{
  "best_metric": 9.01811695098877,
  "best_model_checkpoint": "miner_id_24/checkpoint-100",
  "epoch": 2.6084142394822005,
  "eval_steps": 50,
  "global_step": 100,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.025889967637540454,
      "grad_norm": 11.74299144744873,
      "learning_rate": 5.000000000000001e-07,
      "loss": 9.4965,
      "step": 1
    },
    {
      "epoch": 0.025889967637540454,
      "eval_loss": 9.478606224060059,
      "eval_runtime": 0.4121,
      "eval_samples_per_second": 157.735,
      "eval_steps_per_second": 41.254,
      "step": 1
    },
    {
      "epoch": 0.05177993527508091,
      "grad_norm": 11.671175003051758,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 9.5064,
      "step": 2
    },
    {
      "epoch": 0.07766990291262135,
      "grad_norm": 12.935548782348633,
      "learning_rate": 1.5e-06,
      "loss": 9.3157,
      "step": 3
    },
    {
      "epoch": 0.10355987055016182,
      "grad_norm": 13.709819793701172,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 9.7602,
      "step": 4
    },
    {
      "epoch": 0.12944983818770225,
      "grad_norm": 13.389347076416016,
      "learning_rate": 2.5e-06,
      "loss": 9.5181,
      "step": 5
    },
    {
      "epoch": 0.1553398058252427,
      "grad_norm": 12.952323913574219,
      "learning_rate": 3e-06,
      "loss": 9.5492,
      "step": 6
    },
    {
      "epoch": 0.18122977346278318,
      "grad_norm": 12.68387508392334,
      "learning_rate": 3.5e-06,
      "loss": 9.5063,
      "step": 7
    },
    {
      "epoch": 0.20711974110032363,
      "grad_norm": 12.480040550231934,
      "learning_rate": 4.000000000000001e-06,
      "loss": 9.5266,
      "step": 8
    },
    {
      "epoch": 0.23300970873786409,
      "grad_norm": 12.971917152404785,
      "learning_rate": 4.5e-06,
      "loss": 9.4577,
      "step": 9
    },
    {
      "epoch": 0.2588996763754045,
      "grad_norm": 11.367093086242676,
      "learning_rate": 5e-06,
      "loss": 9.6127,
      "step": 10
    },
    {
      "epoch": 0.284789644012945,
      "grad_norm": 11.295645713806152,
      "learning_rate": 4.99847706754774e-06,
      "loss": 9.3534,
      "step": 11
    },
    {
      "epoch": 0.3106796116504854,
      "grad_norm": 12.023831367492676,
      "learning_rate": 4.993910125649561e-06,
      "loss": 9.055,
      "step": 12
    },
    {
      "epoch": 0.3365695792880259,
      "grad_norm": 12.505583763122559,
      "learning_rate": 4.986304738420684e-06,
      "loss": 9.3667,
      "step": 13
    },
    {
      "epoch": 0.36245954692556637,
      "grad_norm": 12.947257041931152,
      "learning_rate": 4.975670171853926e-06,
      "loss": 9.396,
      "step": 14
    },
    {
      "epoch": 0.3883495145631068,
      "grad_norm": 12.820213317871094,
      "learning_rate": 4.962019382530521e-06,
      "loss": 9.3142,
      "step": 15
    },
    {
      "epoch": 0.41423948220064727,
      "grad_norm": 11.85090160369873,
      "learning_rate": 4.9453690018345144e-06,
      "loss": 9.3381,
      "step": 16
    },
    {
      "epoch": 0.4401294498381877,
      "grad_norm": 14.81877613067627,
      "learning_rate": 4.925739315689991e-06,
      "loss": 9.519,
      "step": 17
    },
    {
      "epoch": 0.46601941747572817,
      "grad_norm": 14.836597442626953,
      "learning_rate": 4.903154239845798e-06,
      "loss": 9.4054,
      "step": 18
    },
    {
      "epoch": 0.4919093851132686,
      "grad_norm": 12.943735122680664,
      "learning_rate": 4.8776412907378845e-06,
      "loss": 9.4375,
      "step": 19
    },
    {
      "epoch": 0.517799352750809,
      "grad_norm": 12.675292015075684,
      "learning_rate": 4.849231551964771e-06,
      "loss": 9.3666,
      "step": 20
    },
    {
      "epoch": 0.5436893203883495,
      "grad_norm": 11.758707046508789,
      "learning_rate": 4.817959636416969e-06,
      "loss": 9.3973,
      "step": 21
    },
    {
      "epoch": 0.56957928802589,
      "grad_norm": 12.52666187286377,
      "learning_rate": 4.783863644106502e-06,
      "loss": 9.3869,
      "step": 22
    },
    {
      "epoch": 0.5954692556634305,
      "grad_norm": 13.23238468170166,
      "learning_rate": 4.746985115747918e-06,
      "loss": 9.423,
      "step": 23
    },
    {
      "epoch": 0.6213592233009708,
      "grad_norm": 11.95095443725586,
      "learning_rate": 4.707368982147318e-06,
      "loss": 9.3316,
      "step": 24
    },
    {
      "epoch": 0.6472491909385113,
      "grad_norm": 12.786251068115234,
      "learning_rate": 4.665063509461098e-06,
      "loss": 9.2699,
      "step": 25
    },
    {
      "epoch": 0.6731391585760518,
      "grad_norm": 11.97091293334961,
      "learning_rate": 4.620120240391065e-06,
      "loss": 9.2894,
      "step": 26
    },
    {
      "epoch": 0.6990291262135923,
      "grad_norm": 13.269583702087402,
      "learning_rate": 4.572593931387604e-06,
      "loss": 9.3116,
      "step": 27
    },
    {
      "epoch": 0.7249190938511327,
      "grad_norm": 11.793617248535156,
      "learning_rate": 4.522542485937369e-06,
      "loss": 9.3654,
      "step": 28
    },
    {
      "epoch": 0.7508090614886731,
      "grad_norm": 11.819991111755371,
      "learning_rate": 4.470026884016805e-06,
      "loss": 9.2624,
      "step": 29
    },
    {
      "epoch": 0.7766990291262136,
      "grad_norm": 13.980379104614258,
      "learning_rate": 4.415111107797445e-06,
      "loss": 9.2752,
      "step": 30
    },
    {
      "epoch": 0.8025889967637541,
      "grad_norm": 15.096542358398438,
      "learning_rate": 4.357862063693486e-06,
      "loss": 9.3893,
      "step": 31
    },
    {
      "epoch": 0.8284789644012945,
      "grad_norm": 12.700614929199219,
      "learning_rate": 4.2983495008466285e-06,
      "loss": 9.2669,
      "step": 32
    },
    {
      "epoch": 0.8543689320388349,
      "grad_norm": 13.240720748901367,
      "learning_rate": 4.236645926147493e-06,
      "loss": 9.1943,
      "step": 33
    },
    {
      "epoch": 0.8802588996763754,
      "grad_norm": 12.622847557067871,
      "learning_rate": 4.172826515897146e-06,
      "loss": 9.2701,
      "step": 34
    },
    {
      "epoch": 0.9061488673139159,
      "grad_norm": 14.08974552154541,
      "learning_rate": 4.106969024216348e-06,
      "loss": 9.4026,
      "step": 35
    },
    {
      "epoch": 0.9320388349514563,
      "grad_norm": 12.45661449432373,
      "learning_rate": 4.039153688314146e-06,
      "loss": 9.2997,
      "step": 36
    },
    {
      "epoch": 0.9579288025889967,
      "grad_norm": 12.231471061706543,
      "learning_rate": 3.969463130731183e-06,
      "loss": 9.1133,
      "step": 37
    },
    {
      "epoch": 0.9838187702265372,
      "grad_norm": 13.44810962677002,
      "learning_rate": 3.897982258676867e-06,
      "loss": 9.3883,
      "step": 38
    },
    {
      "epoch": 1.0194174757281553,
      "grad_norm": 20.87478256225586,
      "learning_rate": 3.824798160583012e-06,
      "loss": 15.709,
      "step": 39
    },
    {
      "epoch": 1.0453074433656957,
      "grad_norm": 11.6742525100708,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 9.1093,
      "step": 40
    },
    {
      "epoch": 1.0711974110032363,
      "grad_norm": 11.33270263671875,
      "learning_rate": 3.6736789069647273e-06,
      "loss": 9.1661,
      "step": 41
    },
    {
      "epoch": 1.0970873786407767,
      "grad_norm": 14.351101875305176,
      "learning_rate": 3.595927866972694e-06,
      "loss": 9.304,
      "step": 42
    },
    {
      "epoch": 1.1229773462783172,
      "grad_norm": 14.262197494506836,
      "learning_rate": 3.516841607689501e-06,
      "loss": 9.3505,
      "step": 43
    },
    {
      "epoch": 1.1488673139158576,
      "grad_norm": 12.826080322265625,
      "learning_rate": 3.436516483539781e-06,
      "loss": 9.0723,
      "step": 44
    },
    {
      "epoch": 1.174757281553398,
      "grad_norm": 13.243085861206055,
      "learning_rate": 3.3550503583141726e-06,
      "loss": 9.1447,
      "step": 45
    },
    {
      "epoch": 1.2006472491909386,
      "grad_norm": 12.890806198120117,
      "learning_rate": 3.272542485937369e-06,
      "loss": 9.1691,
      "step": 46
    },
    {
      "epoch": 1.226537216828479,
      "grad_norm": 14.562002182006836,
      "learning_rate": 3.189093389542498e-06,
      "loss": 9.5301,
      "step": 47
    },
    {
      "epoch": 1.2524271844660193,
      "grad_norm": 11.241775512695312,
      "learning_rate": 3.1048047389991693e-06,
      "loss": 9.257,
      "step": 48
    },
    {
      "epoch": 1.27831715210356,
      "grad_norm": 11.758941650390625,
      "learning_rate": 3.019779227044398e-06,
      "loss": 9.2089,
      "step": 49
    },
    {
      "epoch": 1.3042071197411003,
      "grad_norm": 13.430214881896973,
      "learning_rate": 2.9341204441673267e-06,
      "loss": 9.0806,
      "step": 50
    },
    {
      "epoch": 1.3042071197411003,
      "eval_loss": 9.157350540161133,
      "eval_runtime": 0.4117,
      "eval_samples_per_second": 157.863,
      "eval_steps_per_second": 41.287,
      "step": 50
    },
    {
      "epoch": 1.3300970873786409,
      "grad_norm": 12.582650184631348,
      "learning_rate": 2.847932752400164e-06,
      "loss": 9.1575,
      "step": 51
    },
    {
      "epoch": 1.3559870550161812,
      "grad_norm": 13.364744186401367,
      "learning_rate": 2.761321158169134e-06,
      "loss": 9.4123,
      "step": 52
    },
    {
      "epoch": 1.3818770226537218,
      "grad_norm": 11.977982521057129,
      "learning_rate": 2.6743911843603134e-06,
      "loss": 9.1648,
      "step": 53
    },
    {
      "epoch": 1.4077669902912622,
      "grad_norm": 13.052461624145508,
      "learning_rate": 2.587248741756253e-06,
      "loss": 9.0674,
      "step": 54
    },
    {
      "epoch": 1.4336569579288025,
      "grad_norm": 13.613630294799805,
      "learning_rate": 2.5e-06,
      "loss": 9.1512,
      "step": 55
    },
    {
      "epoch": 1.4595469255663431,
      "grad_norm": 13.225386619567871,
      "learning_rate": 2.4127512582437486e-06,
      "loss": 9.1321,
      "step": 56
    },
    {
      "epoch": 1.4854368932038835,
      "grad_norm": 12.972779273986816,
      "learning_rate": 2.325608815639687e-06,
      "loss": 8.9664,
      "step": 57
    },
    {
      "epoch": 1.5113268608414239,
      "grad_norm": 13.470908164978027,
      "learning_rate": 2.238678841830867e-06,
      "loss": 9.1312,
      "step": 58
    },
    {
      "epoch": 1.5372168284789645,
      "grad_norm": 12.796125411987305,
      "learning_rate": 2.1520672475998374e-06,
      "loss": 8.9491,
      "step": 59
    },
    {
      "epoch": 1.5631067961165048,
      "grad_norm": 11.87038516998291,
      "learning_rate": 2.0658795558326745e-06,
      "loss": 8.9343,
      "step": 60
    },
    {
      "epoch": 1.5889967637540452,
      "grad_norm": 14.072746276855469,
      "learning_rate": 1.9802207729556023e-06,
      "loss": 8.988,
      "step": 61
    },
    {
      "epoch": 1.6148867313915858,
      "grad_norm": 13.03939151763916,
      "learning_rate": 1.895195261000831e-06,
      "loss": 8.9825,
      "step": 62
    },
    {
      "epoch": 1.6407766990291264,
      "grad_norm": 12.500960350036621,
      "learning_rate": 1.8109066104575023e-06,
      "loss": 8.9965,
      "step": 63
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 13.652578353881836,
      "learning_rate": 1.7274575140626318e-06,
      "loss": 9.1256,
      "step": 64
    },
    {
      "epoch": 1.692556634304207,
      "grad_norm": 13.726714134216309,
      "learning_rate": 1.6449496416858285e-06,
      "loss": 8.875,
      "step": 65
    },
    {
      "epoch": 1.7184466019417477,
      "grad_norm": 11.982444763183594,
      "learning_rate": 1.56348351646022e-06,
      "loss": 9.0407,
      "step": 66
    },
    {
      "epoch": 1.744336569579288,
      "grad_norm": 11.878983497619629,
      "learning_rate": 1.4831583923105e-06,
      "loss": 8.9615,
      "step": 67
    },
    {
      "epoch": 1.7702265372168284,
      "grad_norm": 12.483765602111816,
      "learning_rate": 1.4040721330273063e-06,
      "loss": 8.9653,
      "step": 68
    },
    {
      "epoch": 1.796116504854369,
      "grad_norm": 12.622501373291016,
      "learning_rate": 1.3263210930352737e-06,
      "loss": 8.8598,
      "step": 69
    },
    {
      "epoch": 1.8220064724919094,
      "grad_norm": 13.878958702087402,
      "learning_rate": 1.2500000000000007e-06,
      "loss": 9.0102,
      "step": 70
    },
    {
      "epoch": 1.8478964401294498,
      "grad_norm": 12.5238676071167,
      "learning_rate": 1.1752018394169882e-06,
      "loss": 9.0324,
      "step": 71
    },
    {
      "epoch": 1.8737864077669903,
      "grad_norm": 13.529470443725586,
      "learning_rate": 1.1020177413231334e-06,
      "loss": 8.9774,
      "step": 72
    },
    {
      "epoch": 1.8996763754045307,
      "grad_norm": 12.65139389038086,
      "learning_rate": 1.0305368692688175e-06,
      "loss": 8.9267,
      "step": 73
    },
    {
      "epoch": 1.925566343042071,
      "grad_norm": 14.668947219848633,
      "learning_rate": 9.608463116858544e-07,
      "loss": 9.1233,
      "step": 74
    },
    {
      "epoch": 1.9514563106796117,
      "grad_norm": 12.97563362121582,
      "learning_rate": 8.930309757836517e-07,
      "loss": 9.0197,
      "step": 75
    },
    {
      "epoch": 1.9773462783171523,
      "grad_norm": 13.905110359191895,
      "learning_rate": 8.271734841028553e-07,
      "loss": 9.0242,
      "step": 76
    },
    {
      "epoch": 2.01294498381877,
      "grad_norm": 23.526731491088867,
      "learning_rate": 7.633540738525066e-07,
      "loss": 15.7391,
      "step": 77
    },
    {
      "epoch": 2.0388349514563107,
      "grad_norm": 12.419938087463379,
      "learning_rate": 7.016504991533727e-07,
      "loss": 8.9641,
      "step": 78
    },
    {
      "epoch": 2.0647249190938513,
      "grad_norm": 12.351256370544434,
      "learning_rate": 6.421379363065142e-07,
      "loss": 8.9708,
      "step": 79
    },
    {
      "epoch": 2.0906148867313914,
      "grad_norm": 13.134671211242676,
      "learning_rate": 5.848888922025553e-07,
      "loss": 8.6328,
      "step": 80
    },
    {
      "epoch": 2.116504854368932,
      "grad_norm": 13.836918830871582,
      "learning_rate": 5.299731159831953e-07,
      "loss": 9.0935,
      "step": 81
    },
    {
      "epoch": 2.1423948220064726,
      "grad_norm": 14.656155586242676,
      "learning_rate": 4.774575140626317e-07,
      "loss": 9.0881,
      "step": 82
    },
    {
      "epoch": 2.168284789644013,
      "grad_norm": 13.374807357788086,
      "learning_rate": 4.27406068612396e-07,
      "loss": 8.9988,
      "step": 83
    },
    {
      "epoch": 2.1941747572815533,
      "grad_norm": 12.77652645111084,
      "learning_rate": 3.798797596089351e-07,
      "loss": 9.1808,
      "step": 84
    },
    {
      "epoch": 2.220064724919094,
      "grad_norm": 14.176050186157227,
      "learning_rate": 3.3493649053890325e-07,
      "loss": 8.8116,
      "step": 85
    },
    {
      "epoch": 2.2459546925566345,
      "grad_norm": 12.411694526672363,
      "learning_rate": 2.9263101785268253e-07,
      "loss": 9.0052,
      "step": 86
    },
    {
      "epoch": 2.2718446601941746,
      "grad_norm": 11.73877239227295,
      "learning_rate": 2.53014884252083e-07,
      "loss": 8.9109,
      "step": 87
    },
    {
      "epoch": 2.2977346278317152,
      "grad_norm": 12.658856391906738,
      "learning_rate": 2.1613635589349756e-07,
      "loss": 8.909,
      "step": 88
    },
    {
      "epoch": 2.323624595469256,
      "grad_norm": 12.234189987182617,
      "learning_rate": 1.8204036358303173e-07,
      "loss": 8.8621,
      "step": 89
    },
    {
      "epoch": 2.349514563106796,
      "grad_norm": 13.402729988098145,
      "learning_rate": 1.507684480352292e-07,
      "loss": 9.1459,
      "step": 90
    },
    {
      "epoch": 2.3754045307443366,
      "grad_norm": 12.723470687866211,
      "learning_rate": 1.223587092621162e-07,
      "loss": 9.0186,
      "step": 91
    },
    {
      "epoch": 2.401294498381877,
      "grad_norm": 13.641341209411621,
      "learning_rate": 9.684576015420277e-08,
      "loss": 9.2626,
      "step": 92
    },
    {
      "epoch": 2.4271844660194173,
      "grad_norm": 12.847984313964844,
      "learning_rate": 7.426068431000883e-08,
      "loss": 8.6923,
      "step": 93
    },
    {
      "epoch": 2.453074433656958,
      "grad_norm": 13.19314956665039,
      "learning_rate": 5.463099816548578e-08,
      "loss": 9.08,
      "step": 94
    },
    {
      "epoch": 2.4789644012944985,
      "grad_norm": 13.388517379760742,
      "learning_rate": 3.798061746947995e-08,
      "loss": 9.3973,
      "step": 95
    },
    {
      "epoch": 2.5048543689320386,
      "grad_norm": 12.82660961151123,
      "learning_rate": 2.4329828146074096e-08,
      "loss": 9.0044,
      "step": 96
    },
    {
      "epoch": 2.530744336569579,
      "grad_norm": 13.584592819213867,
      "learning_rate": 1.3695261579316776e-08,
      "loss": 9.0041,
      "step": 97
    },
    {
      "epoch": 2.55663430420712,
      "grad_norm": 12.776042938232422,
      "learning_rate": 6.089874350439507e-09,
      "loss": 9.1079,
      "step": 98
    },
    {
      "epoch": 2.58252427184466,
      "grad_norm": 12.881516456604004,
      "learning_rate": 1.5229324522605949e-09,
      "loss": 9.0031,
      "step": 99
    },
    {
      "epoch": 2.6084142394822005,
      "grad_norm": 13.592841148376465,
      "learning_rate": 0.0,
      "loss": 9.0054,
      "step": 100
    },
    {
      "epoch": 2.6084142394822005,
      "eval_loss": 9.01811695098877,
      "eval_runtime": 0.4122,
      "eval_samples_per_second": 157.701,
      "eval_steps_per_second": 41.245,
      "step": 100
    }
  ],
  "logging_steps": 1,
  "max_steps": 100,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "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": 54299040153600.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}