{
  "best_metric": 0.2263110727071762,
  "best_model_checkpoint": "miner_id_24/checkpoint-150",
  "epoch": 3.02212389380531,
  "eval_steps": 50,
  "global_step": 170,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.017699115044247787,
      "grad_norm": 31.40521812438965,
      "learning_rate": 1e-05,
      "loss": 5.6653,
      "step": 1
    },
    {
      "epoch": 0.017699115044247787,
      "eval_loss": 1.8833309412002563,
      "eval_runtime": 8.7556,
      "eval_samples_per_second": 10.85,
      "eval_steps_per_second": 2.741,
      "step": 1
    },
    {
      "epoch": 0.035398230088495575,
      "grad_norm": 34.309478759765625,
      "learning_rate": 2e-05,
      "loss": 5.7857,
      "step": 2
    },
    {
      "epoch": 0.05309734513274336,
      "grad_norm": 31.77597427368164,
      "learning_rate": 3e-05,
      "loss": 5.877,
      "step": 3
    },
    {
      "epoch": 0.07079646017699115,
      "grad_norm": 34.736934661865234,
      "learning_rate": 4e-05,
      "loss": 5.0209,
      "step": 4
    },
    {
      "epoch": 0.08849557522123894,
      "grad_norm": 32.39447021484375,
      "learning_rate": 5e-05,
      "loss": 4.4748,
      "step": 5
    },
    {
      "epoch": 0.10619469026548672,
      "grad_norm": 22.86501121520996,
      "learning_rate": 6e-05,
      "loss": 3.7596,
      "step": 6
    },
    {
      "epoch": 0.12389380530973451,
      "grad_norm": 22.00687026977539,
      "learning_rate": 7e-05,
      "loss": 3.9657,
      "step": 7
    },
    {
      "epoch": 0.1415929203539823,
      "grad_norm": 23.1529598236084,
      "learning_rate": 8e-05,
      "loss": 3.4275,
      "step": 8
    },
    {
      "epoch": 0.1592920353982301,
      "grad_norm": 31.291841506958008,
      "learning_rate": 9e-05,
      "loss": 3.2902,
      "step": 9
    },
    {
      "epoch": 0.17699115044247787,
      "grad_norm": 24.0184326171875,
      "learning_rate": 0.0001,
      "loss": 3.1102,
      "step": 10
    },
    {
      "epoch": 0.19469026548672566,
      "grad_norm": 22.568328857421875,
      "learning_rate": 9.999036202410325e-05,
      "loss": 3.0612,
      "step": 11
    },
    {
      "epoch": 0.21238938053097345,
      "grad_norm": 24.94548797607422,
      "learning_rate": 9.996145181203615e-05,
      "loss": 3.3743,
      "step": 12
    },
    {
      "epoch": 0.23008849557522124,
      "grad_norm": 25.63581657409668,
      "learning_rate": 9.991328050923581e-05,
      "loss": 3.3714,
      "step": 13
    },
    {
      "epoch": 0.24778761061946902,
      "grad_norm": 41.194435119628906,
      "learning_rate": 9.98458666866564e-05,
      "loss": 3.6124,
      "step": 14
    },
    {
      "epoch": 0.26548672566371684,
      "grad_norm": 15.79212760925293,
      "learning_rate": 9.975923633360985e-05,
      "loss": 2.9636,
      "step": 15
    },
    {
      "epoch": 0.2831858407079646,
      "grad_norm": 17.443269729614258,
      "learning_rate": 9.965342284774632e-05,
      "loss": 2.8034,
      "step": 16
    },
    {
      "epoch": 0.3008849557522124,
      "grad_norm": 14.93327522277832,
      "learning_rate": 9.952846702217886e-05,
      "loss": 2.7963,
      "step": 17
    },
    {
      "epoch": 0.3185840707964602,
      "grad_norm": 16.94188117980957,
      "learning_rate": 9.938441702975689e-05,
      "loss": 2.7428,
      "step": 18
    },
    {
      "epoch": 0.336283185840708,
      "grad_norm": 14.261035919189453,
      "learning_rate": 9.922132840449459e-05,
      "loss": 2.5116,
      "step": 19
    },
    {
      "epoch": 0.35398230088495575,
      "grad_norm": 11.297643661499023,
      "learning_rate": 9.903926402016153e-05,
      "loss": 2.4818,
      "step": 20
    },
    {
      "epoch": 0.37168141592920356,
      "grad_norm": 13.465043067932129,
      "learning_rate": 9.883829406604363e-05,
      "loss": 2.4556,
      "step": 21
    },
    {
      "epoch": 0.3893805309734513,
      "grad_norm": 16.393932342529297,
      "learning_rate": 9.861849601988383e-05,
      "loss": 2.4103,
      "step": 22
    },
    {
      "epoch": 0.40707964601769914,
      "grad_norm": 18.413082122802734,
      "learning_rate": 9.837995461801299e-05,
      "loss": 2.4597,
      "step": 23
    },
    {
      "epoch": 0.4247787610619469,
      "grad_norm": 24.529258728027344,
      "learning_rate": 9.812276182268236e-05,
      "loss": 2.9546,
      "step": 24
    },
    {
      "epoch": 0.4424778761061947,
      "grad_norm": 17.656185150146484,
      "learning_rate": 9.784701678661045e-05,
      "loss": 2.1286,
      "step": 25
    },
    {
      "epoch": 0.46017699115044247,
      "grad_norm": 15.885833740234375,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.3271,
      "step": 26
    },
    {
      "epoch": 0.4778761061946903,
      "grad_norm": 19.011276245117188,
      "learning_rate": 9.724030232334391e-05,
      "loss": 2.4495,
      "step": 27
    },
    {
      "epoch": 0.49557522123893805,
      "grad_norm": 17.471874237060547,
      "learning_rate": 9.690956679612421e-05,
      "loss": 1.8381,
      "step": 28
    },
    {
      "epoch": 0.5132743362831859,
      "grad_norm": 12.52908706665039,
      "learning_rate": 9.656074673794018e-05,
      "loss": 2.492,
      "step": 29
    },
    {
      "epoch": 0.5309734513274337,
      "grad_norm": 10.347203254699707,
      "learning_rate": 9.619397662556435e-05,
      "loss": 2.3401,
      "step": 30
    },
    {
      "epoch": 0.5486725663716814,
      "grad_norm": 9.183003425598145,
      "learning_rate": 9.580939785585681e-05,
      "loss": 2.1367,
      "step": 31
    },
    {
      "epoch": 0.5663716814159292,
      "grad_norm": 11.108962059020996,
      "learning_rate": 9.540715869125407e-05,
      "loss": 2.4407,
      "step": 32
    },
    {
      "epoch": 0.584070796460177,
      "grad_norm": 9.463624954223633,
      "learning_rate": 9.498741420261108e-05,
      "loss": 2.0885,
      "step": 33
    },
    {
      "epoch": 0.6017699115044248,
      "grad_norm": 11.597883224487305,
      "learning_rate": 9.45503262094184e-05,
      "loss": 2.1125,
      "step": 34
    },
    {
      "epoch": 0.6194690265486725,
      "grad_norm": 13.664706230163574,
      "learning_rate": 9.409606321741775e-05,
      "loss": 2.6442,
      "step": 35
    },
    {
      "epoch": 0.6371681415929203,
      "grad_norm": 14.292327880859375,
      "learning_rate": 9.362480035363986e-05,
      "loss": 2.362,
      "step": 36
    },
    {
      "epoch": 0.6548672566371682,
      "grad_norm": 9.707732200622559,
      "learning_rate": 9.31367192988896e-05,
      "loss": 1.8003,
      "step": 37
    },
    {
      "epoch": 0.672566371681416,
      "grad_norm": 15.758881568908691,
      "learning_rate": 9.263200821770461e-05,
      "loss": 1.9986,
      "step": 38
    },
    {
      "epoch": 0.6902654867256637,
      "grad_norm": 12.185977935791016,
      "learning_rate": 9.211086168581433e-05,
      "loss": 1.7782,
      "step": 39
    },
    {
      "epoch": 0.7079646017699115,
      "grad_norm": 16.444316864013672,
      "learning_rate": 9.157348061512727e-05,
      "loss": 1.7913,
      "step": 40
    },
    {
      "epoch": 0.7256637168141593,
      "grad_norm": 14.423544883728027,
      "learning_rate": 9.102007217627568e-05,
      "loss": 1.4715,
      "step": 41
    },
    {
      "epoch": 0.7433628318584071,
      "grad_norm": 22.212081909179688,
      "learning_rate": 9.045084971874738e-05,
      "loss": 2.1426,
      "step": 42
    },
    {
      "epoch": 0.7610619469026548,
      "grad_norm": 10.408870697021484,
      "learning_rate": 8.986603268863536e-05,
      "loss": 2.0713,
      "step": 43
    },
    {
      "epoch": 0.7787610619469026,
      "grad_norm": 9.761504173278809,
      "learning_rate": 8.926584654403724e-05,
      "loss": 2.1216,
      "step": 44
    },
    {
      "epoch": 0.7964601769911505,
      "grad_norm": 9.110294342041016,
      "learning_rate": 8.865052266813685e-05,
      "loss": 1.9699,
      "step": 45
    },
    {
      "epoch": 0.8141592920353983,
      "grad_norm": 7.902236461639404,
      "learning_rate": 8.802029828000156e-05,
      "loss": 1.714,
      "step": 46
    },
    {
      "epoch": 0.831858407079646,
      "grad_norm": 10.150836944580078,
      "learning_rate": 8.737541634312985e-05,
      "loss": 2.045,
      "step": 47
    },
    {
      "epoch": 0.8495575221238938,
      "grad_norm": 10.204972267150879,
      "learning_rate": 8.671612547178428e-05,
      "loss": 1.551,
      "step": 48
    },
    {
      "epoch": 0.8672566371681416,
      "grad_norm": 10.702434539794922,
      "learning_rate": 8.604267983514594e-05,
      "loss": 1.6763,
      "step": 49
    },
    {
      "epoch": 0.8849557522123894,
      "grad_norm": 14.998902320861816,
      "learning_rate": 8.535533905932738e-05,
      "loss": 2.2768,
      "step": 50
    },
    {
      "epoch": 0.8849557522123894,
      "eval_loss": 0.39903146028518677,
      "eval_runtime": 8.9539,
      "eval_samples_per_second": 10.61,
      "eval_steps_per_second": 2.68,
      "step": 50
    },
    {
      "epoch": 0.9026548672566371,
      "grad_norm": 9.826573371887207,
      "learning_rate": 8.46543681272818e-05,
      "loss": 1.5854,
      "step": 51
    },
    {
      "epoch": 0.9203539823008849,
      "grad_norm": 11.782295227050781,
      "learning_rate": 8.39400372766471e-05,
      "loss": 1.7079,
      "step": 52
    },
    {
      "epoch": 0.9380530973451328,
      "grad_norm": 13.14542293548584,
      "learning_rate": 8.321262189556409e-05,
      "loss": 1.6517,
      "step": 53
    },
    {
      "epoch": 0.9557522123893806,
      "grad_norm": 10.501534461975098,
      "learning_rate": 8.247240241650918e-05,
      "loss": 1.575,
      "step": 54
    },
    {
      "epoch": 0.9734513274336283,
      "grad_norm": 18.712419509887695,
      "learning_rate": 8.171966420818228e-05,
      "loss": 2.1136,
      "step": 55
    },
    {
      "epoch": 0.9911504424778761,
      "grad_norm": 14.00768756866455,
      "learning_rate": 8.095469746549172e-05,
      "loss": 1.7777,
      "step": 56
    },
    {
      "epoch": 1.0132743362831858,
      "grad_norm": 7.709658622741699,
      "learning_rate": 8.017779709767858e-05,
      "loss": 1.2188,
      "step": 57
    },
    {
      "epoch": 1.0309734513274336,
      "grad_norm": 5.869880676269531,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.0963,
      "step": 58
    },
    {
      "epoch": 1.0486725663716814,
      "grad_norm": 6.3835930824279785,
      "learning_rate": 7.858939801138061e-05,
      "loss": 0.9459,
      "step": 59
    },
    {
      "epoch": 1.0663716814159292,
      "grad_norm": 6.956117153167725,
      "learning_rate": 7.777851165098012e-05,
      "loss": 0.7443,
      "step": 60
    },
    {
      "epoch": 1.084070796460177,
      "grad_norm": 10.194844245910645,
      "learning_rate": 7.695691614555003e-05,
      "loss": 0.7741,
      "step": 61
    },
    {
      "epoch": 1.1017699115044248,
      "grad_norm": 11.043907165527344,
      "learning_rate": 7.612492823579745e-05,
      "loss": 1.0631,
      "step": 62
    },
    {
      "epoch": 1.1194690265486726,
      "grad_norm": 7.981068134307861,
      "learning_rate": 7.528286866889924e-05,
      "loss": 0.6533,
      "step": 63
    },
    {
      "epoch": 1.1371681415929205,
      "grad_norm": 9.238661766052246,
      "learning_rate": 7.443106207484776e-05,
      "loss": 0.8124,
      "step": 64
    },
    {
      "epoch": 1.154867256637168,
      "grad_norm": 6.48264741897583,
      "learning_rate": 7.35698368412999e-05,
      "loss": 0.4049,
      "step": 65
    },
    {
      "epoch": 1.1725663716814159,
      "grad_norm": 7.588451385498047,
      "learning_rate": 7.269952498697734e-05,
      "loss": 0.5313,
      "step": 66
    },
    {
      "epoch": 1.1902654867256637,
      "grad_norm": 8.53221607208252,
      "learning_rate": 7.18204620336671e-05,
      "loss": 0.5832,
      "step": 67
    },
    {
      "epoch": 1.2079646017699115,
      "grad_norm": 7.945702075958252,
      "learning_rate": 7.09329868768714e-05,
      "loss": 0.6402,
      "step": 68
    },
    {
      "epoch": 1.2256637168141593,
      "grad_norm": 11.50102424621582,
      "learning_rate": 7.003744165515705e-05,
      "loss": 0.6571,
      "step": 69
    },
    {
      "epoch": 1.2433628318584071,
      "grad_norm": 8.518128395080566,
      "learning_rate": 6.91341716182545e-05,
      "loss": 0.5089,
      "step": 70
    },
    {
      "epoch": 1.261061946902655,
      "grad_norm": 11.149508476257324,
      "learning_rate": 6.82235249939575e-05,
      "loss": 1.0319,
      "step": 71
    },
    {
      "epoch": 1.2787610619469025,
      "grad_norm": 9.314420700073242,
      "learning_rate": 6.730585285387465e-05,
      "loss": 1.0209,
      "step": 72
    },
    {
      "epoch": 1.2964601769911503,
      "grad_norm": 12.915304183959961,
      "learning_rate": 6.638150897808468e-05,
      "loss": 0.8383,
      "step": 73
    },
    {
      "epoch": 1.3141592920353982,
      "grad_norm": 7.141340732574463,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.7152,
      "step": 74
    },
    {
      "epoch": 1.331858407079646,
      "grad_norm": 6.504374027252197,
      "learning_rate": 6.451423386272312e-05,
      "loss": 0.6845,
      "step": 75
    },
    {
      "epoch": 1.3495575221238938,
      "grad_norm": 5.089697360992432,
      "learning_rate": 6.357202249325371e-05,
      "loss": 0.5919,
      "step": 76
    },
    {
      "epoch": 1.3672566371681416,
      "grad_norm": 5.1988115310668945,
      "learning_rate": 6.26245788507579e-05,
      "loss": 0.6042,
      "step": 77
    },
    {
      "epoch": 1.3849557522123894,
      "grad_norm": 7.5055060386657715,
      "learning_rate": 6.167226819279528e-05,
      "loss": 0.6017,
      "step": 78
    },
    {
      "epoch": 1.4026548672566372,
      "grad_norm": 6.506912708282471,
      "learning_rate": 6.071545765325254e-05,
      "loss": 0.5644,
      "step": 79
    },
    {
      "epoch": 1.420353982300885,
      "grad_norm": 12.563440322875977,
      "learning_rate": 5.9754516100806423e-05,
      "loss": 0.7918,
      "step": 80
    },
    {
      "epoch": 1.4380530973451329,
      "grad_norm": 17.038772583007812,
      "learning_rate": 5.8789813996717736e-05,
      "loss": 0.5758,
      "step": 81
    },
    {
      "epoch": 1.4557522123893805,
      "grad_norm": 9.219198226928711,
      "learning_rate": 5.782172325201155e-05,
      "loss": 0.6938,
      "step": 82
    },
    {
      "epoch": 1.4734513274336283,
      "grad_norm": 11.23676872253418,
      "learning_rate": 5.685061708409841e-05,
      "loss": 0.6351,
      "step": 83
    },
    {
      "epoch": 1.491150442477876,
      "grad_norm": 9.347498893737793,
      "learning_rate": 5.587686987289189e-05,
      "loss": 0.5458,
      "step": 84
    },
    {
      "epoch": 1.508849557522124,
      "grad_norm": 7.985458850860596,
      "learning_rate": 5.490085701647805e-05,
      "loss": 0.8692,
      "step": 85
    },
    {
      "epoch": 1.5265486725663717,
      "grad_norm": 9.773151397705078,
      "learning_rate": 5.392295478639225e-05,
      "loss": 1.0329,
      "step": 86
    },
    {
      "epoch": 1.5442477876106193,
      "grad_norm": 7.224915504455566,
      "learning_rate": 5.294354018255945e-05,
      "loss": 0.8338,
      "step": 87
    },
    {
      "epoch": 1.5619469026548671,
      "grad_norm": 6.956666946411133,
      "learning_rate": 5.196299078795344e-05,
      "loss": 0.6281,
      "step": 88
    },
    {
      "epoch": 1.579646017699115,
      "grad_norm": 7.078762054443359,
      "learning_rate": 5.0981684623031415e-05,
      "loss": 0.5895,
      "step": 89
    },
    {
      "epoch": 1.5973451327433628,
      "grad_norm": 9.647902488708496,
      "learning_rate": 5e-05,
      "loss": 0.7035,
      "step": 90
    },
    {
      "epoch": 1.6150442477876106,
      "grad_norm": 6.713580131530762,
      "learning_rate": 4.901831537696859e-05,
      "loss": 0.5963,
      "step": 91
    },
    {
      "epoch": 1.6327433628318584,
      "grad_norm": 9.912200927734375,
      "learning_rate": 4.8037009212046586e-05,
      "loss": 0.7586,
      "step": 92
    },
    {
      "epoch": 1.6504424778761062,
      "grad_norm": 5.545166969299316,
      "learning_rate": 4.7056459817440544e-05,
      "loss": 0.5426,
      "step": 93
    },
    {
      "epoch": 1.668141592920354,
      "grad_norm": 8.735177993774414,
      "learning_rate": 4.607704521360776e-05,
      "loss": 0.6275,
      "step": 94
    },
    {
      "epoch": 1.6858407079646018,
      "grad_norm": 6.513275623321533,
      "learning_rate": 4.509914298352197e-05,
      "loss": 0.5658,
      "step": 95
    },
    {
      "epoch": 1.7035398230088497,
      "grad_norm": 7.394782066345215,
      "learning_rate": 4.412313012710813e-05,
      "loss": 0.5707,
      "step": 96
    },
    {
      "epoch": 1.7212389380530975,
      "grad_norm": 11.5717191696167,
      "learning_rate": 4.3149382915901606e-05,
      "loss": 0.8114,
      "step": 97
    },
    {
      "epoch": 1.7389380530973453,
      "grad_norm": 9.202227592468262,
      "learning_rate": 4.2178276747988446e-05,
      "loss": 0.765,
      "step": 98
    },
    {
      "epoch": 1.7566371681415929,
      "grad_norm": 6.01152229309082,
      "learning_rate": 4.1210186003282275e-05,
      "loss": 0.7097,
      "step": 99
    },
    {
      "epoch": 1.7743362831858407,
      "grad_norm": 6.051780700683594,
      "learning_rate": 4.0245483899193595e-05,
      "loss": 0.8358,
      "step": 100
    },
    {
      "epoch": 1.7743362831858407,
      "eval_loss": 0.2679302394390106,
      "eval_runtime": 8.9626,
      "eval_samples_per_second": 10.6,
      "eval_steps_per_second": 2.678,
      "step": 100
    },
    {
      "epoch": 1.7920353982300885,
      "grad_norm": 5.6379547119140625,
      "learning_rate": 3.928454234674747e-05,
      "loss": 0.6571,
      "step": 101
    },
    {
      "epoch": 1.8097345132743363,
      "grad_norm": 6.8352885246276855,
      "learning_rate": 3.832773180720475e-05,
      "loss": 0.6342,
      "step": 102
    },
    {
      "epoch": 1.827433628318584,
      "grad_norm": 4.388557434082031,
      "learning_rate": 3.73754211492421e-05,
      "loss": 0.5032,
      "step": 103
    },
    {
      "epoch": 1.8451327433628317,
      "grad_norm": 4.776178359985352,
      "learning_rate": 3.642797750674629e-05,
      "loss": 0.4031,
      "step": 104
    },
    {
      "epoch": 1.8628318584070795,
      "grad_norm": 8.055367469787598,
      "learning_rate": 3.5485766137276894e-05,
      "loss": 0.6053,
      "step": 105
    },
    {
      "epoch": 1.8805309734513274,
      "grad_norm": 6.032299041748047,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.5773,
      "step": 106
    },
    {
      "epoch": 1.8982300884955752,
      "grad_norm": 7.200865745544434,
      "learning_rate": 3.361849102191533e-05,
      "loss": 0.497,
      "step": 107
    },
    {
      "epoch": 1.915929203539823,
      "grad_norm": 6.174206733703613,
      "learning_rate": 3.2694147146125345e-05,
      "loss": 0.4945,
      "step": 108
    },
    {
      "epoch": 1.9336283185840708,
      "grad_norm": 8.304489135742188,
      "learning_rate": 3.177647500604252e-05,
      "loss": 0.5834,
      "step": 109
    },
    {
      "epoch": 1.9513274336283186,
      "grad_norm": 5.608364582061768,
      "learning_rate": 3.086582838174551e-05,
      "loss": 0.4771,
      "step": 110
    },
    {
      "epoch": 1.9690265486725664,
      "grad_norm": 7.852937698364258,
      "learning_rate": 2.996255834484296e-05,
      "loss": 0.4913,
      "step": 111
    },
    {
      "epoch": 1.9867256637168142,
      "grad_norm": 9.713995933532715,
      "learning_rate": 2.9067013123128613e-05,
      "loss": 0.6423,
      "step": 112
    },
    {
      "epoch": 2.0088495575221237,
      "grad_norm": 8.310989379882812,
      "learning_rate": 2.8179537966332887e-05,
      "loss": 0.5832,
      "step": 113
    },
    {
      "epoch": 2.0265486725663715,
      "grad_norm": 3.7431464195251465,
      "learning_rate": 2.7300475013022663e-05,
      "loss": 0.2774,
      "step": 114
    },
    {
      "epoch": 2.0442477876106193,
      "grad_norm": 3.602764844894409,
      "learning_rate": 2.6430163158700115e-05,
      "loss": 0.2663,
      "step": 115
    },
    {
      "epoch": 2.061946902654867,
      "grad_norm": 4.605719089508057,
      "learning_rate": 2.556893792515227e-05,
      "loss": 0.2502,
      "step": 116
    },
    {
      "epoch": 2.079646017699115,
      "grad_norm": 2.8502187728881836,
      "learning_rate": 2.471713133110078e-05,
      "loss": 0.219,
      "step": 117
    },
    {
      "epoch": 2.0973451327433628,
      "grad_norm": 2.774047613143921,
      "learning_rate": 2.3875071764202563e-05,
      "loss": 0.1839,
      "step": 118
    },
    {
      "epoch": 2.1150442477876106,
      "grad_norm": 3.8455257415771484,
      "learning_rate": 2.3043083854449988e-05,
      "loss": 0.2419,
      "step": 119
    },
    {
      "epoch": 2.1327433628318584,
      "grad_norm": 5.175693511962891,
      "learning_rate": 2.2221488349019903e-05,
      "loss": 0.192,
      "step": 120
    },
    {
      "epoch": 2.150442477876106,
      "grad_norm": 3.1044692993164062,
      "learning_rate": 2.1410601988619394e-05,
      "loss": 0.1589,
      "step": 121
    },
    {
      "epoch": 2.168141592920354,
      "grad_norm": 6.46194314956665,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.2275,
      "step": 122
    },
    {
      "epoch": 2.185840707964602,
      "grad_norm": 4.672300815582275,
      "learning_rate": 1.982220290232143e-05,
      "loss": 0.2475,
      "step": 123
    },
    {
      "epoch": 2.2035398230088497,
      "grad_norm": 5.018974304199219,
      "learning_rate": 1.9045302534508297e-05,
      "loss": 0.2665,
      "step": 124
    },
    {
      "epoch": 2.2212389380530975,
      "grad_norm": 8.109318733215332,
      "learning_rate": 1.8280335791817733e-05,
      "loss": 0.4599,
      "step": 125
    },
    {
      "epoch": 2.2389380530973453,
      "grad_norm": 5.888927459716797,
      "learning_rate": 1.7527597583490822e-05,
      "loss": 0.3197,
      "step": 126
    },
    {
      "epoch": 2.256637168141593,
      "grad_norm": 6.209958076477051,
      "learning_rate": 1.678737810443593e-05,
      "loss": 0.4531,
      "step": 127
    },
    {
      "epoch": 2.274336283185841,
      "grad_norm": 3.8774378299713135,
      "learning_rate": 1.605996272335291e-05,
      "loss": 0.2294,
      "step": 128
    },
    {
      "epoch": 2.2920353982300883,
      "grad_norm": 4.786031723022461,
      "learning_rate": 1.5345631872718214e-05,
      "loss": 0.2806,
      "step": 129
    },
    {
      "epoch": 2.309734513274336,
      "grad_norm": 6.043717384338379,
      "learning_rate": 1.4644660940672627e-05,
      "loss": 0.187,
      "step": 130
    },
    {
      "epoch": 2.327433628318584,
      "grad_norm": 4.6054606437683105,
      "learning_rate": 1.3957320164854059e-05,
      "loss": 0.2062,
      "step": 131
    },
    {
      "epoch": 2.3451327433628317,
      "grad_norm": 5.415921211242676,
      "learning_rate": 1.3283874528215733e-05,
      "loss": 0.2537,
      "step": 132
    },
    {
      "epoch": 2.3628318584070795,
      "grad_norm": 2.9057247638702393,
      "learning_rate": 1.2624583656870154e-05,
      "loss": 0.1775,
      "step": 133
    },
    {
      "epoch": 2.3805309734513274,
      "grad_norm": 3.5282516479492188,
      "learning_rate": 1.1979701719998453e-05,
      "loss": 0.1577,
      "step": 134
    },
    {
      "epoch": 2.398230088495575,
      "grad_norm": 5.042332172393799,
      "learning_rate": 1.134947733186315e-05,
      "loss": 0.1807,
      "step": 135
    },
    {
      "epoch": 2.415929203539823,
      "grad_norm": 4.14683198928833,
      "learning_rate": 1.0734153455962765e-05,
      "loss": 0.2114,
      "step": 136
    },
    {
      "epoch": 2.433628318584071,
      "grad_norm": 4.2411980628967285,
      "learning_rate": 1.013396731136465e-05,
      "loss": 0.2662,
      "step": 137
    },
    {
      "epoch": 2.4513274336283186,
      "grad_norm": 2.8422842025756836,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.2022,
      "step": 138
    },
    {
      "epoch": 2.4690265486725664,
      "grad_norm": 5.191140174865723,
      "learning_rate": 8.97992782372432e-06,
      "loss": 0.2508,
      "step": 139
    },
    {
      "epoch": 2.4867256637168142,
      "grad_norm": 7.143102645874023,
      "learning_rate": 8.426519384872733e-06,
      "loss": 0.3408,
      "step": 140
    },
    {
      "epoch": 2.504424778761062,
      "grad_norm": 3.898927927017212,
      "learning_rate": 7.889138314185678e-06,
      "loss": 0.2535,
      "step": 141
    },
    {
      "epoch": 2.52212389380531,
      "grad_norm": 3.8761115074157715,
      "learning_rate": 7.367991782295391e-06,
      "loss": 0.2149,
      "step": 142
    },
    {
      "epoch": 2.5398230088495577,
      "grad_norm": 3.058852434158325,
      "learning_rate": 6.863280701110408e-06,
      "loss": 0.1875,
      "step": 143
    },
    {
      "epoch": 2.557522123893805,
      "grad_norm": 1.8599612712860107,
      "learning_rate": 6.375199646360142e-06,
      "loss": 0.1232,
      "step": 144
    },
    {
      "epoch": 2.5752212389380533,
      "grad_norm": 6.451328277587891,
      "learning_rate": 5.903936782582253e-06,
      "loss": 0.2049,
      "step": 145
    },
    {
      "epoch": 2.5929203539823007,
      "grad_norm": 3.4158663749694824,
      "learning_rate": 5.449673790581611e-06,
      "loss": 0.1678,
      "step": 146
    },
    {
      "epoch": 2.6106194690265485,
      "grad_norm": 3.9194419384002686,
      "learning_rate": 5.012585797388936e-06,
      "loss": 0.1724,
      "step": 147
    },
    {
      "epoch": 2.6283185840707963,
      "grad_norm": 5.876561164855957,
      "learning_rate": 4.592841308745932e-06,
      "loss": 0.2246,
      "step": 148
    },
    {
      "epoch": 2.646017699115044,
      "grad_norm": 2.878607749938965,
      "learning_rate": 4.190602144143207e-06,
      "loss": 0.1746,
      "step": 149
    },
    {
      "epoch": 2.663716814159292,
      "grad_norm": 2.6272075176239014,
      "learning_rate": 3.8060233744356633e-06,
      "loss": 0.1676,
      "step": 150
    },
    {
      "epoch": 2.663716814159292,
      "eval_loss": 0.2263110727071762,
      "eval_runtime": 8.962,
      "eval_samples_per_second": 10.6,
      "eval_steps_per_second": 2.678,
      "step": 150
    },
    {
      "epoch": 2.6814159292035398,
      "grad_norm": 7.257627964019775,
      "learning_rate": 3.4392532620598216e-06,
      "loss": 0.2907,
      "step": 151
    },
    {
      "epoch": 2.6991150442477876,
      "grad_norm": 2.6858301162719727,
      "learning_rate": 3.0904332038757977e-06,
      "loss": 0.2093,
      "step": 152
    },
    {
      "epoch": 2.7168141592920354,
      "grad_norm": 4.403625965118408,
      "learning_rate": 2.759697676656098e-06,
      "loss": 0.2625,
      "step": 153
    },
    {
      "epoch": 2.734513274336283,
      "grad_norm": 5.445953845977783,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.2808,
      "step": 154
    },
    {
      "epoch": 2.752212389380531,
      "grad_norm": 3.8295862674713135,
      "learning_rate": 2.152983213389559e-06,
      "loss": 0.2573,
      "step": 155
    },
    {
      "epoch": 2.769911504424779,
      "grad_norm": 2.5319910049438477,
      "learning_rate": 1.8772381773176417e-06,
      "loss": 0.1505,
      "step": 156
    },
    {
      "epoch": 2.7876106194690267,
      "grad_norm": 2.632370710372925,
      "learning_rate": 1.620045381987012e-06,
      "loss": 0.1901,
      "step": 157
    },
    {
      "epoch": 2.8053097345132745,
      "grad_norm": 4.662055969238281,
      "learning_rate": 1.3815039801161721e-06,
      "loss": 0.2873,
      "step": 158
    },
    {
      "epoch": 2.823008849557522,
      "grad_norm": 2.293377161026001,
      "learning_rate": 1.1617059339563807e-06,
      "loss": 0.1484,
      "step": 159
    },
    {
      "epoch": 2.84070796460177,
      "grad_norm": 2.53718638420105,
      "learning_rate": 9.607359798384785e-07,
      "loss": 0.1503,
      "step": 160
    },
    {
      "epoch": 2.8584070796460175,
      "grad_norm": 2.226994752883911,
      "learning_rate": 7.786715955054203e-07,
      "loss": 0.1639,
      "step": 161
    },
    {
      "epoch": 2.8761061946902657,
      "grad_norm": 2.434995174407959,
      "learning_rate": 6.15582970243117e-07,
      "loss": 0.1509,
      "step": 162
    },
    {
      "epoch": 2.893805309734513,
      "grad_norm": 2.5986759662628174,
      "learning_rate": 4.715329778211375e-07,
      "loss": 0.1388,
      "step": 163
    },
    {
      "epoch": 2.911504424778761,
      "grad_norm": 3.339171886444092,
      "learning_rate": 3.465771522536854e-07,
      "loss": 0.1941,
      "step": 164
    },
    {
      "epoch": 2.9292035398230087,
      "grad_norm": 3.910398006439209,
      "learning_rate": 2.407636663901591e-07,
      "loss": 0.2387,
      "step": 165
    },
    {
      "epoch": 2.9469026548672566,
      "grad_norm": 2.9578335285186768,
      "learning_rate": 1.5413331334360182e-07,
      "loss": 0.234,
      "step": 166
    },
    {
      "epoch": 2.9646017699115044,
      "grad_norm": 3.8173089027404785,
      "learning_rate": 8.671949076420882e-08,
      "loss": 0.2285,
      "step": 167
    },
    {
      "epoch": 2.982300884955752,
      "grad_norm": 4.536461353302002,
      "learning_rate": 3.8548187963854956e-08,
      "loss": 0.2497,
      "step": 168
    },
    {
      "epoch": 3.004424778761062,
      "grad_norm": 4.463059425354004,
      "learning_rate": 9.637975896759077e-09,
      "loss": 0.2892,
      "step": 169
    },
    {
      "epoch": 3.02212389380531,
      "grad_norm": 2.0114989280700684,
      "learning_rate": 0.0,
      "loss": 0.1349,
      "step": 170
    }
  ],
  "logging_steps": 1,
  "max_steps": 170,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "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": 2.4326966152790016e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}