{
  "best_metric": 0.7186845541000366,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 2.150537634408602,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.010752688172043012,
      "grad_norm": 4.536034107208252,
      "learning_rate": 7e-06,
      "loss": 8.1083,
      "step": 1
    },
    {
      "epoch": 0.010752688172043012,
      "eval_loss": 2.7238574028015137,
      "eval_runtime": 14.4588,
      "eval_samples_per_second": 10.858,
      "eval_steps_per_second": 2.766,
      "step": 1
    },
    {
      "epoch": 0.021505376344086023,
      "grad_norm": 5.978432655334473,
      "learning_rate": 1.4e-05,
      "loss": 10.3139,
      "step": 2
    },
    {
      "epoch": 0.03225806451612903,
      "grad_norm": 7.468502998352051,
      "learning_rate": 2.1e-05,
      "loss": 10.007,
      "step": 3
    },
    {
      "epoch": 0.043010752688172046,
      "grad_norm": 12.983536720275879,
      "learning_rate": 2.8e-05,
      "loss": 9.7253,
      "step": 4
    },
    {
      "epoch": 0.053763440860215055,
      "grad_norm": 13.686447143554688,
      "learning_rate": 3.5e-05,
      "loss": 9.548,
      "step": 5
    },
    {
      "epoch": 0.06451612903225806,
      "grad_norm": 9.852103233337402,
      "learning_rate": 4.2e-05,
      "loss": 9.0148,
      "step": 6
    },
    {
      "epoch": 0.07526881720430108,
      "grad_norm": 7.126467704772949,
      "learning_rate": 4.899999999999999e-05,
      "loss": 8.349,
      "step": 7
    },
    {
      "epoch": 0.08602150537634409,
      "grad_norm": 6.550177574157715,
      "learning_rate": 5.6e-05,
      "loss": 7.4971,
      "step": 8
    },
    {
      "epoch": 0.0967741935483871,
      "grad_norm": 10.200550079345703,
      "learning_rate": 6.3e-05,
      "loss": 5.8202,
      "step": 9
    },
    {
      "epoch": 0.10752688172043011,
      "grad_norm": 7.892305850982666,
      "learning_rate": 7e-05,
      "loss": 4.6753,
      "step": 10
    },
    {
      "epoch": 0.11827956989247312,
      "grad_norm": 19.061233520507812,
      "learning_rate": 6.999521567473641e-05,
      "loss": 5.6292,
      "step": 11
    },
    {
      "epoch": 0.12903225806451613,
      "grad_norm": 14.547859191894531,
      "learning_rate": 6.998086400693241e-05,
      "loss": 5.6758,
      "step": 12
    },
    {
      "epoch": 0.13978494623655913,
      "grad_norm": 9.280782699584961,
      "learning_rate": 6.995694892019065e-05,
      "loss": 4.7095,
      "step": 13
    },
    {
      "epoch": 0.15053763440860216,
      "grad_norm": 8.161127090454102,
      "learning_rate": 6.99234769526571e-05,
      "loss": 3.4122,
      "step": 14
    },
    {
      "epoch": 0.16129032258064516,
      "grad_norm": 34.015987396240234,
      "learning_rate": 6.988045725523343e-05,
      "loss": 4.0857,
      "step": 15
    },
    {
      "epoch": 0.17204301075268819,
      "grad_norm": 24.155447006225586,
      "learning_rate": 6.982790158907539e-05,
      "loss": 4.2336,
      "step": 16
    },
    {
      "epoch": 0.1827956989247312,
      "grad_norm": 5.08544397354126,
      "learning_rate": 6.976582432237733e-05,
      "loss": 4.5834,
      "step": 17
    },
    {
      "epoch": 0.1935483870967742,
      "grad_norm": 13.041632652282715,
      "learning_rate": 6.969424242644413e-05,
      "loss": 3.4418,
      "step": 18
    },
    {
      "epoch": 0.20430107526881722,
      "grad_norm": 5.957951545715332,
      "learning_rate": 6.961317547105138e-05,
      "loss": 3.8441,
      "step": 19
    },
    {
      "epoch": 0.21505376344086022,
      "grad_norm": 5.243581771850586,
      "learning_rate": 6.952264561909527e-05,
      "loss": 3.2856,
      "step": 20
    },
    {
      "epoch": 0.22580645161290322,
      "grad_norm": 7.811934471130371,
      "learning_rate": 6.942267762053337e-05,
      "loss": 4.6588,
      "step": 21
    },
    {
      "epoch": 0.23655913978494625,
      "grad_norm": 11.206618309020996,
      "learning_rate": 6.931329880561832e-05,
      "loss": 4.762,
      "step": 22
    },
    {
      "epoch": 0.24731182795698925,
      "grad_norm": 11.149868965148926,
      "learning_rate": 6.919453907742597e-05,
      "loss": 2.4062,
      "step": 23
    },
    {
      "epoch": 0.25806451612903225,
      "grad_norm": 11.476567268371582,
      "learning_rate": 6.90664309036802e-05,
      "loss": 9.3109,
      "step": 24
    },
    {
      "epoch": 0.26881720430107525,
      "grad_norm": 9.604092597961426,
      "learning_rate": 6.892900930787656e-05,
      "loss": 7.829,
      "step": 25
    },
    {
      "epoch": 0.27956989247311825,
      "grad_norm": 6.5685601234436035,
      "learning_rate": 6.87823118597072e-05,
      "loss": 6.88,
      "step": 26
    },
    {
      "epoch": 0.2903225806451613,
      "grad_norm": 9.324129104614258,
      "learning_rate": 6.862637866478969e-05,
      "loss": 7.8981,
      "step": 27
    },
    {
      "epoch": 0.3010752688172043,
      "grad_norm": 8.132095336914062,
      "learning_rate": 6.846125235370252e-05,
      "loss": 5.5197,
      "step": 28
    },
    {
      "epoch": 0.3118279569892473,
      "grad_norm": 5.731638431549072,
      "learning_rate": 6.828697807033038e-05,
      "loss": 6.1606,
      "step": 29
    },
    {
      "epoch": 0.3225806451612903,
      "grad_norm": 3.5606529712677,
      "learning_rate": 6.81036034595222e-05,
      "loss": 5.3749,
      "step": 30
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 3.248960256576538,
      "learning_rate": 6.791117865406564e-05,
      "loss": 5.605,
      "step": 31
    },
    {
      "epoch": 0.34408602150537637,
      "grad_norm": 2.7672736644744873,
      "learning_rate": 6.770975626098112e-05,
      "loss": 3.8032,
      "step": 32
    },
    {
      "epoch": 0.3548387096774194,
      "grad_norm": 2.312937021255493,
      "learning_rate": 6.749939134713974e-05,
      "loss": 2.5618,
      "step": 33
    },
    {
      "epoch": 0.3655913978494624,
      "grad_norm": 2.682342290878296,
      "learning_rate": 6.728014142420846e-05,
      "loss": 3.8125,
      "step": 34
    },
    {
      "epoch": 0.3763440860215054,
      "grad_norm": 3.345135450363159,
      "learning_rate": 6.7052066432927e-05,
      "loss": 3.4451,
      "step": 35
    },
    {
      "epoch": 0.3870967741935484,
      "grad_norm": 2.8873608112335205,
      "learning_rate": 6.681522872672069e-05,
      "loss": 3.7995,
      "step": 36
    },
    {
      "epoch": 0.3978494623655914,
      "grad_norm": 2.9758994579315186,
      "learning_rate": 6.656969305465356e-05,
      "loss": 1.916,
      "step": 37
    },
    {
      "epoch": 0.40860215053763443,
      "grad_norm": 2.9894485473632812,
      "learning_rate": 6.631552654372672e-05,
      "loss": 2.6401,
      "step": 38
    },
    {
      "epoch": 0.41935483870967744,
      "grad_norm": 2.8846404552459717,
      "learning_rate": 6.60527986805264e-05,
      "loss": 3.5161,
      "step": 39
    },
    {
      "epoch": 0.43010752688172044,
      "grad_norm": 2.933262348175049,
      "learning_rate": 6.578158129222711e-05,
      "loss": 3.2549,
      "step": 40
    },
    {
      "epoch": 0.44086021505376344,
      "grad_norm": 3.833869218826294,
      "learning_rate": 6.550194852695469e-05,
      "loss": 3.841,
      "step": 41
    },
    {
      "epoch": 0.45161290322580644,
      "grad_norm": 3.596083402633667,
      "learning_rate": 6.521397683351509e-05,
      "loss": 2.6962,
      "step": 42
    },
    {
      "epoch": 0.46236559139784944,
      "grad_norm": 6.494784355163574,
      "learning_rate": 6.491774494049386e-05,
      "loss": 3.7265,
      "step": 43
    },
    {
      "epoch": 0.4731182795698925,
      "grad_norm": 3.249906063079834,
      "learning_rate": 6.461333383473272e-05,
      "loss": 1.9618,
      "step": 44
    },
    {
      "epoch": 0.4838709677419355,
      "grad_norm": 6.374508380889893,
      "learning_rate": 6.430082673918849e-05,
      "loss": 3.3468,
      "step": 45
    },
    {
      "epoch": 0.4946236559139785,
      "grad_norm": 7.742399215698242,
      "learning_rate": 6.398030909018069e-05,
      "loss": 2.4064,
      "step": 46
    },
    {
      "epoch": 0.5053763440860215,
      "grad_norm": 5.071364879608154,
      "learning_rate": 6.365186851403423e-05,
      "loss": 6.4741,
      "step": 47
    },
    {
      "epoch": 0.5161290322580645,
      "grad_norm": 4.5256571769714355,
      "learning_rate": 6.331559480312315e-05,
      "loss": 6.4723,
      "step": 48
    },
    {
      "epoch": 0.5268817204301075,
      "grad_norm": 3.286426305770874,
      "learning_rate": 6.297157989132236e-05,
      "loss": 5.6636,
      "step": 49
    },
    {
      "epoch": 0.5376344086021505,
      "grad_norm": 2.9677677154541016,
      "learning_rate": 6.261991782887377e-05,
      "loss": 4.5262,
      "step": 50
    },
    {
      "epoch": 0.5376344086021505,
      "eval_loss": 1.031134843826294,
      "eval_runtime": 14.7506,
      "eval_samples_per_second": 10.644,
      "eval_steps_per_second": 2.712,
      "step": 50
    },
    {
      "epoch": 0.5483870967741935,
      "grad_norm": 3.2802512645721436,
      "learning_rate": 6.226070475667393e-05,
      "loss": 5.2625,
      "step": 51
    },
    {
      "epoch": 0.5591397849462365,
      "grad_norm": 3.9532055854797363,
      "learning_rate": 6.189403887999006e-05,
      "loss": 4.5323,
      "step": 52
    },
    {
      "epoch": 0.5698924731182796,
      "grad_norm": 4.0816731452941895,
      "learning_rate": 6.152002044161171e-05,
      "loss": 5.2472,
      "step": 53
    },
    {
      "epoch": 0.5806451612903226,
      "grad_norm": 3.2194178104400635,
      "learning_rate": 6.113875169444539e-05,
      "loss": 4.1174,
      "step": 54
    },
    {
      "epoch": 0.5913978494623656,
      "grad_norm": 3.1189627647399902,
      "learning_rate": 6.0750336873559605e-05,
      "loss": 3.5628,
      "step": 55
    },
    {
      "epoch": 0.6021505376344086,
      "grad_norm": 3.0379552841186523,
      "learning_rate": 6.035488216768811e-05,
      "loss": 2.5643,
      "step": 56
    },
    {
      "epoch": 0.6129032258064516,
      "grad_norm": 2.96771240234375,
      "learning_rate": 5.9952495690198894e-05,
      "loss": 3.8131,
      "step": 57
    },
    {
      "epoch": 0.6236559139784946,
      "grad_norm": 3.369340181350708,
      "learning_rate": 5.954328744953709e-05,
      "loss": 2.6112,
      "step": 58
    },
    {
      "epoch": 0.6344086021505376,
      "grad_norm": 2.662733793258667,
      "learning_rate": 5.91273693191498e-05,
      "loss": 2.678,
      "step": 59
    },
    {
      "epoch": 0.6451612903225806,
      "grad_norm": 2.3960416316986084,
      "learning_rate": 5.870485500690094e-05,
      "loss": 1.9448,
      "step": 60
    },
    {
      "epoch": 0.6559139784946236,
      "grad_norm": 3.0840823650360107,
      "learning_rate": 5.827586002398468e-05,
      "loss": 3.0983,
      "step": 61
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 4.117824077606201,
      "learning_rate": 5.784050165334589e-05,
      "loss": 2.1824,
      "step": 62
    },
    {
      "epoch": 0.6774193548387096,
      "grad_norm": 3.039947271347046,
      "learning_rate": 5.739889891761608e-05,
      "loss": 3.4342,
      "step": 63
    },
    {
      "epoch": 0.6881720430107527,
      "grad_norm": 3.3664510250091553,
      "learning_rate": 5.6951172546573794e-05,
      "loss": 2.7055,
      "step": 64
    },
    {
      "epoch": 0.6989247311827957,
      "grad_norm": 7.528428077697754,
      "learning_rate": 5.6497444944138376e-05,
      "loss": 3.2575,
      "step": 65
    },
    {
      "epoch": 0.7096774193548387,
      "grad_norm": 2.7763924598693848,
      "learning_rate": 5.603784015490587e-05,
      "loss": 2.7032,
      "step": 66
    },
    {
      "epoch": 0.7204301075268817,
      "grad_norm": 4.630373001098633,
      "learning_rate": 5.557248383023655e-05,
      "loss": 3.1262,
      "step": 67
    },
    {
      "epoch": 0.7311827956989247,
      "grad_norm": 4.628628730773926,
      "learning_rate": 5.510150319390302e-05,
      "loss": 2.4391,
      "step": 68
    },
    {
      "epoch": 0.7419354838709677,
      "grad_norm": 6.389301300048828,
      "learning_rate": 5.4625027007308546e-05,
      "loss": 1.6421,
      "step": 69
    },
    {
      "epoch": 0.7526881720430108,
      "grad_norm": 3.937209129333496,
      "learning_rate": 5.414318553428494e-05,
      "loss": 6.887,
      "step": 70
    },
    {
      "epoch": 0.7634408602150538,
      "grad_norm": 3.489776611328125,
      "learning_rate": 5.3656110505479776e-05,
      "loss": 5.776,
      "step": 71
    },
    {
      "epoch": 0.7741935483870968,
      "grad_norm": 3.345602035522461,
      "learning_rate": 5.316393508234253e-05,
      "loss": 5.4254,
      "step": 72
    },
    {
      "epoch": 0.7849462365591398,
      "grad_norm": 2.5286366939544678,
      "learning_rate": 5.266679382071953e-05,
      "loss": 5.4602,
      "step": 73
    },
    {
      "epoch": 0.7956989247311828,
      "grad_norm": 2.6356401443481445,
      "learning_rate": 5.216482263406778e-05,
      "loss": 4.1083,
      "step": 74
    },
    {
      "epoch": 0.8064516129032258,
      "grad_norm": 2.794057607650757,
      "learning_rate": 5.1658158756297576e-05,
      "loss": 4.0295,
      "step": 75
    },
    {
      "epoch": 0.8172043010752689,
      "grad_norm": 3.1053338050842285,
      "learning_rate": 5.114694070425407e-05,
      "loss": 4.2456,
      "step": 76
    },
    {
      "epoch": 0.8279569892473119,
      "grad_norm": 3.4362998008728027,
      "learning_rate": 5.063130823984823e-05,
      "loss": 4.0444,
      "step": 77
    },
    {
      "epoch": 0.8387096774193549,
      "grad_norm": 3.6933822631835938,
      "learning_rate": 5.011140233184724e-05,
      "loss": 3.3562,
      "step": 78
    },
    {
      "epoch": 0.8494623655913979,
      "grad_norm": 2.264709711074829,
      "learning_rate": 4.958736511733516e-05,
      "loss": 1.5203,
      "step": 79
    },
    {
      "epoch": 0.8602150537634409,
      "grad_norm": 3.0925395488739014,
      "learning_rate": 4.905933986285393e-05,
      "loss": 3.4232,
      "step": 80
    },
    {
      "epoch": 0.8709677419354839,
      "grad_norm": 3.5374860763549805,
      "learning_rate": 4.8527470925235824e-05,
      "loss": 2.9371,
      "step": 81
    },
    {
      "epoch": 0.8817204301075269,
      "grad_norm": 2.737464666366577,
      "learning_rate": 4.799190371213772e-05,
      "loss": 2.9893,
      "step": 82
    },
    {
      "epoch": 0.8924731182795699,
      "grad_norm": 2.4057750701904297,
      "learning_rate": 4.745278464228808e-05,
      "loss": 2.4454,
      "step": 83
    },
    {
      "epoch": 0.9032258064516129,
      "grad_norm": 3.7483322620391846,
      "learning_rate": 4.69102611054575e-05,
      "loss": 2.2346,
      "step": 84
    },
    {
      "epoch": 0.9139784946236559,
      "grad_norm": 3.1117513179779053,
      "learning_rate": 4.6364481422163926e-05,
      "loss": 2.3842,
      "step": 85
    },
    {
      "epoch": 0.9247311827956989,
      "grad_norm": 2.581960678100586,
      "learning_rate": 4.581559480312316e-05,
      "loss": 2.3513,
      "step": 86
    },
    {
      "epoch": 0.9354838709677419,
      "grad_norm": 2.8223772048950195,
      "learning_rate": 4.526375130845627e-05,
      "loss": 2.713,
      "step": 87
    },
    {
      "epoch": 0.946236559139785,
      "grad_norm": 3.449551582336426,
      "learning_rate": 4.4709101806664554e-05,
      "loss": 2.1045,
      "step": 88
    },
    {
      "epoch": 0.956989247311828,
      "grad_norm": 3.203859329223633,
      "learning_rate": 4.4151797933383685e-05,
      "loss": 2.2587,
      "step": 89
    },
    {
      "epoch": 0.967741935483871,
      "grad_norm": 2.9277474880218506,
      "learning_rate": 4.359199204992797e-05,
      "loss": 2.6041,
      "step": 90
    },
    {
      "epoch": 0.978494623655914,
      "grad_norm": 3.5323002338409424,
      "learning_rate": 4.30298372016363e-05,
      "loss": 2.2093,
      "step": 91
    },
    {
      "epoch": 0.989247311827957,
      "grad_norm": 4.1092681884765625,
      "learning_rate": 4.246548707603114e-05,
      "loss": 1.9357,
      "step": 92
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.2726848125457764,
      "learning_rate": 4.1899095960801805e-05,
      "loss": 2.8715,
      "step": 93
    },
    {
      "epoch": 1.010752688172043,
      "grad_norm": 2.6090543270111084,
      "learning_rate": 4.133081870162385e-05,
      "loss": 6.4364,
      "step": 94
    },
    {
      "epoch": 1.021505376344086,
      "grad_norm": 3.0788755416870117,
      "learning_rate": 4.076081065982569e-05,
      "loss": 6.1323,
      "step": 95
    },
    {
      "epoch": 1.032258064516129,
      "grad_norm": 3.0953359603881836,
      "learning_rate": 4.018922766991447e-05,
      "loss": 5.4343,
      "step": 96
    },
    {
      "epoch": 1.043010752688172,
      "grad_norm": 2.4865167140960693,
      "learning_rate": 3.961622599697241e-05,
      "loss": 3.752,
      "step": 97
    },
    {
      "epoch": 1.053763440860215,
      "grad_norm": 3.014270305633545,
      "learning_rate": 3.9041962293935516e-05,
      "loss": 3.9712,
      "step": 98
    },
    {
      "epoch": 1.064516129032258,
      "grad_norm": 2.697925329208374,
      "learning_rate": 3.84665935587662e-05,
      "loss": 3.8601,
      "step": 99
    },
    {
      "epoch": 1.075268817204301,
      "grad_norm": 2.772761106491089,
      "learning_rate": 3.7890277091531636e-05,
      "loss": 3.554,
      "step": 100
    },
    {
      "epoch": 1.075268817204301,
      "eval_loss": 0.8277088403701782,
      "eval_runtime": 14.7467,
      "eval_samples_per_second": 10.646,
      "eval_steps_per_second": 2.712,
      "step": 100
    },
    {
      "epoch": 1.086021505376344,
      "grad_norm": 2.8855340480804443,
      "learning_rate": 3.7313170451399475e-05,
      "loss": 2.9625,
      "step": 101
    },
    {
      "epoch": 1.096774193548387,
      "grad_norm": 3.348214864730835,
      "learning_rate": 3.673543141356278e-05,
      "loss": 3.571,
      "step": 102
    },
    {
      "epoch": 1.10752688172043,
      "grad_norm": 2.3443796634674072,
      "learning_rate": 3.6157217926105783e-05,
      "loss": 1.8009,
      "step": 103
    },
    {
      "epoch": 1.118279569892473,
      "grad_norm": 3.143157482147217,
      "learning_rate": 3.557868806682255e-05,
      "loss": 2.5616,
      "step": 104
    },
    {
      "epoch": 1.129032258064516,
      "grad_norm": 2.678039073944092,
      "learning_rate": 3.5e-05,
      "loss": 1.6681,
      "step": 105
    },
    {
      "epoch": 1.139784946236559,
      "grad_norm": 3.7475228309631348,
      "learning_rate": 3.442131193317745e-05,
      "loss": 2.4192,
      "step": 106
    },
    {
      "epoch": 1.1505376344086022,
      "grad_norm": 2.9420273303985596,
      "learning_rate": 3.384278207389421e-05,
      "loss": 1.9677,
      "step": 107
    },
    {
      "epoch": 1.1612903225806452,
      "grad_norm": 3.134910821914673,
      "learning_rate": 3.3264568586437216e-05,
      "loss": 1.8615,
      "step": 108
    },
    {
      "epoch": 1.1720430107526882,
      "grad_norm": 4.73399543762207,
      "learning_rate": 3.268682954860052e-05,
      "loss": 2.2775,
      "step": 109
    },
    {
      "epoch": 1.1827956989247312,
      "grad_norm": 4.327884197235107,
      "learning_rate": 3.210972290846837e-05,
      "loss": 2.0575,
      "step": 110
    },
    {
      "epoch": 1.1935483870967742,
      "grad_norm": 4.249052047729492,
      "learning_rate": 3.15334064412338e-05,
      "loss": 1.999,
      "step": 111
    },
    {
      "epoch": 1.2043010752688172,
      "grad_norm": 4.234005928039551,
      "learning_rate": 3.0958037706064485e-05,
      "loss": 1.7021,
      "step": 112
    },
    {
      "epoch": 1.2150537634408602,
      "grad_norm": 3.8542721271514893,
      "learning_rate": 3.038377400302758e-05,
      "loss": 1.4706,
      "step": 113
    },
    {
      "epoch": 1.2258064516129032,
      "grad_norm": 4.0482892990112305,
      "learning_rate": 2.9810772330085524e-05,
      "loss": 2.0322,
      "step": 114
    },
    {
      "epoch": 1.2365591397849462,
      "grad_norm": 5.8383073806762695,
      "learning_rate": 2.9239189340174306e-05,
      "loss": 2.4762,
      "step": 115
    },
    {
      "epoch": 1.2473118279569892,
      "grad_norm": 4.042543888092041,
      "learning_rate": 2.8669181298376163e-05,
      "loss": 1.7183,
      "step": 116
    },
    {
      "epoch": 1.2580645161290323,
      "grad_norm": 2.37276291847229,
      "learning_rate": 2.8100904039198193e-05,
      "loss": 5.3169,
      "step": 117
    },
    {
      "epoch": 1.2688172043010753,
      "grad_norm": 3.0219969749450684,
      "learning_rate": 2.7534512923968863e-05,
      "loss": 5.8673,
      "step": 118
    },
    {
      "epoch": 1.2795698924731183,
      "grad_norm": 3.038010597229004,
      "learning_rate": 2.6970162798363695e-05,
      "loss": 5.1296,
      "step": 119
    },
    {
      "epoch": 1.2903225806451613,
      "grad_norm": 3.0977330207824707,
      "learning_rate": 2.640800795007203e-05,
      "loss": 3.7753,
      "step": 120
    },
    {
      "epoch": 1.3010752688172043,
      "grad_norm": 3.9190726280212402,
      "learning_rate": 2.5848202066616305e-05,
      "loss": 4.0651,
      "step": 121
    },
    {
      "epoch": 1.3118279569892473,
      "grad_norm": 2.9296584129333496,
      "learning_rate": 2.5290898193335446e-05,
      "loss": 3.2334,
      "step": 122
    },
    {
      "epoch": 1.3225806451612903,
      "grad_norm": 2.870746374130249,
      "learning_rate": 2.4736248691543736e-05,
      "loss": 3.5725,
      "step": 123
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 3.1865155696868896,
      "learning_rate": 2.4184405196876842e-05,
      "loss": 2.4533,
      "step": 124
    },
    {
      "epoch": 1.3440860215053765,
      "grad_norm": 2.683004379272461,
      "learning_rate": 2.363551857783608e-05,
      "loss": 2.8205,
      "step": 125
    },
    {
      "epoch": 1.3548387096774195,
      "grad_norm": 2.89284348487854,
      "learning_rate": 2.308973889454249e-05,
      "loss": 1.5345,
      "step": 126
    },
    {
      "epoch": 1.3655913978494625,
      "grad_norm": 2.9330499172210693,
      "learning_rate": 2.2547215357711918e-05,
      "loss": 3.3656,
      "step": 127
    },
    {
      "epoch": 1.3763440860215055,
      "grad_norm": 3.1823220252990723,
      "learning_rate": 2.2008096287862266e-05,
      "loss": 1.9697,
      "step": 128
    },
    {
      "epoch": 1.3870967741935485,
      "grad_norm": 3.1404953002929688,
      "learning_rate": 2.1472529074764177e-05,
      "loss": 1.8351,
      "step": 129
    },
    {
      "epoch": 1.3978494623655915,
      "grad_norm": 2.9065327644348145,
      "learning_rate": 2.0940660137146074e-05,
      "loss": 1.4858,
      "step": 130
    },
    {
      "epoch": 1.4086021505376345,
      "grad_norm": 2.9487435817718506,
      "learning_rate": 2.041263488266484e-05,
      "loss": 1.7826,
      "step": 131
    },
    {
      "epoch": 1.4193548387096775,
      "grad_norm": 3.2001399993896484,
      "learning_rate": 1.988859766815275e-05,
      "loss": 1.7777,
      "step": 132
    },
    {
      "epoch": 1.4301075268817205,
      "grad_norm": 3.180950403213501,
      "learning_rate": 1.9368691760151773e-05,
      "loss": 1.9536,
      "step": 133
    },
    {
      "epoch": 1.4408602150537635,
      "grad_norm": 3.969357490539551,
      "learning_rate": 1.885305929574593e-05,
      "loss": 1.6653,
      "step": 134
    },
    {
      "epoch": 1.4516129032258065,
      "grad_norm": 3.9064815044403076,
      "learning_rate": 1.8341841243702424e-05,
      "loss": 1.6647,
      "step": 135
    },
    {
      "epoch": 1.4623655913978495,
      "grad_norm": 3.6417884826660156,
      "learning_rate": 1.7835177365932225e-05,
      "loss": 1.8726,
      "step": 136
    },
    {
      "epoch": 1.4731182795698925,
      "grad_norm": 4.675624847412109,
      "learning_rate": 1.7333206179280478e-05,
      "loss": 1.9269,
      "step": 137
    },
    {
      "epoch": 1.4838709677419355,
      "grad_norm": 4.082760810852051,
      "learning_rate": 1.6836064917657478e-05,
      "loss": 1.4504,
      "step": 138
    },
    {
      "epoch": 1.4946236559139785,
      "grad_norm": 3.289503812789917,
      "learning_rate": 1.6343889494520224e-05,
      "loss": 0.7642,
      "step": 139
    },
    {
      "epoch": 1.5053763440860215,
      "grad_norm": 2.7232275009155273,
      "learning_rate": 1.5856814465715064e-05,
      "loss": 6.1022,
      "step": 140
    },
    {
      "epoch": 1.5161290322580645,
      "grad_norm": 3.177217960357666,
      "learning_rate": 1.5374972992691458e-05,
      "loss": 5.1127,
      "step": 141
    },
    {
      "epoch": 1.5268817204301075,
      "grad_norm": 2.949465751647949,
      "learning_rate": 1.4898496806096974e-05,
      "loss": 4.6851,
      "step": 142
    },
    {
      "epoch": 1.5376344086021505,
      "grad_norm": 3.0858075618743896,
      "learning_rate": 1.4427516169763444e-05,
      "loss": 4.389,
      "step": 143
    },
    {
      "epoch": 1.5483870967741935,
      "grad_norm": 3.117891788482666,
      "learning_rate": 1.396215984509412e-05,
      "loss": 4.1926,
      "step": 144
    },
    {
      "epoch": 1.5591397849462365,
      "grad_norm": 3.024880886077881,
      "learning_rate": 1.3502555055861625e-05,
      "loss": 3.3123,
      "step": 145
    },
    {
      "epoch": 1.5698924731182795,
      "grad_norm": 3.528703451156616,
      "learning_rate": 1.3048827453426203e-05,
      "loss": 4.0442,
      "step": 146
    },
    {
      "epoch": 1.5806451612903225,
      "grad_norm": 3.134192943572998,
      "learning_rate": 1.2601101082383917e-05,
      "loss": 2.9924,
      "step": 147
    },
    {
      "epoch": 1.5913978494623655,
      "grad_norm": 3.5123281478881836,
      "learning_rate": 1.2159498346654094e-05,
      "loss": 2.2857,
      "step": 148
    },
    {
      "epoch": 1.6021505376344085,
      "grad_norm": 2.7393951416015625,
      "learning_rate": 1.1724139976015306e-05,
      "loss": 1.1784,
      "step": 149
    },
    {
      "epoch": 1.6129032258064515,
      "grad_norm": 3.7451350688934326,
      "learning_rate": 1.1295144993099068e-05,
      "loss": 2.1274,
      "step": 150
    },
    {
      "epoch": 1.6129032258064515,
      "eval_loss": 0.7497462034225464,
      "eval_runtime": 14.7503,
      "eval_samples_per_second": 10.644,
      "eval_steps_per_second": 2.712,
      "step": 150
    },
    {
      "epoch": 1.6236559139784945,
      "grad_norm": 3.5955772399902344,
      "learning_rate": 1.0872630680850196e-05,
      "loss": 2.3566,
      "step": 151
    },
    {
      "epoch": 1.6344086021505375,
      "grad_norm": 3.7463667392730713,
      "learning_rate": 1.0456712550462898e-05,
      "loss": 1.9469,
      "step": 152
    },
    {
      "epoch": 1.6451612903225805,
      "grad_norm": 2.806169033050537,
      "learning_rate": 1.0047504309801104e-05,
      "loss": 2.1513,
      "step": 153
    },
    {
      "epoch": 1.6559139784946235,
      "grad_norm": 3.7219080924987793,
      "learning_rate": 9.645117832311886e-06,
      "loss": 2.5836,
      "step": 154
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 3.056955337524414,
      "learning_rate": 9.249663126440394e-06,
      "loss": 1.6478,
      "step": 155
    },
    {
      "epoch": 1.6774193548387095,
      "grad_norm": 3.247894525527954,
      "learning_rate": 8.861248305554624e-06,
      "loss": 1.455,
      "step": 156
    },
    {
      "epoch": 1.6881720430107527,
      "grad_norm": 3.3513553142547607,
      "learning_rate": 8.47997955838829e-06,
      "loss": 2.0281,
      "step": 157
    },
    {
      "epoch": 1.6989247311827957,
      "grad_norm": 3.108705759048462,
      "learning_rate": 8.10596112000994e-06,
      "loss": 1.5561,
      "step": 158
    },
    {
      "epoch": 1.7096774193548387,
      "grad_norm": 3.947622299194336,
      "learning_rate": 7.739295243326067e-06,
      "loss": 1.7348,
      "step": 159
    },
    {
      "epoch": 1.7204301075268817,
      "grad_norm": 4.6300530433654785,
      "learning_rate": 7.380082171126228e-06,
      "loss": 1.9787,
      "step": 160
    },
    {
      "epoch": 1.7311827956989247,
      "grad_norm": 4.57441520690918,
      "learning_rate": 7.028420108677635e-06,
      "loss": 1.4791,
      "step": 161
    },
    {
      "epoch": 1.7419354838709677,
      "grad_norm": 4.058063983917236,
      "learning_rate": 6.684405196876842e-06,
      "loss": 0.762,
      "step": 162
    },
    {
      "epoch": 1.7526881720430108,
      "grad_norm": 2.3837802410125732,
      "learning_rate": 6.3481314859657675e-06,
      "loss": 5.4107,
      "step": 163
    },
    {
      "epoch": 1.7634408602150538,
      "grad_norm": 3.205528974533081,
      "learning_rate": 6.019690909819298e-06,
      "loss": 5.3617,
      "step": 164
    },
    {
      "epoch": 1.7741935483870968,
      "grad_norm": 3.044926881790161,
      "learning_rate": 5.6991732608115e-06,
      "loss": 4.1357,
      "step": 165
    },
    {
      "epoch": 1.7849462365591398,
      "grad_norm": 3.4720242023468018,
      "learning_rate": 5.386666165267256e-06,
      "loss": 4.7002,
      "step": 166
    },
    {
      "epoch": 1.7956989247311828,
      "grad_norm": 3.4407799243927,
      "learning_rate": 5.08225505950613e-06,
      "loss": 3.6684,
      "step": 167
    },
    {
      "epoch": 1.8064516129032258,
      "grad_norm": 3.670388698577881,
      "learning_rate": 4.786023166484913e-06,
      "loss": 2.7388,
      "step": 168
    },
    {
      "epoch": 1.817204301075269,
      "grad_norm": 3.6209826469421387,
      "learning_rate": 4.498051473045291e-06,
      "loss": 3.9281,
      "step": 169
    },
    {
      "epoch": 1.827956989247312,
      "grad_norm": 3.7421019077301025,
      "learning_rate": 4.218418707772886e-06,
      "loss": 3.1656,
      "step": 170
    },
    {
      "epoch": 1.838709677419355,
      "grad_norm": 3.480229377746582,
      "learning_rate": 3.947201319473587e-06,
      "loss": 1.5602,
      "step": 171
    },
    {
      "epoch": 1.849462365591398,
      "grad_norm": 2.7893009185791016,
      "learning_rate": 3.684473456273278e-06,
      "loss": 1.3773,
      "step": 172
    },
    {
      "epoch": 1.860215053763441,
      "grad_norm": 3.345553398132324,
      "learning_rate": 3.4303069453464383e-06,
      "loss": 1.6992,
      "step": 173
    },
    {
      "epoch": 1.870967741935484,
      "grad_norm": 4.188620567321777,
      "learning_rate": 3.184771273279312e-06,
      "loss": 2.5594,
      "step": 174
    },
    {
      "epoch": 1.881720430107527,
      "grad_norm": 3.4975194931030273,
      "learning_rate": 2.947933567072987e-06,
      "loss": 2.225,
      "step": 175
    },
    {
      "epoch": 1.89247311827957,
      "grad_norm": 3.273622989654541,
      "learning_rate": 2.719858575791534e-06,
      "loss": 1.1841,
      "step": 176
    },
    {
      "epoch": 1.903225806451613,
      "grad_norm": 3.025174617767334,
      "learning_rate": 2.500608652860256e-06,
      "loss": 1.6974,
      "step": 177
    },
    {
      "epoch": 1.913978494623656,
      "grad_norm": 2.607513427734375,
      "learning_rate": 2.2902437390188737e-06,
      "loss": 1.1319,
      "step": 178
    },
    {
      "epoch": 1.924731182795699,
      "grad_norm": 3.18086576461792,
      "learning_rate": 2.0888213459343587e-06,
      "loss": 1.8406,
      "step": 179
    },
    {
      "epoch": 1.935483870967742,
      "grad_norm": 3.675136089324951,
      "learning_rate": 1.8963965404777875e-06,
      "loss": 2.3657,
      "step": 180
    },
    {
      "epoch": 1.946236559139785,
      "grad_norm": 3.96602725982666,
      "learning_rate": 1.7130219296696263e-06,
      "loss": 1.5121,
      "step": 181
    },
    {
      "epoch": 1.956989247311828,
      "grad_norm": 3.6097054481506348,
      "learning_rate": 1.5387476462974824e-06,
      "loss": 1.5113,
      "step": 182
    },
    {
      "epoch": 1.967741935483871,
      "grad_norm": 3.757528781890869,
      "learning_rate": 1.3736213352103147e-06,
      "loss": 1.5258,
      "step": 183
    },
    {
      "epoch": 1.978494623655914,
      "grad_norm": 4.669776439666748,
      "learning_rate": 1.2176881402928002e-06,
      "loss": 1.3748,
      "step": 184
    },
    {
      "epoch": 1.989247311827957,
      "grad_norm": 3.6529650688171387,
      "learning_rate": 1.0709906921234367e-06,
      "loss": 0.7173,
      "step": 185
    },
    {
      "epoch": 2.0,
      "grad_norm": 3.2291080951690674,
      "learning_rate": 9.33569096319799e-07,
      "loss": 2.3399,
      "step": 186
    },
    {
      "epoch": 2.010752688172043,
      "grad_norm": 2.3444361686706543,
      "learning_rate": 8.054609225740255e-07,
      "loss": 5.6927,
      "step": 187
    },
    {
      "epoch": 2.021505376344086,
      "grad_norm": 2.7011375427246094,
      "learning_rate": 6.867011943816724e-07,
      "loss": 5.3949,
      "step": 188
    },
    {
      "epoch": 2.032258064516129,
      "grad_norm": 2.745737314224243,
      "learning_rate": 5.77322379466617e-07,
      "loss": 4.3841,
      "step": 189
    },
    {
      "epoch": 2.043010752688172,
      "grad_norm": 2.8492581844329834,
      "learning_rate": 4.773543809047186e-07,
      "loss": 3.7209,
      "step": 190
    },
    {
      "epoch": 2.053763440860215,
      "grad_norm": 3.298788547515869,
      "learning_rate": 3.868245289486027e-07,
      "loss": 4.1478,
      "step": 191
    },
    {
      "epoch": 2.064516129032258,
      "grad_norm": 2.956620931625366,
      "learning_rate": 3.0575757355586817e-07,
      "loss": 2.9301,
      "step": 192
    },
    {
      "epoch": 2.075268817204301,
      "grad_norm": 2.8849871158599854,
      "learning_rate": 2.3417567762266497e-07,
      "loss": 3.1787,
      "step": 193
    },
    {
      "epoch": 2.086021505376344,
      "grad_norm": 3.1618738174438477,
      "learning_rate": 1.7209841092460043e-07,
      "loss": 3.161,
      "step": 194
    },
    {
      "epoch": 2.096774193548387,
      "grad_norm": 3.032891273498535,
      "learning_rate": 1.1954274476655534e-07,
      "loss": 2.5093,
      "step": 195
    },
    {
      "epoch": 2.10752688172043,
      "grad_norm": 2.315091133117676,
      "learning_rate": 7.652304734289127e-08,
      "loss": 1.0642,
      "step": 196
    },
    {
      "epoch": 2.118279569892473,
      "grad_norm": 2.705153226852417,
      "learning_rate": 4.30510798093342e-08,
      "loss": 2.0394,
      "step": 197
    },
    {
      "epoch": 2.129032258064516,
      "grad_norm": 2.7641355991363525,
      "learning_rate": 1.9135993067588284e-08,
      "loss": 1.127,
      "step": 198
    },
    {
      "epoch": 2.139784946236559,
      "grad_norm": 3.088827610015869,
      "learning_rate": 4.784325263584854e-09,
      "loss": 1.8748,
      "step": 199
    },
    {
      "epoch": 2.150537634408602,
      "grad_norm": 2.703360080718994,
      "learning_rate": 0.0,
      "loss": 1.2501,
      "step": 200
    },
    {
      "epoch": 2.150537634408602,
      "eval_loss": 0.7186845541000366,
      "eval_runtime": 14.749,
      "eval_samples_per_second": 10.645,
      "eval_steps_per_second": 2.712,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 4,
        "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": 2.659354182502318e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}