{
  "best_metric": 10.496330261230469,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.022584834283778443,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00011292417141889222,
      "grad_norm": 1.3611441850662231,
      "learning_rate": 1.004e-05,
      "loss": 22.2276,
      "step": 1
    },
    {
      "epoch": 0.00011292417141889222,
      "eval_loss": 11.130960464477539,
      "eval_runtime": 30.54,
      "eval_samples_per_second": 122.102,
      "eval_steps_per_second": 30.55,
      "step": 1
    },
    {
      "epoch": 0.00022584834283778443,
      "grad_norm": 2.113603353500366,
      "learning_rate": 2.008e-05,
      "loss": 22.2533,
      "step": 2
    },
    {
      "epoch": 0.0003387725142566766,
      "grad_norm": 1.7150770425796509,
      "learning_rate": 3.012e-05,
      "loss": 22.2395,
      "step": 3
    },
    {
      "epoch": 0.00045169668567556886,
      "grad_norm": 2.2730438709259033,
      "learning_rate": 4.016e-05,
      "loss": 22.281,
      "step": 4
    },
    {
      "epoch": 0.0005646208570944611,
      "grad_norm": 2.303051710128784,
      "learning_rate": 5.02e-05,
      "loss": 22.2371,
      "step": 5
    },
    {
      "epoch": 0.0006775450285133532,
      "grad_norm": 2.2910170555114746,
      "learning_rate": 6.024e-05,
      "loss": 22.2374,
      "step": 6
    },
    {
      "epoch": 0.0007904691999322455,
      "grad_norm": 1.702985167503357,
      "learning_rate": 7.028e-05,
      "loss": 22.2062,
      "step": 7
    },
    {
      "epoch": 0.0009033933713511377,
      "grad_norm": 1.9652076959609985,
      "learning_rate": 8.032e-05,
      "loss": 22.2247,
      "step": 8
    },
    {
      "epoch": 0.00101631754277003,
      "grad_norm": 1.9771640300750732,
      "learning_rate": 9.036000000000001e-05,
      "loss": 22.1742,
      "step": 9
    },
    {
      "epoch": 0.0011292417141889222,
      "grad_norm": 1.9720170497894287,
      "learning_rate": 0.0001004,
      "loss": 22.2208,
      "step": 10
    },
    {
      "epoch": 0.0012421658856078144,
      "grad_norm": 1.9375696182250977,
      "learning_rate": 9.987157894736842e-05,
      "loss": 22.1704,
      "step": 11
    },
    {
      "epoch": 0.0013550900570267065,
      "grad_norm": 2.2164692878723145,
      "learning_rate": 9.934315789473684e-05,
      "loss": 22.2258,
      "step": 12
    },
    {
      "epoch": 0.0014680142284455987,
      "grad_norm": 2.3525643348693848,
      "learning_rate": 9.881473684210525e-05,
      "loss": 22.1919,
      "step": 13
    },
    {
      "epoch": 0.001580938399864491,
      "grad_norm": 2.076218843460083,
      "learning_rate": 9.828631578947369e-05,
      "loss": 22.096,
      "step": 14
    },
    {
      "epoch": 0.0016938625712833832,
      "grad_norm": 2.2245523929595947,
      "learning_rate": 9.77578947368421e-05,
      "loss": 22.0613,
      "step": 15
    },
    {
      "epoch": 0.0018067867427022754,
      "grad_norm": 2.361453056335449,
      "learning_rate": 9.722947368421052e-05,
      "loss": 22.0267,
      "step": 16
    },
    {
      "epoch": 0.0019197109141211677,
      "grad_norm": 2.272428274154663,
      "learning_rate": 9.670105263157895e-05,
      "loss": 22.0847,
      "step": 17
    },
    {
      "epoch": 0.00203263508554006,
      "grad_norm": 2.3448495864868164,
      "learning_rate": 9.617263157894737e-05,
      "loss": 22.0495,
      "step": 18
    },
    {
      "epoch": 0.002145559256958952,
      "grad_norm": 2.503242015838623,
      "learning_rate": 9.564421052631579e-05,
      "loss": 22.0463,
      "step": 19
    },
    {
      "epoch": 0.0022584834283778444,
      "grad_norm": 2.3206300735473633,
      "learning_rate": 9.511578947368421e-05,
      "loss": 22.0854,
      "step": 20
    },
    {
      "epoch": 0.0023714075997967364,
      "grad_norm": 2.207120180130005,
      "learning_rate": 9.458736842105264e-05,
      "loss": 22.0101,
      "step": 21
    },
    {
      "epoch": 0.002484331771215629,
      "grad_norm": 2.4101004600524902,
      "learning_rate": 9.405894736842106e-05,
      "loss": 21.985,
      "step": 22
    },
    {
      "epoch": 0.002597255942634521,
      "grad_norm": 2.3305113315582275,
      "learning_rate": 9.353052631578947e-05,
      "loss": 21.9712,
      "step": 23
    },
    {
      "epoch": 0.002710180114053413,
      "grad_norm": 2.0429296493530273,
      "learning_rate": 9.300210526315789e-05,
      "loss": 21.9259,
      "step": 24
    },
    {
      "epoch": 0.0028231042854723054,
      "grad_norm": 2.037775993347168,
      "learning_rate": 9.247368421052631e-05,
      "loss": 21.8812,
      "step": 25
    },
    {
      "epoch": 0.0029360284568911974,
      "grad_norm": 2.0648608207702637,
      "learning_rate": 9.194526315789473e-05,
      "loss": 21.9428,
      "step": 26
    },
    {
      "epoch": 0.00304895262831009,
      "grad_norm": 2.152031421661377,
      "learning_rate": 9.141684210526316e-05,
      "loss": 21.7899,
      "step": 27
    },
    {
      "epoch": 0.003161876799728982,
      "grad_norm": 2.4206976890563965,
      "learning_rate": 9.088842105263158e-05,
      "loss": 21.8019,
      "step": 28
    },
    {
      "epoch": 0.0032748009711478744,
      "grad_norm": 2.530184030532837,
      "learning_rate": 9.036000000000001e-05,
      "loss": 21.7797,
      "step": 29
    },
    {
      "epoch": 0.0033877251425667664,
      "grad_norm": 2.2082767486572266,
      "learning_rate": 8.983157894736843e-05,
      "loss": 21.8125,
      "step": 30
    },
    {
      "epoch": 0.0035006493139856584,
      "grad_norm": 2.187060832977295,
      "learning_rate": 8.930315789473684e-05,
      "loss": 21.6609,
      "step": 31
    },
    {
      "epoch": 0.003613573485404551,
      "grad_norm": 2.0385313034057617,
      "learning_rate": 8.877473684210526e-05,
      "loss": 21.7933,
      "step": 32
    },
    {
      "epoch": 0.003726497656823443,
      "grad_norm": 1.9034841060638428,
      "learning_rate": 8.824631578947368e-05,
      "loss": 21.7488,
      "step": 33
    },
    {
      "epoch": 0.0038394218282423354,
      "grad_norm": 2.1895627975463867,
      "learning_rate": 8.771789473684211e-05,
      "loss": 21.642,
      "step": 34
    },
    {
      "epoch": 0.003952345999661227,
      "grad_norm": 2.4400441646575928,
      "learning_rate": 8.718947368421053e-05,
      "loss": 21.5303,
      "step": 35
    },
    {
      "epoch": 0.00406527017108012,
      "grad_norm": 2.0822086334228516,
      "learning_rate": 8.666105263157895e-05,
      "loss": 21.7251,
      "step": 36
    },
    {
      "epoch": 0.004178194342499012,
      "grad_norm": 1.9033600091934204,
      "learning_rate": 8.613263157894737e-05,
      "loss": 21.7548,
      "step": 37
    },
    {
      "epoch": 0.004291118513917904,
      "grad_norm": 2.0422472953796387,
      "learning_rate": 8.560421052631578e-05,
      "loss": 21.5744,
      "step": 38
    },
    {
      "epoch": 0.004404042685336796,
      "grad_norm": 2.0038177967071533,
      "learning_rate": 8.50757894736842e-05,
      "loss": 21.5973,
      "step": 39
    },
    {
      "epoch": 0.004516966856755689,
      "grad_norm": 1.8099231719970703,
      "learning_rate": 8.454736842105263e-05,
      "loss": 21.6777,
      "step": 40
    },
    {
      "epoch": 0.00462989102817458,
      "grad_norm": 1.8867391347885132,
      "learning_rate": 8.401894736842106e-05,
      "loss": 21.5857,
      "step": 41
    },
    {
      "epoch": 0.004742815199593473,
      "grad_norm": 2.133280038833618,
      "learning_rate": 8.349052631578948e-05,
      "loss": 21.5483,
      "step": 42
    },
    {
      "epoch": 0.004855739371012365,
      "grad_norm": 1.978566288948059,
      "learning_rate": 8.29621052631579e-05,
      "loss": 21.4628,
      "step": 43
    },
    {
      "epoch": 0.004968663542431258,
      "grad_norm": 2.010080099105835,
      "learning_rate": 8.243368421052632e-05,
      "loss": 21.4539,
      "step": 44
    },
    {
      "epoch": 0.005081587713850149,
      "grad_norm": 2.023642063140869,
      "learning_rate": 8.190526315789474e-05,
      "loss": 21.3488,
      "step": 45
    },
    {
      "epoch": 0.005194511885269042,
      "grad_norm": 2.155339002609253,
      "learning_rate": 8.137684210526315e-05,
      "loss": 21.2927,
      "step": 46
    },
    {
      "epoch": 0.005307436056687934,
      "grad_norm": 1.9990131855010986,
      "learning_rate": 8.084842105263157e-05,
      "loss": 21.2376,
      "step": 47
    },
    {
      "epoch": 0.005420360228106826,
      "grad_norm": 1.915147304534912,
      "learning_rate": 8.032e-05,
      "loss": 21.4163,
      "step": 48
    },
    {
      "epoch": 0.005533284399525718,
      "grad_norm": 2.0212175846099854,
      "learning_rate": 7.979157894736842e-05,
      "loss": 21.348,
      "step": 49
    },
    {
      "epoch": 0.005646208570944611,
      "grad_norm": 1.9949758052825928,
      "learning_rate": 7.926315789473684e-05,
      "loss": 21.4609,
      "step": 50
    },
    {
      "epoch": 0.005646208570944611,
      "eval_loss": 10.721461296081543,
      "eval_runtime": 30.1165,
      "eval_samples_per_second": 123.819,
      "eval_steps_per_second": 30.98,
      "step": 50
    },
    {
      "epoch": 0.005759132742363503,
      "grad_norm": 1.6324496269226074,
      "learning_rate": 7.873473684210526e-05,
      "loss": 21.7549,
      "step": 51
    },
    {
      "epoch": 0.005872056913782395,
      "grad_norm": 2.2080705165863037,
      "learning_rate": 7.820631578947369e-05,
      "loss": 21.6339,
      "step": 52
    },
    {
      "epoch": 0.005984981085201287,
      "grad_norm": 2.8481478691101074,
      "learning_rate": 7.76778947368421e-05,
      "loss": 21.6641,
      "step": 53
    },
    {
      "epoch": 0.00609790525662018,
      "grad_norm": 1.7890151739120483,
      "learning_rate": 7.714947368421052e-05,
      "loss": 21.6296,
      "step": 54
    },
    {
      "epoch": 0.006210829428039071,
      "grad_norm": 2.223647117614746,
      "learning_rate": 7.662105263157896e-05,
      "loss": 21.7086,
      "step": 55
    },
    {
      "epoch": 0.006323753599457964,
      "grad_norm": 2.120537519454956,
      "learning_rate": 7.609263157894737e-05,
      "loss": 21.5439,
      "step": 56
    },
    {
      "epoch": 0.006436677770876856,
      "grad_norm": 2.205350399017334,
      "learning_rate": 7.556421052631579e-05,
      "loss": 21.5929,
      "step": 57
    },
    {
      "epoch": 0.006549601942295749,
      "grad_norm": 2.2164692878723145,
      "learning_rate": 7.503578947368421e-05,
      "loss": 21.6118,
      "step": 58
    },
    {
      "epoch": 0.00666252611371464,
      "grad_norm": 1.7517759799957275,
      "learning_rate": 7.450736842105263e-05,
      "loss": 21.4765,
      "step": 59
    },
    {
      "epoch": 0.006775450285133533,
      "grad_norm": 2.0426087379455566,
      "learning_rate": 7.397894736842105e-05,
      "loss": 21.3901,
      "step": 60
    },
    {
      "epoch": 0.006888374456552425,
      "grad_norm": 1.6669620275497437,
      "learning_rate": 7.345052631578948e-05,
      "loss": 21.4112,
      "step": 61
    },
    {
      "epoch": 0.007001298627971317,
      "grad_norm": 1.3309324979782104,
      "learning_rate": 7.29221052631579e-05,
      "loss": 21.4535,
      "step": 62
    },
    {
      "epoch": 0.007114222799390209,
      "grad_norm": 1.7843608856201172,
      "learning_rate": 7.239368421052631e-05,
      "loss": 21.3178,
      "step": 63
    },
    {
      "epoch": 0.007227146970809102,
      "grad_norm": 1.73214590549469,
      "learning_rate": 7.186526315789474e-05,
      "loss": 21.3145,
      "step": 64
    },
    {
      "epoch": 0.007340071142227994,
      "grad_norm": 1.5890671014785767,
      "learning_rate": 7.133684210526316e-05,
      "loss": 21.3931,
      "step": 65
    },
    {
      "epoch": 0.007452995313646886,
      "grad_norm": 1.6645324230194092,
      "learning_rate": 7.080842105263158e-05,
      "loss": 21.4537,
      "step": 66
    },
    {
      "epoch": 0.007565919485065778,
      "grad_norm": 2.022207260131836,
      "learning_rate": 7.028e-05,
      "loss": 21.2167,
      "step": 67
    },
    {
      "epoch": 0.007678843656484671,
      "grad_norm": 1.2683295011520386,
      "learning_rate": 6.975157894736843e-05,
      "loss": 21.4912,
      "step": 68
    },
    {
      "epoch": 0.007791767827903562,
      "grad_norm": 1.4991168975830078,
      "learning_rate": 6.922315789473685e-05,
      "loss": 21.4556,
      "step": 69
    },
    {
      "epoch": 0.007904691999322455,
      "grad_norm": 1.737343430519104,
      "learning_rate": 6.869473684210527e-05,
      "loss": 21.3404,
      "step": 70
    },
    {
      "epoch": 0.008017616170741346,
      "grad_norm": 1.889341115951538,
      "learning_rate": 6.816631578947368e-05,
      "loss": 21.235,
      "step": 71
    },
    {
      "epoch": 0.00813054034216024,
      "grad_norm": 1.8898837566375732,
      "learning_rate": 6.76378947368421e-05,
      "loss": 21.1924,
      "step": 72
    },
    {
      "epoch": 0.008243464513579131,
      "grad_norm": 1.6848102807998657,
      "learning_rate": 6.710947368421052e-05,
      "loss": 21.1746,
      "step": 73
    },
    {
      "epoch": 0.008356388684998025,
      "grad_norm": 1.798210859298706,
      "learning_rate": 6.658105263157894e-05,
      "loss": 21.2693,
      "step": 74
    },
    {
      "epoch": 0.008469312856416916,
      "grad_norm": 1.9311983585357666,
      "learning_rate": 6.605263157894737e-05,
      "loss": 21.1865,
      "step": 75
    },
    {
      "epoch": 0.008582237027835808,
      "grad_norm": 1.6010700464248657,
      "learning_rate": 6.55242105263158e-05,
      "loss": 21.3468,
      "step": 76
    },
    {
      "epoch": 0.008695161199254701,
      "grad_norm": 1.6437805891036987,
      "learning_rate": 6.499578947368422e-05,
      "loss": 21.1226,
      "step": 77
    },
    {
      "epoch": 0.008808085370673593,
      "grad_norm": 1.63422691822052,
      "learning_rate": 6.446736842105264e-05,
      "loss": 21.2402,
      "step": 78
    },
    {
      "epoch": 0.008921009542092484,
      "grad_norm": 1.6775226593017578,
      "learning_rate": 6.393894736842105e-05,
      "loss": 21.1452,
      "step": 79
    },
    {
      "epoch": 0.009033933713511378,
      "grad_norm": 1.6660118103027344,
      "learning_rate": 6.341052631578947e-05,
      "loss": 21.21,
      "step": 80
    },
    {
      "epoch": 0.00914685788493027,
      "grad_norm": 1.8009381294250488,
      "learning_rate": 6.288210526315789e-05,
      "loss": 21.1508,
      "step": 81
    },
    {
      "epoch": 0.00925978205634916,
      "grad_norm": 1.7096575498580933,
      "learning_rate": 6.235368421052632e-05,
      "loss": 21.2174,
      "step": 82
    },
    {
      "epoch": 0.009372706227768054,
      "grad_norm": 1.5187997817993164,
      "learning_rate": 6.182526315789474e-05,
      "loss": 21.241,
      "step": 83
    },
    {
      "epoch": 0.009485630399186946,
      "grad_norm": 1.7166595458984375,
      "learning_rate": 6.129684210526316e-05,
      "loss": 21.15,
      "step": 84
    },
    {
      "epoch": 0.009598554570605837,
      "grad_norm": 1.6249490976333618,
      "learning_rate": 6.076842105263158e-05,
      "loss": 21.0705,
      "step": 85
    },
    {
      "epoch": 0.00971147874202473,
      "grad_norm": 1.7954708337783813,
      "learning_rate": 6.024e-05,
      "loss": 21.034,
      "step": 86
    },
    {
      "epoch": 0.009824402913443622,
      "grad_norm": 1.746080994606018,
      "learning_rate": 5.971157894736842e-05,
      "loss": 21.1158,
      "step": 87
    },
    {
      "epoch": 0.009937327084862516,
      "grad_norm": 1.6499457359313965,
      "learning_rate": 5.9183157894736835e-05,
      "loss": 21.0956,
      "step": 88
    },
    {
      "epoch": 0.010050251256281407,
      "grad_norm": 1.7913421392440796,
      "learning_rate": 5.8654736842105267e-05,
      "loss": 21.4237,
      "step": 89
    },
    {
      "epoch": 0.010163175427700299,
      "grad_norm": 1.8211878538131714,
      "learning_rate": 5.8126315789473684e-05,
      "loss": 21.1766,
      "step": 90
    },
    {
      "epoch": 0.010276099599119192,
      "grad_norm": 1.7606409788131714,
      "learning_rate": 5.759789473684211e-05,
      "loss": 21.0793,
      "step": 91
    },
    {
      "epoch": 0.010389023770538084,
      "grad_norm": 1.8371472358703613,
      "learning_rate": 5.706947368421053e-05,
      "loss": 21.1063,
      "step": 92
    },
    {
      "epoch": 0.010501947941956975,
      "grad_norm": 1.793402075767517,
      "learning_rate": 5.6541052631578945e-05,
      "loss": 20.9959,
      "step": 93
    },
    {
      "epoch": 0.010614872113375869,
      "grad_norm": 1.8255559206008911,
      "learning_rate": 5.601263157894736e-05,
      "loss": 21.1812,
      "step": 94
    },
    {
      "epoch": 0.01072779628479476,
      "grad_norm": 1.7802122831344604,
      "learning_rate": 5.5484210526315794e-05,
      "loss": 21.2362,
      "step": 95
    },
    {
      "epoch": 0.010840720456213652,
      "grad_norm": 1.7444708347320557,
      "learning_rate": 5.495578947368421e-05,
      "loss": 21.187,
      "step": 96
    },
    {
      "epoch": 0.010953644627632545,
      "grad_norm": 1.6838579177856445,
      "learning_rate": 5.442736842105264e-05,
      "loss": 20.9765,
      "step": 97
    },
    {
      "epoch": 0.011066568799051437,
      "grad_norm": 1.8817654848098755,
      "learning_rate": 5.3898947368421055e-05,
      "loss": 21.1037,
      "step": 98
    },
    {
      "epoch": 0.01117949297047033,
      "grad_norm": 1.7965583801269531,
      "learning_rate": 5.337052631578947e-05,
      "loss": 21.0701,
      "step": 99
    },
    {
      "epoch": 0.011292417141889222,
      "grad_norm": 1.8796077966690063,
      "learning_rate": 5.284210526315789e-05,
      "loss": 21.0952,
      "step": 100
    },
    {
      "epoch": 0.011292417141889222,
      "eval_loss": 10.57370376586914,
      "eval_runtime": 30.1691,
      "eval_samples_per_second": 123.603,
      "eval_steps_per_second": 30.926,
      "step": 100
    },
    {
      "epoch": 0.011405341313308113,
      "grad_norm": 1.898417353630066,
      "learning_rate": 5.231368421052631e-05,
      "loss": 21.5383,
      "step": 101
    },
    {
      "epoch": 0.011518265484727007,
      "grad_norm": 1.7448859214782715,
      "learning_rate": 5.178526315789474e-05,
      "loss": 21.5164,
      "step": 102
    },
    {
      "epoch": 0.011631189656145898,
      "grad_norm": 2.15998911857605,
      "learning_rate": 5.1256842105263165e-05,
      "loss": 21.4158,
      "step": 103
    },
    {
      "epoch": 0.01174411382756479,
      "grad_norm": 1.7991974353790283,
      "learning_rate": 5.072842105263158e-05,
      "loss": 21.544,
      "step": 104
    },
    {
      "epoch": 0.011857037998983683,
      "grad_norm": 1.6782306432724,
      "learning_rate": 5.02e-05,
      "loss": 21.4899,
      "step": 105
    },
    {
      "epoch": 0.011969962170402575,
      "grad_norm": 1.6934300661087036,
      "learning_rate": 4.967157894736842e-05,
      "loss": 21.4828,
      "step": 106
    },
    {
      "epoch": 0.012082886341821466,
      "grad_norm": 1.7622517347335815,
      "learning_rate": 4.914315789473684e-05,
      "loss": 21.5518,
      "step": 107
    },
    {
      "epoch": 0.01219581051324036,
      "grad_norm": 2.0620055198669434,
      "learning_rate": 4.861473684210526e-05,
      "loss": 21.4972,
      "step": 108
    },
    {
      "epoch": 0.012308734684659251,
      "grad_norm": 1.7865617275238037,
      "learning_rate": 4.8086315789473686e-05,
      "loss": 21.1913,
      "step": 109
    },
    {
      "epoch": 0.012421658856078143,
      "grad_norm": 1.6678588390350342,
      "learning_rate": 4.7557894736842104e-05,
      "loss": 21.33,
      "step": 110
    },
    {
      "epoch": 0.012534583027497036,
      "grad_norm": 1.5663342475891113,
      "learning_rate": 4.702947368421053e-05,
      "loss": 21.2791,
      "step": 111
    },
    {
      "epoch": 0.012647507198915928,
      "grad_norm": 1.3401545286178589,
      "learning_rate": 4.6501052631578946e-05,
      "loss": 21.2329,
      "step": 112
    },
    {
      "epoch": 0.012760431370334821,
      "grad_norm": 1.5291308164596558,
      "learning_rate": 4.5972631578947364e-05,
      "loss": 21.0561,
      "step": 113
    },
    {
      "epoch": 0.012873355541753713,
      "grad_norm": 1.5034675598144531,
      "learning_rate": 4.544421052631579e-05,
      "loss": 21.0625,
      "step": 114
    },
    {
      "epoch": 0.012986279713172604,
      "grad_norm": 1.502207636833191,
      "learning_rate": 4.4915789473684213e-05,
      "loss": 20.9603,
      "step": 115
    },
    {
      "epoch": 0.013099203884591498,
      "grad_norm": 1.3683857917785645,
      "learning_rate": 4.438736842105263e-05,
      "loss": 21.2751,
      "step": 116
    },
    {
      "epoch": 0.013212128056010389,
      "grad_norm": 1.3831427097320557,
      "learning_rate": 4.3858947368421056e-05,
      "loss": 21.2536,
      "step": 117
    },
    {
      "epoch": 0.01332505222742928,
      "grad_norm": 1.7075899839401245,
      "learning_rate": 4.3330526315789474e-05,
      "loss": 20.9327,
      "step": 118
    },
    {
      "epoch": 0.013437976398848174,
      "grad_norm": 1.4572417736053467,
      "learning_rate": 4.280210526315789e-05,
      "loss": 21.0398,
      "step": 119
    },
    {
      "epoch": 0.013550900570267066,
      "grad_norm": 1.8132352828979492,
      "learning_rate": 4.2273684210526317e-05,
      "loss": 20.8115,
      "step": 120
    },
    {
      "epoch": 0.013663824741685957,
      "grad_norm": 1.7252711057662964,
      "learning_rate": 4.174526315789474e-05,
      "loss": 20.8063,
      "step": 121
    },
    {
      "epoch": 0.01377674891310485,
      "grad_norm": 1.995081901550293,
      "learning_rate": 4.121684210526316e-05,
      "loss": 20.6842,
      "step": 122
    },
    {
      "epoch": 0.013889673084523742,
      "grad_norm": 1.5869985818862915,
      "learning_rate": 4.068842105263158e-05,
      "loss": 21.1246,
      "step": 123
    },
    {
      "epoch": 0.014002597255942634,
      "grad_norm": 1.448175072669983,
      "learning_rate": 4.016e-05,
      "loss": 21.1664,
      "step": 124
    },
    {
      "epoch": 0.014115521427361527,
      "grad_norm": 1.4581042528152466,
      "learning_rate": 3.963157894736842e-05,
      "loss": 21.1033,
      "step": 125
    },
    {
      "epoch": 0.014228445598780419,
      "grad_norm": 1.358923077583313,
      "learning_rate": 3.9103157894736844e-05,
      "loss": 21.054,
      "step": 126
    },
    {
      "epoch": 0.014341369770199312,
      "grad_norm": 1.5759377479553223,
      "learning_rate": 3.857473684210526e-05,
      "loss": 21.0667,
      "step": 127
    },
    {
      "epoch": 0.014454293941618204,
      "grad_norm": 1.616433024406433,
      "learning_rate": 3.804631578947369e-05,
      "loss": 20.9253,
      "step": 128
    },
    {
      "epoch": 0.014567218113037095,
      "grad_norm": 1.6013528108596802,
      "learning_rate": 3.7517894736842105e-05,
      "loss": 21.0409,
      "step": 129
    },
    {
      "epoch": 0.014680142284455988,
      "grad_norm": 1.7452136278152466,
      "learning_rate": 3.698947368421052e-05,
      "loss": 21.2808,
      "step": 130
    },
    {
      "epoch": 0.01479306645587488,
      "grad_norm": 1.6038225889205933,
      "learning_rate": 3.646105263157895e-05,
      "loss": 21.1862,
      "step": 131
    },
    {
      "epoch": 0.014905990627293772,
      "grad_norm": 1.6077948808670044,
      "learning_rate": 3.593263157894737e-05,
      "loss": 20.9297,
      "step": 132
    },
    {
      "epoch": 0.015018914798712665,
      "grad_norm": 1.8298345804214478,
      "learning_rate": 3.540421052631579e-05,
      "loss": 20.9612,
      "step": 133
    },
    {
      "epoch": 0.015131838970131557,
      "grad_norm": 1.5444914102554321,
      "learning_rate": 3.4875789473684215e-05,
      "loss": 21.0415,
      "step": 134
    },
    {
      "epoch": 0.015244763141550448,
      "grad_norm": 1.6737421751022339,
      "learning_rate": 3.434736842105263e-05,
      "loss": 21.0475,
      "step": 135
    },
    {
      "epoch": 0.015357687312969341,
      "grad_norm": 2.06323504447937,
      "learning_rate": 3.381894736842105e-05,
      "loss": 20.9078,
      "step": 136
    },
    {
      "epoch": 0.015470611484388233,
      "grad_norm": 1.7106398344039917,
      "learning_rate": 3.329052631578947e-05,
      "loss": 20.9534,
      "step": 137
    },
    {
      "epoch": 0.015583535655807125,
      "grad_norm": 1.7233997583389282,
      "learning_rate": 3.27621052631579e-05,
      "loss": 20.9764,
      "step": 138
    },
    {
      "epoch": 0.015696459827226016,
      "grad_norm": 1.6839755773544312,
      "learning_rate": 3.223368421052632e-05,
      "loss": 20.7462,
      "step": 139
    },
    {
      "epoch": 0.01580938399864491,
      "grad_norm": 1.747623085975647,
      "learning_rate": 3.1705263157894736e-05,
      "loss": 20.7554,
      "step": 140
    },
    {
      "epoch": 0.015922308170063803,
      "grad_norm": 1.651952862739563,
      "learning_rate": 3.117684210526316e-05,
      "loss": 21.0266,
      "step": 141
    },
    {
      "epoch": 0.016035232341482693,
      "grad_norm": 1.6088999509811401,
      "learning_rate": 3.064842105263158e-05,
      "loss": 21.1224,
      "step": 142
    },
    {
      "epoch": 0.016148156512901586,
      "grad_norm": 1.762588620185852,
      "learning_rate": 3.012e-05,
      "loss": 20.9005,
      "step": 143
    },
    {
      "epoch": 0.01626108068432048,
      "grad_norm": 1.566470742225647,
      "learning_rate": 2.9591578947368418e-05,
      "loss": 21.1413,
      "step": 144
    },
    {
      "epoch": 0.01637400485573937,
      "grad_norm": 1.6514183282852173,
      "learning_rate": 2.9063157894736842e-05,
      "loss": 20.8771,
      "step": 145
    },
    {
      "epoch": 0.016486929027158263,
      "grad_norm": 1.7934764623641968,
      "learning_rate": 2.8534736842105264e-05,
      "loss": 20.9009,
      "step": 146
    },
    {
      "epoch": 0.016599853198577156,
      "grad_norm": 1.6794406175613403,
      "learning_rate": 2.800631578947368e-05,
      "loss": 20.9681,
      "step": 147
    },
    {
      "epoch": 0.01671277736999605,
      "grad_norm": 1.6893374919891357,
      "learning_rate": 2.7477894736842106e-05,
      "loss": 21.0311,
      "step": 148
    },
    {
      "epoch": 0.01682570154141494,
      "grad_norm": 1.7404531240463257,
      "learning_rate": 2.6949473684210527e-05,
      "loss": 20.9912,
      "step": 149
    },
    {
      "epoch": 0.016938625712833832,
      "grad_norm": 1.9118671417236328,
      "learning_rate": 2.6421052631578945e-05,
      "loss": 21.0618,
      "step": 150
    },
    {
      "epoch": 0.016938625712833832,
      "eval_loss": 10.512722969055176,
      "eval_runtime": 30.1193,
      "eval_samples_per_second": 123.808,
      "eval_steps_per_second": 30.977,
      "step": 150
    },
    {
      "epoch": 0.017051549884252726,
      "grad_norm": 1.5834763050079346,
      "learning_rate": 2.589263157894737e-05,
      "loss": 21.509,
      "step": 151
    },
    {
      "epoch": 0.017164474055671616,
      "grad_norm": 1.5266532897949219,
      "learning_rate": 2.536421052631579e-05,
      "loss": 21.5389,
      "step": 152
    },
    {
      "epoch": 0.01727739822709051,
      "grad_norm": 2.0491244792938232,
      "learning_rate": 2.483578947368421e-05,
      "loss": 21.3882,
      "step": 153
    },
    {
      "epoch": 0.017390322398509402,
      "grad_norm": 1.5211632251739502,
      "learning_rate": 2.430736842105263e-05,
      "loss": 21.5286,
      "step": 154
    },
    {
      "epoch": 0.017503246569928292,
      "grad_norm": 2.0383026599884033,
      "learning_rate": 2.3778947368421052e-05,
      "loss": 21.3977,
      "step": 155
    },
    {
      "epoch": 0.017616170741347185,
      "grad_norm": 2.338569164276123,
      "learning_rate": 2.3250526315789473e-05,
      "loss": 21.2509,
      "step": 156
    },
    {
      "epoch": 0.01772909491276608,
      "grad_norm": 2.0516114234924316,
      "learning_rate": 2.2722105263157894e-05,
      "loss": 21.3403,
      "step": 157
    },
    {
      "epoch": 0.01784201908418497,
      "grad_norm": 1.9918718338012695,
      "learning_rate": 2.2193684210526316e-05,
      "loss": 21.2006,
      "step": 158
    },
    {
      "epoch": 0.017954943255603862,
      "grad_norm": 2.553642749786377,
      "learning_rate": 2.1665263157894737e-05,
      "loss": 21.1381,
      "step": 159
    },
    {
      "epoch": 0.018067867427022755,
      "grad_norm": 1.9166467189788818,
      "learning_rate": 2.1136842105263158e-05,
      "loss": 21.1056,
      "step": 160
    },
    {
      "epoch": 0.018180791598441645,
      "grad_norm": 1.5806626081466675,
      "learning_rate": 2.060842105263158e-05,
      "loss": 20.8517,
      "step": 161
    },
    {
      "epoch": 0.01829371576986054,
      "grad_norm": 1.4952349662780762,
      "learning_rate": 2.008e-05,
      "loss": 20.8729,
      "step": 162
    },
    {
      "epoch": 0.018406639941279432,
      "grad_norm": 1.5474094152450562,
      "learning_rate": 1.9551578947368422e-05,
      "loss": 20.7769,
      "step": 163
    },
    {
      "epoch": 0.01851956411269832,
      "grad_norm": 1.4593998193740845,
      "learning_rate": 1.9023157894736843e-05,
      "loss": 21.062,
      "step": 164
    },
    {
      "epoch": 0.018632488284117215,
      "grad_norm": 1.531658411026001,
      "learning_rate": 1.849473684210526e-05,
      "loss": 20.7112,
      "step": 165
    },
    {
      "epoch": 0.01874541245553611,
      "grad_norm": 1.4082283973693848,
      "learning_rate": 1.7966315789473686e-05,
      "loss": 20.9409,
      "step": 166
    },
    {
      "epoch": 0.018858336626954998,
      "grad_norm": 1.4776971340179443,
      "learning_rate": 1.7437894736842107e-05,
      "loss": 20.8996,
      "step": 167
    },
    {
      "epoch": 0.01897126079837389,
      "grad_norm": 1.5207204818725586,
      "learning_rate": 1.6909473684210525e-05,
      "loss": 20.8288,
      "step": 168
    },
    {
      "epoch": 0.019084184969792785,
      "grad_norm": 1.6669046878814697,
      "learning_rate": 1.638105263157895e-05,
      "loss": 20.6042,
      "step": 169
    },
    {
      "epoch": 0.019197109141211675,
      "grad_norm": 1.6279850006103516,
      "learning_rate": 1.5852631578947368e-05,
      "loss": 20.8314,
      "step": 170
    },
    {
      "epoch": 0.019310033312630568,
      "grad_norm": 1.5464578866958618,
      "learning_rate": 1.532421052631579e-05,
      "loss": 20.8447,
      "step": 171
    },
    {
      "epoch": 0.01942295748404946,
      "grad_norm": 1.5121798515319824,
      "learning_rate": 1.4795789473684209e-05,
      "loss": 20.6078,
      "step": 172
    },
    {
      "epoch": 0.019535881655468355,
      "grad_norm": 1.574658751487732,
      "learning_rate": 1.4267368421052632e-05,
      "loss": 20.7243,
      "step": 173
    },
    {
      "epoch": 0.019648805826887245,
      "grad_norm": 1.4556739330291748,
      "learning_rate": 1.3738947368421053e-05,
      "loss": 20.7603,
      "step": 174
    },
    {
      "epoch": 0.019761729998306138,
      "grad_norm": 1.8167383670806885,
      "learning_rate": 1.3210526315789473e-05,
      "loss": 20.6841,
      "step": 175
    },
    {
      "epoch": 0.01987465416972503,
      "grad_norm": 1.6482831239700317,
      "learning_rate": 1.2682105263157896e-05,
      "loss": 21.1056,
      "step": 176
    },
    {
      "epoch": 0.01998757834114392,
      "grad_norm": 1.7014974355697632,
      "learning_rate": 1.2153684210526315e-05,
      "loss": 20.7359,
      "step": 177
    },
    {
      "epoch": 0.020100502512562814,
      "grad_norm": 1.6174249649047852,
      "learning_rate": 1.1625263157894737e-05,
      "loss": 20.9042,
      "step": 178
    },
    {
      "epoch": 0.020213426683981708,
      "grad_norm": 1.530600905418396,
      "learning_rate": 1.1096842105263158e-05,
      "loss": 20.8118,
      "step": 179
    },
    {
      "epoch": 0.020326350855400598,
      "grad_norm": 1.7124817371368408,
      "learning_rate": 1.0568421052631579e-05,
      "loss": 21.0409,
      "step": 180
    },
    {
      "epoch": 0.02043927502681949,
      "grad_norm": 1.6012896299362183,
      "learning_rate": 1.004e-05,
      "loss": 21.0352,
      "step": 181
    },
    {
      "epoch": 0.020552199198238384,
      "grad_norm": 1.5824925899505615,
      "learning_rate": 9.511578947368422e-06,
      "loss": 20.6592,
      "step": 182
    },
    {
      "epoch": 0.020665123369657274,
      "grad_norm": 1.4896966218948364,
      "learning_rate": 8.983157894736843e-06,
      "loss": 20.9496,
      "step": 183
    },
    {
      "epoch": 0.020778047541076167,
      "grad_norm": 1.5998221635818481,
      "learning_rate": 8.454736842105263e-06,
      "loss": 20.8633,
      "step": 184
    },
    {
      "epoch": 0.02089097171249506,
      "grad_norm": 1.663162350654602,
      "learning_rate": 7.926315789473684e-06,
      "loss": 21.0101,
      "step": 185
    },
    {
      "epoch": 0.02100389588391395,
      "grad_norm": 1.6890687942504883,
      "learning_rate": 7.397894736842104e-06,
      "loss": 20.7532,
      "step": 186
    },
    {
      "epoch": 0.021116820055332844,
      "grad_norm": 1.707810878753662,
      "learning_rate": 6.8694736842105265e-06,
      "loss": 21.1776,
      "step": 187
    },
    {
      "epoch": 0.021229744226751737,
      "grad_norm": 1.679599642753601,
      "learning_rate": 6.341052631578948e-06,
      "loss": 21.1268,
      "step": 188
    },
    {
      "epoch": 0.021342668398170627,
      "grad_norm": 1.676312804222107,
      "learning_rate": 5.812631578947368e-06,
      "loss": 20.7901,
      "step": 189
    },
    {
      "epoch": 0.02145559256958952,
      "grad_norm": 1.652443766593933,
      "learning_rate": 5.2842105263157896e-06,
      "loss": 20.9088,
      "step": 190
    },
    {
      "epoch": 0.021568516741008414,
      "grad_norm": 1.552148461341858,
      "learning_rate": 4.755789473684211e-06,
      "loss": 20.8383,
      "step": 191
    },
    {
      "epoch": 0.021681440912427304,
      "grad_norm": 1.5653036832809448,
      "learning_rate": 4.227368421052631e-06,
      "loss": 21.1144,
      "step": 192
    },
    {
      "epoch": 0.021794365083846197,
      "grad_norm": 1.6634958982467651,
      "learning_rate": 3.698947368421052e-06,
      "loss": 20.8196,
      "step": 193
    },
    {
      "epoch": 0.02190728925526509,
      "grad_norm": 1.586753010749817,
      "learning_rate": 3.170526315789474e-06,
      "loss": 20.9744,
      "step": 194
    },
    {
      "epoch": 0.02202021342668398,
      "grad_norm": 1.57020103931427,
      "learning_rate": 2.6421052631578948e-06,
      "loss": 20.9053,
      "step": 195
    },
    {
      "epoch": 0.022133137598102873,
      "grad_norm": 1.6131221055984497,
      "learning_rate": 2.1136842105263157e-06,
      "loss": 20.8536,
      "step": 196
    },
    {
      "epoch": 0.022246061769521767,
      "grad_norm": 1.749698519706726,
      "learning_rate": 1.585263157894737e-06,
      "loss": 20.9774,
      "step": 197
    },
    {
      "epoch": 0.02235898594094066,
      "grad_norm": 1.8347561359405518,
      "learning_rate": 1.0568421052631578e-06,
      "loss": 20.8986,
      "step": 198
    },
    {
      "epoch": 0.02247191011235955,
      "grad_norm": 1.7269152402877808,
      "learning_rate": 5.284210526315789e-07,
      "loss": 20.9061,
      "step": 199
    },
    {
      "epoch": 0.022584834283778443,
      "grad_norm": 2.184429407119751,
      "learning_rate": 0.0,
      "loss": 20.7804,
      "step": 200
    },
    {
      "epoch": 0.022584834283778443,
      "eval_loss": 10.496330261230469,
      "eval_runtime": 30.1113,
      "eval_samples_per_second": 123.841,
      "eval_steps_per_second": 30.985,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5986779660288.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}