 1.1066821024018737e-07,
      "loss": 0.5575,
      "step": 20959
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.340929795306152,
      "learning_rate": 1.1051711919360442e-07,
      "loss": 0.6377,
      "step": 20960
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.7613701049822925,
      "learning_rate": 1.1036613020426857e-07,
      "loss": 0.5538,
      "step": 20961
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.517978576874103,
      "learning_rate": 1.1021524327533173e-07,
      "loss": 0.5762,
      "step": 20962
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.2094377973948465,
      "learning_rate": 1.1006445840994362e-07,
      "loss": 0.5722,
      "step": 20963
    },
    {
      "epoch": 1.87,
      "grad_norm": 5.1648454765545,
      "learning_rate": 1.0991377561125117e-07,
      "loss": 0.6041,
      "step": 20964
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.6184597610100715,
      "learning_rate": 1.0976319488239962e-07,
      "loss": 0.6285,
      "step": 20965
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.117048501331203,
      "learning_rate": 1.0961271622653146e-07,
      "loss": 0.5369,
      "step": 20966
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.095108293115146,
      "learning_rate": 1.094623396467881e-07,
      "loss": 0.5412,
      "step": 20967
    },
    {
      "epoch": 1.87,
      "grad_norm": 5.926405046707017,
      "learning_rate": 1.0931206514630865e-07,
      "loss": 0.5289,
      "step": 20968
    },
    {
      "epoch": 1.87,
      "grad_norm": 5.513219753036493,
      "learning_rate": 1.0916189272822952e-07,
      "loss": 0.5609,
      "step": 20969
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.2191461126698595,
      "learning_rate": 1.0901182239568486e-07,
      "loss": 0.5208,
      "step": 20970
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.723353851902933,
      "learning_rate": 1.0886185415180772e-07,
      "loss": 0.5385,
      "step": 20971
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.861800870430771,
      "learning_rate": 1.0871198799972727e-07,
      "loss": 0.5556,
      "step": 20972
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.978839196497429,
      "learning_rate": 1.0856222394257266e-07,
      "loss": 0.5457,
      "step": 20973
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.247056373850858,
      "learning_rate": 1.0841256198346972e-07,
      "loss": 0.5456,
      "step": 20974
    },
    {
      "epoch": 1.87,
      "grad_norm": 5.466506386462839,
      "learning_rate": 1.0826300212554152e-07,
      "loss": 0.5465,
      "step": 20975
    },
    {
      "epoch": 1.87,
      "grad_norm": 5.973019177804934,
      "learning_rate": 1.0811354437191112e-07,
      "loss": 0.6181,
      "step": 20976
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.052942396593039,
      "learning_rate": 1.079641887256977e-07,
      "loss": 0.5722,
      "step": 20977
    },
    {
      "epoch": 1.87,
      "grad_norm": 9.88685081089164,
      "learning_rate": 1.0781493519001762e-07,
      "loss": 0.657,
      "step": 20978
    },
    {
      "epoch": 1.87,
      "grad_norm": 8.703250707936471,
      "learning_rate": 1.076657837679873e-07,
      "loss": 0.6004,
      "step": 20979
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.4981857177443105,
      "learning_rate": 1.0751673446271982e-07,
      "loss": 0.5229,
      "step": 20980
    },
    {
      "epoch": 1.87,
      "grad_norm": 5.9910162353900835,
      "learning_rate": 1.0736778727732599e-07,
      "loss": 0.5412,
      "step": 20981
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.811396228691841,
      "learning_rate": 1.0721894221491503e-07,
      "loss": 0.5812,
      "step": 20982
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.135355408093023,
      "learning_rate": 1.0707019927859442e-07,
      "loss": 0.5629,
      "step": 20983
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.618619654228134,
      "learning_rate": 1.0692155847146778e-07,
      "loss": 0.5609,
      "step": 20984
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.008841302458632,
      "learning_rate": 1.0677301979663712e-07,
      "loss": 0.5995,
      "step": 20985
    },
    {
      "epoch": 1.87,
      "grad_norm": 8.076161352596728,
      "learning_rate": 1.0662458325720437e-07,
      "loss": 0.5478,
      "step": 20986
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.39499712359614,
      "learning_rate": 1.0647624885626651e-07,
      "loss": 0.5045,
      "step": 20987
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.227426236675855,
      "learning_rate": 1.0632801659692049e-07,
      "loss": 0.5288,
      "step": 20988
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.673261044738194,
      "learning_rate": 1.0617988648226052e-07,
      "loss": 0.5762,
      "step": 20989
    },
    {
      "epoch": 1.87,
      "grad_norm": 5.52449066866207,
      "learning_rate": 1.0603185851537801e-07,
      "loss": 0.6155,
      "step": 20990
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.871960270340291,
      "learning_rate": 1.0588393269936271e-07,
      "loss": 0.5442,
      "step": 20991
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.360867096397531,
      "learning_rate": 1.057361090373027e-07,
      "loss": 0.6224,
      "step": 20992
    },
    {
      "epoch": 1.87,
      "grad_norm": 8.191812047350124,
      "learning_rate": 1.0558838753228274e-07,
      "loss": 0.5469,
      "step": 20993
    },
    {
      "epoch": 1.87,
      "grad_norm": 8.220967750961645,
      "learning_rate": 1.0544076818738647e-07,
      "loss": 0.567,
      "step": 20994
    },
    {
      "epoch": 1.87,
      "grad_norm": 5.918973395484816,
      "learning_rate": 1.0529325100569587e-07,
      "loss": 0.5688,
      "step": 20995
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.1131889417481675,
      "learning_rate": 1.051458359902896e-07,
      "loss": 0.577,
      "step": 20996
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.500139646554104,
      "learning_rate": 1.0499852314424352e-07,
      "loss": 0.5688,
      "step": 20997
    },
    {
      "epoch": 1.87,
      "grad_norm": 5.784883374748337,
      "learning_rate": 1.0485131247063462e-07,
      "loss": 0.5661,
      "step": 20998
    },
    {
      "epoch": 1.87,
      "grad_norm": 5.105611509148954,
      "learning_rate": 1.0470420397253323e-07,
      "loss": 0.531,
      "step": 20999
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.081659600619727,
      "learning_rate": 1.0455719765301132e-07,
      "loss": 0.5507,
      "step": 21000
    },
    {
      "epoch": 1.87,
      "grad_norm": 5.7648986896866505,
      "learning_rate": 1.0441029351513754e-07,
      "loss": 0.5222,
      "step": 21001
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.882754037045833,
      "learning_rate": 1.0426349156197724e-07,
      "loss": 0.6034,
      "step": 21002
    },
    {
      "epoch": 1.87,
      "grad_norm": 8.564706896905118,
      "learning_rate": 1.0411679179659572e-07,
      "loss": 0.6193,
      "step": 21003
    },
    {
      "epoch": 1.87,
      "grad_norm": 5.9486969852878975,
      "learning_rate": 1.039701942220539e-07,
      "loss": 0.5172,
      "step": 21004
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.182818354273246,
      "learning_rate": 1.0382369884141208e-07,
      "loss": 0.5954,
      "step": 21005
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.117792817763418,
      "learning_rate": 1.0367730565772782e-07,
      "loss": 0.5806,
      "step": 21006
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.744833521114091,
      "learning_rate": 1.0353101467405757e-07,
      "loss": 0.5652,
      "step": 21007
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.043595638035254,
      "learning_rate": 1.0338482589345388e-07,
      "loss": 0.5976,
      "step": 21008
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.92044000804046,
      "learning_rate": 1.0323873931896932e-07,
      "loss": 0.6178,
      "step": 21009
    },
    {
      "epoch": 1.87,
      "grad_norm": 5.756595721755721,
      "learning_rate": 1.0309275495365145e-07,
      "loss": 0.5571,
      "step": 21010
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.8404246638191335,
      "learning_rate": 1.0294687280054893e-07,
      "loss": 0.5876,
      "step": 21011
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.324573143669161,
      "learning_rate": 1.0280109286270545e-07,
      "loss": 0.5683,
      "step": 21012
    },
    {
      "epoch": 1.87,
      "grad_norm": 5.392224068117696,
      "learning_rate": 1.0265541514316524e-07,
      "loss": 0.5924,
      "step": 21013
    },
    {
      "epoch": 1.87,
      "grad_norm": 4.7213037041337795,
      "learning_rate": 1.0250983964496753e-07,
      "loss": 0.5692,
      "step": 21014
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.3262848901877815,
      "learning_rate": 1.023643663711521e-07,
      "loss": 0.5796,
      "step": 21015
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.765760935001783,
      "learning_rate": 1.0221899532475487e-07,
      "loss": 0.5945,
      "step": 21016
    },
    {
      "epoch": 1.87,
      "grad_norm": 4.663834158201626,
      "learning_rate": 1.0207372650881009e-07,
      "loss": 0.5627,
      "step": 21017
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.761356836217155,
      "learning_rate": 1.0192855992634976e-07,
      "loss": 0.5876,
      "step": 21018
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.477685171264411,
      "learning_rate": 1.0178349558040423e-07,
      "loss": 0.5668,
      "step": 21019
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.966789202727508,
      "learning_rate": 1.016385334740011e-07,
      "loss": 0.5409,
      "step": 21020
    },
    {
      "epoch": 1.88,
      "grad_norm": 8.540290575300181,
      "learning_rate": 1.0149367361016681e-07,
      "loss": 0.5866,
      "step": 21021
    },
    {
      "epoch": 1.88,
      "grad_norm": 5.877045220239096,
      "learning_rate": 1.0134891599192454e-07,
      "loss": 0.5554,
      "step": 21022
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.4945530247157635,
      "learning_rate": 1.0120426062229516e-07,
      "loss": 0.6048,
      "step": 21023
    },
    {
      "epoch": 1.88,
      "grad_norm": 5.756123808370641,
      "learning_rate": 1.0105970750429905e-07,
      "loss": 0.5804,
      "step": 21024
    },
    {
      "epoch": 1.88,
      "grad_norm": 5.80553577021573,
      "learning_rate": 1.0091525664095325e-07,
      "loss": 0.5756,
      "step": 21025
    },
    {
      "epoch": 1.88,
      "grad_norm": 8.028187409184863,
      "learning_rate": 1.0077090803527201e-07,
      "loss": 0.5451,
      "step": 21026
    },
    {
      "epoch": 1.88,
      "grad_norm": 5.726121164201468,
      "learning_rate": 1.0062666169027013e-07,
      "loss": 0.5572,
      "step": 21027
    },
    {
      "epoch": 1.88,
      "grad_norm": 8.044147922793677,
      "learning_rate": 1.0048251760895633e-07,
      "loss": 0.573,
      "step": 21028
    },
    {
      "epoch": 1.88,
      "grad_norm": 8.32404009080991,
      "learning_rate": 1.0033847579434097e-07,
      "loss": 0.53,
      "step": 21029
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.017190452028039,
      "learning_rate": 1.0019453624942888e-07,
      "loss": 0.6251,
      "step": 21030
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.984505869247188,
      "learning_rate": 1.0005069897722596e-07,
      "loss": 0.5828,
      "step": 21031
    },
    {
      "epoch": 1.88,
      "grad_norm": 8.251839727816092,
      "learning_rate": 9.990696398073374e-08,
      "loss": 0.5571,
      "step": 21032
    },
    {
      "epoch": 1.88,
      "grad_norm": 9.674990465783313,
      "learning_rate": 9.976333126295257e-08,
      "loss": 0.5807,
      "step": 21033
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.183177169012831,
      "learning_rate": 9.961980082688116e-08,
      "loss": 0.5347,
      "step": 21034
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.2984786773149954,
      "learning_rate": 9.947637267551436e-08,
      "loss": 0.5431,
      "step": 21035
    },
    {
      "epoch": 1.88,
      "grad_norm": 8.213698527413909,
      "learning_rate": 9.933304681184585e-08,
      "loss": 0.5312,
      "step": 21036
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.125524649620422,
      "learning_rate": 9.918982323886828e-08,
      "loss": 0.5488,
      "step": 21037
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.878777547878165,
      "learning_rate": 9.904670195957034e-08,
      "loss": 0.6652,
      "step": 21038
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.154940016007884,
      "learning_rate": 9.890368297693965e-08,
      "loss": 0.5184,
      "step": 21039
    },
    {
      "epoch": 1.88,
      "grad_norm": 9.121495592600011,
      "learning_rate": 9.876076629396158e-08,
      "loss": 0.5908,
      "step": 21040
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.9152238896204805,
      "learning_rate": 9.861795191361933e-08,
      "loss": 0.5123,
      "step": 21041
    },
    {
      "epoch": 1.88,
      "grad_norm": 5.2665991966493175,
      "learning_rate": 9.847523983889273e-08,
      "loss": 0.6131,
      "step": 21042
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.613200426822425,
      "learning_rate": 9.833263007276162e-08,
      "loss": 0.6033,
      "step": 21043
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.690416990376193,
      "learning_rate": 9.819012261820193e-08,
      "loss": 0.5369,
      "step": 21044
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.902464524271597,
      "learning_rate": 9.804771747818909e-08,
      "loss": 0.5978,
      "step": 21045
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.754476106472237,
      "learning_rate": 9.790541465569513e-08,
      "loss": 0.6242,
      "step": 21046
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.929307900018466,
      "learning_rate": 9.77632141536905e-08,
      "loss": 0.5393,
      "step": 21047
    },
    {
      "epoch": 1.88,
      "grad_norm": 8.144310903613064,
      "learning_rate": 9.762111597514224e-08,
      "loss": 0.5392,
      "step": 21048
    },
    {
      "epoch": 1.88,
      "grad_norm": 8.798252309915991,
      "learning_rate": 9.747912012301796e-08,
      "loss": 0.5879,
      "step": 21049
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.087811128086951,
      "learning_rate": 9.733722660028033e-08,
      "loss": 0.5224,
      "step": 21050
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.717106677846806,
      "learning_rate": 9.719543540989084e-08,
      "loss": 0.5735,
      "step": 21051
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.386410185427043,
      "learning_rate": 9.705374655481048e-08,
      "loss": 0.5465,
      "step": 21052
    },
    {
      "epoch": 1.88,
      "grad_norm": 8.342076035618675,
      "learning_rate": 9.691216003799631e-08,
      "loss": 0.6147,
      "step": 21053
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.373354154442819,
      "learning_rate": 9.677067586240208e-08,
      "loss": 0.5481,
      "step": 21054
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.026883464010685,
      "learning_rate": 9.662929403098264e-08,
      "loss": 0.5855,
      "step": 21055
    },
    {
      "epoch": 1.88,
      "grad_norm": 8.948727389901416,
      "learning_rate": 9.6488014546689e-08,
      "loss": 0.5914,
      "step": 21056
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.0119345841496035,
      "learning_rate": 9.634683741246876e-08,
      "loss": 0.512,
      "step": 21057
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.93871094655152,
      "learning_rate": 9.620576263126957e-08,
      "loss": 0.6586,
      "step": 21058
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.693270915639578,
      "learning_rate": 9.606479020603632e-08,
      "loss": 0.6209,
      "step": 21059
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.232523815545704,
      "learning_rate": 9.592392013971108e-08,
      "loss": 0.5739,
      "step": 21060
    },
    {
      "epoch": 1.88,
      "grad_norm": 8.112241687000637,
      "learning_rate": 9.578315243523429e-08,
      "loss": 0.5929,
      "step": 21061
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.042098238482723,
      "learning_rate": 9.564248709554414e-08,
      "loss": 0.5416,
      "step": 21062
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.324781172846449,
      "learning_rate": 9.550192412357606e-08,
      "loss": 0.6331,
      "step": 21063
    },
    {
      "epoch": 1.88,
      "grad_norm": 8.330404089884158,
      "learning_rate": 9.53614635222655e-08,
      "loss": 0.5862,
      "step": 21064
    },
    {
      "epoch": 1.88,
      "grad_norm": 8.357932501359688,
      "learning_rate": 9.522110529454287e-08,
      "loss": 0.5769,
      "step": 21065
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.0861906281064835,
      "learning_rate": 9.508084944333862e-08,
      "loss": 0.5198,
      "step": 21066
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.706221033533223,
      "learning_rate": 9.494069597157985e-08,
      "loss": 0.6321,
      "step": 21067
    },
    {
      "epoch": 1.88,
      "grad_norm": 5.181119666370835,
      "learning_rate": 9.480064488219254e-08,
      "loss": 0.4918,
      "step": 21068
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.457176072222198,
      "learning_rate": 9.466069617809937e-08,
      "loss": 0.6288,
      "step": 21069
    },
    {
      "epoch": 1.88,
      "grad_norm": 9.307425518226696,
      "learning_rate": 9.452084986222187e-08,
      "loss": 0.5613,
      "step": 21070
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.995049484840391,
      "learning_rate": 9.438110593747829e-08,
      "loss": 0.5403,
      "step": 21071
    },
    {
      "epoch": 1.88,
      "grad_norm": 8.358372076397222,
      "learning_rate": 9.424146440678628e-08,
      "loss": 0.4983,
      "step": 21072
    },
    {
      "epoch": 1.88,
      "grad_norm": 10.894386721979007,
      "learning_rate": 9.410192527306072e-08,
      "loss": 0.576,
      "step": 21073
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.358989337853672,
      "learning_rate": 9.39624885392132e-08,
      "loss": 0.5125,
      "step": 21074
    },
    {
      "epoch": 1.88,
      "grad_norm": 9.017168031045633,
      "learning_rate": 9.382315420815469e-08,
      "loss": 0.5543,
      "step": 21075
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.25357257921551,
      "learning_rate": 9.368392228279344e-08,
      "loss": 0.5578,
      "step": 21076
    },
    {
      "epoch": 1.88,
      "grad_norm": 12.40406806151217,
      "learning_rate": 9.354479276603546e-08,
      "loss": 0.6401,
      "step": 21077
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.523919425834857,
      "learning_rate": 9.340576566078563e-08,
      "loss": 0.5109,
      "step": 21078
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.746705850901944,
      "learning_rate": 9.326684096994498e-08,
      "loss": 0.5777,
      "step": 21079
    },
    {
      "epoch": 1.88,
      "grad_norm": 5.720976722641286,
      "learning_rate": 9.312801869641341e-08,
      "loss": 0.5493,
      "step": 21080
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.147691602844768,
      "learning_rate": 9.29892988430886e-08,
      "loss": 0.5294,
      "step": 21081
    },
    {
      "epoch": 1.88,
      "grad_norm": 5.4350075382095335,
      "learning_rate": 9.285068141286547e-08,
      "loss": 0.535,
      "step": 21082
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.994920264991677,
      "learning_rate": 9.271216640863834e-08,
      "loss": 0.5257,
      "step": 21083
    },
    {
      "epoch": 1.88,
      "grad_norm": 10.093781463787423,
      "learning_rate": 9.257375383329825e-08,
      "loss": 0.5624,
      "step": 21084
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.430438183147152,
      "learning_rate": 9.2435443689734e-08,
      "loss": 0.5839,
      "step": 21085
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.982229787678982,
      "learning_rate": 9.229723598083218e-08,
      "loss": 0.5251,
      "step": 21086
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.998294347362701,
      "learning_rate": 9.215913070947769e-08,
      "loss": 0.5367,
      "step": 21087
    },
    {
      "epoch": 1.88,
      "grad_norm": 9.631673910914987,
      "learning_rate": 9.202112787855377e-08,
      "loss": 0.5538,
      "step": 21088
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.076637514098293,
      "learning_rate": 9.188322749094037e-08,
      "loss": 0.5761,
      "step": 21089
    },
    {
      "epoch": 1.88,
      "grad_norm": 5.797484830147023,
      "learning_rate": 9.174542954951571e-08,
      "loss": 0.5927,
      "step": 21090
    },
    {
      "epoch": 1.88,
      "grad_norm": 5.26199837826957,
      "learning_rate": 9.160773405715639e-08,
      "loss": 0.5355,
      "step": 21091
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.311031241913768,
      "learning_rate": 9.14701410167368e-08,
      "loss": 0.5415,
      "step": 21092
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.031135181305323,
      "learning_rate": 9.133265043112848e-08,
      "loss": 0.5719,
      "step": 21093
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.08666209420605,
      "learning_rate": 9.119526230320086e-08,
      "loss": 0.5189,
      "step": 21094
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.278733883808456,
      "learning_rate": 9.105797663582217e-08,
      "loss": 0.5825,
      "step": 21095
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.731023127240319,
      "learning_rate": 9.09207934318579e-08,
      "loss": 0.5376,
      "step": 21096
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.491362047355932,
      "learning_rate": 9.07837126941713e-08,
      "loss": 0.5216,
      "step": 21097
    },
    {
      "epoch": 1.88,
      "grad_norm": 8.135138066493788,
      "learning_rate": 9.064673442562399e-08,
      "loss": 0.5512,
      "step": 21098
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.720871081480444,
      "learning_rate": 9.05098586290748e-08,
      "loss": 0.5985,
      "step": 21099
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.329273175589783,
      "learning_rate": 9.037308530738032e-08,
      "loss": 0.6054,
      "step": 21100
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.983989579848732,
      "learning_rate": 9.023641446339604e-08,
      "loss": 0.5724,
      "step": 21101
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.68364464774362,
      "learning_rate": 9.009984609997414e-08,
      "loss": 0.5728,
      "step": 21102
    },
    {
      "epoch": 1.88,
      "grad_norm": 5.45456930749683,
      "learning_rate": 8.996338021996564e-08,
      "loss": 0.5548,
      "step": 21103
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.485024020850431,
      "learning_rate": 8.982701682621887e-08,
      "loss": 0.5456,
      "step": 21104
    },
    {
      "epoch": 1.88,
      "grad_norm": 11.012736006234121,
      "learning_rate": 8.969075592158039e-08,
      "loss": 0.5319,
      "step": 21105
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.680700642372978,
      "learning_rate": 8.955459750889295e-08,
      "loss": 0.5453,
      "step": 21106
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.97370394914972,
      "learning_rate": 8.941854159100038e-08,
      "loss": 0.6031,
      "step": 21107
    },
    {
      "epoch": 1.88,
      "grad_norm": 10.223889324051077,
      "learning_rate": 8.928258817074153e-08,
      "loss": 0.5341,
      "step": 21108
    },
    {
      "epoch": 1.88,
      "grad_norm": 9.920022119853549,
      "learning_rate": 8.914673725095413e-08,
      "loss": 0.58,
      "step": 21109
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.539416770130542,
      "learning_rate": 8.901098883447478e-08,
      "loss": 0.5349,
      "step": 21110
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.829898130753236,
      "learning_rate": 8.887534292413569e-08,
      "loss": 0.5597,
      "step": 21111
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.405169741408688,
      "learning_rate": 8.873979952276846e-08,
      "loss": 0.5541,
      "step": 21112
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.890574789813791,
      "learning_rate": 8.860435863320249e-08,
      "loss": 0.5733,
      "step": 21113
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.06281027199534,
      "learning_rate": 8.846902025826498e-08,
      "loss": 0.6075,
      "step": 21114
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.769969459543523,
      "learning_rate": 8.833378440078034e-08,
      "loss": 0.6162,
      "step": 21115
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.033262339254884,
      "learning_rate": 8.819865106357183e-08,
      "loss": 0.5981,
      "step": 21116
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.9277828219379085,
      "learning_rate": 8.806362024946003e-08,
      "loss": 0.55,
      "step": 21117
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.531070891734815,
      "learning_rate": 8.792869196126263e-08,
      "loss": 0.4928,
      "step": 21118
    },
    {
      "epoch": 1.88,
      "grad_norm": 8.297239634440928,
      "learning_rate": 8.779386620179687e-08,
      "loss": 0.5872,
      "step": 21119
    },
    {
      "epoch": 1.88,
      "grad_norm": 8.347112691020662,
      "learning_rate": 8.765914297387657e-08,
      "loss": 0.5995,
      "step": 21120
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.840416030735981,
      "learning_rate": 8.752452228031339e-08,
      "loss": 0.5812,
      "step": 21121
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.97885316150236,
      "learning_rate": 8.739000412391785e-08,
      "loss": 0.5771,
      "step": 21122
    },
    {
      "epoch": 1.88,
      "grad_norm": 9.114544796708218,
      "learning_rate": 8.725558850749827e-08,
      "loss": 0.5627,
      "step": 21123
    },
    {
      "epoch": 1.88,
      "grad_norm": 10.219581524215707,
      "learning_rate": 8.712127543385907e-08,
      "loss": 0.5639,
      "step": 21124
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.8106195448937585,
      "learning_rate": 8.698706490580411e-08,
      "loss": 0.5008,
      "step": 21125
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.497132225304615,
      "learning_rate": 8.685295692613448e-08,
      "loss": 0.5673,
      "step": 21126
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.903177687800151,
      "learning_rate": 8.671895149765019e-08,
      "loss": 0.5933,
      "step": 21127
    },
    {
      "epoch": 1.88,
      "grad_norm": 8.34520062040471,
      "learning_rate": 8.658504862314842e-08,
      "loss": 0.58,
      "step": 21128
    },
    {
      "epoch": 1.88,
      "grad_norm": 8.978786452738884,
      "learning_rate": 8.645124830542306e-08,
      "loss": 0.545,
      "step": 21129
    },
    {
      "epoch": 1.89,
      "grad_norm": 8.546217499072739,
      "learning_rate": 8.631755054726742e-08,
      "loss": 0.5541,
      "step": 21130
    },
    {
      "epoch": 1.89,
      "grad_norm": 5.829793517665486,
      "learning_rate": 8.618395535147205e-08,
      "loss": 0.5557,
      "step": 21131
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.180026328352093,
      "learning_rate": 8.605046272082529e-08,
      "loss": 0.5721,
      "step": 21132
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.189375459895696,
      "learning_rate": 8.591707265811433e-08,
      "loss": 0.645,
      "step": 21133
    },
    {
      "epoch": 1.89,
      "grad_norm": 8.758828113363906,
      "learning_rate": 8.578378516612251e-08,
      "loss": 0.5518,
      "step": 21134
    },
    {
      "epoch": 1.89,
      "grad_norm": 8.270019547459423,
      "learning_rate": 8.565060024763205e-08,
      "loss": 0.5654,
      "step": 21135
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.945740115012732,
      "learning_rate": 8.551751790542351e-08,
      "loss": 0.6497,
      "step": 21136
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.48431986513705,
      "learning_rate": 8.53845381422741e-08,
      "loss": 0.5342,
      "step": 21137
    },
    {
      "epoch": 1.89,
      "grad_norm": 5.816646463486906,
      "learning_rate": 8.525166096095938e-08,
      "loss": 0.5494,
      "step": 21138
    },
    {
      "epoch": 1.89,
      "grad_norm": 8.0944829235967,
      "learning_rate": 8.511888636425325e-08,
      "loss": 0.5972,
      "step": 21139
    },
    {
      "epoch": 1.89,
      "grad_norm": 5.087622257138645,
      "learning_rate": 8.498621435492682e-08,
      "loss": 0.5373,
      "step": 21140
    },
    {
      "epoch": 1.89,
      "grad_norm": 9.190088149044145,
      "learning_rate": 8.485364493574954e-08,
      "loss": 0.5279,
      "step": 21141
    },
    {
      "epoch": 1.89,
      "grad_norm": 9.026852965193367,
      "learning_rate": 8.472117810948866e-08,
      "loss": 0.5252,
      "step": 21142
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.460827744204616,
      "learning_rate": 8.458881387890915e-08,
      "loss": 0.5311,
      "step": 21143
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.948383371409069,
      "learning_rate": 8.44565522467733e-08,
      "loss": 0.5511,
      "step": 21144
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.7385698948371235,
      "learning_rate": 8.432439321584163e-08,
      "loss": 0.5371,
      "step": 21145
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.518330506943037,
      "learning_rate": 8.419233678887362e-08,
      "loss": 0.5279,
      "step": 21146
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.740748241442324,
      "learning_rate": 8.406038296862484e-08,
      "loss": 0.6462,
      "step": 21147
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.412072516617177,
      "learning_rate": 8.392853175784976e-08,
      "loss": 0.5529,
      "step": 21148
    },
    {
      "epoch": 1.89,
      "grad_norm": 5.6597061762987915,
      "learning_rate": 8.379678315930118e-08,
      "loss": 0.547,
      "step": 21149
    },
    {
      "epoch": 1.89,
      "grad_norm": 8.81068304869317,
      "learning_rate": 8.366513717572799e-08,
      "loss": 0.5885,
      "step": 21150
    },
    {
      "epoch": 1.89,
      "grad_norm": 8.475895930156181,
      "learning_rate": 8.353359380987858e-08,
      "loss": 0.6308,
      "step": 21151
    },
    {
      "epoch": 1.89,
      "grad_norm": 8.28513526965134,
      "learning_rate": 8.340215306449851e-08,
      "loss": 0.5955,
      "step": 21152
    },
    {
      "epoch": 1.89,
      "grad_norm": 8.592654312280175,
      "learning_rate": 8.327081494233113e-08,
      "loss": 0.524,
      "step": 21153
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.450810856364779,
      "learning_rate": 8.313957944611817e-08,
      "loss": 0.5524,
      "step": 21154
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.181207319846565,
      "learning_rate": 8.300844657859908e-08,
      "loss": 0.5395,
      "step": 21155
    },
    {
      "epoch": 1.89,
      "grad_norm": 8.190229305582598,
      "learning_rate": 8.287741634251001e-08,
      "loss": 0.5736,
      "step": 21156
    },
    {
      "epoch": 1.89,
      "grad_norm": 5.261313972629176,
      "learning_rate": 8.274648874058766e-08,
      "loss": 0.5485,
      "step": 21157
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.364280200376816,
      "learning_rate": 8.261566377556262e-08,
      "loss": 0.6123,
      "step": 21158
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.137916298788042,
      "learning_rate": 8.248494145016717e-08,
      "loss": 0.5289,
      "step": 21159
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.113797752586996,
      "learning_rate": 8.23543217671291e-08,
      "loss": 0.5605,
      "step": 21160
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.287510052170146,
      "learning_rate": 8.222380472917513e-08,
      "loss": 0.5493,
      "step": 21161
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.270832573943363,
      "learning_rate": 8.20933903390303e-08,
      "loss": 0.544,
      "step": 21162
    },
    {
      "epoch": 1.89,
      "grad_norm": 8.439224567632623,
      "learning_rate": 8.196307859941521e-08,
      "loss": 0.5788,
      "step": 21163
    },
    {
      "epoch": 1.89,
      "grad_norm": 8.924388574510065,
      "learning_rate": 8.183286951305047e-08,
      "loss": 0.5449,
      "step": 21164
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.369660143928247,
      "learning_rate": 8.17027630826539e-08,
      "loss": 0.5215,
      "step": 21165
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.029320396172398,
      "learning_rate": 8.157275931094111e-08,
      "loss": 0.6069,
      "step": 21166
    },
    {
      "epoch": 1.89,
      "grad_norm": 9.843167276843955,
      "learning_rate": 8.144285820062658e-08,
      "loss": 0.5663,
      "step": 21167
    },
    {
      "epoch": 1.89,
      "grad_norm": 5.099140189899764,
      "learning_rate": 8.131305975442039e-08,
      "loss": 0.5821,
      "step": 21168
    },
    {
      "epoch": 1.89,
      "grad_norm": 5.727717331201871,
      "learning_rate": 8.118336397503257e-08,
      "loss": 0.5515,
      "step": 21169
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.390398060383969,
      "learning_rate": 8.105377086516986e-08,
      "loss": 0.5671,
      "step": 21170
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.111677188852284,
      "learning_rate": 8.092428042753786e-08,
      "loss": 0.5328,
      "step": 21171
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.418843425236351,
      "learning_rate": 8.079489266483831e-08,
      "loss": 0.5185,
      "step": 21172
    },
    {
      "epoch": 1.89,
      "grad_norm": 10.170038208841909,
      "learning_rate": 8.066560757977238e-08,
      "loss": 0.5309,
      "step": 21173
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.341451944030578,
      "learning_rate": 8.053642517503901e-08,
      "loss": 0.5658,
      "step": 21174
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.008435330145526,
      "learning_rate": 8.04073454533344e-08,
      "loss": 0.6095,
      "step": 21175
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.377374831233594,
      "learning_rate": 8.027836841735304e-08,
      "loss": 0.6157,
      "step": 21176
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.546115291842907,
      "learning_rate": 8.014949406978612e-08,
      "loss": 0.5847,
      "step": 21177
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.312430108215395,
      "learning_rate": 8.002072241332425e-08,
      "loss": 0.5361,
      "step": 21178
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.927394348321175,
      "learning_rate": 7.989205345065532e-08,
      "loss": 0.5488,
      "step": 21179
    },
    {
      "epoch": 1.89,
      "grad_norm": 8.012208391183124,
      "learning_rate": 7.976348718446492e-08,
      "loss": 0.5325,
      "step": 21180
    },
    {
      "epoch": 1.89,
      "grad_norm": 8.398578408294263,
      "learning_rate": 7.963502361743702e-08,
      "loss": 0.5452,
      "step": 21181
    },
    {
      "epoch": 1.89,
      "grad_norm": 8.192189895145455,
      "learning_rate": 7.950666275225227e-08,
      "loss": 0.5998,
      "step": 21182
    },
    {
      "epoch": 1.89,
      "grad_norm": 8.400783734311167,
      "learning_rate": 7.937840459159018e-08,
      "loss": 0.6337,
      "step": 21183
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.838336171114387,
      "learning_rate": 7.925024913812806e-08,
      "loss": 0.6085,
      "step": 21184
    },
    {
      "epoch": 1.89,
      "grad_norm": 8.19701411916909,
      "learning_rate": 7.912219639454044e-08,
      "loss": 0.6112,
      "step": 21185
    },
    {
      "epoch": 1.89,
      "grad_norm": 8.226439058555018,
      "learning_rate": 7.899424636350073e-08,
      "loss": 0.5772,
      "step": 21186
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.138423338776228,
      "learning_rate": 7.886639904767901e-08,
      "loss": 0.5434,
      "step": 21187
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.896775179310554,
      "learning_rate": 7.873865444974483e-08,
      "loss": 0.5994,
      "step": 21188
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.514032696115205,
      "learning_rate": 7.861101257236325e-08,
      "loss": 0.5089,
      "step": 21189
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.995714588032202,
      "learning_rate": 7.84834734181994e-08,
      "loss": 0.5645,
      "step": 21190
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.697463427384625,
      "learning_rate": 7.835603698991445e-08,
      "loss": 0.5965,
      "step": 21191
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.659054167553722,
      "learning_rate": 7.822870329016963e-08,
      "loss": 0.6033,
      "step": 21192
    },
    {
      "epoch": 1.89,
      "grad_norm": 5.943248928584518,
      "learning_rate": 7.810147232162225e-08,
      "loss": 0.5464,
      "step": 21193
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.247359601012381,
      "learning_rate": 7.797434408692739e-08,
      "loss": 0.5598,
      "step": 21194
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.997616606952509,
      "learning_rate": 7.784731858873906e-08,
      "loss": 0.5522,
      "step": 21195
    },
    {
      "epoch": 1.89,
      "grad_norm": 9.034705519507975,
      "learning_rate": 7.772039582970903e-08,
      "loss": 0.5776,
      "step": 21196
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.549796533642617,
      "learning_rate": 7.759357581248572e-08,
      "loss": 0.5341,
      "step": 21197
    },
    {
      "epoch": 1.89,
      "grad_norm": 5.796084991934007,
      "learning_rate": 7.746685853971702e-08,
      "loss": 0.6157,
      "step": 21198
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.386655552388123,
      "learning_rate": 7.73402440140475e-08,
      "loss": 0.5507,
      "step": 21199
    },
    {
      "epoch": 1.89,
      "grad_norm": 9.034859642829629,
      "learning_rate": 7.721373223812e-08,
      "loss": 0.5426,
      "step": 21200
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.850136365421014,
      "learning_rate": 7.708732321457469e-08,
      "loss": 0.5223,
      "step": 21201
    },
    {
      "epoch": 1.89,
      "grad_norm": 5.976969913020208,
      "learning_rate": 7.696101694605052e-08,
      "loss": 0.5359,
      "step": 21202
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.129082694731104,
      "learning_rate": 7.683481343518373e-08,
      "loss": 0.588,
      "step": 21203
    },
    {
      "epoch": 1.89,
      "grad_norm": 8.601553382291334,
      "learning_rate": 7.670871268460889e-08,
      "loss": 0.5444,
      "step": 21204
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.034120440568441,
      "learning_rate": 7.658271469695777e-08,
      "loss": 0.5219,
      "step": 21205
    },
    {
      "epoch": 1.89,
      "grad_norm": 5.619076675569958,
      "learning_rate": 7.645681947486106e-08,
      "loss": 0.51,
      "step": 21206
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.440947258866161,
      "learning_rate": 7.633102702094497e-08,
      "loss": 0.591,
      "step": 21207
    },
    {
      "epoch": 1.89,
      "grad_norm": 12.116005485657649,
      "learning_rate": 7.62053373378363e-08,
      "loss": 0.5893,
      "step": 21208
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.116217436243719,
      "learning_rate": 7.607975042815852e-08,
      "loss": 0.5742,
      "step": 21209
    },
    {
      "epoch": 1.89,
      "grad_norm": 5.953501520362834,
      "learning_rate": 7.595426629453229e-08,
      "loss": 0.5154,
      "step": 21210
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.214152565203953,
      "learning_rate": 7.582888493957774e-08,
      "loss": 0.5891,
      "step": 21211
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.502129342070258,
      "learning_rate": 7.570360636591167e-08,
      "loss": 0.5748,
      "step": 21212
    },
    {
      "epoch": 1.89,
      "grad_norm": 8.327396089627522,
      "learning_rate": 7.55784305761481e-08,
      "loss": 0.5789,
      "step": 21213
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.541653550181685,
      "learning_rate": 7.545335757290106e-08,
      "loss": 0.5964,
      "step": 21214
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.0797506364099165,
      "learning_rate": 7.532838735878067e-08,
      "loss": 0.5404,
      "step": 21215
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.908760230571993,
      "learning_rate": 7.520351993639485e-08,
      "loss": 0.5488,
      "step": 21216
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.8061141081547545,
      "learning_rate": 7.50787553083504e-08,
      "loss": 0.5748,
      "step": 21217
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.544581225643431,
      "learning_rate": 7.495409347725192e-08,
      "loss": 0.5468,
      "step": 21218
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.02726339707892,
      "learning_rate": 7.482953444570119e-08,
      "loss": 0.5672,
      "step": 21219
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.81385148460501,
      "learning_rate": 7.47050782162978e-08,
      "loss": 0.519,
      "step": 21220
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.043106575442452,
      "learning_rate": 7.458072479163914e-08,
      "loss": 0.565,
      "step": 21221
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.9407432509502245,
      "learning_rate": 7.4456474174322e-08,
      "loss": 0.5581,
      "step": 21222
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.557421435801596,
      "learning_rate": 7.433232636693877e-08,
      "loss": 0.5084,
      "step": 21223
    },
    {
      "epoch": 1.89,
      "grad_norm": 8.81752414440379,
      "learning_rate": 7.420828137208124e-08,
      "loss": 0.5961,
      "step": 21224
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.334190301771621,
      "learning_rate": 7.408433919233904e-08,
      "loss": 0.5896,
      "step": 21225
    },
    {
      "epoch": 1.89,
      "grad_norm": 9.1536672033219,
      "learning_rate": 7.39604998302984e-08,
      "loss": 0.5689,
      "step": 21226
    },
    {
      "epoch": 1.89,
      "grad_norm": 8.257805724980575,
      "learning_rate": 7.38367632885445e-08,
      "loss": 0.5815,
      "step": 21227
    },
    {
      "epoch": 1.89,
      "grad_norm": 8.388189220844321,
      "learning_rate": 7.371312956966025e-08,
      "loss": 0.5561,
      "step": 21228
    },
    {
      "epoch": 1.89,
      "grad_norm": 5.649777089405236,
      "learning_rate": 7.358959867622584e-08,
      "loss": 0.6399,
      "step": 21229
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.614265969846501,
      "learning_rate": 7.346617061081973e-08,
      "loss": 0.5408,
      "step": 21230
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.1860568325518415,
      "learning_rate": 7.334284537601877e-08,
      "loss": 0.5397,
      "step": 21231
    },
    {
      "epoch": 1.89,
      "grad_norm": 8.134669127648996,
      "learning_rate": 7.3219622974397e-08,
      "loss": 0.6287,
      "step": 21232
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.1714984310287395,
      "learning_rate": 7.309650340852515e-08,
      "loss": 0.5759,
      "step": 21233
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.6210544328143435,
      "learning_rate": 7.297348668097448e-08,
      "loss": 0.5626,
      "step": 21234
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.209082890118315,
      "learning_rate": 7.28505727943124e-08,
      "loss": 0.5376,
      "step": 21235
    },
    {
      "epoch": 1.89,
      "grad_norm": 5.9306570683673145,
      "learning_rate": 7.272776175110408e-08,
      "loss": 0.4996,
      "step": 21236
    },
    {
      "epoch": 1.89,
      "grad_norm": 5.755382082283328,
      "learning_rate": 7.260505355391356e-08,
      "loss": 0.5984,
      "step": 21237
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.157364927243349,
      "learning_rate": 7.248244820530159e-08,
      "loss": 0.5346,
      "step": 21238
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.819671326413819,
      "learning_rate": 7.235994570782723e-08,
      "loss": 0.5412,
      "step": 21239
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.313787517730441,
      "learning_rate": 7.223754606404788e-08,
      "loss": 0.5412,
      "step": 21240
    },
    {
      "epoch": 1.89,
      "grad_norm": 5.93781517738872,
      "learning_rate": 7.211524927651814e-08,
      "loss": 0.5198,
      "step": 21241
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.298157508584693,
      "learning_rate": 7.199305534779045e-08,
      "loss": 0.5681,
      "step": 21242
    },
    {
      "epoch": 1.9,
      "grad_norm": 8.712756795163743,
      "learning_rate": 7.187096428041606e-08,
      "loss": 0.5654,
      "step": 21243
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.33761487636059,
      "learning_rate": 7.174897607694242e-08,
      "loss": 0.5601,
      "step": 21244
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.338365844825216,
      "learning_rate": 7.162709073991691e-08,
      "loss": 0.6337,
      "step": 21245
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.874363440735824,
      "learning_rate": 7.150530827188196e-08,
      "loss": 0.5564,
      "step": 21246
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.406717761710562,
      "learning_rate": 7.138362867538107e-08,
      "loss": 0.6057,
      "step": 21247
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.950410814305446,
      "learning_rate": 7.126205195295332e-08,
      "loss": 0.5323,
      "step": 21248
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.897296804811561,
      "learning_rate": 7.114057810713615e-08,
      "loss": 0.5789,
      "step": 21249
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.987618005323567,
      "learning_rate": 7.101920714046583e-08,
      "loss": 0.5398,
      "step": 21250
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.7634494272940335,
      "learning_rate": 7.089793905547537e-08,
      "loss": 0.5898,
      "step": 21251
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.288002343068503,
      "learning_rate": 7.077677385469551e-08,
      "loss": 0.4917,
      "step": 21252
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.840368395463145,
      "learning_rate": 7.065571154065587e-08,
      "loss": 0.5097,
      "step": 21253
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.190726768128051,
      "learning_rate": 7.053475211588279e-08,
      "loss": 0.5112,
      "step": 21254
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.374897725960252,
      "learning_rate": 7.041389558290202e-08,
      "loss": 0.5601,
      "step": 21255
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.867212252830151,
      "learning_rate": 7.029314194423486e-08,
      "loss": 0.5702,
      "step": 21256
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.997787466702303,
      "learning_rate": 7.017249120240322e-08,
      "loss": 0.5939,
      "step": 21257
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.389942410504035,
      "learning_rate": 7.005194335992504e-08,
      "loss": 0.571,
      "step": 21258
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.626326598132809,
      "learning_rate": 6.993149841931502e-08,
      "loss": 0.4967,
      "step": 21259
    },
    {
      "epoch": 1.9,
      "grad_norm": 12.723864671590995,
      "learning_rate": 6.981115638308889e-08,
      "loss": 0.6317,
      "step": 21260
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.528606360518601,
      "learning_rate": 6.969091725375798e-08,
      "loss": 0.5446,
      "step": 21261
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.141327160649061,
      "learning_rate": 6.957078103383142e-08,
      "loss": 0.5505,
      "step": 21262
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.510898770733263,
      "learning_rate": 6.945074772581773e-08,
      "loss": 0.5857,
      "step": 21263
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.946808627047982,
      "learning_rate": 6.933081733222269e-08,
      "loss": 0.6042,
      "step": 21264
    },
    {
      "epoch": 1.9,
      "grad_norm": 9.044333302868512,
      "learning_rate": 6.921098985554764e-08,
      "loss": 0.5535,
      "step": 21265
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.021392487641217,
      "learning_rate": 6.909126529829557e-08,
      "loss": 0.598,
      "step": 21266
    },
    {
      "epoch": 1.9,
      "grad_norm": 8.12615345551302,
      "learning_rate": 6.897164366296449e-08,
      "loss": 0.564,
      "step": 21267
    },
    {
      "epoch": 1.9,
      "grad_norm": 8.15088501287265,
      "learning_rate": 6.885212495205185e-08,
      "loss": 0.535,
      "step": 21268
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.521672920163566,
      "learning_rate": 6.87327091680523e-08,
      "loss": 0.5609,
      "step": 21269
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.681788648935735,
      "learning_rate": 6.861339631345776e-08,
      "loss": 0.5505,
      "step": 21270
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.892789616363748,
      "learning_rate": 6.8494186390759e-08,
      "loss": 0.5983,
      "step": 21271
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.40536060634718,
      "learning_rate": 6.837507940244515e-08,
      "loss": 0.6254,
      "step": 21272
    },
    {
      "epoch": 1.9,
      "grad_norm": 10.862060460469634,
      "learning_rate": 6.825607535100032e-08,
      "loss": 0.5531,
      "step": 21273
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.702137927361141,
      "learning_rate": 6.813717423890975e-08,
      "loss": 0.6056,
      "step": 21274
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.981637180233195,
      "learning_rate": 6.801837606865535e-08,
      "loss": 0.5903,
      "step": 21275
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.484555391800655,
      "learning_rate": 6.789968084271681e-08,
      "loss": 0.5461,
      "step": 21276
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.20522418035437,
      "learning_rate": 6.778108856357102e-08,
      "loss": 0.6109,
      "step": 21277
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.465355543409743,
      "learning_rate": 6.766259923369323e-08,
      "loss": 0.5904,
      "step": 21278
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.154501705374097,
      "learning_rate": 6.754421285555757e-08,
      "loss": 0.5889,
      "step": 21279
    },
    {
      "epoch": 1.9,
      "grad_norm": 10.24658216082087,
      "learning_rate": 6.742592943163428e-08,
      "loss": 0.5551,
      "step": 21280
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.727183456471248,
      "learning_rate": 6.730774896439252e-08,
      "loss": 0.6214,
      "step": 21281
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.577313713431559,
      "learning_rate": 6.718967145629974e-08,
      "loss": 0.5864,
      "step": 21282
    },
    {
      "epoch": 1.9,
      "grad_norm": 8.035321119385008,
      "learning_rate": 6.707169690981896e-08,
      "loss": 0.6083,
      "step": 21283
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.934589613123476,
      "learning_rate": 6.695382532741379e-08,
      "loss": 0.5712,
      "step": 21284
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.836845737951196,
      "learning_rate": 6.683605671154448e-08,
      "loss": 0.5675,
      "step": 21285
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.3898090056858265,
      "learning_rate": 6.671839106466904e-08,
      "loss": 0.544,
      "step": 21286
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.189011827662973,
      "learning_rate": 6.660082838924387e-08,
      "loss": 0.5462,
      "step": 21287
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.452300240563997,
      "learning_rate": 6.648336868772254e-08,
      "loss": 0.573,
      "step": 21288
    },
    {
      "epoch": 1.9,
      "grad_norm": 9.657326938255489,
      "learning_rate": 6.636601196255644e-08,
      "loss": 0.5279,
      "step": 21289
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.514370848100157,
      "learning_rate": 6.62487582161947e-08,
      "loss": 0.5402,
      "step": 21290
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.419504622745159,
      "learning_rate": 6.613160745108593e-08,
      "loss": 0.6029,
      "step": 21291
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.623082611581208,
      "learning_rate": 6.601455966967485e-08,
      "loss": 0.6,
      "step": 21292
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.33329908400149,
      "learning_rate": 6.589761487440449e-08,
      "loss": 0.6153,
      "step": 21293
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.231758448682552,
      "learning_rate": 6.578077306771624e-08,
      "loss": 0.6008,
      "step": 21294
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.855029539282295,
      "learning_rate": 6.566403425204814e-08,
      "loss": 0.5209,
      "step": 21295
    },
    {
      "epoch": 1.9,
      "grad_norm": 8.949644031250369,
      "learning_rate": 6.554739842983771e-08,
      "loss": 0.5478,
      "step": 21296
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.958494660912915,
      "learning_rate": 6.543086560351907e-08,
      "loss": 0.5414,
      "step": 21297
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.785996567061647,
      "learning_rate": 6.53144357755242e-08,
      "loss": 0.5419,
      "step": 21298
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.015172539154094,
      "learning_rate": 6.519810894828394e-08,
      "loss": 0.5188,
      "step": 21299
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.565896080790453,
      "learning_rate": 6.508188512422575e-08,
      "loss": 0.583,
      "step": 21300
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.802204822149363,
      "learning_rate": 6.496576430577661e-08,
      "loss": 0.5773,
      "step": 21301
    },
    {
      "epoch": 1.9,
      "grad_norm": 8.462789919547253,
      "learning_rate": 6.484974649535958e-08,
      "loss": 0.6077,
      "step": 21302
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.99504791263156,
      "learning_rate": 6.473383169539605e-08,
      "loss": 0.5829,
      "step": 21303
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.584529988402077,
      "learning_rate": 6.461801990830519e-08,
      "loss": 0.5553,
      "step": 21304
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.389552597786946,
      "learning_rate": 6.450231113650563e-08,
      "loss": 0.6531,
      "step": 21305
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.45104601379624,
      "learning_rate": 6.438670538241099e-08,
      "loss": 0.5332,
      "step": 21306
    },
    {
      "epoch": 1.9,
      "grad_norm": 8.589456007267785,
      "learning_rate": 6.4271202648436e-08,
      "loss": 0.5312,
      "step": 21307
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.022145390691316,
      "learning_rate": 6.415580293699042e-08,
      "loss": 0.5696,
      "step": 21308
    },
    {
      "epoch": 1.9,
      "grad_norm": 8.873129385294643,
      "learning_rate": 6.40405062504823e-08,
      "loss": 0.5769,
      "step": 21309
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.074253775255684,
      "learning_rate": 6.392531259131973e-08,
      "loss": 0.5511,
      "step": 21310
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.067451883498281,
      "learning_rate": 6.381022196190633e-08,
      "loss": 0.5067,
      "step": 21311
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.430163554536063,
      "learning_rate": 6.369523436464409e-08,
      "loss": 0.5934,
      "step": 21312
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.872598635485549,
      "learning_rate": 6.358034980193384e-08,
      "loss": 0.6076,
      "step": 21313
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.607174660533924,
      "learning_rate": 6.346556827617368e-08,
      "loss": 0.5321,
      "step": 21314
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.130659308917549,
      "learning_rate": 6.33508897897589e-08,
      "loss": 0.5716,
      "step": 21315
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.712218575965231,
      "learning_rate": 6.32363143450826e-08,
      "loss": 0.5608,
      "step": 21316
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.873596512060934,
      "learning_rate": 6.31218419445373e-08,
      "loss": 0.6169,
      "step": 21317
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.664751567692232,
      "learning_rate": 6.300747259051165e-08,
      "loss": 0.6056,
      "step": 21318
    },
    {
      "epoch": 1.9,
      "grad_norm": 8.189202080767128,
      "learning_rate": 6.289320628539319e-08,
      "loss": 0.5705,
      "step": 21319
    },
    {
      "epoch": 1.9,
      "grad_norm": 4.8799085852024175,
      "learning_rate": 6.277904303156723e-08,
      "loss": 0.5296,
      "step": 21320
    },
    {
      "epoch": 1.9,
      "grad_norm": 8.023611975770347,
      "learning_rate": 6.266498283141686e-08,
      "loss": 0.5518,
      "step": 21321
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.196427945571334,
      "learning_rate": 6.25510256873213e-08,
      "loss": 0.5439,
      "step": 21322
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.214545165534594,
      "learning_rate": 6.243717160166085e-08,
      "loss": 0.5455,
      "step": 21323
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.317314380266117,
      "learning_rate": 6.23234205768114e-08,
      "loss": 0.5343,
      "step": 21324
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.821251853454146,
      "learning_rate": 6.220977261514716e-08,
      "loss": 0.5453,
      "step": 21325
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.681031012561174,
      "learning_rate": 6.209622771904012e-08,
      "loss": 0.5417,
      "step": 21326
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.002304885837011,
      "learning_rate": 6.198278589086115e-08,
      "loss": 0.4771,
      "step": 21327
    },
    {
      "epoch": 1.9,
      "grad_norm": 9.267428650193255,
      "learning_rate": 6.186944713297671e-08,
      "loss": 0.5496,
      "step": 21328
    },
    {
      "epoch": 1.9,
      "grad_norm": 4.397417725045336,
      "learning_rate": 6.175621144775379e-08,
      "loss": 0.526,
      "step": 21329
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.116603999827927,
      "learning_rate": 6.164307883755494e-08,
      "loss": 0.5945,
      "step": 21330
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.144813342570023,
      "learning_rate": 6.153004930474215e-08,
      "loss": 0.5785,
      "step": 21331
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.171124549102536,
      "learning_rate": 6.141712285167467e-08,
      "loss": 0.6322,
      "step": 21332
    },
    {
      "epoch": 1.9,
      "grad_norm": 9.272236093636987,
      "learning_rate": 6.130429948070948e-08,
      "loss": 0.6029,
      "step": 21333
    },
    {
      "epoch": 1.9,
      "grad_norm": 9.060489877448727,
      "learning_rate": 6.119157919420138e-08,
      "loss": 0.5628,
      "step": 21334
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.744579559068785,
      "learning_rate": 6.107896199450291e-08,
      "loss": 0.5089,
      "step": 21335
    },
    {
      "epoch": 1.9,
      "grad_norm": 8.492473072669513,
      "learning_rate": 6.096644788396555e-08,
      "loss": 0.5773,
      "step": 21336
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.80379316995366,
      "learning_rate": 6.085403686493685e-08,
      "loss": 0.5434,
      "step": 21337
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.357029891972265,
      "learning_rate": 6.074172893976382e-08,
      "loss": 0.5114,
      "step": 21338
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.467090747574635,
      "learning_rate": 6.062952411079015e-08,
      "loss": 0.5694,
      "step": 21339
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.082034016544279,
      "learning_rate": 6.051742238035896e-08,
      "loss": 0.5087,
      "step": 21340
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.679360045424672,
      "learning_rate": 6.040542375080838e-08,
      "loss": 0.5511,
      "step": 21341
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.982653620090274,
      "learning_rate": 6.029352822447765e-08,
      "loss": 0.527,
      "step": 21342
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.502862500067863,
      "learning_rate": 6.018173580370157e-08,
      "loss": 0.6024,
      "step": 21343
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.267060355983231,
      "learning_rate": 6.007004649081383e-08,
      "loss": 0.5524,
      "step": 21344
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.9607593961245575,
      "learning_rate": 5.995846028814534e-08,
      "loss": 0.5205,
      "step": 21345
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.427406781860701,
      "learning_rate": 5.984697719802535e-08,
      "loss": 0.5312,
      "step": 21346
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.696278884635256,
      "learning_rate": 5.973559722278144e-08,
      "loss": 0.5739,
      "step": 21347
    },
    {
      "epoch": 1.9,
      "grad_norm": 4.918737121566061,
      "learning_rate": 5.962432036473787e-08,
      "loss": 0.5852,
      "step": 21348
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.569367538704405,
      "learning_rate": 5.9513146626217214e-08,
      "loss": 0.6018,
      "step": 21349
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.453765576738879,
      "learning_rate": 5.9402076009539843e-08,
      "loss": 0.5531,
      "step": 21350
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.2035410403665185,
      "learning_rate": 5.9291108517024463e-08,
      "loss": 0.5304,
      "step": 21351
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.848260871027334,
      "learning_rate": 5.9180244150988106e-08,
      "loss": 0.6166,
      "step": 21352
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.016532055159449,
      "learning_rate": 5.906948291374281e-08,
      "loss": 0.5567,
      "step": 21353
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.804231185301059,
      "learning_rate": 5.895882480760229e-08,
      "loss": 0.5558,
      "step": 21354
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.336289411796137,
      "learning_rate": 5.884826983487524e-08,
      "loss": 0.53,
      "step": 21355
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.783032466300503,
      "learning_rate": 5.873781799786982e-08,
      "loss": 0.544,
      "step": 21356
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.89802165450593,
      "learning_rate": 5.862746929889085e-08,
      "loss": 0.5845,
      "step": 21357
    },
    {
      "epoch": 1.91,
      "grad_norm": 7.594493626911466,
      "learning_rate": 5.851722374024316e-08,
      "loss": 0.5874,
      "step": 21358
    },
    {
      "epoch": 1.91,
      "grad_norm": 7.59827004375222,
      "learning_rate": 5.8407081324225456e-08,
      "loss": 0.5716,
      "step": 21359
    },
    {
      "epoch": 1.91,
      "grad_norm": 7.992995498437647,
      "learning_rate": 5.829704205313924e-08,
      "loss": 0.5296,
      "step": 21360
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.046078495054364,
      "learning_rate": 5.8187105929279344e-08,
      "loss": 0.5806,
      "step": 21361
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.069622644270423,
      "learning_rate": 5.8077272954941145e-08,
      "loss": 0.5795,
      "step": 21362
    },
    {
      "epoch": 1.91,
      "grad_norm": 8.989380039572852,
      "learning_rate": 5.796754313241726e-08,
      "loss": 0.6492,
      "step": 21363
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.837520267272899,
      "learning_rate": 5.7857916463998634e-08,
      "loss": 0.5435,
      "step": 21364
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.557491161467952,
      "learning_rate": 5.7748392951972324e-08,
      "loss": 0.5344,
      "step": 21365
    },
    {
      "epoch": 1.91,
      "grad_norm": 9.805066633644588,
      "learning_rate": 5.763897259862539e-08,
      "loss": 0.5481,
      "step": 21366
    },
    {
      "epoch": 1.91,
      "grad_norm": 8.262325253351532,
      "learning_rate": 5.7529655406241004e-08,
      "loss": 0.5781,
      "step": 21367
    },
    {
      "epoch": 1.91,
      "grad_norm": 7.249298985547511,
      "learning_rate": 5.742044137710123e-08,
      "loss": 0.6067,
      "step": 21368
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.597185459926209,
      "learning_rate": 5.7311330513485364e-08,
      "loss": 0.6243,
      "step": 21369
    },
    {
      "epoch": 1.91,
      "grad_norm": 7.3657708157723105,
      "learning_rate": 5.720232281767157e-08,
      "loss": 0.537,
      "step": 21370
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.815613225613898,
      "learning_rate": 5.709341829193527e-08,
      "loss": 0.5288,
      "step": 21371
    },
    {
      "epoch": 1.91,
      "grad_norm": 5.721985573575241,
      "learning_rate": 5.698461693854851e-08,
      "loss": 0.546,
      "step": 21372
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.181521317986416,
      "learning_rate": 5.687591875978282e-08,
      "loss": 0.561,
      "step": 21373
    },
    {
      "epoch": 1.91,
      "grad_norm": 5.517124678400668,
      "learning_rate": 5.67673237579075e-08,
      "loss": 0.6032,
      "step": 21374
    },
    {
      "epoch": 1.91,
      "grad_norm": 7.12765622400804,
      "learning_rate": 5.6658831935187955e-08,
      "loss": 0.5604,
      "step": 21375
    },
    {
      "epoch": 1.91,
      "grad_norm": 9.256872768865327,
      "learning_rate": 5.6550443293890144e-08,
      "loss": 0.6227,
      "step": 21376
    },
    {
      "epoch": 1.91,
      "grad_norm": 8.220233972439093,
      "learning_rate": 5.64421578362756e-08,
      "loss": 0.6111,
      "step": 21377
    },
    {
      "epoch": 1.91,
      "grad_norm": 8.352521048898353,
      "learning_rate": 5.633397556460529e-08,
      "loss": 0.5812,
      "step": 21378
    },
    {
      "epoch": 1.91,
      "grad_norm": 8.163902218221494,
      "learning_rate": 5.62258964811363e-08,
      "loss": 0.5029,
      "step": 21379
    },
    {
      "epoch": 1.91,
      "grad_norm": 5.96831042558661,
      "learning_rate": 5.611792058812515e-08,
      "loss": 0.5482,
      "step": 21380
    },
    {
      "epoch": 1.91,
      "grad_norm": 8.710456687774462,
      "learning_rate": 5.6010047887825606e-08,
      "loss": 0.4915,
      "step": 21381
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.1750553161035535,
      "learning_rate": 5.5902278382489185e-08,
      "loss": 0.5619,
      "step": 21382
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.994320616012002,
      "learning_rate": 5.5794612074365204e-08,
      "loss": 0.5382,
      "step": 21383
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.680109490805696,
      "learning_rate": 5.568704896570132e-08,
      "loss": 0.544,
      "step": 21384
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.9411810572235675,
      "learning_rate": 5.557958905874183e-08,
      "loss": 0.6136,
      "step": 21385
    },
    {
      "epoch": 1.91,
      "grad_norm": 5.951999430672837,
      "learning_rate": 5.547223235573107e-08,
      "loss": 0.5463,
      "step": 21386
    },
    {
      "epoch": 1.91,
      "grad_norm": 5.5146187773066995,
      "learning_rate": 5.5364978858908904e-08,
      "loss": 0.5515,
      "step": 21387
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.735158946381258,
      "learning_rate": 5.5257828570514094e-08,
      "loss": 0.548,
      "step": 21388
    },
    {
      "epoch": 1.91,
      "grad_norm": 8.470210319870674,
      "learning_rate": 5.515078149278319e-08,
      "loss": 0.5784,
      "step": 21389
    },
    {
      "epoch": 1.91,
      "grad_norm": 7.691901289769762,
      "learning_rate": 5.504383762795051e-08,
      "loss": 0.549,
      "step": 21390
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.505807358698942,
      "learning_rate": 5.493699697824928e-08,
      "loss": 0.5669,
      "step": 21391
    },
    {
      "epoch": 1.91,
      "grad_norm": 8.512321133325601,
      "learning_rate": 5.483025954590771e-08,
      "loss": 0.6421,
      "step": 21392
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.528042288440155,
      "learning_rate": 5.4723625333155115e-08,
      "loss": 0.5607,
      "step": 21393
    },
    {
      "epoch": 1.91,
      "grad_norm": 8.442187578827639,
      "learning_rate": 5.4617094342216956e-08,
      "loss": 0.5517,
      "step": 21394
    },
    {
      "epoch": 1.91,
      "grad_norm": 9.040884402851328,
      "learning_rate": 5.4510666575317004e-08,
      "loss": 0.561,
      "step": 21395
    },
    {
      "epoch": 1.91,
      "grad_norm": 8.616599415130706,
      "learning_rate": 5.4404342034676263e-08,
      "loss": 0.5776,
      "step": 21396
    },
    {
      "epoch": 1.91,
      "grad_norm": 8.095612286648642,
      "learning_rate": 5.429812072251406e-08,
      "loss": 0.5204,
      "step": 21397
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.4600437453868755,
      "learning_rate": 5.419200264104807e-08,
      "loss": 0.4865,
      "step": 21398
    },
    {
      "epoch": 1.91,
      "grad_norm": 5.6556540501034425,
      "learning_rate": 5.408598779249263e-08,
      "loss": 0.5782,
      "step": 21399
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.142954764627255,
      "learning_rate": 5.3980076179060424e-08,
      "loss": 0.5729,
      "step": 21400
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.205970862928872,
      "learning_rate": 5.3874267802963006e-08,
      "loss": 0.6124,
      "step": 21401
    },
    {
      "epoch": 1.91,
      "grad_norm": 8.409992918834801,
      "learning_rate": 5.376856266640862e-08,
      "loss": 0.6009,
      "step": 21402
    },
    {
      "epoch": 1.91,
      "grad_norm": 5.8881805684984,
      "learning_rate": 5.3662960771602714e-08,
      "loss": 0.5127,
      "step": 21403
    },
    {
      "epoch": 1.91,
      "grad_norm": 10.258702135296515,
      "learning_rate": 5.35574621207513e-08,
      "loss": 0.5414,
      "step": 21404
    },
    {
      "epoch": 1.91,
      "grad_norm": 8.739939932591984,
      "learning_rate": 5.34520667160543e-08,
      "loss": 0.602,
      "step": 21405
    },
    {
      "epoch": 1.91,
      "grad_norm": 8.22835968568237,
      "learning_rate": 5.3346774559713266e-08,
      "loss": 0.498,
      "step": 21406
    },
    {
      "epoch": 1.91,
      "grad_norm": 8.653556087134426,
      "learning_rate": 5.3241585653924786e-08,
      "loss": 0.532,
      "step": 21407
    },
    {
      "epoch": 1.91,
      "grad_norm": 5.551220000413304,
      "learning_rate": 5.313650000088544e-08,
      "loss": 0.6128,
      "step": 21408
    },
    {
      "epoch": 1.91,
      "grad_norm": 7.102984350773631,
      "learning_rate": 5.303151760278791e-08,
      "loss": 0.5175,
      "step": 21409
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.894878455958238,
      "learning_rate": 5.292663846182433e-08,
      "loss": 0.4853,
      "step": 21410
    },
    {
      "epoch": 1.91,
      "grad_norm": 5.027828369218479,
      "learning_rate": 5.28218625801824e-08,
      "loss": 0.5399,
      "step": 21411
    },
    {
      "epoch": 1.91,
      "grad_norm": 8.141860519107901,
      "learning_rate": 5.2717189960050375e-08,
      "loss": 0.5814,
      "step": 21412
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.236755803706455,
      "learning_rate": 5.2612620603612606e-08,
      "loss": 0.5286,
      "step": 21413
    },
    {
      "epoch": 1.91,
      "grad_norm": 7.149004708070925,
      "learning_rate": 5.25081545130518e-08,
      "loss": 0.5422,
      "step": 21414
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.238914681953426,
      "learning_rate": 5.2403791690548435e-08,
      "loss": 0.5371,
      "step": 21415
    },
    {
      "epoch": 1.91,
      "grad_norm": 7.531508990558202,
      "learning_rate": 5.229953213828132e-08,
      "loss": 0.5872,
      "step": 21416
    },
    {
      "epoch": 1.91,
      "grad_norm": 5.395012435633081,
      "learning_rate": 5.219537585842593e-08,
      "loss": 0.576,
      "step": 21417
    },
    {
      "epoch": 1.91,
      "grad_norm": 9.213480699418618,
      "learning_rate": 5.2091322853156654e-08,
      "loss": 0.5054,
      "step": 21418
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.7647381904990995,
      "learning_rate": 5.198737312464508e-08,
      "loss": 0.5128,
      "step": 21419
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.591314604813206,
      "learning_rate": 5.1883526675061136e-08,
      "loss": 0.5732,
      "step": 21420
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.031386774914491,
      "learning_rate": 5.1779783506571976e-08,
      "loss": 0.5158,
      "step": 21421
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.061250427727376,
      "learning_rate": 5.1676143621343654e-08,
      "loss": 0.5844,
      "step": 21422
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.692733908468737,
      "learning_rate": 5.157260702153999e-08,
      "loss": 0.6054,
      "step": 21423
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.44427224083828,
      "learning_rate": 5.146917370932037e-08,
      "loss": 0.5067,
      "step": 21424
    },
    {
      "epoch": 1.91,
      "grad_norm": 8.371100207749128,
      "learning_rate": 5.136584368684472e-08,
      "loss": 0.5893,
      "step": 21425
    },
    {
      "epoch": 1.91,
      "grad_norm": 7.057685765642618,
      "learning_rate": 5.126261695627022e-08,
      "loss": 0.5507,
      "step": 21426
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.950222248270137,
      "learning_rate": 5.115949351975069e-08,
      "loss": 0.5393,
      "step": 21427
    },
    {
      "epoch": 1.91,
      "grad_norm": 7.278229544900507,
      "learning_rate": 5.1056473379438864e-08,
      "loss": 0.563,
      "step": 21428
    },
    {
      "epoch": 1.91,
      "grad_norm": 5.037304083051875,
      "learning_rate": 5.095355653748524e-08,
      "loss": 0.5067,
      "step": 21429
    },
    {
      "epoch": 1.91,
      "grad_norm": 8.682240359100392,
      "learning_rate": 5.0850742996038093e-08,
      "loss": 0.5439,
      "step": 21430
    },
    {
      "epoch": 1.91,
      "grad_norm": 7.152042617980153,
      "learning_rate": 5.0748032757243493e-08,
      "loss": 0.5783,
      "step": 21431
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.891760321447729,
      "learning_rate": 5.064542582324472e-08,
      "loss": 0.5748,
      "step": 21432
    },
    {
      "epoch": 1.91,
      "grad_norm": 9.286852631776844,
      "learning_rate": 5.05429221961834e-08,
      "loss": 0.5606,
      "step": 21433
    },
    {
      "epoch": 1.91,
      "grad_norm": 11.047150444275132,
      "learning_rate": 5.044052187820003e-08,
      "loss": 0.5432,
      "step": 21434
    },
    {
      "epoch": 1.91,
      "grad_norm": 7.996704218233253,
      "learning_rate": 5.0338224871430695e-08,
      "loss": 0.5333,
      "step": 21435
    },
    {
      "epoch": 1.91,
      "grad_norm": 5.418592172421419,
      "learning_rate": 5.0236031178012015e-08,
      "loss": 0.5729,
      "step": 21436
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.14752762352502,
      "learning_rate": 5.0133940800076165e-08,
      "loss": 0.5165,
      "step": 21437
    },
    {
      "epoch": 1.91,
      "grad_norm": 5.516788036842529,
      "learning_rate": 5.003195373975422e-08,
      "loss": 0.5083,
      "step": 21438
    },
    {
      "epoch": 1.91,
      "grad_norm": 5.9269353088502745,
      "learning_rate": 4.993006999917504e-08,
      "loss": 0.5784,
      "step": 21439
    },
    {
      "epoch": 1.91,
      "grad_norm": 8.591672249272733,
      "learning_rate": 4.9828289580465264e-08,
      "loss": 0.5623,
      "step": 21440
    },
    {
      "epoch": 1.91,
      "grad_norm": 10.18279867544072,
      "learning_rate": 4.972661248574928e-08,
      "loss": 0.5946,
      "step": 21441
    },
    {
      "epoch": 1.91,
      "grad_norm": 10.50019249855371,
      "learning_rate": 4.96250387171493e-08,
      "loss": 0.6232,
      "step": 21442
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.949859513460178,
      "learning_rate": 4.952356827678528e-08,
      "loss": 0.5296,
      "step": 21443
    },
    {
      "epoch": 1.91,
      "grad_norm": 8.426314693945141,
      "learning_rate": 4.942220116677554e-08,
      "loss": 0.5627,
      "step": 21444
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.861115161514378,
      "learning_rate": 4.9320937389236155e-08,
      "loss": 0.6362,
      "step": 21445
    },
    {
      "epoch": 1.91,
      "grad_norm": 7.10884018689446,
      "learning_rate": 4.921977694628044e-08,
      "loss": 0.5119,
      "step": 21446
    },
    {
      "epoch": 1.91,
      "grad_norm": 7.014274833860725,
      "learning_rate": 4.911871984001892e-08,
      "loss": 0.5639,
      "step": 21447
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.464375744355344,
      "learning_rate": 4.90177660725627e-08,
      "loss": 0.6046,
      "step": 21448
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.351580761099356,
      "learning_rate": 4.891691564601786e-08,
      "loss": 0.6123,
      "step": 21449
    },
    {
      "epoch": 1.91,
      "grad_norm": 11.42200622993026,
      "learning_rate": 4.881616856248994e-08,
      "loss": 0.4941,
      "step": 21450
    },
    {
      "epoch": 1.91,
      "grad_norm": 9.427342972913612,
      "learning_rate": 4.871552482408115e-08,
      "loss": 0.6114,
      "step": 21451
    },
    {
      "epoch": 1.91,
      "grad_norm": 5.058803591463046,
      "learning_rate": 4.861498443289314e-08,
      "loss": 0.5302,
      "step": 21452
    },
    {
      "epoch": 1.91,
      "grad_norm": 5.758235382033623,
      "learning_rate": 4.851454739102368e-08,
      "loss": 0.5724,
      "step": 21453
    },
    {
      "epoch": 1.91,
      "grad_norm": 8.381568892531039,
      "learning_rate": 4.8414213700569977e-08,
      "loss": 0.5786,
      "step": 21454
    },
    {
      "epoch": 1.91,
      "grad_norm": 9.504062094259034,
      "learning_rate": 4.831398336362481e-08,
      "loss": 0.5523,
      "step": 21455
    },
    {
      "epoch": 1.91,
      "grad_norm": 7.72514166448945,
      "learning_rate": 4.821385638228204e-08,
      "loss": 0.5077,
      "step": 21456
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.024034845244637,
      "learning_rate": 4.8113832758630576e-08,
      "loss": 0.579,
      "step": 21457
    },
    {
      "epoch": 1.91,
      "grad_norm": 8.77898673563874,
      "learning_rate": 4.801391249475817e-08,
      "loss": 0.5078,
      "step": 21458
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.703302777873915,
      "learning_rate": 4.7914095592750396e-08,
      "loss": 0.5673,
      "step": 21459
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.921905457940107,
      "learning_rate": 4.7814382054692246e-08,
      "loss": 0.6212,
      "step": 21460
    },
    {
      "epoch": 1.91,
      "grad_norm": 5.43475290795116,
      "learning_rate": 4.771477188266316e-08,
      "loss": 0.5468,
      "step": 21461
    },
    {
      "epoch": 1.91,
      "grad_norm": 8.91399680225555,
      "learning_rate": 4.7615265078742587e-08,
      "loss": 0.5989,
      "step": 21462
    },
    {
      "epoch": 1.91,
      "grad_norm": 7.080605791604,
      "learning_rate": 4.7515861645007766e-08,
      "loss": 0.5705,
      "step": 21463
    },
    {
      "epoch": 1.91,
      "grad_norm": 7.704995036258877,
      "learning_rate": 4.741656158353369e-08,
      "loss": 0.5679,
      "step": 21464
    },
    {
      "epoch": 1.91,
      "grad_norm": 7.9916507776348835,
      "learning_rate": 4.731736489639316e-08,
      "loss": 0.5662,
      "step": 21465
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.031492703454426,
      "learning_rate": 4.721827158565617e-08,
      "loss": 0.5438,
      "step": 21466
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.025983980676027,
      "learning_rate": 4.7119281653391634e-08,
      "loss": 0.5511,
      "step": 21467
    },
    {
      "epoch": 1.92,
      "grad_norm": 7.345638638719595,
      "learning_rate": 4.702039510166512e-08,
      "loss": 0.6195,
      "step": 21468
    },
    {
      "epoch": 1.92,
      "grad_norm": 8.759056928000941,
      "learning_rate": 4.6921611932541075e-08,
      "loss": 0.5354,
      "step": 21469
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.98172661570194,
      "learning_rate": 4.6822932148081205e-08,
      "loss": 0.6002,
      "step": 21470
    },
    {
      "epoch": 1.92,
      "grad_norm": 8.432431445860681,
      "learning_rate": 4.6724355750345506e-08,
      "loss": 0.5951,
      "step": 21471
    },
    {
      "epoch": 1.92,
      "grad_norm": 8.720835186220297,
      "learning_rate": 4.662588274139179e-08,
      "loss": 0.6468,
      "step": 21472
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.678829404336922,
      "learning_rate": 4.6527513123274527e-08,
      "loss": 0.5411,
      "step": 21473
    },
    {
      "epoch": 1.92,
      "grad_norm": 7.44426436200547,
      "learning_rate": 4.642924689804818e-08,
      "loss": 0.571,
      "step": 21474
    },
    {
      "epoch": 1.92,
      "grad_norm": 9.04607724235734,
      "learning_rate": 4.633108406776277e-08,
      "loss": 0.5613,
      "step": 21475
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.476541718782039,
      "learning_rate": 4.623302463446722e-08,
      "loss": 0.5785,
      "step": 21476
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.195136852920775,
      "learning_rate": 4.613506860020933e-08,
      "loss": 0.5323,
      "step": 21477
    },
    {
      "epoch": 1.92,
      "grad_norm": 7.880610263491197,
      "learning_rate": 4.603721596703248e-08,
      "loss": 0.5586,
      "step": 21478
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.660385437303607,
      "learning_rate": 4.593946673698002e-08,
      "loss": 0.5793,
      "step": 21479
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.778083395066137,
      "learning_rate": 4.584182091209255e-08,
      "loss": 0.5499,
      "step": 21480
    },
    {
      "epoch": 1.92,
      "grad_norm": 8.44357857163459,
      "learning_rate": 4.5744278494407326e-08,
      "loss": 0.5226,
      "step": 21481
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.465154941421891,
      "learning_rate": 4.5646839485960494e-08,
      "loss": 0.6021,
      "step": 21482
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.999017891262438,
      "learning_rate": 4.5549503888785985e-08,
      "loss": 0.5881,
      "step": 21483
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.853043337448896,
      "learning_rate": 4.545227170491551e-08,
      "loss": 0.528,
      "step": 21484
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.550572688197819,
      "learning_rate": 4.535514293637855e-08,
      "loss": 0.5455,
      "step": 21485
    },
    {
      "epoch": 1.92,
      "grad_norm": 7.326586952613605,
      "learning_rate": 4.5258117585202935e-08,
      "loss": 0.5466,
      "step": 21486
    },
    {
      "epoch": 1.92,
      "grad_norm": 8.660951757315528,
      "learning_rate": 4.5161195653412593e-08,
      "loss": 0.5245,
      "step": 21487
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.254171752919263,
      "learning_rate": 4.506437714303258e-08,
      "loss": 0.5583,
      "step": 21488
    },
    {
      "epoch": 1.92,
      "grad_norm": 7.425863039150528,
      "learning_rate": 4.496766205608127e-08,
      "loss": 0.5573,
      "step": 21489
    },
    {
      "epoch": 1.92,
      "grad_norm": 8.550187285960789,
      "learning_rate": 4.487105039457929e-08,
      "loss": 0.5926,
      "step": 21490
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.724978060580899,
      "learning_rate": 4.4774542160542775e-08,
      "loss": 0.5758,
      "step": 21491
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.07922892125418,
      "learning_rate": 4.46781373559857e-08,
      "loss": 0.5677,
      "step": 21492
    },
    {
      "epoch": 1.92,
      "grad_norm": 9.307440651001514,
      "learning_rate": 4.4581835982920873e-08,
      "loss": 0.528,
      "step": 21493
    },
    {
      "epoch": 1.92,
      "grad_norm": 7.944192586435162,
      "learning_rate": 4.4485638043357257e-08,
      "loss": 0.573,
      "step": 21494
    },
    {
      "epoch": 1.92,
      "grad_norm": 7.798318080138457,
      "learning_rate": 4.438954353930436e-08,
      "loss": 0.5328,
      "step": 21495
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.959631056083193,
      "learning_rate": 4.429355247276668e-08,
      "loss": 0.5894,
      "step": 21496
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.664137042958867,
      "learning_rate": 4.4197664845748166e-08,
      "loss": 0.6024,
      "step": 21497
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.828269261544087,
      "learning_rate": 4.4101880660250564e-08,
      "loss": 0.5818,
      "step": 21498
    },
    {
      "epoch": 1.92,
      "grad_norm": 7.241043594227954,
      "learning_rate": 4.400619991827337e-08,
      "loss": 0.5302,
      "step": 21499
    },
    {
      "epoch": 1.92,
      "grad_norm": 8.125741292204633,
      "learning_rate": 4.391062262181278e-08,
      "loss": 0.5744,
      "step": 21500
    },
    {
      "epoch": 1.92,
      "grad_norm": 7.48875296169228,
      "learning_rate": 4.381514877286441e-08,
      "loss": 0.5362,
      "step": 21501
    },
    {
      "epoch": 1.92,
      "grad_norm": 8.755288449846281,
      "learning_rate": 4.371977837342112e-08,
      "loss": 0.5725,
      "step": 21502
    },
    {
      "epoch": 1.92,
      "grad_norm": 8.324566394596554,
      "learning_rate": 4.3624511425473525e-08,
      "loss": 0.5692,
      "step": 21503
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.059433196692797,
      "learning_rate": 4.352934793100949e-08,
      "loss": 0.5397,
      "step": 21504
    },
    {
      "epoch": 1.92,
      "grad_norm": 10.370703237258407,
      "learning_rate": 4.343428789201631e-08,
      "loss": 0.5871,
      "step": 21505
    },
    {
      "epoch": 1.92,
      "grad_norm": 8.297872022742396,
      "learning_rate": 4.3339331310477404e-08,
      "loss": 0.5656,
      "step": 21506
    },
    {
      "epoch": 1.92,
      "grad_norm": 8.052184348689103,
      "learning_rate": 4.324447818837507e-08,
      "loss": 0.5831,
      "step": 21507
    },
    {
      "epoch": 1.92,
      "grad_norm": 8.171907712791318,
      "learning_rate": 4.3149728527689395e-08,
      "loss": 0.6148,
      "step": 21508
    },
    {
      "epoch": 1.92,
      "grad_norm": 7.850065732979883,
      "learning_rate": 4.305508233039768e-08,
      "loss": 0.5781,
      "step": 21509
    },
    {
      "epoch": 1.92,
      "grad_norm": 8.969325801468154,
      "learning_rate": 4.296053959847501e-08,
      "loss": 0.5648,
      "step": 21510
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.847464223227468,
      "learning_rate": 4.2866100333895935e-08,
      "loss": 0.5812,
      "step": 21511
    },
    {
      "epoch": 1.92,
      "grad_norm": 8.924605067062105,
      "learning_rate": 4.2771764538631075e-08,
      "loss": 0.6016,
      "step": 21512
    },
    {
      "epoch": 1.92,
      "grad_norm": 7.880240859062239,
      "learning_rate": 4.267753221464943e-08,
      "loss": 0.5827,
      "step": 21513
    },
    {
      "epoch": 1.92,
      "grad_norm": 7.410131362128329,
      "learning_rate": 4.258340336391775e-08,
      "loss": 0.5387,
      "step": 21514
    },
    {
      "epoch": 1.92,
      "grad_norm": 10.087773348752298,
      "learning_rate": 4.248937798840114e-08,
      "loss": 0.515,
      "step": 21515
    },
    {
      "epoch": 1.92,
      "grad_norm": 7.363472980203379,
      "learning_rate": 4.2395456090061906e-08,
      "loss": 0.6186,
      "step": 21516
    },
    {
      "epoch": 1.92,
      "grad_norm": 8.432445162405642,
      "learning_rate": 4.230163767086126e-08,
      "loss": 0.5803,
      "step": 21517
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.550571310040353,
      "learning_rate": 4.220792273275598e-08,
      "loss": 0.5414,
      "step": 21518
    },
    {
      "epoch": 1.92,
      "grad_norm": 8.528200996264104,
      "learning_rate": 4.211431127770338e-08,
      "loss": 0.5896,
      "step": 21519
    },
    {
      "epoch": 1.92,
      "grad_norm": 7.298528099268789,
      "learning_rate": 4.2020803307656345e-08,
      "loss": 0.5954,
      "step": 21520
    },
    {
      "epoch": 1.92,
      "grad_norm": 7.08489776765719,
      "learning_rate": 4.192739882456776e-08,
      "loss": 0.5333,
      "step": 21521
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.126554713572835,
      "learning_rate": 4.183409783038661e-08,
      "loss": 0.6367,
      "step": 21522
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.872196860474327,
      "learning_rate": 4.1740900327060244e-08,
      "loss": 0.6218,
      "step": 21523
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.823065036902385,
      "learning_rate": 4.1647806316534314e-08,
      "loss": 0.5782,
      "step": 21524
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.570855585598809,
      "learning_rate": 4.1554815800752266e-08,
      "loss": 0.5702,
      "step": 21525
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.476622272937005,
      "learning_rate": 4.146192878165367e-08,
      "loss": 0.5109,
      "step": 21526
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.399102059529293,
      "learning_rate": 4.136914526117919e-08,
      "loss": 0.5611,
      "step": 21527
    },
    {
      "epoch": 1.92,
      "grad_norm": 7.184404373269967,
      "learning_rate": 4.127646524126394e-08,
      "loss": 0.6016,
      "step": 21528
    },
    {
      "epoch": 1.92,
      "grad_norm": 7.224993888869598,
      "learning_rate": 4.11838887238436e-08,
      "loss": 0.5254,
      "step": 21529
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.826091851265598,
      "learning_rate": 4.109141571084996e-08,
      "loss": 0.5528,
      "step": 21530
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.160898513943865,
      "learning_rate": 4.099904620421258e-08,
      "loss": 0.5545,
      "step": 21531
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.790212366595633,
      "learning_rate": 4.090678020586047e-08,
      "loss": 0.5792,
      "step": 21532
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.989191100380436,
      "learning_rate": 4.081461771771933e-08,
      "loss": 0.5791,
      "step": 21533
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.120979997113866,
      "learning_rate": 4.072255874171205e-08,
      "loss": 0.5399,
      "step": 21534
    },
    {
      "epoch": 1.92,
      "grad_norm": 10.070739244562043,
      "learning_rate": 4.0630603279760984e-08,
      "loss": 0.5094,
      "step": 21535
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.961891859429244,
      "learning_rate": 4.053875133378515e-08,
      "loss": 0.5759,
      "step": 21536
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.3305089832471735,
      "learning_rate": 4.044700290570247e-08,
      "loss": 0.5693,
      "step": 21537
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.7909224249042035,
      "learning_rate": 4.035535799742696e-08,
      "loss": 0.507,
      "step": 21538
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.271415838537933,
      "learning_rate": 4.026381661087153e-08,
      "loss": 0.5892,
      "step": 21539
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.99852639158797,
      "learning_rate": 4.0172378747947435e-08,
      "loss": 0.552,
      "step": 21540
    },
    {
      "epoch": 1.92,
      "grad_norm": 8.085641378889687,
      "learning_rate": 4.0081044410563706e-08,
      "loss": 0.5753,
      "step": 21541
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.074212159974657,
      "learning_rate": 3.9989813600625484e-08,
      "loss": 0.5796,
      "step": 21542
    },
    {
      "epoch": 1.92,
      "grad_norm": 7.108685634730223,
      "learning_rate": 3.9898686320037904e-08,
      "loss": 0.5949,
      "step": 21543
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.245838488940144,
      "learning_rate": 3.980766257070279e-08,
      "loss": 0.5921,
      "step": 21544
    },
    {
      "epoch": 1.92,
      "grad_norm": 7.358307532348258,
      "learning_rate": 3.9716742354520275e-08,
      "loss": 0.5874,
      "step": 21545
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.937147453757614,
      "learning_rate": 3.9625925673387745e-08,
      "loss": 0.5012,
      "step": 21546
    },
    {
      "epoch": 1.92,
      "grad_norm": 9.033436253883677,
      "learning_rate": 3.95352125292009e-08,
      "loss": 0.518,
      "step": 21547
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.808512314667065,
      "learning_rate": 3.9444602923853216e-08,
      "loss": 0.5232,
      "step": 21548
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.9495362315369364,
      "learning_rate": 3.9354096859235966e-08,
      "loss": 0.6129,
      "step": 21549
    },
    {
      "epoch": 1.92,
      "grad_norm": 7.641297029476702,
      "learning_rate": 3.926369433723876e-08,
      "loss": 0.598,
      "step": 21550
    },
    {
      "epoch": 1.92,
      "grad_norm": 7.438317151052993,
      "learning_rate": 3.917339535974784e-08,
      "loss": 0.5645,
      "step": 21551
    },
    {
      "epoch": 1.92,
      "grad_norm": 8.403117260392882,
      "learning_rate": 3.90831999286484e-08,
      "loss": 0.5959,
      "step": 21552
    },
    {
      "epoch": 1.92,
      "grad_norm": 8.466231752601669,
      "learning_rate": 3.8993108045822794e-08,
      "loss": 0.564,
      "step": 21553
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.051660504254396,
      "learning_rate": 3.8903119713151193e-08,
      "loss": 0.5098,
      "step": 21554
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.9552887083378065,
      "learning_rate": 3.8813234932512654e-08,
      "loss": 0.5252,
      "step": 21555
    },
    {
      "epoch": 1.92,
      "grad_norm": 7.018026794275556,
      "learning_rate": 3.872345370578345e-08,
      "loss": 0.5249,
      "step": 21556
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.124714095828262,
      "learning_rate": 3.863377603483709e-08,
      "loss": 0.635,
      "step": 21557
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.8491280939702515,
      "learning_rate": 3.8544201921544846e-08,
      "loss": 0.5562,
      "step": 21558
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.909527367738023,
      "learning_rate": 3.8454731367777444e-08,
      "loss": 0.558,
      "step": 21559
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.1765290007692615,
      "learning_rate": 3.836536437540172e-08,
      "loss": 0.5249,
      "step": 21560
    },
    {
      "epoch": 1.92,
      "grad_norm": 8.85935228599553,
      "learning_rate": 3.827610094628342e-08,
      "loss": 0.5987,
      "step": 21561
    },
    {
      "epoch": 1.92,
      "grad_norm": 7.163932264280189,
      "learning_rate": 3.8186941082285486e-08,
      "loss": 0.5668,
      "step": 21562
    },
    {
      "epoch": 1.92,
      "grad_norm": 8.343469255376862,
      "learning_rate": 3.8097884785269215e-08,
      "loss": 0.5896,
      "step": 21563
    },
    {
      "epoch": 1.92,
      "grad_norm": 7.357124716862172,
      "learning_rate": 3.800893205709311e-08,
      "loss": 0.5753,
      "step": 21564
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.955465049010952,
      "learning_rate": 3.792008289961346e-08,
      "loss": 0.5732,
      "step": 21565
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.5909924509584235,
      "learning_rate": 3.7831337314685467e-08,
      "loss": 0.5972,
      "step": 21566
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.9820844862360865,
      "learning_rate": 3.7742695304161526e-08,
      "loss": 0.5802,
      "step": 21567
    },
    {
      "epoch": 1.92,
      "grad_norm": 7.036138003107151,
      "learning_rate": 3.765415686989182e-08,
      "loss": 0.5451,
      "step": 21568
    },
    {
      "epoch": 1.92,
      "grad_norm": 7.406601796515372,
      "learning_rate": 3.756572201372433e-08,
      "loss": 0.5448,
      "step": 21569
    },
    {
      "epoch": 1.92,
      "grad_norm": 8.320624998793535,
      "learning_rate": 3.7477390737504226e-08,
      "loss": 0.5821,
      "step": 21570
    },
    {
      "epoch": 1.92,
      "grad_norm": 7.1867340698653885,
      "learning_rate": 3.738916304307616e-08,
      "loss": 0.5395,
      "step": 21571
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.996262084880464,
      "learning_rate": 3.730103893228143e-08,
      "loss": 0.5724,
      "step": 21572
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.021966686049023,
      "learning_rate": 3.721301840695912e-08,
      "loss": 0.5769,
      "step": 21573
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.810927163032706,
      "learning_rate": 3.712510146894721e-08,
      "loss": 0.6046,
      "step": 21574
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.673354388763369,
      "learning_rate": 3.703728812008034e-08,
      "loss": 0.4794,
      "step": 21575
    },
    {
      "epoch": 1.92,
      "grad_norm": 8.421699061009063,
      "learning_rate": 3.694957836219149e-08,
      "loss": 0.5765,
      "step": 21576
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.427353297524863,
      "learning_rate": 3.6861972197110854e-08,
      "loss": 0.4986,
      "step": 21577
    },
    {
      "epoch": 1.92,
      "grad_norm": 9.250760911872892,
      "learning_rate": 3.677446962666697e-08,
      "loss": 0.5092,
      "step": 21578
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.5174971949680875,
      "learning_rate": 3.6687070652687285e-08,
      "loss": 0.5696,
      "step": 21579
    },
    {
      "epoch": 1.93,
      "grad_norm": 8.93650891446575,
      "learning_rate": 3.6599775276995875e-08,
      "loss": 0.5698,
      "step": 21580
    },
    {
      "epoch": 1.93,
      "grad_norm": 8.074392682373993,
      "learning_rate": 3.6512583501414064e-08,
      "loss": 0.578,
      "step": 21581
    },
    {
      "epoch": 1.93,
      "grad_norm": 5.64411955867253,
      "learning_rate": 3.642549532776263e-08,
      "loss": 0.4571,
      "step": 21582
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.691337208689975,
      "learning_rate": 3.633851075785844e-08,
      "loss": 0.5146,
      "step": 21583
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.312885166270491,
      "learning_rate": 3.625162979351782e-08,
      "loss": 0.529,
      "step": 21584
    },
    {
      "epoch": 1.93,
      "grad_norm": 5.69524241239654,
      "learning_rate": 3.6164852436553766e-08,
      "loss": 0.5647,
      "step": 21585
    },
    {
      "epoch": 1.93,
      "grad_norm": 5.72597093820198,
      "learning_rate": 3.6078178688778166e-08,
      "loss": 0.5545,
      "step": 21586
    },
    {
      "epoch": 1.93,
      "grad_norm": 5.871367210911551,
      "learning_rate": 3.599160855199957e-08,
      "loss": 0.5692,
      "step": 21587
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.460081650978581,
      "learning_rate": 3.590514202802542e-08,
      "loss": 0.5838,
      "step": 21588
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.46711196920435,
      "learning_rate": 3.5818779118660386e-08,
      "loss": 0.5592,
      "step": 21589
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.31204337456974,
      "learning_rate": 3.573251982570636e-08,
      "loss": 0.6142,
      "step": 21590
    },
    {
      "epoch": 1.93,
      "grad_norm": 9.238336857412623,
      "learning_rate": 3.5646364150964674e-08,
      "loss": 0.5831,
      "step": 21591
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.948616494561954,
      "learning_rate": 3.556031209623334e-08,
      "loss": 0.5709,
      "step": 21592
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.045384683841549,
      "learning_rate": 3.54743636633087e-08,
      "loss": 0.5542,
      "step": 21593
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.610896682560669,
      "learning_rate": 3.5388518853984314e-08,
      "loss": 0.5888,
      "step": 21594
    },
    {
      "epoch": 1.93,
      "grad_norm": 8.697616993052401,
      "learning_rate": 3.530277767005263e-08,
      "loss": 0.5909,
      "step": 21595
    },
    {
      "epoch": 1.93,
      "grad_norm": 9.934471946191781,
      "learning_rate": 3.5217140113302795e-08,
      "loss": 0.5275,
      "step": 21596
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.371504483906515,
      "learning_rate": 3.513160618552281e-08,
      "loss": 0.6846,
      "step": 21597
    },
    {
      "epoch": 1.93,
      "grad_norm": 5.734714735315072,
      "learning_rate": 3.504617588849735e-08,
      "loss": 0.5929,
      "step": 21598
    },
    {
      "epoch": 1.93,
      "grad_norm": 8.454139423799212,
      "learning_rate": 3.496084922400944e-08,
      "loss": 0.559,
      "step": 21599
    },
    {
      "epoch": 1.93,
      "grad_norm": 5.647400243937806,
      "learning_rate": 3.487562619384155e-08,
      "loss": 0.4977,
      "step": 21600
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.313605619347463,
      "learning_rate": 3.4790506799770585e-08,
      "loss": 0.5688,
      "step": 21601
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.607160250221201,
      "learning_rate": 3.470549104357457e-08,
      "loss": 0.5965,
      "step": 21602
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.0142550249419235,
      "learning_rate": 3.462057892702764e-08,
      "loss": 0.5575,
      "step": 21603
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.191625118834433,
      "learning_rate": 3.453577045190226e-08,
      "loss": 0.5955,
      "step": 21604
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.186947799277279,
      "learning_rate": 3.445106561996869e-08,
      "loss": 0.5877,
      "step": 21605
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.400534298116333,
      "learning_rate": 3.436646443299441e-08,
      "loss": 0.5414,
      "step": 21606
    },
    {
      "epoch": 1.93,
      "grad_norm": 8.603893842729763,
      "learning_rate": 3.4281966892745765e-08,
      "loss": 0.5622,
      "step": 21607
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.2124252213330315,
      "learning_rate": 3.4197573000985806e-08,
      "loss": 0.5134,
      "step": 21608
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.173512831432326,
      "learning_rate": 3.4113282759477004e-08,
      "loss": 0.5753,
      "step": 21609
    },
    {
      "epoch": 1.93,
      "grad_norm": 11.688112270190844,
      "learning_rate": 3.402909616997851e-08,
      "loss": 0.6355,
      "step": 21610
    },
    {
      "epoch": 1.93,
      "grad_norm": 8.503052469381904,
      "learning_rate": 3.39450132342467e-08,
      "loss": 0.5535,
      "step": 21611
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.318610798722964,
      "learning_rate": 3.3861033954037946e-08,
      "loss": 0.5419,
      "step": 21612
    },
    {
      "epoch": 1.93,
      "grad_norm": 8.103343772384733,
      "learning_rate": 3.3777158331104176e-08,
      "loss": 0.5372,
      "step": 21613
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.895471075512797,
      "learning_rate": 3.369338636719677e-08,
      "loss": 0.5408,
      "step": 21614
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.4670121857546645,
      "learning_rate": 3.360971806406321e-08,
      "loss": 0.5508,
      "step": 21615
    },
    {
      "epoch": 1.93,
      "grad_norm": 5.20245207504106,
      "learning_rate": 3.3526153423451005e-08,
      "loss": 0.6136,
      "step": 21616
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.592729101963866,
      "learning_rate": 3.344269244710374e-08,
      "loss": 0.5844,
      "step": 21617
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.077439577221369,
      "learning_rate": 3.3359335136763924e-08,
      "loss": 0.5805,
      "step": 21618
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.657542859086552,
      "learning_rate": 3.327608149417127e-08,
      "loss": 0.557,
      "step": 21619
    },
    {
      "epoch": 1.93,
      "grad_norm": 11.887703522680765,
      "learning_rate": 3.319293152106329e-08,
      "loss": 0.5676,
      "step": 21620
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.387175168808456,
      "learning_rate": 3.31098852191758e-08,
      "loss": 0.465,
      "step": 21621
    },
    {
      "epoch": 1.93,
      "grad_norm": 9.076237623559106,
      "learning_rate": 3.302694259024186e-08,
      "loss": 0.5635,
      "step": 21622
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.689398449218368,
      "learning_rate": 3.294410363599343e-08,
      "loss": 0.5615,
      "step": 21623
    },
    {
      "epoch": 1.93,
      "grad_norm": 8.563658185164009,
      "learning_rate": 3.286136835815912e-08,
      "loss": 0.5131,
      "step": 21624
    },
    {
      "epoch": 1.93,
      "grad_norm": 5.9235884010767945,
      "learning_rate": 3.277873675846532e-08,
      "loss": 0.5522,
      "step": 21625
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.4171803145443,
      "learning_rate": 3.2696208838638444e-08,
      "loss": 0.5333,
      "step": 21626
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.3285573835640045,
      "learning_rate": 3.261378460039932e-08,
      "loss": 0.5335,
      "step": 21627
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.555243017226559,
      "learning_rate": 3.2531464045468806e-08,
      "loss": 0.5063,
      "step": 21628
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.431481238485403,
      "learning_rate": 3.244924717556552e-08,
      "loss": 0.56,
      "step": 21629
    },
    {
      "epoch": 1.93,
      "grad_norm": 8.070535263731713,
      "learning_rate": 3.236713399240588e-08,
      "loss": 0.5657,
      "step": 21630
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.04229734061394,
      "learning_rate": 3.2285124497702935e-08,
      "loss": 0.5684,
      "step": 21631
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.37767618504824,
      "learning_rate": 3.2203218693169224e-08,
      "loss": 0.5228,
      "step": 21632
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.9367457962268215,
      "learning_rate": 3.212141658051449e-08,
      "loss": 0.6148,
      "step": 21633
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.260695546082759,
      "learning_rate": 3.203971816144513e-08,
      "loss": 0.55,
      "step": 21634
    },
    {
      "epoch": 1.93,
      "grad_norm": 9.28486018788722,
      "learning_rate": 3.195812343766702e-08,
      "loss": 0.5923,
      "step": 21635
    },
    {
      "epoch": 1.93,
      "grad_norm": 5.12744097849967,
      "learning_rate": 3.187663241088323e-08,
      "loss": 0.5729,
      "step": 21636
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.405021340508431,
      "learning_rate": 3.1795245082794635e-08,
      "loss": 0.5998,
      "step": 21637
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.3479507943118705,
      "learning_rate": 3.1713961455099863e-08,
      "loss": 0.5603,
      "step": 21638
    },
    {
      "epoch": 1.93,
      "grad_norm": 5.92341285103511,
      "learning_rate": 3.163278152949645e-08,
      "loss": 0.6319,
      "step": 21639
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.140741345856287,
      "learning_rate": 3.155170530767804e-08,
      "loss": 0.6084,
      "step": 21640
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.81086848424523,
      "learning_rate": 3.147073279133661e-08,
      "loss": 0.6211,
      "step": 21641
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.838572680155705,
      "learning_rate": 3.1389863982163036e-08,
      "loss": 0.6266,
      "step": 21642
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.777214921172168,
      "learning_rate": 3.1309098881844854e-08,
      "loss": 0.5654,
      "step": 21643
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.0289884298797,
      "learning_rate": 3.1228437492068495e-08,
      "loss": 0.5649,
      "step": 21644
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.925940534185544,
      "learning_rate": 3.114787981451706e-08,
      "loss": 0.6166,
      "step": 21645
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.5584978115646155,
      "learning_rate": 3.106742585087141e-08,
      "loss": 0.5393,
      "step": 21646
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.431395716915658,
      "learning_rate": 3.098707560281189e-08,
      "loss": 0.5826,
      "step": 21647
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.275768686059172,
      "learning_rate": 3.090682907201492e-08,
      "loss": 0.5595,
      "step": 21648
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.463363317051256,
      "learning_rate": 3.0826686260155834e-08,
      "loss": 0.589,
      "step": 21649
    },
    {
      "epoch": 1.93,
      "grad_norm": 5.418174401169631,
      "learning_rate": 3.074664716890774e-08,
      "loss": 0.6099,
      "step": 21650
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.279435585677287,
      "learning_rate": 3.0666711799940407e-08,
      "loss": 0.6163,
      "step": 21651
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.990277697158665,
      "learning_rate": 3.0586880154923057e-08,
      "loss": 0.5729,
      "step": 21652
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.398321651888566,
      "learning_rate": 3.050715223552214e-08,
      "loss": 0.6426,
      "step": 21653
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.681328515219577,
      "learning_rate": 3.0427528043400765e-08,
      "loss": 0.5796,
      "step": 21654
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.984316225684207,
      "learning_rate": 3.0348007580221493e-08,
      "loss": 0.5341,
      "step": 21655
    },
    {
      "epoch": 1.93,
      "grad_norm": 8.956535136888618,
      "learning_rate": 3.0268590847644663e-08,
      "loss": 0.5599,
      "step": 21656
    },
    {
      "epoch": 1.93,
      "grad_norm": 5.894537014662968,
      "learning_rate": 3.0189277847327283e-08,
      "loss": 0.5736,
      "step": 21657
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.588749045979971,
      "learning_rate": 3.011006858092524e-08,
      "loss": 0.5685,
      "step": 21658
    },
    {
      "epoch": 1.93,
      "grad_norm": 8.237382776860303,
      "learning_rate": 3.003096305009168e-08,
      "loss": 0.587,
      "step": 21659
    },
    {
      "epoch": 1.93,
      "grad_norm": 5.745459412511875,
      "learning_rate": 2.995196125647748e-08,
      "loss": 0.575,
      "step": 21660
    },
    {
      "epoch": 1.93,
      "grad_norm": 5.75090511598569,
      "learning_rate": 2.987306320173189e-08,
      "loss": 0.565,
      "step": 21661
    },
    {
      "epoch": 1.93,
      "grad_norm": 8.671715078018352,
      "learning_rate": 2.979426888750192e-08,
      "loss": 0.6179,
      "step": 21662
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.574905651049755,
      "learning_rate": 2.971557831543126e-08,
      "loss": 0.5837,
      "step": 21663
    },
    {
      "epoch": 1.93,
      "grad_norm": 8.578212737622344,
      "learning_rate": 2.9636991487164147e-08,
      "loss": 0.492,
      "step": 21664
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.01954210654065,
      "learning_rate": 2.955850840433927e-08,
      "loss": 0.5608,
      "step": 21665
    },
    {
      "epoch": 1.93,
      "grad_norm": 9.184294196952886,
      "learning_rate": 2.9480129068595876e-08,
      "loss": 0.5868,
      "step": 21666
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.684982419124985,
      "learning_rate": 2.940185348156932e-08,
      "loss": 0.5377,
      "step": 21667
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.626160557581862,
      "learning_rate": 2.9323681644893854e-08,
      "loss": 0.4892,
      "step": 21668
    },
    {
      "epoch": 1.93,
      "grad_norm": 8.159275097691499,
      "learning_rate": 2.9245613560200392e-08,
      "loss": 0.5682,
      "step": 21669
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.3972225065802375,
      "learning_rate": 2.91676492291193e-08,
      "loss": 0.5504,
      "step": 21670
    },
    {
      "epoch": 1.93,
      "grad_norm": 8.336891794623213,
      "learning_rate": 2.908978865327816e-08,
      "loss": 0.6123,
      "step": 21671
    },
    {
      "epoch": 1.93,
      "grad_norm": 9.962378175026108,
      "learning_rate": 2.901203183430068e-08,
      "loss": 0.5785,
      "step": 21672
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.0873731368302755,
      "learning_rate": 2.8934378773811112e-08,
      "loss": 0.4597,
      "step": 21673
    },
    {
      "epoch": 1.93,
      "grad_norm": 5.770400705308239,
      "learning_rate": 2.8856829473429826e-08,
      "loss": 0.5584,
      "step": 21674
    },
    {
      "epoch": 1.93,
      "grad_norm": 5.98537114280287,
      "learning_rate": 2.8779383934776083e-08,
      "loss": 0.5543,
      "step": 21675
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.26360424480145,
      "learning_rate": 2.8702042159465815e-08,
      "loss": 0.5161,
      "step": 21676
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.903211358658218,
      "learning_rate": 2.862480414911273e-08,
      "loss": 0.5362,
      "step": 21677
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.334159169442731,
      "learning_rate": 2.854766990533053e-08,
      "loss": 0.5241,
      "step": 21678
    },
    {
      "epoch": 1.93,
      "grad_norm": 5.155062263961702,
      "learning_rate": 2.8470639429727943e-08,
      "loss": 0.5638,
      "step": 21679
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.013764467140999,
      "learning_rate": 2.8393712723913668e-08,
      "loss": 0.5653,
      "step": 21680
    },
    {
      "epoch": 1.93,
      "grad_norm": 8.62147477717422,
      "learning_rate": 2.8316889789492542e-08,
      "loss": 0.6257,
      "step": 21681
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.48504150792668,
      "learning_rate": 2.824017062806883e-08,
      "loss": 0.4972,
      "step": 21682
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.229000105739291,
      "learning_rate": 2.8163555241244033e-08,
      "loss": 0.5538,
      "step": 21683
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.105741468743016,
      "learning_rate": 2.8087043630616316e-08,
      "loss": 0.5743,
      "step": 21684
    },
    {
      "epoch": 1.93,
      "grad_norm": 9.143191289090243,
      "learning_rate": 2.8010635797783293e-08,
      "loss": 0.5613,
      "step": 21685
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.188738294250861,
      "learning_rate": 2.7934331744339795e-08,
      "loss": 0.6004,
      "step": 21686
    },
    {
      "epoch": 1.93,
      "grad_norm": 8.037861413711314,
      "learning_rate": 2.7858131471878434e-08,
      "loss": 0.5724,
      "step": 21687
    },
    {
      "epoch": 1.93,
      "grad_norm": 5.978432286737291,
      "learning_rate": 2.7782034981989613e-08,
      "loss": 0.5129,
      "step": 21688
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.283690544180308,
      "learning_rate": 2.7706042276262612e-08,
      "loss": 0.5543,
      "step": 21689
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.063971222062769,
      "learning_rate": 2.7630153356282273e-08,
      "loss": 0.5699,
      "step": 21690
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.564173195424937,
      "learning_rate": 2.755436822363344e-08,
      "loss": 0.5426,
      "step": 21691
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.813720855508775,
      "learning_rate": 2.7478686879897078e-08,
      "loss": 0.4985,
      "step": 21692
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.237062708309648,
      "learning_rate": 2.7403109326654132e-08,
      "loss": 0.6252,
      "step": 21693
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.906552619506226,
      "learning_rate": 2.7327635565481126e-08,
      "loss": 0.5749,
      "step": 21694
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.164837117661732,
      "learning_rate": 2.725226559795402e-08,
      "loss": 0.5352,
      "step": 21695
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.521367107648826,
      "learning_rate": 2.7176999425645443e-08,
      "loss": 0.5708,
      "step": 21696
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.956684702550687,
      "learning_rate": 2.710183705012692e-08,
      "loss": 0.4719,
      "step": 21697
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.9417550637537735,
      "learning_rate": 2.7026778472967197e-08,
      "loss": 0.5652,
      "step": 21698
    },
    {
      "epoch": 1.94,
      "grad_norm": 9.011630165752578,
      "learning_rate": 2.6951823695732794e-08,
      "loss": 0.5084,
      "step": 21699
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.928921285653225,
      "learning_rate": 2.6876972719988015e-08,
      "loss": 0.5273,
      "step": 21700
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.795206335142996,
      "learning_rate": 2.6802225547295502e-08,
      "loss": 0.6469,
      "step": 21701
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.303985755631059,
      "learning_rate": 2.6727582179215673e-08,
      "loss": 0.5531,
      "step": 21702
    },
    {
      "epoch": 1.94,
      "grad_norm": 8.163472142916998,
      "learning_rate": 2.6653042617306168e-08,
      "loss": 0.5323,
      "step": 21703
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.856084808659494,
      "learning_rate": 2.657860686312297e-08,
      "loss": 0.5508,
      "step": 21704
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.57423562291821,
      "learning_rate": 2.650427491821983e-08,
      "loss": 0.5463,
      "step": 21705
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.12631226421573,
      "learning_rate": 2.643004678414829e-08,
      "loss": 0.5873,
      "step": 21706
    },
    {
      "epoch": 1.94,
      "grad_norm": 10.032383195913038,
      "learning_rate": 2.6355922462457106e-08,
      "loss": 0.5439,
      "step": 21707
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.705325154399892,
      "learning_rate": 2.6281901954693933e-08,
      "loss": 0.5455,
      "step": 21708
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.88599536969444,
      "learning_rate": 2.6207985262404756e-08,
      "loss": 0.541,
      "step": 21709
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.035639995313974,
      "learning_rate": 2.613417238713056e-08,
      "loss": 0.5448,
      "step": 21710
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.849341234135785,
      "learning_rate": 2.60604633304129e-08,
      "loss": 0.5634,
      "step": 21711
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.616220993979164,
      "learning_rate": 2.5986858093790533e-08,
      "loss": 0.5436,
      "step": 21712
    },
    {
      "epoch": 1.94,
      "grad_norm": 9.013412023019955,
      "learning_rate": 2.5913356678800017e-08,
      "loss": 0.564,
      "step": 21713
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.690049020012235,
      "learning_rate": 2.5839959086974566e-08,
      "loss": 0.5867,
      "step": 21714
    },
    {
      "epoch": 1.94,
      "grad_norm": 9.324827819300607,
      "learning_rate": 2.5766665319846838e-08,
      "loss": 0.5619,
      "step": 21715
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.499834939180944,
      "learning_rate": 2.569347537894673e-08,
      "loss": 0.5772,
      "step": 21716
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.403407924147132,
      "learning_rate": 2.5620389265801902e-08,
      "loss": 0.6118,
      "step": 21717
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.318616992698283,
      "learning_rate": 2.5547406981937252e-08,
      "loss": 0.5703,
      "step": 21718
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.413373511667841,
      "learning_rate": 2.547452852887711e-08,
      "loss": 0.6351,
      "step": 21719
    },
    {
      "epoch": 1.94,
      "grad_norm": 8.38946448377567,
      "learning_rate": 2.5401753908141923e-08,
      "loss": 0.5962,
      "step": 21720
    },
    {
      "epoch": 1.94,
      "grad_norm": 8.033886722127164,
      "learning_rate": 2.532908312125104e-08,
      "loss": 0.6198,
      "step": 21721
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.635767556240944,
      "learning_rate": 2.525651616972158e-08,
      "loss": 0.6014,
      "step": 21722
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.048111145833499,
      "learning_rate": 2.5184053055067325e-08,
      "loss": 0.5288,
      "step": 21723
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.139698722291098,
      "learning_rate": 2.5111693778801517e-08,
      "loss": 0.5694,
      "step": 21724
    },
    {
      "epoch": 1.94,
      "grad_norm": 9.55859996798095,
      "learning_rate": 2.5039438342434054e-08,
      "loss": 0.5734,
      "step": 21725
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.6864111351829445,
      "learning_rate": 2.4967286747473174e-08,
      "loss": 0.5306,
      "step": 21726
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.270800718907912,
      "learning_rate": 2.4895238995425452e-08,
      "loss": 0.5533,
      "step": 21727
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.596191210622362,
      "learning_rate": 2.482329508779413e-08,
      "loss": 0.5254,
      "step": 21728
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.719597001417965,
      "learning_rate": 2.475145502608134e-08,
      "loss": 0.5675,
      "step": 21729
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.610054334619293,
      "learning_rate": 2.467971881178588e-08,
      "loss": 0.5595,
      "step": 21730
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.433317693823554,
      "learning_rate": 2.4608086446406e-08,
      "loss": 0.6024,
      "step": 21731
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.3088969381000455,
      "learning_rate": 2.4536557931436056e-08,
      "loss": 0.5494,
      "step": 21732
    },
    {
      "epoch": 1.94,
      "grad_norm": 9.325905916468637,
      "learning_rate": 2.44651332683693e-08,
      "loss": 0.569,
      "step": 21733
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.943450737940932,
      "learning_rate": 2.4393812458697318e-08,
      "loss": 0.5482,
      "step": 21734
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.886559314369704,
      "learning_rate": 2.432259550390781e-08,
      "loss": 0.5226,
      "step": 21735
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.279691845873403,
      "learning_rate": 2.4251482405486803e-08,
      "loss": 0.5251,
      "step": 21736
    },
    {
      "epoch": 1.94,
      "grad_norm": 8.146489938986656,
      "learning_rate": 2.4180473164920336e-08,
      "loss": 0.4777,
      "step": 21737
    },
    {
      "epoch": 1.94,
      "grad_norm": 8.611729753779853,
      "learning_rate": 2.4109567783688892e-08,
      "loss": 0.5269,
      "step": 21738
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.499968072533326,
      "learning_rate": 2.4038766263273506e-08,
      "loss": 0.557,
      "step": 21739
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.387552872696248,
      "learning_rate": 2.3968068605151885e-08,
      "loss": 0.531,
      "step": 21740
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.582810447730976,
      "learning_rate": 2.3897474810799516e-08,
      "loss": 0.505,
      "step": 21741
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.207494358130621,
      "learning_rate": 2.3826984881690217e-08,
      "loss": 0.5507,
      "step": 21742
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.0469528797532615,
      "learning_rate": 2.3756598819294484e-08,
      "loss": 0.5196,
      "step": 21743
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.744261255695491,
      "learning_rate": 2.3686316625081697e-08,
      "loss": 0.6048,
      "step": 21744
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.5594763097313304,
      "learning_rate": 2.3616138300519566e-08,
      "loss": 0.5272,
      "step": 21745
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.747032405768726,
      "learning_rate": 2.3546063847071922e-08,
      "loss": 0.5161,
      "step": 21746
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.847796168836106,
      "learning_rate": 2.3476093266202594e-08,
      "loss": 0.5579,
      "step": 21747
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.429880582454862,
      "learning_rate": 2.3406226559371526e-08,
      "loss": 0.5411,
      "step": 21748
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.72429520966671,
      "learning_rate": 2.3336463728036442e-08,
      "loss": 0.5687,
      "step": 21749
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.101917867828132,
      "learning_rate": 2.326680477365395e-08,
      "loss": 0.5768,
      "step": 21750
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.5558094214494265,
      "learning_rate": 2.319724969767845e-08,
      "loss": 0.4945,
      "step": 21751
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.569155119608232,
      "learning_rate": 2.3127798501560993e-08,
      "loss": 0.5771,
      "step": 21752
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.9547347612579,
      "learning_rate": 2.305845118675154e-08,
      "loss": 0.6309,
      "step": 21753
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.89059520103603,
      "learning_rate": 2.2989207754697818e-08,
      "loss": 0.5517,
      "step": 21754
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.933713526928987,
      "learning_rate": 2.2920068206844782e-08,
      "loss": 0.5618,
      "step": 21755
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.755314729988431,
      "learning_rate": 2.285103254463572e-08,
      "loss": 0.531,
      "step": 21756
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.495774901204524,
      "learning_rate": 2.2782100769511705e-08,
      "loss": 0.4857,
      "step": 21757
    },
    {
      "epoch": 1.94,
      "grad_norm": 8.669293251838015,
      "learning_rate": 2.271327288291103e-08,
      "loss": 0.5916,
      "step": 21758
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.145806786956795,
      "learning_rate": 2.2644548886271434e-08,
      "loss": 0.6065,
      "step": 21759
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.836509566957288,
      "learning_rate": 2.2575928781026214e-08,
      "loss": 0.6268,
      "step": 21760
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.20792060934523,
      "learning_rate": 2.2507412568608112e-08,
      "loss": 0.5513,
      "step": 21761
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.259909689467585,
      "learning_rate": 2.243900025044765e-08,
      "loss": 0.5766,
      "step": 21762
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.320544831942432,
      "learning_rate": 2.237069182797147e-08,
      "loss": 0.5363,
      "step": 21763
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.1207956274185875,
      "learning_rate": 2.2302487302606755e-08,
      "loss": 0.5388,
      "step": 21764
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.953254225168298,
      "learning_rate": 2.223438667577682e-08,
      "loss": 0.539,
      "step": 21765
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.237611795456866,
      "learning_rate": 2.2166389948902744e-08,
      "loss": 0.6134,
      "step": 21766
    },
    {
      "epoch": 1.94,
      "grad_norm": 8.85496628603445,
      "learning_rate": 2.2098497123404505e-08,
      "loss": 0.5434,
      "step": 21767
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.669754841773753,
      "learning_rate": 2.2030708200698194e-08,
      "loss": 0.5986,
      "step": 21768
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.664422438985362,
      "learning_rate": 2.1963023182199338e-08,
      "loss": 0.5951,
      "step": 21769
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.006506347296762,
      "learning_rate": 2.1895442069320704e-08,
      "loss": 0.5758,
      "step": 21770
    },
    {
      "epoch": 1.94,
      "grad_norm": 8.788327276227715,
      "learning_rate": 2.1827964863472828e-08,
      "loss": 0.5903,
      "step": 21771
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.246397108608395,
      "learning_rate": 2.1760591566064026e-08,
      "loss": 0.5403,
      "step": 21772
    },
    {
      "epoch": 1.94,
      "grad_norm": 8.172458586331109,
      "learning_rate": 2.1693322178500952e-08,
      "loss": 0.5824,
      "step": 21773
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.264459942800589,
      "learning_rate": 2.1626156702187486e-08,
      "loss": 0.5647,
      "step": 21774
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.144130890767491,
      "learning_rate": 2.1559095138525276e-08,
      "loss": 0.5734,
      "step": 21775
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.704682392144669,
      "learning_rate": 2.1492137488914877e-08,
      "loss": 0.5421,
      "step": 21776
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.882854072911726,
      "learning_rate": 2.1425283754752945e-08,
      "loss": 0.5387,
      "step": 21777
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.127466073571384,
      "learning_rate": 2.135853393743559e-08,
      "loss": 0.593,
      "step": 21778
    },
    {
      "epoch": 1.94,
      "grad_norm": 9.047675678272398,
      "learning_rate": 2.1291888038355578e-08,
      "loss": 0.5452,
      "step": 21779
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.543412211712778,
      "learning_rate": 2.1225346058904585e-08,
      "loss": 0.5344,
      "step": 21780
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.855689262814037,
      "learning_rate": 2.115890800047038e-08,
      "loss": 0.6017,
      "step": 21781
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.580160082893052,
      "learning_rate": 2.1092573864441303e-08,
      "loss": 0.5061,
      "step": 21782
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.173886110891978,
      "learning_rate": 2.1026343652200688e-08,
      "loss": 0.5717,
      "step": 21783
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.219702878251234,
      "learning_rate": 2.096021736513132e-08,
      "loss": 0.5475,
      "step": 21784
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.644428623241483,
      "learning_rate": 2.089419500461376e-08,
      "loss": 0.5215,
      "step": 21785
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.34236938578635,
      "learning_rate": 2.0828276572025796e-08,
      "loss": 0.5471,
      "step": 21786
    },
    {
      "epoch": 1.94,
      "grad_norm": 8.824664217231707,
      "learning_rate": 2.0762462068742993e-08,
      "loss": 0.5277,
      "step": 21787
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.338586047857313,
      "learning_rate": 2.069675149613981e-08,
      "loss": 0.5594,
      "step": 21788
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.177266013250777,
      "learning_rate": 2.0631144855587372e-08,
      "loss": 0.5313,
      "step": 21789
    },
    {
      "epoch": 1.94,
      "grad_norm": 9.316164731579128,
      "learning_rate": 2.0565642148455132e-08,
      "loss": 0.5409,
      "step": 21790
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.359617511991421,
      "learning_rate": 2.050024337611034e-08,
      "loss": 0.5653,
      "step": 21791
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.446679184579295,
      "learning_rate": 2.0434948539918564e-08,
      "loss": 0.5375,
      "step": 21792
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.8127614413638735,
      "learning_rate": 2.0369757641241493e-08,
      "loss": 0.5156,
      "step": 21793
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.677454040081512,
      "learning_rate": 2.0304670681440818e-08,
      "loss": 0.4932,
      "step": 21794
    },
    {
      "epoch": 1.94,
      "grad_norm": 8.754411831207612,
      "learning_rate": 2.023968766187434e-08,
      "loss": 0.5777,
      "step": 21795
    },
    {
      "epoch": 1.94,
      "grad_norm": 9.648765497180351,
      "learning_rate": 2.017480858389931e-08,
      "loss": 0.5648,
      "step": 21796
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.8009615121102,
      "learning_rate": 2.0110033448869636e-08,
      "loss": 0.5529,
      "step": 21797
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.770213002282771,
      "learning_rate": 2.0045362258137024e-08,
      "loss": 0.5726,
      "step": 21798
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.980917992615495,
      "learning_rate": 1.99807950130515e-08,
      "loss": 0.6071,
      "step": 21799
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.684523089804364,
      "learning_rate": 1.991633171496088e-08,
      "loss": 0.4982,
      "step": 21800
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.531041802077858,
      "learning_rate": 1.9851972365210748e-08,
      "loss": 0.521,
      "step": 21801
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.459834074668601,
      "learning_rate": 1.9787716965144476e-08,
      "loss": 0.582,
      "step": 21802
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.117973298394805,
      "learning_rate": 1.972356551610266e-08,
      "loss": 0.5482,
      "step": 21803
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.558900468054927,
      "learning_rate": 1.9659518019425338e-08,
      "loss": 0.5656,
      "step": 21804
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.55750153360635,
      "learning_rate": 1.9595574476448663e-08,
      "loss": 0.5835,
      "step": 21805
    },
    {
      "epoch": 1.95,
      "grad_norm": 5.607593148334502,
      "learning_rate": 1.953173488850768e-08,
      "loss": 0.5316,
      "step": 21806
    },
    {
      "epoch": 1.95,
      "grad_norm": 10.303104718267381,
      "learning_rate": 1.94679992569341e-08,
      "loss": 0.5835,
      "step": 21807
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.011168858637809,
      "learning_rate": 1.940436758305908e-08,
      "loss": 0.5564,
      "step": 21808
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.163093958432611,
      "learning_rate": 1.9340839868210448e-08,
      "loss": 0.5985,
      "step": 21809
    },
    {
      "epoch": 1.95,
      "grad_norm": 9.001056973566198,
      "learning_rate": 1.927741611371492e-08,
      "loss": 0.5901,
      "step": 21810
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.715968098574154,
      "learning_rate": 1.9214096320895325e-08,
      "loss": 0.5595,
      "step": 21811
    },
    {
      "epoch": 1.95,
      "grad_norm": 8.890050944620311,
      "learning_rate": 1.9150880491073388e-08,
      "loss": 0.5914,
      "step": 21812
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.0531972736292765,
      "learning_rate": 1.9087768625569712e-08,
      "loss": 0.588,
      "step": 21813
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.7830220939560295,
      "learning_rate": 1.902476072569992e-08,
      "loss": 0.5451,
      "step": 21814
    },
    {
      "epoch": 1.95,
      "grad_norm": 8.068092357125058,
      "learning_rate": 1.896185679278073e-08,
      "loss": 0.5356,
      "step": 21815
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.753766068435255,
      "learning_rate": 1.889905682812443e-08,
      "loss": 0.5271,
      "step": 21816
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.689748498497799,
      "learning_rate": 1.883636083304219e-08,
      "loss": 0.5753,
      "step": 21817
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.2196375728296465,
      "learning_rate": 1.8773768808841852e-08,
      "loss": 0.5776,
      "step": 21818
    },
    {
      "epoch": 1.95,
      "grad_norm": 8.654680098237652,
      "learning_rate": 1.8711280756830153e-08,
      "loss": 0.576,
      "step": 21819
    },
    {
      "epoch": 1.95,
      "grad_norm": 8.237970805249667,
      "learning_rate": 1.864889667831271e-08,
      "loss": 0.5286,
      "step": 21820
    },
    {
      "epoch": 1.95,
      "grad_norm": 5.763275750601125,
      "learning_rate": 1.85866165745896e-08,
      "loss": 0.6088,
      "step": 21821
    },
    {
      "epoch": 1.95,
      "grad_norm": 8.257102946968551,
      "learning_rate": 1.8524440446962e-08,
      "loss": 0.5224,
      "step": 21822
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.7462269593744,
      "learning_rate": 1.8462368296727207e-08,
      "loss": 0.5268,
      "step": 21823
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.38745605621588,
      "learning_rate": 1.840040012518196e-08,
      "loss": 0.5516,
      "step": 21824
    },
    {
      "epoch": 1.95,
      "grad_norm": 9.16932150495451,
      "learning_rate": 1.8338535933618005e-08,
      "loss": 0.5818,
      "step": 21825
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.448250294110211,
      "learning_rate": 1.8276775723327643e-08,
      "loss": 0.6393,
      "step": 21826
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.208496487501599,
      "learning_rate": 1.8215119495599844e-08,
      "loss": 0.5226,
      "step": 21827
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.964615951074387,
      "learning_rate": 1.8153567251721906e-08,
      "loss": 0.4891,
      "step": 21828
    },
    {
      "epoch": 1.95,
      "grad_norm": 5.236145817163048,
      "learning_rate": 1.809211899297725e-08,
      "loss": 0.4984,
      "step": 21829
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.076123933909918,
      "learning_rate": 1.8030774720650402e-08,
      "loss": 0.5441,
      "step": 21830
    },
    {
      "epoch": 1.95,
      "grad_norm": 9.268996661913597,
      "learning_rate": 1.7969534436020343e-08,
      "loss": 0.5142,
      "step": 21831
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.1965831933105315,
      "learning_rate": 1.7908398140366047e-08,
      "loss": 0.5797,
      "step": 21832
    },
    {
      "epoch": 1.95,
      "grad_norm": 8.596723438921137,
      "learning_rate": 1.784736583496316e-08,
      "loss": 0.5821,
      "step": 21833
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.483842133285127,
      "learning_rate": 1.7786437521085664e-08,
      "loss": 0.5852,
      "step": 21834
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.76343300403562,
      "learning_rate": 1.772561320000532e-08,
      "loss": 0.5282,
      "step": 21835
    },
    {
      "epoch": 1.95,
      "grad_norm": 11.697366139650574,
      "learning_rate": 1.7664892872992222e-08,
      "loss": 0.5285,
      "step": 21836
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.096883579913403,
      "learning_rate": 1.7604276541313136e-08,
      "loss": 0.5902,
      "step": 21837
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.213362822315107,
      "learning_rate": 1.7543764206233715e-08,
      "loss": 0.5237,
      "step": 21838
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.585610807893996,
      "learning_rate": 1.7483355869016284e-08,
      "loss": 0.5794,
      "step": 21839
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.701287970274538,
      "learning_rate": 1.7423051530922607e-08,
      "loss": 0.5773,
      "step": 21840
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.450886686953002,
      "learning_rate": 1.7362851193211127e-08,
      "loss": 0.5627,
      "step": 21841
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.9322337074016325,
      "learning_rate": 1.7302754857138617e-08,
      "loss": 0.5425,
      "step": 21842
    },
    {
      "epoch": 1.95,
      "grad_norm": 13.16693081989037,
      "learning_rate": 1.7242762523959068e-08,
      "loss": 0.6186,
      "step": 21843
    },
    {
      "epoch": 1.95,
      "grad_norm": 5.969367568409202,
      "learning_rate": 1.718287419492426e-08,
      "loss": 0.5421,
      "step": 21844
    },
    {
      "epoch": 1.95,
      "grad_norm": 5.692319397150576,
      "learning_rate": 1.7123089871285415e-08,
      "loss": 0.5786,
      "step": 21845
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.86953665772673,
      "learning_rate": 1.706340955428931e-08,
      "loss": 0.5781,
      "step": 21846
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.438864089678426,
      "learning_rate": 1.7003833245182178e-08,
      "loss": 0.5807,
      "step": 21847
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.176698042411698,
      "learning_rate": 1.6944360945207462e-08,
      "loss": 0.5524,
      "step": 21848
    },
    {
      "epoch": 1.95,
      "grad_norm": 5.597929948814989,
      "learning_rate": 1.6884992655606392e-08,
      "loss": 0.622,
      "step": 21849
    },
    {
      "epoch": 1.95,
      "grad_norm": 12.396916505862915,
      "learning_rate": 1.6825728377618535e-08,
      "loss": 0.5468,
      "step": 21850
    },
    {
      "epoch": 1.95,
      "grad_norm": 11.085768458520963,
      "learning_rate": 1.6766568112480673e-08,
      "loss": 0.5746,
      "step": 21851
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.356648510615586,
      "learning_rate": 1.6707511861427385e-08,
      "loss": 0.5636,
      "step": 21852
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.933014765746647,
      "learning_rate": 1.6648559625691562e-08,
      "loss": 0.5763,
      "step": 21853
    },
    {
      "epoch": 1.95,
      "grad_norm": 5.993230392773871,
      "learning_rate": 1.6589711406503895e-08,
      "loss": 0.5999,
      "step": 21854
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.859258525825007,
      "learning_rate": 1.6530967205092286e-08,
      "loss": 0.5565,
      "step": 21855
    },
    {
      "epoch": 1.95,
      "grad_norm": 5.854174751798261,
      "learning_rate": 1.647232702268353e-08,
      "loss": 0.5685,
      "step": 21856
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.94363133675424,
      "learning_rate": 1.6413790860500546e-08,
      "loss": 0.5324,
      "step": 21857
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.610979952608761,
      "learning_rate": 1.6355358719766235e-08,
      "loss": 0.5577,
      "step": 21858
    },
    {
      "epoch": 1.95,
      "grad_norm": 5.870284716571663,
      "learning_rate": 1.6297030601699625e-08,
      "loss": 0.5575,
      "step": 21859
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.494656682444522,
      "learning_rate": 1.6238806507518635e-08,
      "loss": 0.6091,
      "step": 21860
    },
    {
      "epoch": 1.95,
      "grad_norm": 8.567903681096633,
      "learning_rate": 1.61806864384384e-08,
      "loss": 0.5535,
      "step": 21861
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.061082844157451,
      "learning_rate": 1.6122670395671836e-08,
      "loss": 0.53,
      "step": 21862
    },
    {
      "epoch": 1.95,
      "grad_norm": 8.024174585562154,
      "learning_rate": 1.6064758380429646e-08,
      "loss": 0.5766,
      "step": 21863
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.0795547998402135,
      "learning_rate": 1.6006950393921417e-08,
      "loss": 0.6001,
      "step": 21864
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.190653695346521,
      "learning_rate": 1.5949246437353404e-08,
      "loss": 0.5302,
      "step": 21865
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.780315010440447,
      "learning_rate": 1.5891646511929648e-08,
      "loss": 0.5711,
      "step": 21866
    },
    {
      "epoch": 1.95,
      "grad_norm": 5.969889115321199,
      "learning_rate": 1.5834150618852518e-08,
      "loss": 0.5071,
      "step": 21867
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.839385771997263,
      "learning_rate": 1.5776758759323273e-08,
      "loss": 0.5763,
      "step": 21868
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.401906382816063,
      "learning_rate": 1.5719470934538185e-08,
      "loss": 0.5585,
      "step": 21869
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.975523783224906,
      "learning_rate": 1.5662287145694065e-08,
      "loss": 0.5851,
      "step": 21870
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.611396542000379,
      "learning_rate": 1.5605207393983856e-08,
      "loss": 0.547,
      "step": 21871
    },
    {
      "epoch": 1.95,
      "grad_norm": 8.383525109905737,
      "learning_rate": 1.554823168059938e-08,
      "loss": 0.5657,
      "step": 21872
    },
    {
      "epoch": 1.95,
      "grad_norm": 9.83954082212749,
      "learning_rate": 1.5491360006729682e-08,
      "loss": 0.6231,
      "step": 21873
    },
    {
      "epoch": 1.95,
      "grad_norm": 5.938681088547739,
      "learning_rate": 1.5434592373562707e-08,
      "loss": 0.5453,
      "step": 21874
    },
    {
      "epoch": 1.95,
      "grad_norm": 8.937416017657387,
      "learning_rate": 1.5377928782281948e-08,
      "loss": 0.5836,
      "step": 21875
    },
    {
      "epoch": 1.95,
      "grad_norm": 8.044072807474645,
      "learning_rate": 1.5321369234070905e-08,
      "loss": 0.5321,
      "step": 21876
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.508049724076611,
      "learning_rate": 1.52649137301103e-08,
      "loss": 0.5884,
      "step": 21877
    },
    {
      "epoch": 1.95,
      "grad_norm": 5.719405714296873,
      "learning_rate": 1.5208562271578076e-08,
      "loss": 0.5799,
      "step": 21878
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.485684231468871,
      "learning_rate": 1.515231485965052e-08,
      "loss": 0.5772,
      "step": 21879
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.04411687318923,
      "learning_rate": 1.5096171495501688e-08,
      "loss": 0.5612,
      "step": 21880
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.583075991355989,
      "learning_rate": 1.5040132180303424e-08,
      "loss": 0.5576,
      "step": 21881
    },
    {
      "epoch": 1.95,
      "grad_norm": 8.184883709745376,
      "learning_rate": 1.4984196915225902e-08,
      "loss": 0.5517,
      "step": 21882
    },
    {
      "epoch": 1.95,
      "grad_norm": 5.265927244356894,
      "learning_rate": 1.4928365701435965e-08,
      "loss": 0.5295,
      "step": 21883
    },
    {
      "epoch": 1.95,
      "grad_norm": 9.212112351105205,
      "learning_rate": 1.487263854009935e-08,
      "loss": 0.5708,
      "step": 21884
    },
    {
      "epoch": 1.95,
      "grad_norm": 8.140673258708263,
      "learning_rate": 1.4817015432379012e-08,
      "loss": 0.5908,
      "step": 21885
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.906077343135555,
      "learning_rate": 1.4761496379436247e-08,
      "loss": 0.5951,
      "step": 21886
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.563371754420981,
      "learning_rate": 1.4706081382430127e-08,
      "loss": 0.5574,
      "step": 21887
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.579715636400873,
      "learning_rate": 1.4650770442516949e-08,
      "loss": 0.61,
      "step": 21888
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.2398307942044395,
      "learning_rate": 1.459556356085079e-08,
      "loss": 0.6473,
      "step": 21889
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.775421381021399,
      "learning_rate": 1.454046073858406e-08,
      "loss": 0.5844,
      "step": 21890
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.1549005551265035,
      "learning_rate": 1.4485461976868065e-08,
      "loss": 0.563,
      "step": 21891
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.363805849071634,
      "learning_rate": 1.4430567276849661e-08,
      "loss": 0.5915,
      "step": 21892
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.920287215275853,
      "learning_rate": 1.4375776639674599e-08,
      "loss": 0.5584,
      "step": 21893
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.2360692838449925,
      "learning_rate": 1.4321090066487519e-08,
      "loss": 0.5505,
      "step": 21894
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.999990945676075,
      "learning_rate": 1.4266507558428622e-08,
      "loss": 0.6143,
      "step": 21895
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.774138504279165,
      "learning_rate": 1.4212029116638104e-08,
      "loss": 0.4765,
      "step": 21896
    },
    {
      "epoch": 1.95,
      "grad_norm": 5.632184774719297,
      "learning_rate": 1.4157654742252835e-08,
      "loss": 0.5642,
      "step": 21897
    },
    {
      "epoch": 1.95,
      "grad_norm": 5.302781226546555,
      "learning_rate": 1.4103384436407464e-08,
      "loss": 0.556,
      "step": 21898
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.930624931829395,
      "learning_rate": 1.4049218200234971e-08,
      "loss": 0.5532,
      "step": 21899
    },
    {
      "epoch": 1.95,
      "grad_norm": 10.957703276202325,
      "learning_rate": 1.399515603486612e-08,
      "loss": 0.556,
      "step": 21900
    },
    {
      "epoch": 1.95,
      "grad_norm": 10.351634009908246,
      "learning_rate": 1.3941197941429453e-08,
      "loss": 0.6198,
      "step": 21901
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.924738325696259,
      "learning_rate": 1.3887343921050733e-08,
      "loss": 0.5878,
      "step": 21902
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.572627870313145,
      "learning_rate": 1.383359397485462e-08,
      "loss": 0.5988,
      "step": 21903
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.290889158622327,
      "learning_rate": 1.3779948103962437e-08,
      "loss": 0.5889,
      "step": 21904
    },
    {
      "epoch": 1.95,
      "grad_norm": 9.103948705340398,
      "learning_rate": 1.37264063094944e-08,
      "loss": 0.5782,
      "step": 21905
    },
    {
      "epoch": 1.95,
      "grad_norm": 5.78117735924223,
      "learning_rate": 1.3672968592567393e-08,
      "loss": 0.5442,
      "step": 21906
    },
    {
      "epoch": 1.95,
      "grad_norm": 5.697354723469791,
      "learning_rate": 1.36196349542983e-08,
      "loss": 0.6252,
      "step": 21907
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.148684828778051,
      "learning_rate": 1.3566405395798454e-08,
      "loss": 0.545,
      "step": 21908
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.772800692487624,
      "learning_rate": 1.3513279918179744e-08,
      "loss": 0.5668,
      "step": 21909
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.8589217326617575,
      "learning_rate": 1.3460258522551284e-08,
      "loss": 0.5387,
      "step": 21910
    },
    {
      "epoch": 1.95,
      "grad_norm": 9.67318223070198,
      "learning_rate": 1.3407341210019965e-08,
      "loss": 0.5173,
      "step": 21911
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.80416538307019,
      "learning_rate": 1.3354527981689348e-08,
      "loss": 0.6291,
      "step": 21912
    },
    {
      "epoch": 1.95,
      "grad_norm": 5.896716495438963,
      "learning_rate": 1.3301818838662995e-08,
      "loss": 0.5804,
      "step": 21913
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.750176277455353,
      "learning_rate": 1.3249213782040027e-08,
      "loss": 0.5737,
      "step": 21914
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.351146238956724,
      "learning_rate": 1.3196712812919565e-08,
      "loss": 0.5814,
      "step": 21915
    },
    {
      "epoch": 1.96,
      "grad_norm": 8.865917051951245,
      "learning_rate": 1.314431593239629e-08,
      "loss": 0.6412,
      "step": 21916
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.480307598431985,
      "learning_rate": 1.3092023141564325e-08,
      "loss": 0.5933,
      "step": 21917
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.60316344967438,
      "learning_rate": 1.3039834441515576e-08,
      "loss": 0.5988,
      "step": 21918
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.878466552420484,
      "learning_rate": 1.2987749833338614e-08,
      "loss": 0.5489,
      "step": 21919
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.2431266300648405,
      "learning_rate": 1.2935769318120905e-08,
      "loss": 0.5266,
      "step": 21920
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.659191266935583,
      "learning_rate": 1.2883892896948247e-08,
      "loss": 0.5183,
      "step": 21921
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.959152571270004,
      "learning_rate": 1.2832120570901995e-08,
      "loss": 0.544,
      "step": 21922
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.29953525912357,
      "learning_rate": 1.2780452341064064e-08,
      "loss": 0.5781,
      "step": 21923
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.453785950370717,
      "learning_rate": 1.2728888208511925e-08,
      "loss": 0.4874,
      "step": 21924
    },
    {
      "epoch": 1.96,
      "grad_norm": 9.130338173795755,
      "learning_rate": 1.2677428174322493e-08,
      "loss": 0.5881,
      "step": 21925
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.00744428559856,
      "learning_rate": 1.2626072239569908e-08,
      "loss": 0.5298,
      "step": 21926
    },
    {
      "epoch": 1.96,
      "grad_norm": 8.621002265769834,
      "learning_rate": 1.2574820405325538e-08,
      "loss": 0.5615,
      "step": 21927
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.771571616857394,
      "learning_rate": 1.2523672672659638e-08,
      "loss": 0.5534,
      "step": 21928
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.214323309320008,
      "learning_rate": 1.2472629042639683e-08,
      "loss": 0.5551,
      "step": 21929
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.514961527279013,
      "learning_rate": 1.2421689516331492e-08,
      "loss": 0.6106,
      "step": 21930
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.450719022597871,
      "learning_rate": 1.2370854094797547e-08,
      "loss": 0.6151,
      "step": 21931
    },
    {
      "epoch": 1.96,
      "grad_norm": 9.470032104902227,
      "learning_rate": 1.2320122779099219e-08,
      "loss": 0.602,
      "step": 21932
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.12069260525122,
      "learning_rate": 1.2269495570295664e-08,
      "loss": 0.5986,
      "step": 21933
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.984260729477265,
      "learning_rate": 1.2218972469443258e-08,
      "loss": 0.5669,
      "step": 21934
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.405520364997448,
      "learning_rate": 1.2168553477597267e-08,
      "loss": 0.6238,
      "step": 21935
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.08074038735939,
      "learning_rate": 1.2118238595809073e-08,
      "loss": 0.5624,
      "step": 21936
    },
    {
      "epoch": 1.96,
      "grad_norm": 8.809412100331906,
      "learning_rate": 1.2068027825129502e-08,
      "loss": 0.5379,
      "step": 21937
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.359428103599272,
      "learning_rate": 1.2017921166606605e-08,
      "loss": 0.5884,
      "step": 21938
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.442890664021551,
      "learning_rate": 1.1967918621285657e-08,
      "loss": 0.544,
      "step": 21939
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.626399745658119,
      "learning_rate": 1.1918020190210821e-08,
      "loss": 0.5766,
      "step": 21940
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.520841490804482,
      "learning_rate": 1.1868225874424044e-08,
      "loss": 0.5489,
      "step": 21941
    },
    {
      "epoch": 1.96,
      "grad_norm": 8.180498599058517,
      "learning_rate": 1.1818535674963939e-08,
      "loss": 0.5524,
      "step": 21942
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.551170576438809,
      "learning_rate": 1.1768949592868006e-08,
      "loss": 0.5453,
      "step": 21943
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.704618513275409,
      "learning_rate": 1.1719467629171533e-08,
      "loss": 0.5512,
      "step": 21944
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.190979503938292,
      "learning_rate": 1.1670089784907024e-08,
      "loss": 0.5634,
      "step": 21945
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.134477201072489,
      "learning_rate": 1.1620816061104768e-08,
      "loss": 0.5301,
      "step": 21946
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.4272431658843185,
      "learning_rate": 1.1571646458793385e-08,
      "loss": 0.5736,
      "step": 21947
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.688449350966891,
      "learning_rate": 1.1522580978999832e-08,
      "loss": 0.4847,
      "step": 21948
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.100686894418062,
      "learning_rate": 1.1473619622747733e-08,
      "loss": 0.574,
      "step": 21949
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.500174383408303,
      "learning_rate": 1.1424762391059053e-08,
      "loss": 0.5488,
      "step": 21950
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.709011784445153,
      "learning_rate": 1.1376009284954081e-08,
      "loss": 0.5143,
      "step": 21951
    },
    {
      "epoch": 1.96,
      "grad_norm": 8.236468120039437,
      "learning_rate": 1.1327360305449785e-08,
      "loss": 0.5351,
      "step": 21952
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.617655715063117,
      "learning_rate": 1.1278815453561465e-08,
      "loss": 0.577,
      "step": 21953
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.2863836223556,
      "learning_rate": 1.1230374730303306e-08,
      "loss": 0.5201,
      "step": 21954
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.4856289421205275,
      "learning_rate": 1.1182038136685613e-08,
      "loss": 0.5407,
      "step": 21955
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.050657053953674,
      "learning_rate": 1.113380567371758e-08,
      "loss": 0.5825,
      "step": 21956
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.605896617275991,
      "learning_rate": 1.1085677342405621e-08,
      "loss": 0.5642,
      "step": 21957
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.8627772192426955,
      "learning_rate": 1.1037653143755045e-08,
      "loss": 0.5835,
      "step": 21958
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.383109838268051,
      "learning_rate": 1.0989733078767273e-08,
      "loss": 0.5273,
      "step": 21959
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.568363244460694,
      "learning_rate": 1.0941917148443726e-08,
      "loss": 0.5162,
      "step": 21960
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.820935038919527,
      "learning_rate": 1.0894205353781383e-08,
      "loss": 0.5486,
      "step": 21961
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.812980642933535,
      "learning_rate": 1.0846597695776672e-08,
      "loss": 0.5525,
      "step": 21962
    },
    {
      "epoch": 1.96,
      "grad_norm": 11.44807904690461,
      "learning_rate": 1.0799094175423241e-08,
      "loss": 0.5695,
      "step": 21963
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.244917354329995,
      "learning_rate": 1.0751694793712519e-08,
      "loss": 0.589,
      "step": 21964
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.54626715591563,
      "learning_rate": 1.0704399551634271e-08,
      "loss": 0.557,
      "step": 21965
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.3111808969347045,
      "learning_rate": 1.0657208450174372e-08,
      "loss": 0.5661,
      "step": 21966
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.142644238461012,
      "learning_rate": 1.0610121490319814e-08,
      "loss": 0.5323,
      "step": 21967
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.3400509717002915,
      "learning_rate": 1.0563138673051476e-08,
      "loss": 0.5849,
      "step": 21968
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.912608167487643,
      "learning_rate": 1.0516259999351353e-08,
      "loss": 0.5418,
      "step": 21969
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.051678444441906,
      "learning_rate": 1.046948547019755e-08,
      "loss": 0.582,
      "step": 21970
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.554760681012585,
      "learning_rate": 1.0422815086566506e-08,
      "loss": 0.5642,
      "step": 21971
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.148481409068394,
      "learning_rate": 1.0376248849431891e-08,
      "loss": 0.6103,
      "step": 21972
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.337231408671628,
      "learning_rate": 1.0329786759766258e-08,
      "loss": 0.5275,
      "step": 21973
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.636411907614623,
      "learning_rate": 1.0283428818538833e-08,
      "loss": 0.5015,
      "step": 21974
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.037040493504931,
      "learning_rate": 1.0237175026717727e-08,
      "loss": 0.6455,
      "step": 21975
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.2340828448229,
      "learning_rate": 1.0191025385268282e-08,
      "loss": 0.6243,
      "step": 21976
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.1624675177344646,
      "learning_rate": 1.0144979895153618e-08,
      "loss": 0.6369,
      "step": 21977
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.80273814806708,
      "learning_rate": 1.0099038557335183e-08,
      "loss": 0.597,
      "step": 21978
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.2661584165257995,
      "learning_rate": 1.0053201372771104e-08,
      "loss": 0.5906,
      "step": 21979
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.975178625973742,
      "learning_rate": 1.0007468342418947e-08,
      "loss": 0.495,
      "step": 21980
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.438040678075386,
      "learning_rate": 9.961839467232948e-09,
      "loss": 0.5808,
      "step": 21981
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.903951127484068,
      "learning_rate": 9.916314748165123e-09,
      "loss": 0.6227,
      "step": 21982
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.732914267140765,
      "learning_rate": 9.870894186166935e-09,
      "loss": 0.5197,
      "step": 21983
    },
    {
      "epoch": 1.96,
      "grad_norm": 8.184619130243997,
      "learning_rate": 9.825577782184848e-09,
      "loss": 0.5359,
      "step": 21984
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.1911825675196495,
      "learning_rate": 9.780365537165881e-09,
      "loss": 0.5595,
      "step": 21985
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.297835262637355,
      "learning_rate": 9.735257452053725e-09,
      "loss": 0.5323,
      "step": 21986
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.073531597009415,
      "learning_rate": 9.690253527789296e-09,
      "loss": 0.5556,
      "step": 21987
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.783862439026037,
      "learning_rate": 9.645353765311839e-09,
      "loss": 0.5541,
      "step": 21988
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.0340979156831285,
      "learning_rate": 9.600558165559492e-09,
      "loss": 0.5949,
      "step": 21989
    },
    {
      "epoch": 1.96,
      "grad_norm": 8.154113051325519,
      "learning_rate": 9.555866729466511e-09,
      "loss": 0.5995,
      "step": 21990
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.2145683659446975,
      "learning_rate": 9.511279457966038e-09,
      "loss": 0.5701,
      "step": 21991
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.068402135801149,
      "learning_rate": 9.466796351988438e-09,
      "loss": 0.5793,
      "step": 21992
    },
    {
      "epoch": 1.96,
      "grad_norm": 9.010461106531999,
      "learning_rate": 9.422417412462415e-09,
      "loss": 0.562,
      "step": 21993
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.378639819411811,
      "learning_rate": 9.378142640314448e-09,
      "loss": 0.5336,
      "step": 21994
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.576083084200753,
      "learning_rate": 9.333972036468241e-09,
      "loss": 0.5058,
      "step": 21995
    },
    {
      "epoch": 1.96,
      "grad_norm": 8.795325550128819,
      "learning_rate": 9.289905601845839e-09,
      "loss": 0.6121,
      "step": 21996
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.92419746254786,
      "learning_rate": 9.245943337367614e-09,
      "loss": 0.5292,
      "step": 21997
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.132480552442316,
      "learning_rate": 9.20208524395061e-09,
      "loss": 0.5152,
      "step": 21998
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.116611416200535,
      "learning_rate": 9.158331322510205e-09,
      "loss": 0.5376,
      "step": 21999
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.531544254283831,
      "learning_rate": 9.114681573960116e-09,
      "loss": 0.5249,
      "step": 22000
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.343820613449625,
      "learning_rate": 9.071135999211278e-09,
      "loss": 0.6509,
      "step": 22001
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.372368469351335,
      "learning_rate": 9.02769459917241e-09,
      "loss": 0.5691,
      "step": 22002
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.22842472211101,
      "learning_rate": 8.984357374750008e-09,
      "loss": 0.5823,
      "step": 22003
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.400809834530073,
      "learning_rate": 8.94112432684946e-09,
      "loss": 0.581,
      "step": 22004
    },
    {
      "epoch": 1.96,
      "grad_norm": 9.135431674862053,
      "learning_rate": 8.897995456372266e-09,
      "loss": 0.5412,
      "step": 22005
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.377397032466092,
      "learning_rate": 8.854970764219372e-09,
      "loss": 0.5297,
      "step": 22006
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.4647485219523,
      "learning_rate": 8.812050251288951e-09,
      "loss": 0.5913,
      "step": 22007
    },
    {
      "epoch": 1.96,
      "grad_norm": 8.31079381063605,
      "learning_rate": 8.769233918475839e-09,
      "loss": 0.5965,
      "step": 22008
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.824730585814763,
      "learning_rate": 8.726521766674877e-09,
      "loss": 0.5571,
      "step": 22009
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.244154070236688,
      "learning_rate": 8.683913796777022e-09,
      "loss": 0.5675,
      "step": 22010
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.134743156787807,
      "learning_rate": 8.641410009671003e-09,
      "loss": 0.5498,
      "step": 22011
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.9586463174400155,
      "learning_rate": 8.599010406245002e-09,
      "loss": 0.5856,
      "step": 22012
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.701266580456617,
      "learning_rate": 8.556714987383863e-09,
      "loss": 0.5769,
      "step": 22013
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.40350960994866,
      "learning_rate": 8.514523753970217e-09,
      "loss": 0.5461,
      "step": 22014
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.804541448174119,
      "learning_rate": 8.472436706885023e-09,
      "loss": 0.5859,
      "step": 22015
    },
    {
      "epoch": 1.96,
      "grad_norm": 11.94933422203528,
      "learning_rate": 8.430453847005916e-09,
      "loss": 0.5091,
      "step": 22016
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.453164063403593,
      "learning_rate": 8.388575175209968e-09,
      "loss": 0.5248,
      "step": 22017
    },
    {
      "epoch": 1.96,
      "grad_norm": 8.573766905352766,
      "learning_rate": 8.346800692370926e-09,
      "loss": 0.6115,
      "step": 22018
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.393434245607433,
      "learning_rate": 8.305130399360873e-09,
      "loss": 0.5202,
      "step": 22019
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.704705040274676,
      "learning_rate": 8.263564297049664e-09,
      "loss": 0.5587,
      "step": 22020
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.742572134135562,
      "learning_rate": 8.222102386304942e-09,
      "loss": 0.5616,
      "step": 22021
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.6779874646595205,
      "learning_rate": 8.180744667992124e-09,
      "loss": 0.6313,
      "step": 22022
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.251860091288871,
      "learning_rate": 8.139491142973855e-09,
      "loss": 0.6073,
      "step": 22023
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.020825589095411,
      "learning_rate": 8.098341812112221e-09,
      "loss": 0.5919,
      "step": 22024
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.018372498063385,
      "learning_rate": 8.057296676265425e-09,
      "loss": 0.5513,
      "step": 22025
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.431005277798216,
      "learning_rate": 8.016355736290005e-09,
      "loss": 0.5429,
      "step": 22026
    },
    {
      "epoch": 1.97,
      "grad_norm": 8.686257589237979,
      "learning_rate": 7.975518993041386e-09,
      "loss": 0.5749,
      "step": 22027
    },
    {
      "epoch": 1.97,
      "grad_norm": 6.110633938157378,
      "learning_rate": 7.934786447371667e-09,
      "loss": 0.5277,
      "step": 22028
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.757301984026148,
      "learning_rate": 7.894158100130723e-09,
      "loss": 0.5717,
      "step": 22029
    },
    {
      "epoch": 1.97,
      "grad_norm": 8.792961602503867,
      "learning_rate": 7.853633952166206e-09,
      "loss": 0.5548,
      "step": 22030
    },
    {
      "epoch": 1.97,
      "grad_norm": 6.5129451004531775,
      "learning_rate": 7.813214004324665e-09,
      "loss": 0.6364,
      "step": 22031
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.265300075983106,
      "learning_rate": 7.772898257449868e-09,
      "loss": 0.6038,
      "step": 22032
    },
    {
      "epoch": 1.97,
      "grad_norm": 8.025821154123895,
      "learning_rate": 7.73268671238281e-09,
      "loss": 0.5812,
      "step": 22033
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.002264569700141,
      "learning_rate": 7.692579369963371e-09,
      "loss": 0.5503,
      "step": 22034
    },
    {
      "epoch": 1.97,
      "grad_norm": 10.310596925211497,
      "learning_rate": 7.652576231028109e-09,
      "loss": 0.5853,
      "step": 22035
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.557389622751604,
      "learning_rate": 7.612677296411907e-09,
      "loss": 0.5898,
      "step": 22036
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.516050771078672,
      "learning_rate": 7.572882566948548e-09,
      "loss": 0.5575,
      "step": 22037
    },
    {
      "epoch": 1.97,
      "grad_norm": 6.3501230225520215,
      "learning_rate": 7.533192043467918e-09,
      "loss": 0.5622,
      "step": 22038
    },
    {
      "epoch": 1.97,
      "grad_norm": 8.652017036242453,
      "learning_rate": 7.493605726798247e-09,
      "loss": 0.6065,
      "step": 22039
    },
    {
      "epoch": 1.97,
      "grad_norm": 6.16884416942004,
      "learning_rate": 7.454123617766652e-09,
      "loss": 0.5219,
      "step": 22040
    },
    {
      "epoch": 1.97,
      "grad_norm": 10.382283209383926,
      "learning_rate": 7.414745717196359e-09,
      "loss": 0.5503,
      "step": 22041
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.23880044193648,
      "learning_rate": 7.375472025909491e-09,
      "loss": 0.5181,
      "step": 22042
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.283413367735502,
      "learning_rate": 7.3363025447259485e-09,
      "loss": 0.5467,
      "step": 22043
    },
    {
      "epoch": 1.97,
      "grad_norm": 10.221160789992052,
      "learning_rate": 7.2972372744639645e-09,
      "loss": 0.6034,
      "step": 22044
    },
    {
      "epoch": 1.97,
      "grad_norm": 8.448855123323428,
      "learning_rate": 7.258276215937887e-09,
      "loss": 0.5427,
      "step": 22045
    },
    {
      "epoch": 1.97,
      "grad_norm": 9.09320517833088,
      "learning_rate": 7.2194193699615115e-09,
      "loss": 0.5904,
      "step": 22046
    },
    {
      "epoch": 1.97,
      "grad_norm": 6.342640159012774,
      "learning_rate": 7.180666737345854e-09,
      "loss": 0.5692,
      "step": 22047
    },
    {
      "epoch": 1.97,
      "grad_norm": 6.63727606114676,
      "learning_rate": 7.1420183188991575e-09,
      "loss": 0.5591,
      "step": 22048
    },
    {
      "epoch": 1.97,
      "grad_norm": 11.150524553316345,
      "learning_rate": 7.10347411542911e-09,
      "loss": 0.579,
      "step": 22049
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.518713834743456,
      "learning_rate": 7.065034127740067e-09,
      "loss": 0.5462,
      "step": 22050
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.693131706775559,
      "learning_rate": 7.02669835663361e-09,
      "loss": 0.5867,
      "step": 22051
    },
    {
      "epoch": 1.97,
      "grad_norm": 8.014706095167034,
      "learning_rate": 6.988466802910765e-09,
      "loss": 0.547,
      "step": 22052
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.595365422177443,
      "learning_rate": 6.950339467368672e-09,
      "loss": 0.5175,
      "step": 22053
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.431034081272035,
      "learning_rate": 6.912316350803916e-09,
      "loss": 0.6336,
      "step": 22054
    },
    {
      "epoch": 1.97,
      "grad_norm": 6.881191852867751,
      "learning_rate": 6.8743974540097516e-09,
      "loss": 0.5238,
      "step": 22055
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.789835769212451,
      "learning_rate": 6.836582777778322e-09,
      "loss": 0.4972,
      "step": 22056
    },
    {
      "epoch": 1.97,
      "grad_norm": 8.451533203167667,
      "learning_rate": 6.798872322897887e-09,
      "loss": 0.5873,
      "step": 22057
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.716605302533124,
      "learning_rate": 6.761266090156149e-09,
      "loss": 0.5064,
      "step": 22058
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.335588820362175,
      "learning_rate": 6.7237640803380356e-09,
      "loss": 0.5569,
      "step": 22059
    },
    {
      "epoch": 1.97,
      "grad_norm": 6.09852899925173,
      "learning_rate": 6.686366294226254e-09,
      "loss": 0.5977,
      "step": 22060
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.507497077362959,
      "learning_rate": 6.649072732601292e-09,
      "loss": 0.5771,
      "step": 22061
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.469035775836862,
      "learning_rate": 6.6118833962414145e-09,
      "loss": 0.5542,
      "step": 22062
    },
    {
      "epoch": 1.97,
      "grad_norm": 8.990233419545133,
      "learning_rate": 6.574798285923778e-09,
      "loss": 0.5836,
      "step": 22063
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.9064707162611185,
      "learning_rate": 6.537817402421098e-09,
      "loss": 0.5781,
      "step": 22064
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.75215925933674,
      "learning_rate": 6.500940746506646e-09,
      "loss": 0.5857,
      "step": 22065
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.675528301910791,
      "learning_rate": 6.4641683189492486e-09,
      "loss": 0.5337,
      "step": 22066
    },
    {
      "epoch": 1.97,
      "grad_norm": 9.639202209277736,
      "learning_rate": 6.427500120516628e-09,
      "loss": 0.5596,
      "step": 22067
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.9512503189924795,
      "learning_rate": 6.390936151974281e-09,
      "loss": 0.5554,
      "step": 22068
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.832156784313221,
      "learning_rate": 6.354476414085486e-09,
      "loss": 0.4934,
      "step": 22069
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.106877573812042,
      "learning_rate": 6.318120907610748e-09,
      "loss": 0.5489,
      "step": 22070
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.657840992272151,
      "learning_rate": 6.281869633309457e-09,
      "loss": 0.5655,
      "step": 22071
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.561311194892741,
      "learning_rate": 6.245722591937675e-09,
      "loss": 0.5917,
      "step": 22072
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.599502076491484,
      "learning_rate": 6.209679784250911e-09,
      "loss": 0.5175,
      "step": 22073
    },
    {
      "epoch": 1.97,
      "grad_norm": 8.502509330463846,
      "learning_rate": 6.173741211000783e-09,
      "loss": 0.5923,
      "step": 22074
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.378741671077124,
      "learning_rate": 6.1379068729372496e-09,
      "loss": 0.612,
      "step": 22075
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.302700102833208,
      "learning_rate": 6.102176770808599e-09,
      "loss": 0.5873,
      "step": 22076
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.65063508385373,
      "learning_rate": 6.066550905360347e-09,
      "loss": 0.565,
      "step": 22077
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.3096121766280975,
      "learning_rate": 6.031029277336897e-09,
      "loss": 0.5853,
      "step": 22078
    },
    {
      "epoch": 1.97,
      "grad_norm": 8.535624149628614,
      "learning_rate": 5.995611887478769e-09,
      "loss": 0.5703,
      "step": 22079
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.84459751838388,
      "learning_rate": 5.960298736525372e-09,
      "loss": 0.609,
      "step": 22080
    },
    {
      "epoch": 1.97,
      "grad_norm": 8.838705985100281,
      "learning_rate": 5.925089825214448e-09,
      "loss": 0.5645,
      "step": 22081
    },
    {
      "epoch": 1.97,
      "grad_norm": 9.004230324001808,
      "learning_rate": 5.8899851542798535e-09,
      "loss": 0.5338,
      "step": 22082
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.540797378063944,
      "learning_rate": 5.854984724455448e-09,
      "loss": 0.5237,
      "step": 22083
    },
    {
      "epoch": 1.97,
      "grad_norm": 8.179243630508909,
      "learning_rate": 5.820088536470647e-09,
      "loss": 0.5327,
      "step": 22084
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.722476560829915,
      "learning_rate": 5.785296591054868e-09,
      "loss": 0.5715,
      "step": 22085
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.807088307366163,
      "learning_rate": 5.750608888933639e-09,
      "loss": 0.5596,
      "step": 22086
    },
    {
      "epoch": 1.97,
      "grad_norm": 6.992228054125155,
      "learning_rate": 5.716025430831384e-09,
      "loss": 0.515,
      "step": 22087
    },
    {
      "epoch": 1.97,
      "grad_norm": 6.293570108518441,
      "learning_rate": 5.681546217469747e-09,
      "loss": 0.5733,
      "step": 22088
    },
    {
      "epoch": 1.97,
      "grad_norm": 6.43114049042933,
      "learning_rate": 5.6471712495687056e-09,
      "loss": 0.5876,
      "step": 22089
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.358056408875348,
      "learning_rate": 5.612900527845466e-09,
      "loss": 0.5827,
      "step": 22090
    },
    {
      "epoch": 1.97,
      "grad_norm": 6.728805194743041,
      "learning_rate": 5.5787340530150116e-09,
      "loss": 0.5631,
      "step": 22091
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.436269068743016,
      "learning_rate": 5.544671825791215e-09,
      "loss": 0.5314,
      "step": 22092
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.8127889538873685,
      "learning_rate": 5.510713846884619e-09,
      "loss": 0.565,
      "step": 22093
    },
    {
      "epoch": 1.97,
      "grad_norm": 6.4775666874461555,
      "learning_rate": 5.476860117004101e-09,
      "loss": 0.5529,
      "step": 22094
    },
    {
      "epoch": 1.97,
      "grad_norm": 8.962880985950262,
      "learning_rate": 5.443110636856319e-09,
      "loss": 0.5591,
      "step": 22095
    },
    {
      "epoch": 1.97,
      "grad_norm": 8.080982523960413,
      "learning_rate": 5.409465407146264e-09,
      "loss": 0.536,
      "step": 22096
    },
    {
      "epoch": 1.97,
      "grad_norm": 11.968851020415416,
      "learning_rate": 5.3759244285750415e-09,
      "loss": 0.5732,
      "step": 22097
    },
    {
      "epoch": 1.97,
      "grad_norm": 6.823497698329042,
      "learning_rate": 5.342487701843202e-09,
      "loss": 0.6448,
      "step": 22098
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.882155755839524,
      "learning_rate": 5.3091552276490765e-09,
      "loss": 0.5811,
      "step": 22099
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.888363314160799,
      "learning_rate": 5.2759270066882196e-09,
      "loss": 0.6211,
      "step": 22100
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.51614011410341,
      "learning_rate": 5.242803039653965e-09,
      "loss": 0.6267,
      "step": 22101
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.361870228347744,
      "learning_rate": 5.209783327237983e-09,
      "loss": 0.5357,
      "step": 22102
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.4147805167896,
      "learning_rate": 5.1768678701297205e-09,
      "loss": 0.5448,
      "step": 22103
    },
    {
      "epoch": 1.97,
      "grad_norm": 8.002976013383211,
      "learning_rate": 5.144056669015296e-09,
      "loss": 0.6133,
      "step": 22104
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.892810994609749,
      "learning_rate": 5.1113497245802725e-09,
      "loss": 0.5062,
      "step": 22105
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.005982352958882,
      "learning_rate": 5.078747037507437e-09,
      "loss": 0.5532,
      "step": 22106
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.4732680292932425,
      "learning_rate": 5.0462486084768005e-09,
      "loss": 0.5697,
      "step": 22107
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.242304588383606,
      "learning_rate": 5.013854438167265e-09,
      "loss": 0.5819,
      "step": 22108
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.463927517965537,
      "learning_rate": 4.9815645272544015e-09,
      "loss": 0.5824,
      "step": 22109
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.0761084158727865,
      "learning_rate": 4.9493788764126696e-09,
      "loss": 0.5349,
      "step": 22110
    },
    {
      "epoch": 1.97,
      "grad_norm": 6.601976587011446,
      "learning_rate": 4.917297486313199e-09,
      "loss": 0.6146,
      "step": 22111
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.2609533688160255,
      "learning_rate": 4.885320357626566e-09,
      "loss": 0.5715,
      "step": 22112
    },
    {
      "epoch": 1.97,
      "grad_norm": 6.749371350676115,
      "learning_rate": 4.853447491019458e-09,
      "loss": 0.6309,
      "step": 22113
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.68412372668891,
      "learning_rate": 4.82167888715801e-09,
      "loss": 0.5612,
      "step": 22114
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.322023494860424,
      "learning_rate": 4.790014546703914e-09,
      "loss": 0.6145,
      "step": 22115
    },
    {
      "epoch": 1.97,
      "grad_norm": 8.192507522074091,
      "learning_rate": 4.7584544703199734e-09,
      "loss": 0.5535,
      "step": 22116
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.857025100381078,
      "learning_rate": 4.726998658662885e-09,
      "loss": 0.5628,
      "step": 22117
    },
    {
      "epoch": 1.97,
      "grad_norm": 8.11183746949163,
      "learning_rate": 4.69564711239101e-09,
      "loss": 0.5653,
      "step": 22118
    },
    {
      "epoch": 1.97,
      "grad_norm": 6.578953261384625,
      "learning_rate": 4.664399832157718e-09,
      "loss": 0.5474,
      "step": 22119
    },
    {
      "epoch": 1.97,
      "grad_norm": 6.225527538521193,
      "learning_rate": 4.633256818615261e-09,
      "loss": 0.6185,
      "step": 22120
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.021118665588359,
      "learning_rate": 4.602218072414233e-09,
      "loss": 0.5916,
      "step": 22121
    },
    {
      "epoch": 1.97,
      "grad_norm": 9.745294281567059,
      "learning_rate": 4.571283594201892e-09,
      "loss": 0.5757,
      "step": 22122
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.584740850532303,
      "learning_rate": 4.5404533846243885e-09,
      "loss": 0.5733,
      "step": 22123
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.966059588790516,
      "learning_rate": 4.509727444325096e-09,
      "loss": 0.5046,
      "step": 22124
    },
    {
      "epoch": 1.97,
      "grad_norm": 8.809857586478492,
      "learning_rate": 4.4791057739451695e-09,
      "loss": 0.6573,
      "step": 22125
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.274036380878682,
      "learning_rate": 4.4485883741235415e-09,
      "loss": 0.5631,
      "step": 22126
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.187021233481863,
      "learning_rate": 4.4181752454980355e-09,
      "loss": 0.5951,
      "step": 22127
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.3683257144403855,
      "learning_rate": 4.387866388703144e-09,
      "loss": 0.5447,
      "step": 22128
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.332328173405535,
      "learning_rate": 4.357661804371138e-09,
      "loss": 0.5525,
      "step": 22129
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.956887469671657,
      "learning_rate": 4.327561493132626e-09,
      "loss": 0.569,
      "step": 22130
    },
    {
      "epoch": 1.97,
      "grad_norm": 6.044841952625606,
      "learning_rate": 4.2975654556154375e-09,
      "loss": 0.575,
      "step": 22131
    },
    {
      "epoch": 1.97,
      "grad_norm": 9.136452939140426,
      "learning_rate": 4.26767369244685e-09,
      "loss": 0.4979,
      "step": 22132
    },
    {
      "epoch": 1.97,
      "grad_norm": 9.204335569801762,
      "learning_rate": 4.237886204249697e-09,
      "loss": 0.6336,
      "step": 22133
    },
    {
      "epoch": 1.97,
      "grad_norm": 8.276441138100301,
      "learning_rate": 4.20820299164626e-09,
      "loss": 0.6158,
      "step": 22134
    },
    {
      "epoch": 1.97,
      "grad_norm": 8.602306666521141,
      "learning_rate": 4.178624055256597e-09,
      "loss": 0.5555,
      "step": 22135
    },
    {
      "epoch": 1.97,
      "grad_norm": 10.158657946212927,
      "learning_rate": 4.149149395696883e-09,
      "loss": 0.5609,
      "step": 22136
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.486343534985826,
      "learning_rate": 4.119779013582736e-09,
      "loss": 0.5071,
      "step": 22137
    },
    {
      "epoch": 1.97,
      "grad_norm": 6.259039308174903,
      "learning_rate": 4.09051290952811e-09,
      "loss": 0.5455,
      "step": 22138
    },
    {
      "epoch": 1.98,
      "grad_norm": 8.289409232534155,
      "learning_rate": 4.061351084142518e-09,
      "loss": 0.532,
      "step": 22139
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.824789652604377,
      "learning_rate": 4.03229353803547e-09,
      "loss": 0.5362,
      "step": 22140
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.01866749185576,
      "learning_rate": 4.003340271813705e-09,
      "loss": 0.5537,
      "step": 22141
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.451386480236745,
      "learning_rate": 3.974491286080628e-09,
      "loss": 0.5816,
      "step": 22142
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.065811389958894,
      "learning_rate": 3.945746581439092e-09,
      "loss": 0.5254,
      "step": 22143
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.432251834196403,
      "learning_rate": 3.91710615848917e-09,
      "loss": 0.4919,
      "step": 22144
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.18559907862879,
      "learning_rate": 3.888570017828164e-09,
      "loss": 0.5412,
      "step": 22145
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.4948210962509485,
      "learning_rate": 3.8601381600522625e-09,
      "loss": 0.5826,
      "step": 22146
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.821059562655254,
      "learning_rate": 3.831810585754325e-09,
      "loss": 0.4955,
      "step": 22147
    },
    {
      "epoch": 1.98,
      "grad_norm": 10.332970755238398,
      "learning_rate": 3.803587295526101e-09,
      "loss": 0.5782,
      "step": 22148
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.012364968000539,
      "learning_rate": 3.775468289956008e-09,
      "loss": 0.601,
      "step": 22149
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.69440640144949,
      "learning_rate": 3.7474535696319095e-09,
      "loss": 0.6077,
      "step": 22150
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.063399284310162,
      "learning_rate": 3.719543135138337e-09,
      "loss": 0.5714,
      "step": 22151
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.8253735070080825,
      "learning_rate": 3.6917369870576035e-09,
      "loss": 0.5786,
      "step": 22152
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.959584968020636,
      "learning_rate": 3.6640351259698004e-09,
      "loss": 0.5251,
      "step": 22153
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.273784351740838,
      "learning_rate": 3.6364375524533535e-09,
      "loss": 0.5567,
      "step": 22154
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.703030261907116,
      "learning_rate": 3.6089442670844687e-09,
      "loss": 0.5568,
      "step": 22155
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.087197639356992,
      "learning_rate": 3.5815552704371315e-09,
      "loss": 0.5561,
      "step": 22156
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.38210605886417,
      "learning_rate": 3.5542705630831064e-09,
      "loss": 0.5525,
      "step": 22157
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.531382011163246,
      "learning_rate": 3.527090145591383e-09,
      "loss": 0.6081,
      "step": 22158
    },
    {
      "epoch": 1.98,
      "grad_norm": 8.110667124787977,
      "learning_rate": 3.500014018529285e-09,
      "loss": 0.6429,
      "step": 22159
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.7490661180437765,
      "learning_rate": 3.4730421824630267e-09,
      "loss": 0.5447,
      "step": 22160
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.688546769864429,
      "learning_rate": 3.44617463795438e-09,
      "loss": 0.5698,
      "step": 22161
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.335784763102597,
      "learning_rate": 3.4194113855645637e-09,
      "loss": 0.5577,
      "step": 22162
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.273497132442035,
      "learning_rate": 3.3927524258520194e-09,
      "loss": 0.5738,
      "step": 22163
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.223947758277503,
      "learning_rate": 3.366197759373524e-09,
      "loss": 0.5463,
      "step": 22164
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.709585087684073,
      "learning_rate": 3.339747386683079e-09,
      "loss": 0.54,
      "step": 22165
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.267381325808152,
      "learning_rate": 3.3134013083330197e-09,
      "loss": 0.5497,
      "step": 22166
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.213439007137727,
      "learning_rate": 3.2871595248734624e-09,
      "loss": 0.5837,
      "step": 22167
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.386012550700586,
      "learning_rate": 3.2610220368511914e-09,
      "loss": 0.5637,
      "step": 22168
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.365631344937806,
      "learning_rate": 3.2349888448129918e-09,
      "loss": 0.5867,
      "step": 22169
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.606292904383465,
      "learning_rate": 3.2090599493012074e-09,
      "loss": 0.5983,
      "step": 22170
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.3191073737328125,
      "learning_rate": 3.1832353508581825e-09,
      "loss": 0.5538,
      "step": 22171
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.09358528335141,
      "learning_rate": 3.15751505002182e-09,
      "loss": 0.5479,
      "step": 22172
    },
    {
      "epoch": 1.98,
      "grad_norm": 8.077366036389796,
      "learning_rate": 3.1318990473294677e-09,
      "loss": 0.5714,
      "step": 22173
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.759599016384823,
      "learning_rate": 3.106387343315698e-09,
      "loss": 0.5531,
      "step": 22174
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.371223976891939,
      "learning_rate": 3.0809799385128624e-09,
      "loss": 0.5615,
      "step": 22175
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.712771022138083,
      "learning_rate": 3.0556768334516486e-09,
      "loss": 0.6137,
      "step": 22176
    },
    {
      "epoch": 1.98,
      "grad_norm": 4.740216674499526,
      "learning_rate": 3.030478028660522e-09,
      "loss": 0.5822,
      "step": 22177
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.900454153541293,
      "learning_rate": 3.005383524664618e-09,
      "loss": 0.5791,
      "step": 22178
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.277377046920171,
      "learning_rate": 2.980393321987962e-09,
      "loss": 0.5466,
      "step": 22179
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.926660300689095,
      "learning_rate": 2.9555074211529144e-09,
      "loss": 0.5306,
      "step": 22180
    },
    {
      "epoch": 1.98,
      "grad_norm": 9.450147703442616,
      "learning_rate": 2.930725822677949e-09,
      "loss": 0.602,
      "step": 22181
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.23072795635163,
      "learning_rate": 2.9060485270804294e-09,
      "loss": 0.4974,
      "step": 22182
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.139245127956328,
      "learning_rate": 2.8814755348760548e-09,
      "loss": 0.5216,
      "step": 22183
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.072932794699601,
      "learning_rate": 2.857006846577193e-09,
      "loss": 0.5472,
      "step": 22184
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.014172715628268,
      "learning_rate": 2.832642462694546e-09,
      "loss": 0.5403,
      "step": 22185
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.755293346563621,
      "learning_rate": 2.8083823837371516e-09,
      "loss": 0.5922,
      "step": 22186
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.210067357817848,
      "learning_rate": 2.7842266102112714e-09,
      "loss": 0.5268,
      "step": 22187
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.433993340813551,
      "learning_rate": 2.7601751426203917e-09,
      "loss": 0.5418,
      "step": 22188
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.739319235087379,
      "learning_rate": 2.736227981467443e-09,
      "loss": 0.6033,
      "step": 22189
    },
    {
      "epoch": 1.98,
      "grad_norm": 11.828829244397937,
      "learning_rate": 2.712385127252026e-09,
      "loss": 0.5959,
      "step": 22190
    },
    {
      "epoch": 1.98,
      "grad_norm": 8.080743749123702,
      "learning_rate": 2.6886465804715214e-09,
      "loss": 0.5779,
      "step": 22191
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.436903819573825,
      "learning_rate": 2.6650123416216423e-09,
      "loss": 0.5763,
      "step": 22192
    },
    {
      "epoch": 1.98,
      "grad_norm": 4.911249957389032,
      "learning_rate": 2.6414824111958836e-09,
      "loss": 0.6041,
      "step": 22193
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.232144292057686,
      "learning_rate": 2.618056789684964e-09,
      "loss": 0.6618,
      "step": 22194
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.413317533312572,
      "learning_rate": 2.5947354775779364e-09,
      "loss": 0.5561,
      "step": 22195
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.6933992529764295,
      "learning_rate": 2.571518475361634e-09,
      "loss": 0.5807,
      "step": 22196
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.5778530256922085,
      "learning_rate": 2.548405783520669e-09,
      "loss": 0.5852,
      "step": 22197
    },
    {
      "epoch": 1.98,
      "grad_norm": 9.504704421050342,
      "learning_rate": 2.525397402537988e-09,
      "loss": 0.5824,
      "step": 22198
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.472714847819385,
      "learning_rate": 2.5024933328926525e-09,
      "loss": 0.6358,
      "step": 22199
    },
    {
      "epoch": 1.98,
      "grad_norm": 8.686526010731642,
      "learning_rate": 2.479693575064279e-09,
      "loss": 0.4889,
      "step": 22200
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.660111327322217,
      "learning_rate": 2.4569981295269328e-09,
      "loss": 0.4986,
      "step": 22201
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.1897590493941355,
      "learning_rate": 2.4344069967557893e-09,
      "loss": 0.5649,
      "step": 22202
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.165602658702933,
      "learning_rate": 2.4119201772221377e-09,
      "loss": 0.5355,
      "step": 22203
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.628992049957699,
      "learning_rate": 2.389537671395048e-09,
      "loss": 0.574,
      "step": 22204
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.829832212205508,
      "learning_rate": 2.3672594797413684e-09,
      "loss": 0.5225,
      "step": 22205
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.515512038807537,
      "learning_rate": 2.3450856027268378e-09,
      "loss": 0.552,
      "step": 22206
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.585175524009155,
      "learning_rate": 2.3230160408138637e-09,
      "loss": 0.544,
      "step": 22207
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.345445328426475,
      "learning_rate": 2.3010507944637437e-09,
      "loss": 0.6015,
      "step": 22208
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.495279014794338,
      "learning_rate": 2.2791898641338906e-09,
      "loss": 0.6049,
      "step": 22209
    },
    {
      "epoch": 1.98,
      "grad_norm": 12.291813328485503,
      "learning_rate": 2.2574332502811603e-09,
      "loss": 0.6349,
      "step": 22210
    },
    {
      "epoch": 1.98,
      "grad_norm": 8.08497310401628,
      "learning_rate": 2.2357809533596344e-09,
      "loss": 0.5677,
      "step": 22211
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.919938005389356,
      "learning_rate": 2.214232973821173e-09,
      "loss": 0.5685,
      "step": 22212
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.343612501917077,
      "learning_rate": 2.1927893121159725e-09,
      "loss": 0.5938,
      "step": 22213
    },
    {
      "epoch": 1.98,
      "grad_norm": 10.344741235101608,
      "learning_rate": 2.1714499686908974e-09,
      "loss": 0.5364,
      "step": 22214
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.825223256582493,
      "learning_rate": 2.1502149439917017e-09,
      "loss": 0.5892,
      "step": 22215
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.764974202400133,
      "learning_rate": 2.1290842384619202e-09,
      "loss": 0.5035,
      "step": 22216
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.768045893454325,
      "learning_rate": 2.108057852542311e-09,
      "loss": 0.5486,
      "step": 22217
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.972623124734009,
      "learning_rate": 2.087135786671968e-09,
      "loss": 0.5835,
      "step": 22218
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.97417744155347,
      "learning_rate": 2.0663180412872075e-09,
      "loss": 0.5661,
      "step": 22219
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.641611624202848,
      "learning_rate": 2.045604616822683e-09,
      "loss": 0.5063,
      "step": 22220
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.628411831617069,
      "learning_rate": 2.0249955137108257e-09,
      "loss": 0.5359,
      "step": 22221
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.043586395512921,
      "learning_rate": 2.0044907323824027e-09,
      "loss": 0.5438,
      "step": 22222
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.945288377132015,
      "learning_rate": 1.9840902732642943e-09,
      "loss": 0.6092,
      "step": 22223
    },
    {
      "epoch": 1.98,
      "grad_norm": 10.037462558735797,
      "learning_rate": 1.9637941367828263e-09,
      "loss": 0.5533,
      "step": 22224
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.857549463318732,
      "learning_rate": 1.9436023233615485e-09,
      "loss": 0.5999,
      "step": 22225
    },
    {
      "epoch": 1.98,
      "grad_norm": 8.21846382219361,
      "learning_rate": 1.9235148334223464e-09,
      "loss": 0.5613,
      "step": 22226
    },
    {
      "epoch": 1.98,
      "grad_norm": 8.293212826670509,
      "learning_rate": 1.9035316673837734e-09,
      "loss": 0.5662,
      "step": 22227
    },
    {
      "epoch": 1.98,
      "grad_norm": 8.136819790584159,
      "learning_rate": 1.883652825663829e-09,
      "loss": 0.5767,
      "step": 22228
    },
    {
      "epoch": 1.98,
      "grad_norm": 8.574258718161198,
      "learning_rate": 1.8638783086766253e-09,
      "loss": 0.6195,
      "step": 22229
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.783385237031454,
      "learning_rate": 1.8442081168351667e-09,
      "loss": 0.573,
      "step": 22230
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.935179416352692,
      "learning_rate": 1.8246422505502349e-09,
      "loss": 0.5704,
      "step": 22231
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.68243050718667,
      "learning_rate": 1.805180710229837e-09,
      "loss": 0.6041,
      "step": 22232
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.324448294348646,
      "learning_rate": 1.7858234962808696e-09,
      "loss": 0.5946,
      "step": 22233
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.498361189339322,
      "learning_rate": 1.766570609106899e-09,
      "loss": 0.5711,
      "step": 22234
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.648677367647626,
      "learning_rate": 1.7474220491092708e-09,
      "loss": 0.6099,
      "step": 22235
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.7046150412222145,
      "learning_rate": 1.7283778166887755e-09,
      "loss": 0.585,
      "step": 22236
    },
    {
      "epoch": 1.98,
      "grad_norm": 11.710284260365597,
      "learning_rate": 1.7094379122423177e-09,
      "loss": 0.6785,
      "step": 22237
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.999304942109115,
      "learning_rate": 1.690602336165137e-09,
      "loss": 0.5459,
      "step": 22238
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.5749067549380085,
      "learning_rate": 1.671871088850252e-09,
      "loss": 0.5015,
      "step": 22239
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.963760518397808,
      "learning_rate": 1.6532441706890168e-09,
      "loss": 0.5222,
      "step": 22240
    },
    {
      "epoch": 1.98,
      "grad_norm": 4.886729681365275,
      "learning_rate": 1.634721582070564e-09,
      "loss": 0.5023,
      "step": 22241
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.901001368099837,
      "learning_rate": 1.6163033233806969e-09,
      "loss": 0.5121,
      "step": 22242
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.280376982148588,
      "learning_rate": 1.597989395004107e-09,
      "loss": 0.5578,
      "step": 22243
    },
    {
      "epoch": 1.98,
      "grad_norm": 4.962712680704794,
      "learning_rate": 1.5797797973232665e-09,
      "loss": 0.5569,
      "step": 22244
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.467670250219082,
      "learning_rate": 1.5616745307184266e-09,
      "loss": 0.5509,
      "step": 22245
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.9459056192842885,
      "learning_rate": 1.5436735955670634e-09,
      "loss": 0.5772,
      "step": 22246
    },
    {
      "epoch": 1.98,
      "grad_norm": 8.20946625054735,
      "learning_rate": 1.5257769922449871e-09,
      "loss": 0.5482,
      "step": 22247
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.089134057213708,
      "learning_rate": 1.5079847211263432e-09,
      "loss": 0.5544,
      "step": 22248
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.841388483297915,
      "learning_rate": 1.4902967825813909e-09,
      "loss": 0.567,
      "step": 22249
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.479539293737445,
      "learning_rate": 1.4727131769803892e-09,
      "loss": 0.5478,
      "step": 22250
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.927137508014311,
      "learning_rate": 1.4552339046891572e-09,
      "loss": 0.5597,
      "step": 22251
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.126243134725354,
      "learning_rate": 1.4378589660740683e-09,
      "loss": 0.5536,
      "step": 22252
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.278325038199034,
      "learning_rate": 1.4205883614970551e-09,
      "loss": 0.5958,
      "step": 22253
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.141252265722902,
      "learning_rate": 1.4034220913178297e-09,
      "loss": 0.512,
      "step": 22254
    },
    {
      "epoch": 1.99,
      "grad_norm": 4.45279660630959,
      "learning_rate": 1.3863601558961047e-09,
      "loss": 0.5593,
      "step": 22255
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.557801985810364,
      "learning_rate": 1.3694025555871516e-09,
      "loss": 0.5869,
      "step": 22256
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.04424509941981,
      "learning_rate": 1.352549290745131e-09,
      "loss": 0.5611,
      "step": 22257
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.332266655060409,
      "learning_rate": 1.3358003617214289e-09,
      "loss": 0.6042,
      "step": 22258
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.257661228968681,
      "learning_rate": 1.3191557688663204e-09,
      "loss": 0.5523,
      "step": 22259
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.012924967951881,
      "learning_rate": 1.3026155125267504e-09,
      "loss": 0.5298,
      "step": 22260
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.399939545851276,
      "learning_rate": 1.2861795930485533e-09,
      "loss": 0.5823,
      "step": 22261
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.053853205971235,
      "learning_rate": 1.2698480107736777e-09,
      "loss": 0.5411,
      "step": 22262
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.250067634357125,
      "learning_rate": 1.2536207660440724e-09,
      "loss": 0.5044,
      "step": 22263
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.710928004119726,
      "learning_rate": 1.2374978591983556e-09,
      "loss": 0.5704,
      "step": 22264
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.205464468927915,
      "learning_rate": 1.2214792905723694e-09,
      "loss": 0.6067,
      "step": 22265
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.9075908969721755,
      "learning_rate": 1.2055650605008462e-09,
      "loss": 0.4714,
      "step": 22266
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.1937000282384655,
      "learning_rate": 1.1897551693162978e-09,
      "loss": 0.5307,
      "step": 22267
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.376717580199231,
      "learning_rate": 1.1740496173479054e-09,
      "loss": 0.6077,
      "step": 22268
    },
    {
      "epoch": 1.99,
      "grad_norm": 5.317375885253524,
      "learning_rate": 1.158448404924295e-09,
      "loss": 0.5225,
      "step": 22269
    },
    {
      "epoch": 1.99,
      "grad_norm": 5.042049362703318,
      "learning_rate": 1.142951532370762e-09,
      "loss": 0.5424,
      "step": 22270
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.83602617519204,
      "learning_rate": 1.127559000010936e-09,
      "loss": 0.5723,
      "step": 22271
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.802615294239427,
      "learning_rate": 1.112270808166227e-09,
      "loss": 0.5449,
      "step": 22272
    },
    {
      "epoch": 1.99,
      "grad_norm": 9.202207906290852,
      "learning_rate": 1.0970869571547138e-09,
      "loss": 0.5837,
      "step": 22273
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.265828290557696,
      "learning_rate": 1.0820074472944753e-09,
      "loss": 0.5396,
      "step": 22274
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.881601205138442,
      "learning_rate": 1.0670322789002597e-09,
      "loss": 0.5168,
      "step": 22275
    },
    {
      "epoch": 1.99,
      "grad_norm": 5.931135708019771,
      "learning_rate": 1.0521614522840395e-09,
      "loss": 0.6316,
      "step": 22276
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.242427372973868,
      "learning_rate": 1.0373949677561223e-09,
      "loss": 0.55,
      "step": 22277
    },
    {
      "epoch": 1.99,
      "grad_norm": 8.920528474313231,
      "learning_rate": 1.0227328256251501e-09,
      "loss": 0.553,
      "step": 22278
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.186142473633845,
      "learning_rate": 1.008175026196989e-09,
      "loss": 0.5513,
      "step": 22279
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.788405926784339,
      "learning_rate": 9.937215697758406e-10,
      "loss": 0.5029,
      "step": 22280
    },
    {
      "epoch": 1.99,
      "grad_norm": 10.79056993594378,
      "learning_rate": 9.793724566631301e-10,
      "loss": 0.5983,
      "step": 22281
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.791232439571525,
      "learning_rate": 9.651276871580628e-10,
      "loss": 0.5864,
      "step": 22282
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.065739833620824,
      "learning_rate": 9.509872615587334e-10,
      "loss": 0.567,
      "step": 22283
    },
    {
      "epoch": 1.99,
      "grad_norm": 4.932181675816728,
      "learning_rate": 9.369511801593512e-10,
      "loss": 0.6063,
      "step": 22284
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.715989144499419,
      "learning_rate": 9.230194432535699e-10,
      "loss": 0.532,
      "step": 22285
    },
    {
      "epoch": 1.99,
      "grad_norm": 5.982465027337448,
      "learning_rate": 9.091920511317131e-10,
      "loss": 0.5802,
      "step": 22286
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.671407215443881,
      "learning_rate": 8.954690040829939e-10,
      "loss": 0.5336,
      "step": 22287
    },
    {
      "epoch": 1.99,
      "grad_norm": 5.492344413048058,
      "learning_rate": 8.818503023932945e-10,
      "loss": 0.519,
      "step": 22288
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.405840338861232,
      "learning_rate": 8.683359463473873e-10,
      "loss": 0.5873,
      "step": 22289
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.450353987258691,
      "learning_rate": 8.549259362267138e-10,
      "loss": 0.5332,
      "step": 22290
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.714702304148322,
      "learning_rate": 8.416202723116051e-10,
      "loss": 0.5497,
      "step": 22291
    },
    {
      "epoch": 1.99,
      "grad_norm": 5.442051105345391,
      "learning_rate": 8.284189548796173e-10,
      "loss": 0.5782,
      "step": 22292
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.370225491270838,
      "learning_rate": 8.153219842066406e-10,
      "loss": 0.5651,
      "step": 22293
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.828418350659571,
      "learning_rate": 8.0232936056579e-10,
      "loss": 0.5253,
      "step": 22294
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.670642525389698,
      "learning_rate": 7.894410842279599e-10,
      "loss": 0.5811,
      "step": 22295
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.278670727596919,
      "learning_rate": 7.766571554623792e-10,
      "loss": 0.5846,
      "step": 22296
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.605596781918805,
      "learning_rate": 7.639775745366118e-10,
      "loss": 0.5681,
      "step": 22297
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.166738500166348,
      "learning_rate": 7.514023417143356e-10,
      "loss": 0.5779,
      "step": 22298
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.902794973528408,
      "learning_rate": 7.389314572581185e-10,
      "loss": 0.5002,
      "step": 22299
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.597344245559136,
      "learning_rate": 7.265649214283077e-10,
      "loss": 0.5255,
      "step": 22300
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.435326122372914,
      "learning_rate": 7.143027344841403e-10,
      "loss": 0.6062,
      "step": 22301
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.349023318259661,
      "learning_rate": 7.021448966798572e-10,
      "loss": 0.5843,
      "step": 22302
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.318057371022918,
      "learning_rate": 6.900914082702548e-10,
      "loss": 0.5766,
      "step": 22303
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.999874908401292,
      "learning_rate": 6.781422695067985e-10,
      "loss": 0.5532,
      "step": 22304
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.528025393665032,
      "learning_rate": 6.662974806387335e-10,
      "loss": 0.5757,
      "step": 22305
    },
    {
      "epoch": 1.99,
      "grad_norm": 5.985371905465533,
      "learning_rate": 6.545570419130842e-10,
      "loss": 0.5148,
      "step": 22306
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.385778714980955,
      "learning_rate": 6.429209535757652e-10,
      "loss": 0.5717,
      "step": 22307
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.213825595346466,
      "learning_rate": 6.313892158688051e-10,
      "loss": 0.5196,
      "step": 22308
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.309890260913602,
      "learning_rate": 6.199618290331222e-10,
      "loss": 0.582,
      "step": 22309
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.980942061758254,
      "learning_rate": 6.086387933068594e-10,
      "loss": 0.5758,
      "step": 22310
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.956185146405218,
      "learning_rate": 5.974201089276043e-10,
      "loss": 0.5031,
      "step": 22311
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.6842002281884945,
      "learning_rate": 5.863057761279489e-10,
      "loss": 0.5921,
      "step": 22312
    },
    {
      "epoch": 1.99,
      "grad_norm": 8.798489501381205,
      "learning_rate": 5.752957951410398e-10,
      "loss": 0.5907,
      "step": 22313
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.339016468784381,
      "learning_rate": 5.643901661961382e-10,
      "loss": 0.591,
      "step": 22314
    },
    {
      "epoch": 1.99,
      "grad_norm": 8.114318059968355,
      "learning_rate": 5.535888895213948e-10,
      "loss": 0.6101,
      "step": 22315
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.21171990190858,
      "learning_rate": 5.428919653410747e-10,
      "loss": 0.5087,
      "step": 22316
    },
    {
      "epoch": 1.99,
      "grad_norm": 5.770672321506481,
      "learning_rate": 5.322993938799981e-10,
      "loss": 0.5194,
      "step": 22317
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.0048908511743155,
      "learning_rate": 5.218111753579891e-10,
      "loss": 0.6396,
      "step": 22318
    },
    {
      "epoch": 1.99,
      "grad_norm": 9.026122080328829,
      "learning_rate": 5.114273099948719e-10,
      "loss": 0.5753,
      "step": 22319
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.976888163200731,
      "learning_rate": 5.0114779800714e-10,
      "loss": 0.5295,
      "step": 22320
    },
    {
      "epoch": 1.99,
      "grad_norm": 8.39176762603536,
      "learning_rate": 4.909726396090664e-10,
      "loss": 0.5576,
      "step": 22321
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.770527919307246,
      "learning_rate": 4.809018350132588e-10,
      "loss": 0.5481,
      "step": 22322
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.746287697568165,
      "learning_rate": 4.709353844295494e-10,
      "loss": 0.6207,
      "step": 22323
    },
    {
      "epoch": 1.99,
      "grad_norm": 5.5358198566113135,
      "learning_rate": 4.610732880666602e-10,
      "loss": 0.5902,
      "step": 22324
    },
    {
      "epoch": 1.99,
      "grad_norm": 8.052025380772362,
      "learning_rate": 4.513155461299823e-10,
      "loss": 0.5657,
      "step": 22325
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.276108048712554,
      "learning_rate": 4.416621588226866e-10,
      "loss": 0.5547,
      "step": 22326
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.0606784068956925,
      "learning_rate": 4.321131263473888e-10,
      "loss": 0.5261,
      "step": 22327
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.849666064705663,
      "learning_rate": 4.2266844890281877e-10,
      "loss": 0.6201,
      "step": 22328
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.120413400383429,
      "learning_rate": 4.133281266865963e-10,
      "loss": 0.5803,
      "step": 22329
    },
    {
      "epoch": 1.99,
      "grad_norm": 4.420417292552385,
      "learning_rate": 4.040921598924552e-10,
      "loss": 0.5109,
      "step": 22330
    },
    {
      "epoch": 1.99,
      "grad_norm": 5.719843020374216,
      "learning_rate": 3.9496054871412946e-10,
      "loss": 0.56,
      "step": 22331
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.582330392989247,
      "learning_rate": 3.8593329334257746e-10,
      "loss": 0.6524,
      "step": 22332
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.91985564661322,
      "learning_rate": 3.7701039396542683e-10,
      "loss": 0.6175,
      "step": 22333
    },
    {
      "epoch": 1.99,
      "grad_norm": 5.779922628269397,
      "learning_rate": 3.6819185076919506e-10,
      "loss": 0.592,
      "step": 22334
    },
    {
      "epoch": 1.99,
      "grad_norm": 8.293256949136987,
      "learning_rate": 3.5947766393762405e-10,
      "loss": 0.5785,
      "step": 22335
    },
    {
      "epoch": 1.99,
      "grad_norm": 9.05589642566504,
      "learning_rate": 3.5086783365334554e-10,
      "loss": 0.5371,
      "step": 22336
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.9634224402748215,
      "learning_rate": 3.4236236009510536e-10,
      "loss": 0.502,
      "step": 22337
    },
    {
      "epoch": 1.99,
      "grad_norm": 8.884155631538732,
      "learning_rate": 3.339612434416495e-10,
      "loss": 0.585,
      "step": 22338
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.009075989522639,
      "learning_rate": 3.2566448386728287e-10,
      "loss": 0.5285,
      "step": 22339
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.387055988646861,
      "learning_rate": 3.1747208154575546e-10,
      "loss": 0.4872,
      "step": 22340
    },
    {
      "epoch": 1.99,
      "grad_norm": 9.108964611415423,
      "learning_rate": 3.093840366480416e-10,
      "loss": 0.6097,
      "step": 22341
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.148411506460795,
      "learning_rate": 3.014003493423401e-10,
      "loss": 0.5504,
      "step": 22342
    },
    {
      "epoch": 1.99,
      "grad_norm": 8.599427651303984,
      "learning_rate": 2.935210197957394e-10,
      "loss": 0.5592,
      "step": 22343
    },
    {
      "epoch": 1.99,
      "grad_norm": 5.5391121830018495,
      "learning_rate": 2.857460481731078e-10,
      "loss": 0.5769,
      "step": 22344
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.326672883376822,
      "learning_rate": 2.7807543463598265e-10,
      "loss": 0.6348,
      "step": 22345
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.550443305800905,
      "learning_rate": 2.7050917934534625e-10,
      "loss": 0.5963,
      "step": 22346
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.606377003823034,
      "learning_rate": 2.6304728245829524e-10,
      "loss": 0.4975,
      "step": 22347
    },
    {
      "epoch": 1.99,
      "grad_norm": 10.035261457528115,
      "learning_rate": 2.5568974413081596e-10,
      "loss": 0.5611,
      "step": 22348
    },
    {
      "epoch": 1.99,
      "grad_norm": 8.283573293595728,
      "learning_rate": 2.484365645166742e-10,
      "loss": 0.5977,
      "step": 22349
    },
    {
      "epoch": 1.99,
      "grad_norm": 4.655153255367234,
      "learning_rate": 2.412877437668604e-10,
      "loss": 0.5872,
      "step": 22350
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.809887664082592,
      "learning_rate": 2.342432820312546e-10,
      "loss": 0.5875,
      "step": 22351
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.091482825061737,
      "learning_rate": 2.2730317945640622e-10,
      "loss": 0.5707,
      "step": 22352
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.684219507773038,
      "learning_rate": 2.2046743618719946e-10,
      "loss": 0.5636,
      "step": 22353
    },
    {
      "epoch": 1.99,
      "grad_norm": 5.346664814013866,
      "learning_rate": 2.1373605236685302e-10,
      "loss": 0.5844,
      "step": 22354
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.058181236891613,
      "learning_rate": 2.071090281346999e-10,
      "loss": 0.533,
      "step": 22355
    },
    {
      "epoch": 1.99,
      "grad_norm": 10.791933835727232,
      "learning_rate": 2.0058636363007312e-10,
      "loss": 0.5618,
      "step": 22356
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.141190109263506,
      "learning_rate": 1.9416805898897496e-10,
      "loss": 0.5381,
      "step": 22357
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.284056051986168,
      "learning_rate": 1.878541143451873e-10,
      "loss": 0.5887,
      "step": 22358
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.5037453415255415,
      "learning_rate": 1.816445298308267e-10,
      "loss": 0.5679,
      "step": 22359
    },
    {
      "epoch": 1.99,
      "grad_norm": 5.61125549703276,
      "learning_rate": 1.7553930557467903e-10,
      "loss": 0.5106,
      "step": 22360
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.461146545693215,
      "learning_rate": 1.6953844170497503e-10,
      "loss": 0.5776,
      "step": 22361
    },
    {
      "epoch": 1.99,
      "grad_norm": 9.008774372504956,
      "learning_rate": 1.636419383466148e-10,
      "loss": 0.6448,
      "step": 22362
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.877827375155695,
      "learning_rate": 1.5784979562283308e-10,
      "loss": 0.5749,
      "step": 22363
    },
    {
      "epoch": 2.0,
      "grad_norm": 7.221856642801281,
      "learning_rate": 1.5216201365464422e-10,
      "loss": 0.5199,
      "step": 22364
    },
    {
      "epoch": 2.0,
      "grad_norm": 7.256023992441402,
      "learning_rate": 1.4657859256028695e-10,
      "loss": 0.5573,
      "step": 22365
    },
    {
      "epoch": 2.0,
      "grad_norm": 8.71890946017384,
      "learning_rate": 1.410995324568898e-10,
      "loss": 0.6086,
      "step": 22366
    },
    {
      "epoch": 2.0,
      "grad_norm": 6.0002069364918444,
      "learning_rate": 1.357248334582506e-10,
      "loss": 0.573,
      "step": 22367
    },
    {
      "epoch": 2.0,
      "grad_norm": 6.678948267962412,
      "learning_rate": 1.3045449567705704e-10,
      "loss": 0.6239,
      "step": 22368
    },
    {
      "epoch": 2.0,
      "grad_norm": 9.297306240328377,
      "learning_rate": 1.2528851922322117e-10,
      "loss": 0.4796,
      "step": 22369
    },
    {
      "epoch": 2.0,
      "grad_norm": 6.295557165961889,
      "learning_rate": 1.202269042038795e-10,
      "loss": 0.5714,
      "step": 22370
    },
    {
      "epoch": 2.0,
      "grad_norm": 7.8568982674559775,
      "learning_rate": 1.1526965072561347e-10,
      "loss": 0.5684,
      "step": 22371
    },
    {
      "epoch": 2.0,
      "grad_norm": 5.644673644257143,
      "learning_rate": 1.104167588916738e-10,
      "loss": 0.4973,
      "step": 22372
    },
    {
      "epoch": 2.0,
      "grad_norm": 6.967873596400147,
      "learning_rate": 1.056682288030908e-10,
      "loss": 0.5184,
      "step": 22373
    },
    {
      "epoch": 2.0,
      "grad_norm": 8.820118735951032,
      "learning_rate": 1.0102406055922942e-10,
      "loss": 0.5802,
      "step": 22374
    },
    {
      "epoch": 2.0,
      "grad_norm": 6.92771371377611,
      "learning_rate": 9.648425425667907e-11,
      "loss": 0.5705,
      "step": 22375
    },
    {
      "epoch": 2.0,
      "grad_norm": 7.284148147656708,
      "learning_rate": 9.20488099903638e-11,
      "loss": 0.5394,
      "step": 22376
    },
    {
      "epoch": 2.0,
      "grad_norm": 8.102470022045958,
      "learning_rate": 8.771772785354238e-11,
      "loss": 0.5259,
      "step": 22377
    },
    {
      "epoch": 2.0,
      "grad_norm": 5.404947199674627,
      "learning_rate": 8.349100793503262e-11,
      "loss": 0.4875,
      "step": 22378
    },
    {
      "epoch": 2.0,
      "grad_norm": 5.317725236868967,
      "learning_rate": 7.93686503247626e-11,
      "loss": 0.6658,
      "step": 22379
    },
    {
      "epoch": 2.0,
      "grad_norm": 7.041970215851293,
      "learning_rate": 7.535065510766437e-11,
      "loss": 0.5965,
      "step": 22380
    },
    {
      "epoch": 2.0,
      "grad_norm": 7.099479591242325,
      "learning_rate": 7.143702236811489e-11,
      "loss": 0.5327,
      "step": 22381
    },
    {
      "epoch": 2.0,
      "grad_norm": 5.766664031887119,
      "learning_rate": 6.762775218771556e-11,
      "loss": 0.5871,
      "step": 22382
    },
    {
      "epoch": 2.0,
      "grad_norm": 7.430349653011086,
      "learning_rate": 6.39228446458473e-11,
      "loss": 0.5581,
      "step": 22383
    },
    {
      "epoch": 2.0,
      "grad_norm": 6.355883248614428,
      "learning_rate": 6.032229981967063e-11,
      "loss": 0.5807,
      "step": 22384
    },
    {
      "epoch": 2.0,
      "grad_norm": 6.012090487980204,
      "learning_rate": 5.6826117784680725e-11,
      "loss": 0.6077,
      "step": 22385
    },
    {
      "epoch": 2.0,
      "grad_norm": 6.2942664284429135,
      "learning_rate": 5.343429861415228e-11,
      "loss": 0.5158,
      "step": 22386
    },
    {
      "epoch": 2.0,
      "grad_norm": 6.935991773709688,
      "learning_rate": 5.014684237802936e-11,
      "loss": 0.6201,
      "step": 22387
    },
    {
      "epoch": 2.0,
      "grad_norm": 11.412210861086258,
      "learning_rate": 4.6963749145700896e-11,
      "loss": 0.5305,
      "step": 22388
    },
    {
      "epoch": 2.0,
      "grad_norm": 5.9518147321920045,
      "learning_rate": 4.388501898322517e-11,
      "loss": 0.5432,
      "step": 22389
    },
    {
      "epoch": 2.0,
      "grad_norm": 8.402078870438466,
      "learning_rate": 4.0910651954995105e-11,
      "loss": 0.5094,
      "step": 22390
    },
    {
      "epoch": 2.0,
      "grad_norm": 5.893806644745177,
      "learning_rate": 3.804064812262809e-11,
      "loss": 0.563,
      "step": 22391
    },
    {
      "epoch": 2.0,
      "grad_norm": 6.367471351414142,
      "learning_rate": 3.527500754663127e-11,
      "loss": 0.5852,
      "step": 22392
    },
    {
      "epoch": 2.0,
      "grad_norm": 8.015465006262144,
      "learning_rate": 3.2613730284736245e-11,
      "loss": 0.5592,
      "step": 22393
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.705097361994824,
      "learning_rate": 3.005681639245417e-11,
      "loss": 0.5928,
      "step": 22394
    },
    {
      "epoch": 2.0,
      "grad_norm": 8.775501331283172,
      "learning_rate": 2.7604265922520636e-11,
      "loss": 0.564,
      "step": 22395
    },
    {
      "epoch": 2.0,
      "grad_norm": 7.073701683811733,
      "learning_rate": 2.5256078926561012e-11,
      "loss": 0.614,
      "step": 22396
    },
    {
      "epoch": 2.0,
      "grad_norm": 6.815057691720495,
      "learning_rate": 2.3012255453980226e-11,
      "loss": 0.5899,
      "step": 22397
    },
    {
      "epoch": 2.0,
      "grad_norm": 9.389336987418718,
      "learning_rate": 2.0872795551407643e-11,
      "loss": 0.5649,
      "step": 22398
    },
    {
      "epoch": 2.0,
      "grad_norm": 6.328919634755875,
      "learning_rate": 1.8837699262697075e-11,
      "loss": 0.5788,
      "step": 22399
    },
    {
      "epoch": 2.0,
      "grad_norm": 6.483283933560163,
      "learning_rate": 1.6906966631702327e-11,
      "loss": 0.5674,
      "step": 22400
    },
    {
      "epoch": 2.0,
      "grad_norm": 7.691343781323884,
      "learning_rate": 1.5080597697281206e-11,
      "loss": 0.5634,
      "step": 22401
    },
    {
      "epoch": 2.0,
      "grad_norm": 6.101840360271895,
      "learning_rate": 1.3358592498846634e-11,
      "loss": 0.5501,
      "step": 22402
    },
    {
      "epoch": 2.0,
      "grad_norm": 7.053449336032607,
      "learning_rate": 1.1740951071370632e-11,
      "loss": 0.5978,
      "step": 22403
    },
    {
      "epoch": 2.0,
      "grad_norm": 6.891303774073509,
      "learning_rate": 1.0227673449270114e-11,
      "loss": 0.5771,
      "step": 22404
    },
    {
      "epoch": 2.0,
      "grad_norm": 7.17420441629277,
      "learning_rate": 8.818759663631327e-12,
      "loss": 0.5934,
      "step": 22405
    },
    {
      "epoch": 2.0,
      "grad_norm": 5.358482228732947,
      "learning_rate": 7.51420974443029e-12,
      "loss": 0.4902,
      "step": 22406
    },
    {
      "epoch": 2.0,
      "grad_norm": 9.069459840481255,
      "learning_rate": 6.314023718312356e-12,
      "loss": 0.4829,
      "step": 22407
    },
    {
      "epoch": 2.0,
      "grad_norm": 7.216518776848019,
      "learning_rate": 5.218201610812657e-12,
      "loss": 0.6042,
      "step": 22408
    },
    {
      "epoch": 2.0,
      "grad_norm": 7.80248427150608,
      "learning_rate": 4.22674344413565e-12,
      "loss": 0.5938,
      "step": 22409
    },
    {
      "epoch": 2.0,
      "grad_norm": 8.167349009318386,
      "learning_rate": 3.339649239930687e-12,
      "loss": 0.5479,
      "step": 22410
    },
    {
      "epoch": 2.0,
      "grad_norm": 6.506836064545649,
      "learning_rate": 2.5569190159613343e-12,
      "loss": 0.5285,
      "step": 22411
    },
    {
      "epoch": 2.0,
      "grad_norm": 6.0534463018330795,
      "learning_rate": 1.8785527888809384e-12,
      "loss": 0.5861,
      "step": 22412
    },
    {
      "epoch": 2.0,
      "grad_norm": 5.022926734891331,
      "learning_rate": 1.3045505731223984e-12,
      "loss": 0.5834,
      "step": 22413
    },
    {
      "epoch": 2.0,
      "grad_norm": 5.976569029159346,
      "learning_rate": 8.349123797879444e-13,
      "loss": 0.5953,
      "step": 22414
    },
    {
      "epoch": 2.0,
      "grad_norm": 6.229401353629282,
      "learning_rate": 4.696382194246951e-13,
      "loss": 0.561,
      "step": 22415
    },
    {
      "epoch": 2.0,
      "grad_norm": 10.141267721370903,
      "learning_rate": 2.0872809924910032e-13,
      "loss": 0.5424,
      "step": 22416
    },
    {
      "epoch": 2.0,
      "grad_norm": 6.615625312178967,
      "learning_rate": 5.218202536738659e-14,
      "loss": 0.5515,
      "step": 22417
    },
    {
      "epoch": 2.0,
      "grad_norm": 6.552558411704704,
      "learning_rate": 0.0,
      "loss": 0.5172,
      "step": 22418
    },
    {
      "epoch": 2.0,
      "step": 22418,
      "total_flos": 3.696053433321652e+19,
      "train_loss": 0.7196700681658665,
      "train_runtime": 405070.881,
      "train_samples_per_second": 7.084,
      "train_steps_per_second": 0.055
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 22418,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 1000,
  "total_flos": 3.696053433321652e+19,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}