{
  "best_metric": 4.5732855796813965,
  "best_model_checkpoint": "miner_id_24/checkpoint-100",
  "epoch": 1.3071895424836601,
  "eval_steps": 50,
  "global_step": 100,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.013071895424836602,
      "grad_norm": 7.257218360900879,
      "learning_rate": 5.000000000000001e-07,
      "loss": 2.9113,
      "step": 1
    },
    {
      "epoch": 0.013071895424836602,
      "eval_loss": 8.609292030334473,
      "eval_runtime": 4.5671,
      "eval_samples_per_second": 28.245,
      "eval_steps_per_second": 7.226,
      "step": 1
    },
    {
      "epoch": 0.026143790849673203,
      "grad_norm": 8.341460227966309,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 3.8712,
      "step": 2
    },
    {
      "epoch": 0.0392156862745098,
      "grad_norm": 14.018800735473633,
      "learning_rate": 1.5e-06,
      "loss": 5.3191,
      "step": 3
    },
    {
      "epoch": 0.05228758169934641,
      "grad_norm": 19.952796936035156,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 6.9201,
      "step": 4
    },
    {
      "epoch": 0.06535947712418301,
      "grad_norm": 20.201190948486328,
      "learning_rate": 2.5e-06,
      "loss": 6.9843,
      "step": 5
    },
    {
      "epoch": 0.0784313725490196,
      "grad_norm": 21.60859489440918,
      "learning_rate": 3e-06,
      "loss": 7.7391,
      "step": 6
    },
    {
      "epoch": 0.0915032679738562,
      "grad_norm": 24.700639724731445,
      "learning_rate": 3.5e-06,
      "loss": 8.2004,
      "step": 7
    },
    {
      "epoch": 0.10457516339869281,
      "grad_norm": 22.307209014892578,
      "learning_rate": 4.000000000000001e-06,
      "loss": 7.6429,
      "step": 8
    },
    {
      "epoch": 0.11764705882352941,
      "grad_norm": 25.259756088256836,
      "learning_rate": 4.5e-06,
      "loss": 8.005,
      "step": 9
    },
    {
      "epoch": 0.13071895424836602,
      "grad_norm": 23.363004684448242,
      "learning_rate": 5e-06,
      "loss": 7.4913,
      "step": 10
    },
    {
      "epoch": 0.1437908496732026,
      "grad_norm": 22.28295135498047,
      "learning_rate": 4.99847706754774e-06,
      "loss": 7.4698,
      "step": 11
    },
    {
      "epoch": 0.1568627450980392,
      "grad_norm": 25.160226821899414,
      "learning_rate": 4.993910125649561e-06,
      "loss": 7.8227,
      "step": 12
    },
    {
      "epoch": 0.16993464052287582,
      "grad_norm": 26.892242431640625,
      "learning_rate": 4.986304738420684e-06,
      "loss": 9.0196,
      "step": 13
    },
    {
      "epoch": 0.1830065359477124,
      "grad_norm": 27.60188865661621,
      "learning_rate": 4.975670171853926e-06,
      "loss": 8.9304,
      "step": 14
    },
    {
      "epoch": 0.19607843137254902,
      "grad_norm": 25.04266929626465,
      "learning_rate": 4.962019382530521e-06,
      "loss": 8.8492,
      "step": 15
    },
    {
      "epoch": 0.20915032679738563,
      "grad_norm": 25.570220947265625,
      "learning_rate": 4.9453690018345144e-06,
      "loss": 9.2126,
      "step": 16
    },
    {
      "epoch": 0.2222222222222222,
      "grad_norm": 23.66245460510254,
      "learning_rate": 4.925739315689991e-06,
      "loss": 8.7148,
      "step": 17
    },
    {
      "epoch": 0.23529411764705882,
      "grad_norm": 28.739566802978516,
      "learning_rate": 4.903154239845798e-06,
      "loss": 10.0651,
      "step": 18
    },
    {
      "epoch": 0.24836601307189543,
      "grad_norm": 31.82867431640625,
      "learning_rate": 4.8776412907378845e-06,
      "loss": 11.8835,
      "step": 19
    },
    {
      "epoch": 0.26143790849673204,
      "grad_norm": 7.091309070587158,
      "learning_rate": 4.849231551964771e-06,
      "loss": 2.8035,
      "step": 20
    },
    {
      "epoch": 0.27450980392156865,
      "grad_norm": 9.146745681762695,
      "learning_rate": 4.817959636416969e-06,
      "loss": 4.0063,
      "step": 21
    },
    {
      "epoch": 0.2875816993464052,
      "grad_norm": 14.22527027130127,
      "learning_rate": 4.783863644106502e-06,
      "loss": 5.5576,
      "step": 22
    },
    {
      "epoch": 0.3006535947712418,
      "grad_norm": 18.277191162109375,
      "learning_rate": 4.746985115747918e-06,
      "loss": 6.0938,
      "step": 23
    },
    {
      "epoch": 0.3137254901960784,
      "grad_norm": 19.253273010253906,
      "learning_rate": 4.707368982147318e-06,
      "loss": 6.4934,
      "step": 24
    },
    {
      "epoch": 0.32679738562091504,
      "grad_norm": 22.427473068237305,
      "learning_rate": 4.665063509461098e-06,
      "loss": 6.9725,
      "step": 25
    },
    {
      "epoch": 0.33986928104575165,
      "grad_norm": 21.749441146850586,
      "learning_rate": 4.620120240391065e-06,
      "loss": 6.7396,
      "step": 26
    },
    {
      "epoch": 0.35294117647058826,
      "grad_norm": 22.339990615844727,
      "learning_rate": 4.572593931387604e-06,
      "loss": 6.6181,
      "step": 27
    },
    {
      "epoch": 0.3660130718954248,
      "grad_norm": 25.734874725341797,
      "learning_rate": 4.522542485937369e-06,
      "loss": 6.3757,
      "step": 28
    },
    {
      "epoch": 0.3790849673202614,
      "grad_norm": 23.32753562927246,
      "learning_rate": 4.470026884016805e-06,
      "loss": 6.9307,
      "step": 29
    },
    {
      "epoch": 0.39215686274509803,
      "grad_norm": 22.613615036010742,
      "learning_rate": 4.415111107797445e-06,
      "loss": 7.2312,
      "step": 30
    },
    {
      "epoch": 0.40522875816993464,
      "grad_norm": 23.05530548095703,
      "learning_rate": 4.357862063693486e-06,
      "loss": 6.3992,
      "step": 31
    },
    {
      "epoch": 0.41830065359477125,
      "grad_norm": 24.813129425048828,
      "learning_rate": 4.2983495008466285e-06,
      "loss": 6.8739,
      "step": 32
    },
    {
      "epoch": 0.43137254901960786,
      "grad_norm": 25.304807662963867,
      "learning_rate": 4.236645926147493e-06,
      "loss": 7.928,
      "step": 33
    },
    {
      "epoch": 0.4444444444444444,
      "grad_norm": 25.62441635131836,
      "learning_rate": 4.172826515897146e-06,
      "loss": 7.3664,
      "step": 34
    },
    {
      "epoch": 0.45751633986928103,
      "grad_norm": 24.42110824584961,
      "learning_rate": 4.106969024216348e-06,
      "loss": 7.7416,
      "step": 35
    },
    {
      "epoch": 0.47058823529411764,
      "grad_norm": 25.079715728759766,
      "learning_rate": 4.039153688314146e-06,
      "loss": 8.1472,
      "step": 36
    },
    {
      "epoch": 0.48366013071895425,
      "grad_norm": 30.301733016967773,
      "learning_rate": 3.969463130731183e-06,
      "loss": 9.266,
      "step": 37
    },
    {
      "epoch": 0.49673202614379086,
      "grad_norm": 28.850906372070312,
      "learning_rate": 3.897982258676867e-06,
      "loss": 8.429,
      "step": 38
    },
    {
      "epoch": 0.5098039215686274,
      "grad_norm": 5.361105918884277,
      "learning_rate": 3.824798160583012e-06,
      "loss": 2.6361,
      "step": 39
    },
    {
      "epoch": 0.5228758169934641,
      "grad_norm": 6.960734844207764,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 3.2406,
      "step": 40
    },
    {
      "epoch": 0.5359477124183006,
      "grad_norm": 12.633935928344727,
      "learning_rate": 3.6736789069647273e-06,
      "loss": 5.5256,
      "step": 41
    },
    {
      "epoch": 0.5490196078431373,
      "grad_norm": 15.817687034606934,
      "learning_rate": 3.595927866972694e-06,
      "loss": 5.1769,
      "step": 42
    },
    {
      "epoch": 0.5620915032679739,
      "grad_norm": 19.869218826293945,
      "learning_rate": 3.516841607689501e-06,
      "loss": 5.3096,
      "step": 43
    },
    {
      "epoch": 0.5751633986928104,
      "grad_norm": 16.122995376586914,
      "learning_rate": 3.436516483539781e-06,
      "loss": 4.771,
      "step": 44
    },
    {
      "epoch": 0.5882352941176471,
      "grad_norm": 19.55790138244629,
      "learning_rate": 3.3550503583141726e-06,
      "loss": 4.5633,
      "step": 45
    },
    {
      "epoch": 0.6013071895424836,
      "grad_norm": 20.046592712402344,
      "learning_rate": 3.272542485937369e-06,
      "loss": 5.5604,
      "step": 46
    },
    {
      "epoch": 0.6143790849673203,
      "grad_norm": 19.207847595214844,
      "learning_rate": 3.189093389542498e-06,
      "loss": 5.6587,
      "step": 47
    },
    {
      "epoch": 0.6274509803921569,
      "grad_norm": 18.39757537841797,
      "learning_rate": 3.1048047389991693e-06,
      "loss": 5.0743,
      "step": 48
    },
    {
      "epoch": 0.6405228758169934,
      "grad_norm": 22.641244888305664,
      "learning_rate": 3.019779227044398e-06,
      "loss": 6.5356,
      "step": 49
    },
    {
      "epoch": 0.6535947712418301,
      "grad_norm": 21.692182540893555,
      "learning_rate": 2.9341204441673267e-06,
      "loss": 4.7347,
      "step": 50
    },
    {
      "epoch": 0.6535947712418301,
      "eval_loss": 5.583864212036133,
      "eval_runtime": 4.6346,
      "eval_samples_per_second": 27.834,
      "eval_steps_per_second": 7.12,
      "step": 50
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 20.410966873168945,
      "learning_rate": 2.847932752400164e-06,
      "loss": 5.2445,
      "step": 51
    },
    {
      "epoch": 0.6797385620915033,
      "grad_norm": 22.301136016845703,
      "learning_rate": 2.761321158169134e-06,
      "loss": 6.6047,
      "step": 52
    },
    {
      "epoch": 0.6928104575163399,
      "grad_norm": 26.021312713623047,
      "learning_rate": 2.6743911843603134e-06,
      "loss": 7.225,
      "step": 53
    },
    {
      "epoch": 0.7058823529411765,
      "grad_norm": 25.37548065185547,
      "learning_rate": 2.587248741756253e-06,
      "loss": 7.1886,
      "step": 54
    },
    {
      "epoch": 0.7189542483660131,
      "grad_norm": 24.194847106933594,
      "learning_rate": 2.5e-06,
      "loss": 7.4705,
      "step": 55
    },
    {
      "epoch": 0.7320261437908496,
      "grad_norm": 31.78634262084961,
      "learning_rate": 2.4127512582437486e-06,
      "loss": 7.6195,
      "step": 56
    },
    {
      "epoch": 0.7450980392156863,
      "grad_norm": 33.419776916503906,
      "learning_rate": 2.325608815639687e-06,
      "loss": 8.226,
      "step": 57
    },
    {
      "epoch": 0.7581699346405228,
      "grad_norm": 5.473607063293457,
      "learning_rate": 2.238678841830867e-06,
      "loss": 2.4786,
      "step": 58
    },
    {
      "epoch": 0.7712418300653595,
      "grad_norm": 7.471803665161133,
      "learning_rate": 2.1520672475998374e-06,
      "loss": 3.0997,
      "step": 59
    },
    {
      "epoch": 0.7843137254901961,
      "grad_norm": 11.024681091308594,
      "learning_rate": 2.0658795558326745e-06,
      "loss": 3.7581,
      "step": 60
    },
    {
      "epoch": 0.7973856209150327,
      "grad_norm": 16.229074478149414,
      "learning_rate": 1.9802207729556023e-06,
      "loss": 4.2436,
      "step": 61
    },
    {
      "epoch": 0.8104575163398693,
      "grad_norm": 16.74787139892578,
      "learning_rate": 1.895195261000831e-06,
      "loss": 4.3475,
      "step": 62
    },
    {
      "epoch": 0.8235294117647058,
      "grad_norm": 15.532852172851562,
      "learning_rate": 1.8109066104575023e-06,
      "loss": 4.1585,
      "step": 63
    },
    {
      "epoch": 0.8366013071895425,
      "grad_norm": 17.108413696289062,
      "learning_rate": 1.7274575140626318e-06,
      "loss": 4.5843,
      "step": 64
    },
    {
      "epoch": 0.8496732026143791,
      "grad_norm": 19.03700065612793,
      "learning_rate": 1.6449496416858285e-06,
      "loss": 4.5228,
      "step": 65
    },
    {
      "epoch": 0.8627450980392157,
      "grad_norm": 18.226158142089844,
      "learning_rate": 1.56348351646022e-06,
      "loss": 4.3519,
      "step": 66
    },
    {
      "epoch": 0.8758169934640523,
      "grad_norm": 17.857391357421875,
      "learning_rate": 1.4831583923105e-06,
      "loss": 4.3457,
      "step": 67
    },
    {
      "epoch": 0.8888888888888888,
      "grad_norm": 18.92527198791504,
      "learning_rate": 1.4040721330273063e-06,
      "loss": 4.9343,
      "step": 68
    },
    {
      "epoch": 0.9019607843137255,
      "grad_norm": 21.539527893066406,
      "learning_rate": 1.3263210930352737e-06,
      "loss": 4.5055,
      "step": 69
    },
    {
      "epoch": 0.9150326797385621,
      "grad_norm": 18.269908905029297,
      "learning_rate": 1.2500000000000007e-06,
      "loss": 4.9476,
      "step": 70
    },
    {
      "epoch": 0.9281045751633987,
      "grad_norm": 23.49378204345703,
      "learning_rate": 1.1752018394169882e-06,
      "loss": 5.0107,
      "step": 71
    },
    {
      "epoch": 0.9411764705882353,
      "grad_norm": 25.287702560424805,
      "learning_rate": 1.1020177413231334e-06,
      "loss": 5.5347,
      "step": 72
    },
    {
      "epoch": 0.954248366013072,
      "grad_norm": 21.355749130249023,
      "learning_rate": 1.0305368692688175e-06,
      "loss": 5.5062,
      "step": 73
    },
    {
      "epoch": 0.9673202614379085,
      "grad_norm": 25.698564529418945,
      "learning_rate": 9.608463116858544e-07,
      "loss": 6.4095,
      "step": 74
    },
    {
      "epoch": 0.9803921568627451,
      "grad_norm": 27.62260627746582,
      "learning_rate": 8.930309757836517e-07,
      "loss": 6.5146,
      "step": 75
    },
    {
      "epoch": 0.9934640522875817,
      "grad_norm": 28.68571662902832,
      "learning_rate": 8.271734841028553e-07,
      "loss": 6.5249,
      "step": 76
    },
    {
      "epoch": 1.0065359477124183,
      "grad_norm": 18.36124610900879,
      "learning_rate": 7.633540738525066e-07,
      "loss": 5.1321,
      "step": 77
    },
    {
      "epoch": 1.0196078431372548,
      "grad_norm": 6.567688941955566,
      "learning_rate": 7.016504991533727e-07,
      "loss": 2.5439,
      "step": 78
    },
    {
      "epoch": 1.0326797385620916,
      "grad_norm": 8.866209983825684,
      "learning_rate": 6.421379363065142e-07,
      "loss": 3.0478,
      "step": 79
    },
    {
      "epoch": 1.0457516339869282,
      "grad_norm": 13.741599082946777,
      "learning_rate": 5.848888922025553e-07,
      "loss": 3.9493,
      "step": 80
    },
    {
      "epoch": 1.0588235294117647,
      "grad_norm": 14.087339401245117,
      "learning_rate": 5.299731159831953e-07,
      "loss": 3.115,
      "step": 81
    },
    {
      "epoch": 1.0718954248366013,
      "grad_norm": 18.528522491455078,
      "learning_rate": 4.774575140626317e-07,
      "loss": 4.9909,
      "step": 82
    },
    {
      "epoch": 1.0849673202614378,
      "grad_norm": 17.856454849243164,
      "learning_rate": 4.27406068612396e-07,
      "loss": 4.0747,
      "step": 83
    },
    {
      "epoch": 1.0980392156862746,
      "grad_norm": 19.685808181762695,
      "learning_rate": 3.798797596089351e-07,
      "loss": 4.2566,
      "step": 84
    },
    {
      "epoch": 1.1111111111111112,
      "grad_norm": 18.7196102142334,
      "learning_rate": 3.3493649053890325e-07,
      "loss": 4.0926,
      "step": 85
    },
    {
      "epoch": 1.1241830065359477,
      "grad_norm": 18.23643684387207,
      "learning_rate": 2.9263101785268253e-07,
      "loss": 4.174,
      "step": 86
    },
    {
      "epoch": 1.1372549019607843,
      "grad_norm": 19.65151596069336,
      "learning_rate": 2.53014884252083e-07,
      "loss": 4.9878,
      "step": 87
    },
    {
      "epoch": 1.1503267973856208,
      "grad_norm": 20.80632781982422,
      "learning_rate": 2.1613635589349756e-07,
      "loss": 4.5691,
      "step": 88
    },
    {
      "epoch": 1.1633986928104576,
      "grad_norm": 19.64034652709961,
      "learning_rate": 1.8204036358303173e-07,
      "loss": 4.4191,
      "step": 89
    },
    {
      "epoch": 1.1764705882352942,
      "grad_norm": 20.260753631591797,
      "learning_rate": 1.507684480352292e-07,
      "loss": 4.5841,
      "step": 90
    },
    {
      "epoch": 1.1895424836601307,
      "grad_norm": 23.664470672607422,
      "learning_rate": 1.223587092621162e-07,
      "loss": 5.0259,
      "step": 91
    },
    {
      "epoch": 1.2026143790849673,
      "grad_norm": 23.812976837158203,
      "learning_rate": 9.684576015420277e-08,
      "loss": 5.6728,
      "step": 92
    },
    {
      "epoch": 1.215686274509804,
      "grad_norm": 29.3427791595459,
      "learning_rate": 7.426068431000883e-08,
      "loss": 7.135,
      "step": 93
    },
    {
      "epoch": 1.2287581699346406,
      "grad_norm": 26.24110221862793,
      "learning_rate": 5.463099816548578e-08,
      "loss": 6.2265,
      "step": 94
    },
    {
      "epoch": 1.2418300653594772,
      "grad_norm": 32.70111846923828,
      "learning_rate": 3.798061746947995e-08,
      "loss": 6.5576,
      "step": 95
    },
    {
      "epoch": 1.2549019607843137,
      "grad_norm": 19.617277145385742,
      "learning_rate": 2.4329828146074096e-08,
      "loss": 4.6098,
      "step": 96
    },
    {
      "epoch": 1.2679738562091503,
      "grad_norm": 7.279455661773682,
      "learning_rate": 1.3695261579316776e-08,
      "loss": 2.8293,
      "step": 97
    },
    {
      "epoch": 1.2810457516339868,
      "grad_norm": 9.745218276977539,
      "learning_rate": 6.089874350439507e-09,
      "loss": 3.1096,
      "step": 98
    },
    {
      "epoch": 1.2941176470588236,
      "grad_norm": 16.444101333618164,
      "learning_rate": 1.5229324522605949e-09,
      "loss": 4.1709,
      "step": 99
    },
    {
      "epoch": 1.3071895424836601,
      "grad_norm": 16.58370018005371,
      "learning_rate": 0.0,
      "loss": 4.0866,
      "step": 100
    },
    {
      "epoch": 1.3071895424836601,
      "eval_loss": 4.5732855796813965,
      "eval_runtime": 4.6201,
      "eval_samples_per_second": 27.921,
      "eval_steps_per_second": 7.143,
      "step": 100
    }
  ],
  "logging_steps": 1,
  "max_steps": 100,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "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": 3.931412223963955e+16,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}