{
  "best_metric": 1.9446079730987549,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.005479527116809819,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 2.7397635584049098e-05,
      "grad_norm": 17.083011627197266,
      "learning_rate": 1.0110000000000001e-05,
      "loss": 3.6833,
      "step": 1
    },
    {
      "epoch": 2.7397635584049098e-05,
      "eval_loss": 5.066332817077637,
      "eval_runtime": 907.7509,
      "eval_samples_per_second": 16.931,
      "eval_steps_per_second": 4.234,
      "step": 1
    },
    {
      "epoch": 5.4795271168098196e-05,
      "grad_norm": 30.948528289794922,
      "learning_rate": 2.0220000000000003e-05,
      "loss": 5.5955,
      "step": 2
    },
    {
      "epoch": 8.219290675214729e-05,
      "grad_norm": 24.4798641204834,
      "learning_rate": 3.033e-05,
      "loss": 5.2317,
      "step": 3
    },
    {
      "epoch": 0.00010959054233619639,
      "grad_norm": 26.41942596435547,
      "learning_rate": 4.0440000000000006e-05,
      "loss": 6.3092,
      "step": 4
    },
    {
      "epoch": 0.00013698817792024548,
      "grad_norm": 31.87883949279785,
      "learning_rate": 5.055e-05,
      "loss": 5.9147,
      "step": 5
    },
    {
      "epoch": 0.00016438581350429457,
      "grad_norm": 33.11574172973633,
      "learning_rate": 6.066e-05,
      "loss": 6.6052,
      "step": 6
    },
    {
      "epoch": 0.00019178344908834366,
      "grad_norm": 21.89510726928711,
      "learning_rate": 7.077e-05,
      "loss": 4.7486,
      "step": 7
    },
    {
      "epoch": 0.00021918108467239278,
      "grad_norm": 20.02174949645996,
      "learning_rate": 8.088000000000001e-05,
      "loss": 3.4227,
      "step": 8
    },
    {
      "epoch": 0.00024657872025644185,
      "grad_norm": 19.635520935058594,
      "learning_rate": 9.099000000000001e-05,
      "loss": 3.8081,
      "step": 9
    },
    {
      "epoch": 0.00027397635584049096,
      "grad_norm": 23.233644485473633,
      "learning_rate": 0.0001011,
      "loss": 3.0193,
      "step": 10
    },
    {
      "epoch": 0.0003013739914245401,
      "grad_norm": 22.40667724609375,
      "learning_rate": 0.00010056789473684211,
      "loss": 2.9783,
      "step": 11
    },
    {
      "epoch": 0.00032877162700858915,
      "grad_norm": 15.5116548538208,
      "learning_rate": 0.00010003578947368421,
      "loss": 2.7802,
      "step": 12
    },
    {
      "epoch": 0.00035616926259263826,
      "grad_norm": 12.772533416748047,
      "learning_rate": 9.950368421052632e-05,
      "loss": 2.3746,
      "step": 13
    },
    {
      "epoch": 0.00038356689817668733,
      "grad_norm": 12.892407417297363,
      "learning_rate": 9.897157894736842e-05,
      "loss": 1.5691,
      "step": 14
    },
    {
      "epoch": 0.00041096453376073645,
      "grad_norm": 12.087071418762207,
      "learning_rate": 9.843947368421053e-05,
      "loss": 1.8777,
      "step": 15
    },
    {
      "epoch": 0.00043836216934478556,
      "grad_norm": 19.26116943359375,
      "learning_rate": 9.790736842105264e-05,
      "loss": 2.8353,
      "step": 16
    },
    {
      "epoch": 0.00046575980492883463,
      "grad_norm": 11.111328125,
      "learning_rate": 9.737526315789474e-05,
      "loss": 1.8071,
      "step": 17
    },
    {
      "epoch": 0.0004931574405128837,
      "grad_norm": 12.628350257873535,
      "learning_rate": 9.684315789473684e-05,
      "loss": 2.1004,
      "step": 18
    },
    {
      "epoch": 0.0005205550760969328,
      "grad_norm": 14.211240768432617,
      "learning_rate": 9.631105263157895e-05,
      "loss": 2.2592,
      "step": 19
    },
    {
      "epoch": 0.0005479527116809819,
      "grad_norm": 15.376936912536621,
      "learning_rate": 9.577894736842105e-05,
      "loss": 1.8852,
      "step": 20
    },
    {
      "epoch": 0.000575350347265031,
      "grad_norm": 12.521246910095215,
      "learning_rate": 9.524684210526317e-05,
      "loss": 2.0885,
      "step": 21
    },
    {
      "epoch": 0.0006027479828490802,
      "grad_norm": 13.282447814941406,
      "learning_rate": 9.471473684210526e-05,
      "loss": 2.1103,
      "step": 22
    },
    {
      "epoch": 0.0006301456184331292,
      "grad_norm": 11.5529203414917,
      "learning_rate": 9.418263157894737e-05,
      "loss": 1.9263,
      "step": 23
    },
    {
      "epoch": 0.0006575432540171783,
      "grad_norm": 11.215520858764648,
      "learning_rate": 9.365052631578948e-05,
      "loss": 1.9848,
      "step": 24
    },
    {
      "epoch": 0.0006849408896012274,
      "grad_norm": 14.77686882019043,
      "learning_rate": 9.311842105263157e-05,
      "loss": 2.4333,
      "step": 25
    },
    {
      "epoch": 0.0007123385251852765,
      "grad_norm": 12.779730796813965,
      "learning_rate": 9.258631578947368e-05,
      "loss": 2.5709,
      "step": 26
    },
    {
      "epoch": 0.0007397361607693256,
      "grad_norm": 11.351499557495117,
      "learning_rate": 9.20542105263158e-05,
      "loss": 2.2625,
      "step": 27
    },
    {
      "epoch": 0.0007671337963533747,
      "grad_norm": 13.397592544555664,
      "learning_rate": 9.15221052631579e-05,
      "loss": 2.7227,
      "step": 28
    },
    {
      "epoch": 0.0007945314319374238,
      "grad_norm": 10.616507530212402,
      "learning_rate": 9.099000000000001e-05,
      "loss": 1.8181,
      "step": 29
    },
    {
      "epoch": 0.0008219290675214729,
      "grad_norm": 11.262557983398438,
      "learning_rate": 9.045789473684212e-05,
      "loss": 1.7244,
      "step": 30
    },
    {
      "epoch": 0.000849326703105522,
      "grad_norm": 11.611719131469727,
      "learning_rate": 8.992578947368421e-05,
      "loss": 1.7876,
      "step": 31
    },
    {
      "epoch": 0.0008767243386895711,
      "grad_norm": 12.462687492370605,
      "learning_rate": 8.939368421052632e-05,
      "loss": 2.3846,
      "step": 32
    },
    {
      "epoch": 0.0009041219742736201,
      "grad_norm": 11.689743995666504,
      "learning_rate": 8.886157894736841e-05,
      "loss": 1.893,
      "step": 33
    },
    {
      "epoch": 0.0009315196098576693,
      "grad_norm": 15.821490287780762,
      "learning_rate": 8.832947368421054e-05,
      "loss": 2.277,
      "step": 34
    },
    {
      "epoch": 0.0009589172454417184,
      "grad_norm": 11.826910018920898,
      "learning_rate": 8.779736842105264e-05,
      "loss": 1.8275,
      "step": 35
    },
    {
      "epoch": 0.0009863148810257674,
      "grad_norm": 10.841914176940918,
      "learning_rate": 8.726526315789474e-05,
      "loss": 1.553,
      "step": 36
    },
    {
      "epoch": 0.0010137125166098165,
      "grad_norm": 11.523122787475586,
      "learning_rate": 8.673315789473685e-05,
      "loss": 1.9103,
      "step": 37
    },
    {
      "epoch": 0.0010411101521938656,
      "grad_norm": 11.838396072387695,
      "learning_rate": 8.620105263157896e-05,
      "loss": 1.5697,
      "step": 38
    },
    {
      "epoch": 0.0010685077877779147,
      "grad_norm": 11.571324348449707,
      "learning_rate": 8.566894736842105e-05,
      "loss": 2.0003,
      "step": 39
    },
    {
      "epoch": 0.0010959054233619639,
      "grad_norm": 10.516695022583008,
      "learning_rate": 8.513684210526316e-05,
      "loss": 1.5583,
      "step": 40
    },
    {
      "epoch": 0.001123303058946013,
      "grad_norm": 9.6254301071167,
      "learning_rate": 8.460473684210527e-05,
      "loss": 1.3097,
      "step": 41
    },
    {
      "epoch": 0.001150700694530062,
      "grad_norm": 9.205260276794434,
      "learning_rate": 8.407263157894738e-05,
      "loss": 1.32,
      "step": 42
    },
    {
      "epoch": 0.0011780983301141112,
      "grad_norm": 12.572107315063477,
      "learning_rate": 8.354052631578948e-05,
      "loss": 2.1812,
      "step": 43
    },
    {
      "epoch": 0.0012054959656981603,
      "grad_norm": 12.90071964263916,
      "learning_rate": 8.300842105263158e-05,
      "loss": 1.3577,
      "step": 44
    },
    {
      "epoch": 0.0012328936012822094,
      "grad_norm": 13.483455657958984,
      "learning_rate": 8.247631578947369e-05,
      "loss": 2.3667,
      "step": 45
    },
    {
      "epoch": 0.0012602912368662584,
      "grad_norm": 15.868042945861816,
      "learning_rate": 8.19442105263158e-05,
      "loss": 2.6061,
      "step": 46
    },
    {
      "epoch": 0.0012876888724503075,
      "grad_norm": 13.639692306518555,
      "learning_rate": 8.141210526315789e-05,
      "loss": 2.9595,
      "step": 47
    },
    {
      "epoch": 0.0013150865080343566,
      "grad_norm": 11.425675392150879,
      "learning_rate": 8.088000000000001e-05,
      "loss": 1.7875,
      "step": 48
    },
    {
      "epoch": 0.0013424841436184057,
      "grad_norm": 11.319607734680176,
      "learning_rate": 8.03478947368421e-05,
      "loss": 2.12,
      "step": 49
    },
    {
      "epoch": 0.0013698817792024548,
      "grad_norm": 13.456693649291992,
      "learning_rate": 7.981578947368421e-05,
      "loss": 1.9008,
      "step": 50
    },
    {
      "epoch": 0.0013698817792024548,
      "eval_loss": 2.246959686279297,
      "eval_runtime": 901.5117,
      "eval_samples_per_second": 17.048,
      "eval_steps_per_second": 4.263,
      "step": 50
    },
    {
      "epoch": 0.001397279414786504,
      "grad_norm": 9.39171314239502,
      "learning_rate": 7.928368421052632e-05,
      "loss": 2.5679,
      "step": 51
    },
    {
      "epoch": 0.001424677050370553,
      "grad_norm": 10.350808143615723,
      "learning_rate": 7.875157894736842e-05,
      "loss": 2.6965,
      "step": 52
    },
    {
      "epoch": 0.0014520746859546022,
      "grad_norm": 11.915090560913086,
      "learning_rate": 7.821947368421053e-05,
      "loss": 2.7901,
      "step": 53
    },
    {
      "epoch": 0.0014794723215386513,
      "grad_norm": 14.726212501525879,
      "learning_rate": 7.768736842105263e-05,
      "loss": 3.894,
      "step": 54
    },
    {
      "epoch": 0.0015068699571227004,
      "grad_norm": 14.440069198608398,
      "learning_rate": 7.715526315789474e-05,
      "loss": 3.8703,
      "step": 55
    },
    {
      "epoch": 0.0015342675927067493,
      "grad_norm": 16.271181106567383,
      "learning_rate": 7.662315789473685e-05,
      "loss": 3.6617,
      "step": 56
    },
    {
      "epoch": 0.0015616652282907984,
      "grad_norm": 13.526236534118652,
      "learning_rate": 7.609105263157895e-05,
      "loss": 2.7699,
      "step": 57
    },
    {
      "epoch": 0.0015890628638748476,
      "grad_norm": 9.286958694458008,
      "learning_rate": 7.555894736842105e-05,
      "loss": 2.8028,
      "step": 58
    },
    {
      "epoch": 0.0016164604994588967,
      "grad_norm": 14.175232887268066,
      "learning_rate": 7.502684210526316e-05,
      "loss": 3.3656,
      "step": 59
    },
    {
      "epoch": 0.0016438581350429458,
      "grad_norm": 9.617558479309082,
      "learning_rate": 7.449473684210526e-05,
      "loss": 1.9919,
      "step": 60
    },
    {
      "epoch": 0.001671255770626995,
      "grad_norm": 9.61012077331543,
      "learning_rate": 7.396263157894738e-05,
      "loss": 1.9658,
      "step": 61
    },
    {
      "epoch": 0.001698653406211044,
      "grad_norm": 10.791799545288086,
      "learning_rate": 7.343052631578949e-05,
      "loss": 2.7421,
      "step": 62
    },
    {
      "epoch": 0.0017260510417950931,
      "grad_norm": 11.175983428955078,
      "learning_rate": 7.289842105263158e-05,
      "loss": 1.7803,
      "step": 63
    },
    {
      "epoch": 0.0017534486773791423,
      "grad_norm": 21.124692916870117,
      "learning_rate": 7.236631578947369e-05,
      "loss": 1.9238,
      "step": 64
    },
    {
      "epoch": 0.0017808463129631914,
      "grad_norm": 11.414015769958496,
      "learning_rate": 7.183421052631579e-05,
      "loss": 1.9238,
      "step": 65
    },
    {
      "epoch": 0.0018082439485472403,
      "grad_norm": 9.583854675292969,
      "learning_rate": 7.13021052631579e-05,
      "loss": 2.2229,
      "step": 66
    },
    {
      "epoch": 0.0018356415841312894,
      "grad_norm": 11.926726341247559,
      "learning_rate": 7.077e-05,
      "loss": 1.9964,
      "step": 67
    },
    {
      "epoch": 0.0018630392197153385,
      "grad_norm": 13.71705436706543,
      "learning_rate": 7.023789473684211e-05,
      "loss": 2.9112,
      "step": 68
    },
    {
      "epoch": 0.0018904368552993876,
      "grad_norm": 9.148221969604492,
      "learning_rate": 6.970578947368422e-05,
      "loss": 2.0433,
      "step": 69
    },
    {
      "epoch": 0.0019178344908834368,
      "grad_norm": 11.769927024841309,
      "learning_rate": 6.917368421052633e-05,
      "loss": 1.9223,
      "step": 70
    },
    {
      "epoch": 0.0019452321264674859,
      "grad_norm": 12.037910461425781,
      "learning_rate": 6.864157894736842e-05,
      "loss": 1.8342,
      "step": 71
    },
    {
      "epoch": 0.0019726297620515348,
      "grad_norm": 9.010817527770996,
      "learning_rate": 6.810947368421053e-05,
      "loss": 1.9317,
      "step": 72
    },
    {
      "epoch": 0.002000027397635584,
      "grad_norm": 12.990135192871094,
      "learning_rate": 6.757736842105264e-05,
      "loss": 2.7586,
      "step": 73
    },
    {
      "epoch": 0.002027425033219633,
      "grad_norm": 9.70896053314209,
      "learning_rate": 6.704526315789473e-05,
      "loss": 1.9642,
      "step": 74
    },
    {
      "epoch": 0.002054822668803682,
      "grad_norm": 10.410635948181152,
      "learning_rate": 6.651315789473685e-05,
      "loss": 1.9407,
      "step": 75
    },
    {
      "epoch": 0.0020822203043877312,
      "grad_norm": 11.185718536376953,
      "learning_rate": 6.598105263157895e-05,
      "loss": 1.7685,
      "step": 76
    },
    {
      "epoch": 0.0021096179399717804,
      "grad_norm": 12.08330249786377,
      "learning_rate": 6.544894736842106e-05,
      "loss": 2.5273,
      "step": 77
    },
    {
      "epoch": 0.0021370155755558295,
      "grad_norm": 10.257572174072266,
      "learning_rate": 6.491684210526317e-05,
      "loss": 1.6406,
      "step": 78
    },
    {
      "epoch": 0.0021644132111398786,
      "grad_norm": 8.938873291015625,
      "learning_rate": 6.438473684210526e-05,
      "loss": 1.4209,
      "step": 79
    },
    {
      "epoch": 0.0021918108467239277,
      "grad_norm": 12.932609558105469,
      "learning_rate": 6.385263157894737e-05,
      "loss": 2.4091,
      "step": 80
    },
    {
      "epoch": 0.002219208482307977,
      "grad_norm": 13.582636833190918,
      "learning_rate": 6.332052631578948e-05,
      "loss": 2.4494,
      "step": 81
    },
    {
      "epoch": 0.002246606117892026,
      "grad_norm": 10.341235160827637,
      "learning_rate": 6.278842105263159e-05,
      "loss": 1.9182,
      "step": 82
    },
    {
      "epoch": 0.002274003753476075,
      "grad_norm": 9.691158294677734,
      "learning_rate": 6.22563157894737e-05,
      "loss": 1.6645,
      "step": 83
    },
    {
      "epoch": 0.002301401389060124,
      "grad_norm": 9.134404182434082,
      "learning_rate": 6.172421052631579e-05,
      "loss": 1.5419,
      "step": 84
    },
    {
      "epoch": 0.0023287990246441733,
      "grad_norm": 13.0061674118042,
      "learning_rate": 6.11921052631579e-05,
      "loss": 2.0695,
      "step": 85
    },
    {
      "epoch": 0.0023561966602282224,
      "grad_norm": 11.673482894897461,
      "learning_rate": 6.066e-05,
      "loss": 1.6456,
      "step": 86
    },
    {
      "epoch": 0.0023835942958122715,
      "grad_norm": 12.235639572143555,
      "learning_rate": 6.012789473684211e-05,
      "loss": 2.0552,
      "step": 87
    },
    {
      "epoch": 0.0024109919313963207,
      "grad_norm": 12.531293869018555,
      "learning_rate": 5.959578947368421e-05,
      "loss": 1.7928,
      "step": 88
    },
    {
      "epoch": 0.0024383895669803698,
      "grad_norm": 11.760167121887207,
      "learning_rate": 5.9063684210526324e-05,
      "loss": 1.9041,
      "step": 89
    },
    {
      "epoch": 0.002465787202564419,
      "grad_norm": 11.60911750793457,
      "learning_rate": 5.8531578947368425e-05,
      "loss": 1.7488,
      "step": 90
    },
    {
      "epoch": 0.0024931848381484676,
      "grad_norm": 12.436223983764648,
      "learning_rate": 5.7999473684210527e-05,
      "loss": 1.8447,
      "step": 91
    },
    {
      "epoch": 0.0025205824737325167,
      "grad_norm": 10.578962326049805,
      "learning_rate": 5.7467368421052635e-05,
      "loss": 1.7382,
      "step": 92
    },
    {
      "epoch": 0.002547980109316566,
      "grad_norm": 11.361544609069824,
      "learning_rate": 5.6935263157894736e-05,
      "loss": 1.8159,
      "step": 93
    },
    {
      "epoch": 0.002575377744900615,
      "grad_norm": 12.437856674194336,
      "learning_rate": 5.640315789473684e-05,
      "loss": 1.9681,
      "step": 94
    },
    {
      "epoch": 0.002602775380484664,
      "grad_norm": 8.324384689331055,
      "learning_rate": 5.587105263157895e-05,
      "loss": 1.3098,
      "step": 95
    },
    {
      "epoch": 0.002630173016068713,
      "grad_norm": 10.963112831115723,
      "learning_rate": 5.533894736842106e-05,
      "loss": 1.3977,
      "step": 96
    },
    {
      "epoch": 0.0026575706516527623,
      "grad_norm": 11.730253219604492,
      "learning_rate": 5.480684210526316e-05,
      "loss": 2.0322,
      "step": 97
    },
    {
      "epoch": 0.0026849682872368114,
      "grad_norm": 10.242620468139648,
      "learning_rate": 5.4274736842105264e-05,
      "loss": 1.5608,
      "step": 98
    },
    {
      "epoch": 0.0027123659228208605,
      "grad_norm": 8.740520477294922,
      "learning_rate": 5.374263157894737e-05,
      "loss": 1.5199,
      "step": 99
    },
    {
      "epoch": 0.0027397635584049096,
      "grad_norm": 11.723816871643066,
      "learning_rate": 5.3210526315789474e-05,
      "loss": 2.1569,
      "step": 100
    },
    {
      "epoch": 0.0027397635584049096,
      "eval_loss": 2.1092565059661865,
      "eval_runtime": 908.8001,
      "eval_samples_per_second": 16.911,
      "eval_steps_per_second": 4.229,
      "step": 100
    },
    {
      "epoch": 0.0027671611939889588,
      "grad_norm": 8.458086967468262,
      "learning_rate": 5.2678421052631576e-05,
      "loss": 2.8784,
      "step": 101
    },
    {
      "epoch": 0.002794558829573008,
      "grad_norm": 12.095956802368164,
      "learning_rate": 5.214631578947369e-05,
      "loss": 3.8121,
      "step": 102
    },
    {
      "epoch": 0.002821956465157057,
      "grad_norm": 12.647064208984375,
      "learning_rate": 5.161421052631579e-05,
      "loss": 3.4101,
      "step": 103
    },
    {
      "epoch": 0.002849354100741106,
      "grad_norm": 13.06585693359375,
      "learning_rate": 5.10821052631579e-05,
      "loss": 2.9773,
      "step": 104
    },
    {
      "epoch": 0.0028767517363251552,
      "grad_norm": 14.944098472595215,
      "learning_rate": 5.055e-05,
      "loss": 3.6446,
      "step": 105
    },
    {
      "epoch": 0.0029041493719092044,
      "grad_norm": 14.845952033996582,
      "learning_rate": 5.0017894736842104e-05,
      "loss": 4.0916,
      "step": 106
    },
    {
      "epoch": 0.0029315470074932535,
      "grad_norm": 15.211709976196289,
      "learning_rate": 4.948578947368421e-05,
      "loss": 4.2204,
      "step": 107
    },
    {
      "epoch": 0.0029589446430773026,
      "grad_norm": 8.81589412689209,
      "learning_rate": 4.895368421052632e-05,
      "loss": 1.4566,
      "step": 108
    },
    {
      "epoch": 0.0029863422786613517,
      "grad_norm": 12.095672607421875,
      "learning_rate": 4.842157894736842e-05,
      "loss": 2.6067,
      "step": 109
    },
    {
      "epoch": 0.003013739914245401,
      "grad_norm": 11.675946235656738,
      "learning_rate": 4.7889473684210523e-05,
      "loss": 2.4116,
      "step": 110
    },
    {
      "epoch": 0.0030411375498294495,
      "grad_norm": 11.829989433288574,
      "learning_rate": 4.735736842105263e-05,
      "loss": 2.6115,
      "step": 111
    },
    {
      "epoch": 0.0030685351854134986,
      "grad_norm": 9.828085899353027,
      "learning_rate": 4.682526315789474e-05,
      "loss": 2.2881,
      "step": 112
    },
    {
      "epoch": 0.0030959328209975477,
      "grad_norm": 9.46169376373291,
      "learning_rate": 4.629315789473684e-05,
      "loss": 1.9979,
      "step": 113
    },
    {
      "epoch": 0.003123330456581597,
      "grad_norm": 9.22454833984375,
      "learning_rate": 4.576105263157895e-05,
      "loss": 1.9874,
      "step": 114
    },
    {
      "epoch": 0.003150728092165646,
      "grad_norm": 10.429281234741211,
      "learning_rate": 4.522894736842106e-05,
      "loss": 2.2162,
      "step": 115
    },
    {
      "epoch": 0.003178125727749695,
      "grad_norm": 11.570110321044922,
      "learning_rate": 4.469684210526316e-05,
      "loss": 2.4278,
      "step": 116
    },
    {
      "epoch": 0.0032055233633337442,
      "grad_norm": 9.045998573303223,
      "learning_rate": 4.416473684210527e-05,
      "loss": 1.808,
      "step": 117
    },
    {
      "epoch": 0.0032329209989177933,
      "grad_norm": 10.189197540283203,
      "learning_rate": 4.363263157894737e-05,
      "loss": 1.539,
      "step": 118
    },
    {
      "epoch": 0.0032603186345018425,
      "grad_norm": 8.157354354858398,
      "learning_rate": 4.310052631578948e-05,
      "loss": 1.7214,
      "step": 119
    },
    {
      "epoch": 0.0032877162700858916,
      "grad_norm": 8.792228698730469,
      "learning_rate": 4.256842105263158e-05,
      "loss": 1.3858,
      "step": 120
    },
    {
      "epoch": 0.0033151139056699407,
      "grad_norm": 10.97755241394043,
      "learning_rate": 4.203631578947369e-05,
      "loss": 1.77,
      "step": 121
    },
    {
      "epoch": 0.00334251154125399,
      "grad_norm": 10.724143981933594,
      "learning_rate": 4.150421052631579e-05,
      "loss": 1.8881,
      "step": 122
    },
    {
      "epoch": 0.003369909176838039,
      "grad_norm": 12.109813690185547,
      "learning_rate": 4.09721052631579e-05,
      "loss": 2.1987,
      "step": 123
    },
    {
      "epoch": 0.003397306812422088,
      "grad_norm": 11.524436950683594,
      "learning_rate": 4.0440000000000006e-05,
      "loss": 1.9404,
      "step": 124
    },
    {
      "epoch": 0.003424704448006137,
      "grad_norm": 9.770040512084961,
      "learning_rate": 3.990789473684211e-05,
      "loss": 1.5231,
      "step": 125
    },
    {
      "epoch": 0.0034521020835901863,
      "grad_norm": 14.594894409179688,
      "learning_rate": 3.937578947368421e-05,
      "loss": 2.1621,
      "step": 126
    },
    {
      "epoch": 0.0034794997191742354,
      "grad_norm": 9.148670196533203,
      "learning_rate": 3.884368421052632e-05,
      "loss": 1.9675,
      "step": 127
    },
    {
      "epoch": 0.0035068973547582845,
      "grad_norm": 9.530057907104492,
      "learning_rate": 3.8311578947368426e-05,
      "loss": 1.8147,
      "step": 128
    },
    {
      "epoch": 0.0035342949903423336,
      "grad_norm": 10.43026065826416,
      "learning_rate": 3.777947368421053e-05,
      "loss": 2.0498,
      "step": 129
    },
    {
      "epoch": 0.0035616926259263828,
      "grad_norm": 15.587285041809082,
      "learning_rate": 3.724736842105263e-05,
      "loss": 2.4641,
      "step": 130
    },
    {
      "epoch": 0.0035890902615104314,
      "grad_norm": 12.40803337097168,
      "learning_rate": 3.6715263157894744e-05,
      "loss": 2.7419,
      "step": 131
    },
    {
      "epoch": 0.0036164878970944806,
      "grad_norm": 10.415118217468262,
      "learning_rate": 3.6183157894736845e-05,
      "loss": 2.1787,
      "step": 132
    },
    {
      "epoch": 0.0036438855326785297,
      "grad_norm": 9.937010765075684,
      "learning_rate": 3.565105263157895e-05,
      "loss": 1.8409,
      "step": 133
    },
    {
      "epoch": 0.003671283168262579,
      "grad_norm": 10.181679725646973,
      "learning_rate": 3.5118947368421055e-05,
      "loss": 1.873,
      "step": 134
    },
    {
      "epoch": 0.003698680803846628,
      "grad_norm": 8.590680122375488,
      "learning_rate": 3.458684210526316e-05,
      "loss": 1.5075,
      "step": 135
    },
    {
      "epoch": 0.003726078439430677,
      "grad_norm": 10.049001693725586,
      "learning_rate": 3.4054736842105265e-05,
      "loss": 1.6106,
      "step": 136
    },
    {
      "epoch": 0.003753476075014726,
      "grad_norm": 8.262350082397461,
      "learning_rate": 3.3522631578947366e-05,
      "loss": 1.2253,
      "step": 137
    },
    {
      "epoch": 0.0037808737105987753,
      "grad_norm": 10.25830078125,
      "learning_rate": 3.2990526315789475e-05,
      "loss": 1.6037,
      "step": 138
    },
    {
      "epoch": 0.0038082713461828244,
      "grad_norm": 10.215044975280762,
      "learning_rate": 3.245842105263158e-05,
      "loss": 1.8698,
      "step": 139
    },
    {
      "epoch": 0.0038356689817668735,
      "grad_norm": 9.670408248901367,
      "learning_rate": 3.1926315789473685e-05,
      "loss": 1.6544,
      "step": 140
    },
    {
      "epoch": 0.0038630666173509226,
      "grad_norm": 13.060770988464355,
      "learning_rate": 3.139421052631579e-05,
      "loss": 2.1578,
      "step": 141
    },
    {
      "epoch": 0.0038904642529349717,
      "grad_norm": 10.066153526306152,
      "learning_rate": 3.0862105263157894e-05,
      "loss": 1.7932,
      "step": 142
    },
    {
      "epoch": 0.00391786188851902,
      "grad_norm": 9.591915130615234,
      "learning_rate": 3.033e-05,
      "loss": 1.8039,
      "step": 143
    },
    {
      "epoch": 0.0039452595241030695,
      "grad_norm": 9.10180950164795,
      "learning_rate": 2.9797894736842104e-05,
      "loss": 1.2231,
      "step": 144
    },
    {
      "epoch": 0.003972657159687119,
      "grad_norm": 8.49804401397705,
      "learning_rate": 2.9265789473684213e-05,
      "loss": 1.2112,
      "step": 145
    },
    {
      "epoch": 0.004000054795271168,
      "grad_norm": 10.35539436340332,
      "learning_rate": 2.8733684210526317e-05,
      "loss": 1.397,
      "step": 146
    },
    {
      "epoch": 0.004027452430855217,
      "grad_norm": 12.644351959228516,
      "learning_rate": 2.820157894736842e-05,
      "loss": 1.6601,
      "step": 147
    },
    {
      "epoch": 0.004054850066439266,
      "grad_norm": 10.167571067810059,
      "learning_rate": 2.766947368421053e-05,
      "loss": 1.6444,
      "step": 148
    },
    {
      "epoch": 0.004082247702023315,
      "grad_norm": 11.871386528015137,
      "learning_rate": 2.7137368421052632e-05,
      "loss": 1.9473,
      "step": 149
    },
    {
      "epoch": 0.004109645337607364,
      "grad_norm": 9.358609199523926,
      "learning_rate": 2.6605263157894737e-05,
      "loss": 1.6992,
      "step": 150
    },
    {
      "epoch": 0.004109645337607364,
      "eval_loss": 2.0133330821990967,
      "eval_runtime": 908.0735,
      "eval_samples_per_second": 16.925,
      "eval_steps_per_second": 4.232,
      "step": 150
    },
    {
      "epoch": 0.004137042973191413,
      "grad_norm": 10.242443084716797,
      "learning_rate": 2.6073157894736845e-05,
      "loss": 2.8813,
      "step": 151
    },
    {
      "epoch": 0.0041644406087754625,
      "grad_norm": 11.027935981750488,
      "learning_rate": 2.554105263157895e-05,
      "loss": 3.2701,
      "step": 152
    },
    {
      "epoch": 0.004191838244359512,
      "grad_norm": 9.913683891296387,
      "learning_rate": 2.5008947368421052e-05,
      "loss": 2.8378,
      "step": 153
    },
    {
      "epoch": 0.004219235879943561,
      "grad_norm": 12.305196762084961,
      "learning_rate": 2.447684210526316e-05,
      "loss": 3.3966,
      "step": 154
    },
    {
      "epoch": 0.00424663351552761,
      "grad_norm": 14.736776351928711,
      "learning_rate": 2.3944736842105262e-05,
      "loss": 2.936,
      "step": 155
    },
    {
      "epoch": 0.004274031151111659,
      "grad_norm": 15.716378211975098,
      "learning_rate": 2.341263157894737e-05,
      "loss": 3.3793,
      "step": 156
    },
    {
      "epoch": 0.004301428786695708,
      "grad_norm": 11.29692554473877,
      "learning_rate": 2.2880526315789475e-05,
      "loss": 2.5899,
      "step": 157
    },
    {
      "epoch": 0.004328826422279757,
      "grad_norm": 8.92076587677002,
      "learning_rate": 2.234842105263158e-05,
      "loss": 1.5897,
      "step": 158
    },
    {
      "epoch": 0.004356224057863806,
      "grad_norm": 10.286359786987305,
      "learning_rate": 2.1816315789473685e-05,
      "loss": 2.0679,
      "step": 159
    },
    {
      "epoch": 0.004383621693447855,
      "grad_norm": 10.094871520996094,
      "learning_rate": 2.128421052631579e-05,
      "loss": 2.2023,
      "step": 160
    },
    {
      "epoch": 0.0044110193290319046,
      "grad_norm": 8.398151397705078,
      "learning_rate": 2.0752105263157895e-05,
      "loss": 1.7198,
      "step": 161
    },
    {
      "epoch": 0.004438416964615954,
      "grad_norm": 9.89030647277832,
      "learning_rate": 2.0220000000000003e-05,
      "loss": 1.8957,
      "step": 162
    },
    {
      "epoch": 0.004465814600200003,
      "grad_norm": 10.257231712341309,
      "learning_rate": 1.9687894736842104e-05,
      "loss": 1.8288,
      "step": 163
    },
    {
      "epoch": 0.004493212235784052,
      "grad_norm": 10.972249984741211,
      "learning_rate": 1.9155789473684213e-05,
      "loss": 2.3046,
      "step": 164
    },
    {
      "epoch": 0.004520609871368101,
      "grad_norm": 9.473671913146973,
      "learning_rate": 1.8623684210526314e-05,
      "loss": 1.5979,
      "step": 165
    },
    {
      "epoch": 0.00454800750695215,
      "grad_norm": 8.295541763305664,
      "learning_rate": 1.8091578947368423e-05,
      "loss": 1.6066,
      "step": 166
    },
    {
      "epoch": 0.004575405142536199,
      "grad_norm": 9.380114555358887,
      "learning_rate": 1.7559473684210528e-05,
      "loss": 1.382,
      "step": 167
    },
    {
      "epoch": 0.004602802778120248,
      "grad_norm": 10.417853355407715,
      "learning_rate": 1.7027368421052632e-05,
      "loss": 2.454,
      "step": 168
    },
    {
      "epoch": 0.0046302004137042975,
      "grad_norm": 10.452378273010254,
      "learning_rate": 1.6495263157894737e-05,
      "loss": 1.9223,
      "step": 169
    },
    {
      "epoch": 0.004657598049288347,
      "grad_norm": 9.896018981933594,
      "learning_rate": 1.5963157894736842e-05,
      "loss": 1.8119,
      "step": 170
    },
    {
      "epoch": 0.004684995684872396,
      "grad_norm": 9.311156272888184,
      "learning_rate": 1.5431052631578947e-05,
      "loss": 1.5882,
      "step": 171
    },
    {
      "epoch": 0.004712393320456445,
      "grad_norm": 10.566994667053223,
      "learning_rate": 1.4898947368421052e-05,
      "loss": 1.7865,
      "step": 172
    },
    {
      "epoch": 0.004739790956040494,
      "grad_norm": 13.512009620666504,
      "learning_rate": 1.4366842105263159e-05,
      "loss": 3.2459,
      "step": 173
    },
    {
      "epoch": 0.004767188591624543,
      "grad_norm": 9.658906936645508,
      "learning_rate": 1.3834736842105265e-05,
      "loss": 1.6884,
      "step": 174
    },
    {
      "epoch": 0.004794586227208592,
      "grad_norm": 9.381290435791016,
      "learning_rate": 1.3302631578947369e-05,
      "loss": 1.6764,
      "step": 175
    },
    {
      "epoch": 0.004821983862792641,
      "grad_norm": 11.110421180725098,
      "learning_rate": 1.2770526315789475e-05,
      "loss": 2.4533,
      "step": 176
    },
    {
      "epoch": 0.0048493814983766904,
      "grad_norm": 9.552355766296387,
      "learning_rate": 1.223842105263158e-05,
      "loss": 1.587,
      "step": 177
    },
    {
      "epoch": 0.0048767791339607396,
      "grad_norm": 10.369064331054688,
      "learning_rate": 1.1706315789473685e-05,
      "loss": 1.5635,
      "step": 178
    },
    {
      "epoch": 0.004904176769544789,
      "grad_norm": 10.4714994430542,
      "learning_rate": 1.117421052631579e-05,
      "loss": 1.3165,
      "step": 179
    },
    {
      "epoch": 0.004931574405128838,
      "grad_norm": 10.99134635925293,
      "learning_rate": 1.0642105263157895e-05,
      "loss": 1.9887,
      "step": 180
    },
    {
      "epoch": 0.004958972040712886,
      "grad_norm": 11.243077278137207,
      "learning_rate": 1.0110000000000001e-05,
      "loss": 2.0445,
      "step": 181
    },
    {
      "epoch": 0.004986369676296935,
      "grad_norm": 9.404143333435059,
      "learning_rate": 9.577894736842106e-06,
      "loss": 1.9587,
      "step": 182
    },
    {
      "epoch": 0.005013767311880984,
      "grad_norm": 8.410197257995605,
      "learning_rate": 9.045789473684211e-06,
      "loss": 1.8646,
      "step": 183
    },
    {
      "epoch": 0.005041164947465033,
      "grad_norm": 7.286149501800537,
      "learning_rate": 8.513684210526316e-06,
      "loss": 1.2129,
      "step": 184
    },
    {
      "epoch": 0.0050685625830490825,
      "grad_norm": 10.829649925231934,
      "learning_rate": 7.981578947368421e-06,
      "loss": 2.1754,
      "step": 185
    },
    {
      "epoch": 0.005095960218633132,
      "grad_norm": 9.724632263183594,
      "learning_rate": 7.449473684210526e-06,
      "loss": 1.3908,
      "step": 186
    },
    {
      "epoch": 0.005123357854217181,
      "grad_norm": 9.229701042175293,
      "learning_rate": 6.917368421052633e-06,
      "loss": 1.6759,
      "step": 187
    },
    {
      "epoch": 0.00515075548980123,
      "grad_norm": 9.659087181091309,
      "learning_rate": 6.385263157894738e-06,
      "loss": 1.6113,
      "step": 188
    },
    {
      "epoch": 0.005178153125385279,
      "grad_norm": 10.842117309570312,
      "learning_rate": 5.8531578947368425e-06,
      "loss": 2.3648,
      "step": 189
    },
    {
      "epoch": 0.005205550760969328,
      "grad_norm": 10.12624740600586,
      "learning_rate": 5.321052631578947e-06,
      "loss": 1.6541,
      "step": 190
    },
    {
      "epoch": 0.005232948396553377,
      "grad_norm": 9.375452041625977,
      "learning_rate": 4.788947368421053e-06,
      "loss": 1.9107,
      "step": 191
    },
    {
      "epoch": 0.005260346032137426,
      "grad_norm": 10.81548023223877,
      "learning_rate": 4.256842105263158e-06,
      "loss": 1.9993,
      "step": 192
    },
    {
      "epoch": 0.0052877436677214755,
      "grad_norm": 8.572677612304688,
      "learning_rate": 3.724736842105263e-06,
      "loss": 1.7091,
      "step": 193
    },
    {
      "epoch": 0.005315141303305525,
      "grad_norm": 11.520868301391602,
      "learning_rate": 3.192631578947369e-06,
      "loss": 2.1969,
      "step": 194
    },
    {
      "epoch": 0.005342538938889574,
      "grad_norm": 10.084508895874023,
      "learning_rate": 2.6605263157894737e-06,
      "loss": 1.4943,
      "step": 195
    },
    {
      "epoch": 0.005369936574473623,
      "grad_norm": 10.452617645263672,
      "learning_rate": 2.128421052631579e-06,
      "loss": 1.5151,
      "step": 196
    },
    {
      "epoch": 0.005397334210057672,
      "grad_norm": 8.461989402770996,
      "learning_rate": 1.5963157894736844e-06,
      "loss": 1.3936,
      "step": 197
    },
    {
      "epoch": 0.005424731845641721,
      "grad_norm": 8.975988388061523,
      "learning_rate": 1.0642105263157895e-06,
      "loss": 1.3449,
      "step": 198
    },
    {
      "epoch": 0.00545212948122577,
      "grad_norm": 9.221831321716309,
      "learning_rate": 5.321052631578948e-07,
      "loss": 1.0431,
      "step": 199
    },
    {
      "epoch": 0.005479527116809819,
      "grad_norm": 9.018491744995117,
      "learning_rate": 0.0,
      "loss": 1.1321,
      "step": 200
    },
    {
      "epoch": 0.005479527116809819,
      "eval_loss": 1.9446079730987549,
      "eval_runtime": 906.8244,
      "eval_samples_per_second": 16.948,
      "eval_steps_per_second": 4.238,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.303878615793664e+16,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}