{
  "best_metric": 2.710070848464966,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.06210215804999224,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0003105107902499612,
      "grad_norm": 29.29452133178711,
      "learning_rate": 1.003e-05,
      "loss": 6.4691,
      "step": 1
    },
    {
      "epoch": 0.0003105107902499612,
      "eval_loss": 3.4762182235717773,
      "eval_runtime": 174.2602,
      "eval_samples_per_second": 7.781,
      "eval_steps_per_second": 1.945,
      "step": 1
    },
    {
      "epoch": 0.0006210215804999224,
      "grad_norm": 33.738006591796875,
      "learning_rate": 2.006e-05,
      "loss": 6.5997,
      "step": 2
    },
    {
      "epoch": 0.0009315323707498836,
      "grad_norm": 21.22829818725586,
      "learning_rate": 3.0089999999999998e-05,
      "loss": 6.521,
      "step": 3
    },
    {
      "epoch": 0.0012420431609998447,
      "grad_norm": 19.39741325378418,
      "learning_rate": 4.012e-05,
      "loss": 6.243,
      "step": 4
    },
    {
      "epoch": 0.0015525539512498058,
      "grad_norm": 18.032846450805664,
      "learning_rate": 5.015e-05,
      "loss": 6.0875,
      "step": 5
    },
    {
      "epoch": 0.0018630647414997672,
      "grad_norm": 4.453341960906982,
      "learning_rate": 6.0179999999999996e-05,
      "loss": 6.0478,
      "step": 6
    },
    {
      "epoch": 0.002173575531749728,
      "grad_norm": 31.29482650756836,
      "learning_rate": 7.021e-05,
      "loss": 6.1593,
      "step": 7
    },
    {
      "epoch": 0.0024840863219996894,
      "grad_norm": 5.259471893310547,
      "learning_rate": 8.024e-05,
      "loss": 6.2802,
      "step": 8
    },
    {
      "epoch": 0.0027945971122496508,
      "grad_norm": 28.345125198364258,
      "learning_rate": 9.027e-05,
      "loss": 6.4323,
      "step": 9
    },
    {
      "epoch": 0.0031051079024996117,
      "grad_norm": 1.65045964717865,
      "learning_rate": 0.0001003,
      "loss": 5.9546,
      "step": 10
    },
    {
      "epoch": 0.003415618692749573,
      "grad_norm": 13.602124214172363,
      "learning_rate": 9.97721052631579e-05,
      "loss": 6.123,
      "step": 11
    },
    {
      "epoch": 0.0037261294829995344,
      "grad_norm": 11.951309204101562,
      "learning_rate": 9.924421052631578e-05,
      "loss": 5.9864,
      "step": 12
    },
    {
      "epoch": 0.004036640273249495,
      "grad_norm": 36.824344635009766,
      "learning_rate": 9.871631578947368e-05,
      "loss": 6.1966,
      "step": 13
    },
    {
      "epoch": 0.004347151063499456,
      "grad_norm": 8.036162376403809,
      "learning_rate": 9.818842105263158e-05,
      "loss": 6.3172,
      "step": 14
    },
    {
      "epoch": 0.004657661853749418,
      "grad_norm": 21.814775466918945,
      "learning_rate": 9.766052631578948e-05,
      "loss": 6.2402,
      "step": 15
    },
    {
      "epoch": 0.004968172643999379,
      "grad_norm": 12.35395336151123,
      "learning_rate": 9.713263157894736e-05,
      "loss": 6.067,
      "step": 16
    },
    {
      "epoch": 0.00527868343424934,
      "grad_norm": 3.0644829273223877,
      "learning_rate": 9.660473684210526e-05,
      "loss": 5.9228,
      "step": 17
    },
    {
      "epoch": 0.0055891942244993015,
      "grad_norm": 7.201811790466309,
      "learning_rate": 9.607684210526316e-05,
      "loss": 5.9131,
      "step": 18
    },
    {
      "epoch": 0.0058997050147492625,
      "grad_norm": 4.769659042358398,
      "learning_rate": 9.554894736842104e-05,
      "loss": 6.0645,
      "step": 19
    },
    {
      "epoch": 0.006210215804999223,
      "grad_norm": 1.7823907136917114,
      "learning_rate": 9.502105263157894e-05,
      "loss": 5.857,
      "step": 20
    },
    {
      "epoch": 0.006520726595249185,
      "grad_norm": 1.8709537982940674,
      "learning_rate": 9.449315789473684e-05,
      "loss": 6.0435,
      "step": 21
    },
    {
      "epoch": 0.006831237385499146,
      "grad_norm": 3.4857585430145264,
      "learning_rate": 9.396526315789474e-05,
      "loss": 5.9443,
      "step": 22
    },
    {
      "epoch": 0.007141748175749107,
      "grad_norm": 16.777128219604492,
      "learning_rate": 9.343736842105264e-05,
      "loss": 6.1941,
      "step": 23
    },
    {
      "epoch": 0.007452258965999069,
      "grad_norm": 4.266728401184082,
      "learning_rate": 9.290947368421052e-05,
      "loss": 6.0035,
      "step": 24
    },
    {
      "epoch": 0.00776276975624903,
      "grad_norm": 7.070564270019531,
      "learning_rate": 9.238157894736842e-05,
      "loss": 6.0351,
      "step": 25
    },
    {
      "epoch": 0.00807328054649899,
      "grad_norm": 31.87989044189453,
      "learning_rate": 9.18536842105263e-05,
      "loss": 6.2323,
      "step": 26
    },
    {
      "epoch": 0.008383791336748951,
      "grad_norm": 2.6480438709259033,
      "learning_rate": 9.132578947368422e-05,
      "loss": 6.0649,
      "step": 27
    },
    {
      "epoch": 0.008694302126998912,
      "grad_norm": 2.978304386138916,
      "learning_rate": 9.07978947368421e-05,
      "loss": 5.9796,
      "step": 28
    },
    {
      "epoch": 0.009004812917248875,
      "grad_norm": 2.9000942707061768,
      "learning_rate": 9.027e-05,
      "loss": 5.9194,
      "step": 29
    },
    {
      "epoch": 0.009315323707498836,
      "grad_norm": 16.504867553710938,
      "learning_rate": 8.97421052631579e-05,
      "loss": 6.226,
      "step": 30
    },
    {
      "epoch": 0.009625834497748797,
      "grad_norm": 2.9846396446228027,
      "learning_rate": 8.921421052631578e-05,
      "loss": 5.9803,
      "step": 31
    },
    {
      "epoch": 0.009936345287998758,
      "grad_norm": 1.9557408094406128,
      "learning_rate": 8.868631578947368e-05,
      "loss": 5.9391,
      "step": 32
    },
    {
      "epoch": 0.010246856078248719,
      "grad_norm": 5.9934821128845215,
      "learning_rate": 8.815842105263157e-05,
      "loss": 5.9945,
      "step": 33
    },
    {
      "epoch": 0.01055736686849868,
      "grad_norm": 3.079267978668213,
      "learning_rate": 8.763052631578948e-05,
      "loss": 5.8665,
      "step": 34
    },
    {
      "epoch": 0.010867877658748642,
      "grad_norm": 2.353879928588867,
      "learning_rate": 8.710263157894737e-05,
      "loss": 5.9495,
      "step": 35
    },
    {
      "epoch": 0.011178388448998603,
      "grad_norm": 2.3978257179260254,
      "learning_rate": 8.657473684210526e-05,
      "loss": 5.9834,
      "step": 36
    },
    {
      "epoch": 0.011488899239248564,
      "grad_norm": 1.9142018556594849,
      "learning_rate": 8.604684210526316e-05,
      "loss": 6.1172,
      "step": 37
    },
    {
      "epoch": 0.011799410029498525,
      "grad_norm": 4.943174839019775,
      "learning_rate": 8.551894736842105e-05,
      "loss": 5.8814,
      "step": 38
    },
    {
      "epoch": 0.012109920819748486,
      "grad_norm": 2.114778518676758,
      "learning_rate": 8.499105263157895e-05,
      "loss": 6.1513,
      "step": 39
    },
    {
      "epoch": 0.012420431609998447,
      "grad_norm": 6.381146430969238,
      "learning_rate": 8.446315789473683e-05,
      "loss": 6.0642,
      "step": 40
    },
    {
      "epoch": 0.01273094240024841,
      "grad_norm": 2.5670583248138428,
      "learning_rate": 8.393526315789474e-05,
      "loss": 5.934,
      "step": 41
    },
    {
      "epoch": 0.01304145319049837,
      "grad_norm": 3.5908706188201904,
      "learning_rate": 8.340736842105263e-05,
      "loss": 5.9664,
      "step": 42
    },
    {
      "epoch": 0.013351963980748331,
      "grad_norm": 4.755733013153076,
      "learning_rate": 8.287947368421053e-05,
      "loss": 5.9252,
      "step": 43
    },
    {
      "epoch": 0.013662474770998292,
      "grad_norm": 3.1427011489868164,
      "learning_rate": 8.235157894736842e-05,
      "loss": 5.8733,
      "step": 44
    },
    {
      "epoch": 0.013972985561248253,
      "grad_norm": 4.566623687744141,
      "learning_rate": 8.182368421052631e-05,
      "loss": 5.9674,
      "step": 45
    },
    {
      "epoch": 0.014283496351498214,
      "grad_norm": 1.8994089365005493,
      "learning_rate": 8.129578947368421e-05,
      "loss": 5.8891,
      "step": 46
    },
    {
      "epoch": 0.014594007141748177,
      "grad_norm": 2.4542181491851807,
      "learning_rate": 8.07678947368421e-05,
      "loss": 6.0727,
      "step": 47
    },
    {
      "epoch": 0.014904517931998137,
      "grad_norm": 3.8890998363494873,
      "learning_rate": 8.024e-05,
      "loss": 5.7988,
      "step": 48
    },
    {
      "epoch": 0.015215028722248098,
      "grad_norm": 6.967671871185303,
      "learning_rate": 7.97121052631579e-05,
      "loss": 5.9372,
      "step": 49
    },
    {
      "epoch": 0.01552553951249806,
      "grad_norm": 3.1538615226745605,
      "learning_rate": 7.918421052631579e-05,
      "loss": 6.0131,
      "step": 50
    },
    {
      "epoch": 0.01552553951249806,
      "eval_loss": 2.966737747192383,
      "eval_runtime": 174.432,
      "eval_samples_per_second": 7.774,
      "eval_steps_per_second": 1.943,
      "step": 50
    },
    {
      "epoch": 0.015836050302748022,
      "grad_norm": 1.8466829061508179,
      "learning_rate": 7.865631578947369e-05,
      "loss": 5.5449,
      "step": 51
    },
    {
      "epoch": 0.01614656109299798,
      "grad_norm": 2.522819995880127,
      "learning_rate": 7.812842105263157e-05,
      "loss": 5.6468,
      "step": 52
    },
    {
      "epoch": 0.016457071883247944,
      "grad_norm": 1.65534508228302,
      "learning_rate": 7.760052631578947e-05,
      "loss": 5.7346,
      "step": 53
    },
    {
      "epoch": 0.016767582673497903,
      "grad_norm": 3.5220515727996826,
      "learning_rate": 7.707263157894737e-05,
      "loss": 5.8865,
      "step": 54
    },
    {
      "epoch": 0.017078093463747866,
      "grad_norm": 2.1265854835510254,
      "learning_rate": 7.654473684210527e-05,
      "loss": 5.8832,
      "step": 55
    },
    {
      "epoch": 0.017388604253997825,
      "grad_norm": 2.1006107330322266,
      "learning_rate": 7.601684210526316e-05,
      "loss": 5.9314,
      "step": 56
    },
    {
      "epoch": 0.017699115044247787,
      "grad_norm": 3.3040337562561035,
      "learning_rate": 7.548894736842105e-05,
      "loss": 6.0435,
      "step": 57
    },
    {
      "epoch": 0.01800962583449775,
      "grad_norm": 4.658459186553955,
      "learning_rate": 7.496105263157895e-05,
      "loss": 6.0449,
      "step": 58
    },
    {
      "epoch": 0.01832013662474771,
      "grad_norm": 5.105825901031494,
      "learning_rate": 7.443315789473683e-05,
      "loss": 6.1849,
      "step": 59
    },
    {
      "epoch": 0.018630647414997672,
      "grad_norm": 2.3449723720550537,
      "learning_rate": 7.390526315789473e-05,
      "loss": 6.2216,
      "step": 60
    },
    {
      "epoch": 0.01894115820524763,
      "grad_norm": 3.2831008434295654,
      "learning_rate": 7.337736842105263e-05,
      "loss": 6.5902,
      "step": 61
    },
    {
      "epoch": 0.019251668995497594,
      "grad_norm": 3.941854238510132,
      "learning_rate": 7.284947368421053e-05,
      "loss": 5.8694,
      "step": 62
    },
    {
      "epoch": 0.019562179785747556,
      "grad_norm": 2.985642433166504,
      "learning_rate": 7.232157894736843e-05,
      "loss": 6.0422,
      "step": 63
    },
    {
      "epoch": 0.019872690575997515,
      "grad_norm": 3.05426287651062,
      "learning_rate": 7.179368421052631e-05,
      "loss": 6.0217,
      "step": 64
    },
    {
      "epoch": 0.020183201366247478,
      "grad_norm": 3.321915626525879,
      "learning_rate": 7.126578947368421e-05,
      "loss": 5.789,
      "step": 65
    },
    {
      "epoch": 0.020493712156497437,
      "grad_norm": 3.6718320846557617,
      "learning_rate": 7.07378947368421e-05,
      "loss": 5.8304,
      "step": 66
    },
    {
      "epoch": 0.0208042229467474,
      "grad_norm": 6.508426189422607,
      "learning_rate": 7.021e-05,
      "loss": 6.1865,
      "step": 67
    },
    {
      "epoch": 0.02111473373699736,
      "grad_norm": 3.99029278755188,
      "learning_rate": 6.968210526315789e-05,
      "loss": 5.6967,
      "step": 68
    },
    {
      "epoch": 0.02142524452724732,
      "grad_norm": 2.4960126876831055,
      "learning_rate": 6.915421052631579e-05,
      "loss": 6.0667,
      "step": 69
    },
    {
      "epoch": 0.021735755317497284,
      "grad_norm": 2.0924904346466064,
      "learning_rate": 6.862631578947369e-05,
      "loss": 5.8275,
      "step": 70
    },
    {
      "epoch": 0.022046266107747244,
      "grad_norm": 4.346602916717529,
      "learning_rate": 6.809842105263157e-05,
      "loss": 6.0083,
      "step": 71
    },
    {
      "epoch": 0.022356776897997206,
      "grad_norm": 4.088897228240967,
      "learning_rate": 6.757052631578947e-05,
      "loss": 5.8602,
      "step": 72
    },
    {
      "epoch": 0.022667287688247165,
      "grad_norm": 6.562732696533203,
      "learning_rate": 6.704263157894737e-05,
      "loss": 5.6424,
      "step": 73
    },
    {
      "epoch": 0.022977798478497128,
      "grad_norm": 2.6807758808135986,
      "learning_rate": 6.651473684210526e-05,
      "loss": 6.1966,
      "step": 74
    },
    {
      "epoch": 0.02328830926874709,
      "grad_norm": 4.0222930908203125,
      "learning_rate": 6.598684210526317e-05,
      "loss": 5.4458,
      "step": 75
    },
    {
      "epoch": 0.02359882005899705,
      "grad_norm": 2.724647045135498,
      "learning_rate": 6.545894736842105e-05,
      "loss": 5.7738,
      "step": 76
    },
    {
      "epoch": 0.023909330849247012,
      "grad_norm": 2.0762064456939697,
      "learning_rate": 6.493105263157895e-05,
      "loss": 5.895,
      "step": 77
    },
    {
      "epoch": 0.02421984163949697,
      "grad_norm": 2.2231438159942627,
      "learning_rate": 6.440315789473684e-05,
      "loss": 5.7483,
      "step": 78
    },
    {
      "epoch": 0.024530352429746934,
      "grad_norm": 19.185461044311523,
      "learning_rate": 6.387526315789473e-05,
      "loss": 6.1952,
      "step": 79
    },
    {
      "epoch": 0.024840863219996893,
      "grad_norm": 5.888047695159912,
      "learning_rate": 6.334736842105263e-05,
      "loss": 5.8966,
      "step": 80
    },
    {
      "epoch": 0.025151374010246856,
      "grad_norm": 6.141130447387695,
      "learning_rate": 6.281947368421052e-05,
      "loss": 5.4849,
      "step": 81
    },
    {
      "epoch": 0.02546188480049682,
      "grad_norm": 2.750977039337158,
      "learning_rate": 6.229157894736843e-05,
      "loss": 5.6882,
      "step": 82
    },
    {
      "epoch": 0.025772395590746778,
      "grad_norm": 3.1710402965545654,
      "learning_rate": 6.176368421052631e-05,
      "loss": 5.9581,
      "step": 83
    },
    {
      "epoch": 0.02608290638099674,
      "grad_norm": 5.4258809089660645,
      "learning_rate": 6.123578947368421e-05,
      "loss": 5.9507,
      "step": 84
    },
    {
      "epoch": 0.0263934171712467,
      "grad_norm": 2.3955416679382324,
      "learning_rate": 6.0707894736842105e-05,
      "loss": 5.4458,
      "step": 85
    },
    {
      "epoch": 0.026703927961496662,
      "grad_norm": 2.2369306087493896,
      "learning_rate": 6.0179999999999996e-05,
      "loss": 5.3165,
      "step": 86
    },
    {
      "epoch": 0.02701443875174662,
      "grad_norm": 3.8684492111206055,
      "learning_rate": 5.965210526315789e-05,
      "loss": 5.9835,
      "step": 87
    },
    {
      "epoch": 0.027324949541996584,
      "grad_norm": 4.250136852264404,
      "learning_rate": 5.912421052631578e-05,
      "loss": 6.125,
      "step": 88
    },
    {
      "epoch": 0.027635460332246547,
      "grad_norm": 2.880558729171753,
      "learning_rate": 5.8596315789473685e-05,
      "loss": 5.3855,
      "step": 89
    },
    {
      "epoch": 0.027945971122496506,
      "grad_norm": 15.121377944946289,
      "learning_rate": 5.8068421052631583e-05,
      "loss": 5.6324,
      "step": 90
    },
    {
      "epoch": 0.02825648191274647,
      "grad_norm": 4.951239109039307,
      "learning_rate": 5.7540526315789475e-05,
      "loss": 5.9545,
      "step": 91
    },
    {
      "epoch": 0.028566992702996428,
      "grad_norm": 4.406779766082764,
      "learning_rate": 5.701263157894737e-05,
      "loss": 5.7006,
      "step": 92
    },
    {
      "epoch": 0.02887750349324639,
      "grad_norm": 11.799044609069824,
      "learning_rate": 5.648473684210526e-05,
      "loss": 5.2981,
      "step": 93
    },
    {
      "epoch": 0.029188014283496353,
      "grad_norm": 6.532238483428955,
      "learning_rate": 5.595684210526315e-05,
      "loss": 6.2733,
      "step": 94
    },
    {
      "epoch": 0.029498525073746312,
      "grad_norm": 5.7031450271606445,
      "learning_rate": 5.5428947368421055e-05,
      "loss": 5.9861,
      "step": 95
    },
    {
      "epoch": 0.029809035863996275,
      "grad_norm": 3.811985492706299,
      "learning_rate": 5.490105263157895e-05,
      "loss": 5.9362,
      "step": 96
    },
    {
      "epoch": 0.030119546654246234,
      "grad_norm": 5.6020636558532715,
      "learning_rate": 5.4373157894736846e-05,
      "loss": 4.9524,
      "step": 97
    },
    {
      "epoch": 0.030430057444496197,
      "grad_norm": 5.865322113037109,
      "learning_rate": 5.384526315789474e-05,
      "loss": 5.6313,
      "step": 98
    },
    {
      "epoch": 0.030740568234746156,
      "grad_norm": 6.841184139251709,
      "learning_rate": 5.331736842105263e-05,
      "loss": 5.7727,
      "step": 99
    },
    {
      "epoch": 0.03105107902499612,
      "grad_norm": 8.684503555297852,
      "learning_rate": 5.278947368421052e-05,
      "loss": 5.4442,
      "step": 100
    },
    {
      "epoch": 0.03105107902499612,
      "eval_loss": 2.834892988204956,
      "eval_runtime": 174.3575,
      "eval_samples_per_second": 7.777,
      "eval_steps_per_second": 1.944,
      "step": 100
    },
    {
      "epoch": 0.03136158981524608,
      "grad_norm": 3.085808277130127,
      "learning_rate": 5.226157894736842e-05,
      "loss": 5.5206,
      "step": 101
    },
    {
      "epoch": 0.031672100605496044,
      "grad_norm": 3.0579793453216553,
      "learning_rate": 5.173368421052632e-05,
      "loss": 5.6094,
      "step": 102
    },
    {
      "epoch": 0.031982611395746,
      "grad_norm": 11.420687675476074,
      "learning_rate": 5.1205789473684216e-05,
      "loss": 5.846,
      "step": 103
    },
    {
      "epoch": 0.03229312218599596,
      "grad_norm": 3.8784804344177246,
      "learning_rate": 5.067789473684211e-05,
      "loss": 5.8975,
      "step": 104
    },
    {
      "epoch": 0.032603632976245925,
      "grad_norm": 4.274180889129639,
      "learning_rate": 5.015e-05,
      "loss": 5.8573,
      "step": 105
    },
    {
      "epoch": 0.03291414376649589,
      "grad_norm": 4.395181179046631,
      "learning_rate": 4.962210526315789e-05,
      "loss": 6.031,
      "step": 106
    },
    {
      "epoch": 0.03322465455674585,
      "grad_norm": 4.031984806060791,
      "learning_rate": 4.909421052631579e-05,
      "loss": 6.0785,
      "step": 107
    },
    {
      "epoch": 0.033535165346995806,
      "grad_norm": 3.628544807434082,
      "learning_rate": 4.856631578947368e-05,
      "loss": 5.479,
      "step": 108
    },
    {
      "epoch": 0.03384567613724577,
      "grad_norm": 3.298604965209961,
      "learning_rate": 4.803842105263158e-05,
      "loss": 6.2932,
      "step": 109
    },
    {
      "epoch": 0.03415618692749573,
      "grad_norm": 6.361236095428467,
      "learning_rate": 4.751052631578947e-05,
      "loss": 6.5487,
      "step": 110
    },
    {
      "epoch": 0.034466697717745694,
      "grad_norm": 4.709749698638916,
      "learning_rate": 4.698263157894737e-05,
      "loss": 5.7731,
      "step": 111
    },
    {
      "epoch": 0.03477720850799565,
      "grad_norm": 5.268786907196045,
      "learning_rate": 4.645473684210526e-05,
      "loss": 5.9129,
      "step": 112
    },
    {
      "epoch": 0.03508771929824561,
      "grad_norm": 2.7873456478118896,
      "learning_rate": 4.592684210526315e-05,
      "loss": 6.0852,
      "step": 113
    },
    {
      "epoch": 0.035398230088495575,
      "grad_norm": 3.7218127250671387,
      "learning_rate": 4.539894736842105e-05,
      "loss": 5.2699,
      "step": 114
    },
    {
      "epoch": 0.03570874087874554,
      "grad_norm": 6.234576225280762,
      "learning_rate": 4.487105263157895e-05,
      "loss": 5.7415,
      "step": 115
    },
    {
      "epoch": 0.0360192516689955,
      "grad_norm": 4.063146114349365,
      "learning_rate": 4.434315789473684e-05,
      "loss": 5.8919,
      "step": 116
    },
    {
      "epoch": 0.036329762459245456,
      "grad_norm": 2.7963621616363525,
      "learning_rate": 4.381526315789474e-05,
      "loss": 5.8933,
      "step": 117
    },
    {
      "epoch": 0.03664027324949542,
      "grad_norm": 1.9288524389266968,
      "learning_rate": 4.328736842105263e-05,
      "loss": 5.9503,
      "step": 118
    },
    {
      "epoch": 0.03695078403974538,
      "grad_norm": 8.17712688446045,
      "learning_rate": 4.2759473684210523e-05,
      "loss": 5.5091,
      "step": 119
    },
    {
      "epoch": 0.037261294829995344,
      "grad_norm": 3.089323043823242,
      "learning_rate": 4.2231578947368415e-05,
      "loss": 5.8097,
      "step": 120
    },
    {
      "epoch": 0.037571805620245306,
      "grad_norm": 3.7228176593780518,
      "learning_rate": 4.1703684210526314e-05,
      "loss": 5.96,
      "step": 121
    },
    {
      "epoch": 0.03788231641049526,
      "grad_norm": 3.5552074909210205,
      "learning_rate": 4.117578947368421e-05,
      "loss": 5.8232,
      "step": 122
    },
    {
      "epoch": 0.038192827200745225,
      "grad_norm": 5.932594299316406,
      "learning_rate": 4.0647894736842104e-05,
      "loss": 5.8557,
      "step": 123
    },
    {
      "epoch": 0.03850333799099519,
      "grad_norm": 5.419013977050781,
      "learning_rate": 4.012e-05,
      "loss": 5.9569,
      "step": 124
    },
    {
      "epoch": 0.03881384878124515,
      "grad_norm": 7.140962600708008,
      "learning_rate": 3.9592105263157894e-05,
      "loss": 5.9876,
      "step": 125
    },
    {
      "epoch": 0.03912435957149511,
      "grad_norm": 2.0772342681884766,
      "learning_rate": 3.9064210526315785e-05,
      "loss": 5.9403,
      "step": 126
    },
    {
      "epoch": 0.03943487036174507,
      "grad_norm": 2.7006912231445312,
      "learning_rate": 3.8536315789473684e-05,
      "loss": 5.8678,
      "step": 127
    },
    {
      "epoch": 0.03974538115199503,
      "grad_norm": 15.127945899963379,
      "learning_rate": 3.800842105263158e-05,
      "loss": 5.5416,
      "step": 128
    },
    {
      "epoch": 0.040055891942244994,
      "grad_norm": 6.996811389923096,
      "learning_rate": 3.7480526315789474e-05,
      "loss": 5.7005,
      "step": 129
    },
    {
      "epoch": 0.040366402732494956,
      "grad_norm": 7.073221683502197,
      "learning_rate": 3.6952631578947366e-05,
      "loss": 5.2987,
      "step": 130
    },
    {
      "epoch": 0.04067691352274492,
      "grad_norm": 6.710120677947998,
      "learning_rate": 3.6424736842105264e-05,
      "loss": 5.3538,
      "step": 131
    },
    {
      "epoch": 0.040987424312994875,
      "grad_norm": 2.666349411010742,
      "learning_rate": 3.5896842105263156e-05,
      "loss": 5.3141,
      "step": 132
    },
    {
      "epoch": 0.04129793510324484,
      "grad_norm": 10.526546478271484,
      "learning_rate": 3.536894736842105e-05,
      "loss": 4.1289,
      "step": 133
    },
    {
      "epoch": 0.0416084458934948,
      "grad_norm": 9.269521713256836,
      "learning_rate": 3.4841052631578946e-05,
      "loss": 5.2633,
      "step": 134
    },
    {
      "epoch": 0.04191895668374476,
      "grad_norm": 7.547076225280762,
      "learning_rate": 3.4313157894736844e-05,
      "loss": 5.3717,
      "step": 135
    },
    {
      "epoch": 0.04222946747399472,
      "grad_norm": 6.621392726898193,
      "learning_rate": 3.3785263157894736e-05,
      "loss": 5.5902,
      "step": 136
    },
    {
      "epoch": 0.04253997826424468,
      "grad_norm": 4.0747575759887695,
      "learning_rate": 3.325736842105263e-05,
      "loss": 5.9148,
      "step": 137
    },
    {
      "epoch": 0.04285048905449464,
      "grad_norm": 5.389759063720703,
      "learning_rate": 3.2729473684210526e-05,
      "loss": 5.6964,
      "step": 138
    },
    {
      "epoch": 0.043160999844744606,
      "grad_norm": 13.09533405303955,
      "learning_rate": 3.220157894736842e-05,
      "loss": 4.9712,
      "step": 139
    },
    {
      "epoch": 0.04347151063499457,
      "grad_norm": 5.605279445648193,
      "learning_rate": 3.1673684210526316e-05,
      "loss": 5.8146,
      "step": 140
    },
    {
      "epoch": 0.043782021425244524,
      "grad_norm": 9.980596542358398,
      "learning_rate": 3.1145789473684215e-05,
      "loss": 4.6328,
      "step": 141
    },
    {
      "epoch": 0.04409253221549449,
      "grad_norm": 12.333852767944336,
      "learning_rate": 3.0617894736842107e-05,
      "loss": 5.8281,
      "step": 142
    },
    {
      "epoch": 0.04440304300574445,
      "grad_norm": 4.8984880447387695,
      "learning_rate": 3.0089999999999998e-05,
      "loss": 4.6038,
      "step": 143
    },
    {
      "epoch": 0.04471355379599441,
      "grad_norm": 9.472086906433105,
      "learning_rate": 2.956210526315789e-05,
      "loss": 4.6674,
      "step": 144
    },
    {
      "epoch": 0.045024064586244375,
      "grad_norm": 11.404544830322266,
      "learning_rate": 2.9034210526315792e-05,
      "loss": 4.543,
      "step": 145
    },
    {
      "epoch": 0.04533457537649433,
      "grad_norm": 6.597491264343262,
      "learning_rate": 2.8506315789473683e-05,
      "loss": 5.2974,
      "step": 146
    },
    {
      "epoch": 0.04564508616674429,
      "grad_norm": 7.570100784301758,
      "learning_rate": 2.7978421052631575e-05,
      "loss": 4.5947,
      "step": 147
    },
    {
      "epoch": 0.045955596956994256,
      "grad_norm": 3.8692166805267334,
      "learning_rate": 2.7450526315789474e-05,
      "loss": 5.8644,
      "step": 148
    },
    {
      "epoch": 0.04626610774724422,
      "grad_norm": 8.363037109375,
      "learning_rate": 2.692263157894737e-05,
      "loss": 4.8226,
      "step": 149
    },
    {
      "epoch": 0.04657661853749418,
      "grad_norm": 3.570312023162842,
      "learning_rate": 2.639473684210526e-05,
      "loss": 5.8956,
      "step": 150
    },
    {
      "epoch": 0.04657661853749418,
      "eval_loss": 2.761136770248413,
      "eval_runtime": 174.3307,
      "eval_samples_per_second": 7.778,
      "eval_steps_per_second": 1.945,
      "step": 150
    },
    {
      "epoch": 0.04688712932774414,
      "grad_norm": 8.627881050109863,
      "learning_rate": 2.586684210526316e-05,
      "loss": 5.0342,
      "step": 151
    },
    {
      "epoch": 0.0471976401179941,
      "grad_norm": 3.0994157791137695,
      "learning_rate": 2.5338947368421054e-05,
      "loss": 5.6128,
      "step": 152
    },
    {
      "epoch": 0.04750815090824406,
      "grad_norm": 7.781631946563721,
      "learning_rate": 2.4811052631578945e-05,
      "loss": 5.6382,
      "step": 153
    },
    {
      "epoch": 0.047818661698494025,
      "grad_norm": 8.706255912780762,
      "learning_rate": 2.428315789473684e-05,
      "loss": 5.369,
      "step": 154
    },
    {
      "epoch": 0.04812917248874398,
      "grad_norm": 10.600417137145996,
      "learning_rate": 2.3755263157894736e-05,
      "loss": 5.1998,
      "step": 155
    },
    {
      "epoch": 0.04843968327899394,
      "grad_norm": 3.304457664489746,
      "learning_rate": 2.322736842105263e-05,
      "loss": 5.8636,
      "step": 156
    },
    {
      "epoch": 0.048750194069243906,
      "grad_norm": 3.3973822593688965,
      "learning_rate": 2.2699473684210526e-05,
      "loss": 5.9097,
      "step": 157
    },
    {
      "epoch": 0.04906070485949387,
      "grad_norm": 6.2585039138793945,
      "learning_rate": 2.217157894736842e-05,
      "loss": 6.0155,
      "step": 158
    },
    {
      "epoch": 0.04937121564974383,
      "grad_norm": 4.313554286956787,
      "learning_rate": 2.1643684210526316e-05,
      "loss": 5.977,
      "step": 159
    },
    {
      "epoch": 0.04968172643999379,
      "grad_norm": 6.310354232788086,
      "learning_rate": 2.1115789473684208e-05,
      "loss": 5.8565,
      "step": 160
    },
    {
      "epoch": 0.04999223723024375,
      "grad_norm": 18.50254249572754,
      "learning_rate": 2.0587894736842106e-05,
      "loss": 5.9798,
      "step": 161
    },
    {
      "epoch": 0.05030274802049371,
      "grad_norm": 11.727089881896973,
      "learning_rate": 2.006e-05,
      "loss": 6.3748,
      "step": 162
    },
    {
      "epoch": 0.050613258810743675,
      "grad_norm": 4.172285556793213,
      "learning_rate": 1.9532105263157893e-05,
      "loss": 6.3266,
      "step": 163
    },
    {
      "epoch": 0.05092376960099364,
      "grad_norm": 3.4230823516845703,
      "learning_rate": 1.900421052631579e-05,
      "loss": 5.8562,
      "step": 164
    },
    {
      "epoch": 0.05123428039124359,
      "grad_norm": 3.503084421157837,
      "learning_rate": 1.8476315789473683e-05,
      "loss": 5.3421,
      "step": 165
    },
    {
      "epoch": 0.051544791181493556,
      "grad_norm": 3.9270567893981934,
      "learning_rate": 1.7948421052631578e-05,
      "loss": 5.8137,
      "step": 166
    },
    {
      "epoch": 0.05185530197174352,
      "grad_norm": 5.9435882568359375,
      "learning_rate": 1.7420526315789473e-05,
      "loss": 5.5091,
      "step": 167
    },
    {
      "epoch": 0.05216581276199348,
      "grad_norm": 3.307124137878418,
      "learning_rate": 1.6892631578947368e-05,
      "loss": 6.1505,
      "step": 168
    },
    {
      "epoch": 0.052476323552243444,
      "grad_norm": 3.08250093460083,
      "learning_rate": 1.6364736842105263e-05,
      "loss": 5.857,
      "step": 169
    },
    {
      "epoch": 0.0527868343424934,
      "grad_norm": 3.2450950145721436,
      "learning_rate": 1.5836842105263158e-05,
      "loss": 5.7891,
      "step": 170
    },
    {
      "epoch": 0.05309734513274336,
      "grad_norm": 3.96562123298645,
      "learning_rate": 1.5308947368421053e-05,
      "loss": 5.5198,
      "step": 171
    },
    {
      "epoch": 0.053407855922993325,
      "grad_norm": 3.0473573207855225,
      "learning_rate": 1.4781052631578945e-05,
      "loss": 5.7508,
      "step": 172
    },
    {
      "epoch": 0.05371836671324329,
      "grad_norm": 6.961787223815918,
      "learning_rate": 1.4253157894736842e-05,
      "loss": 5.6795,
      "step": 173
    },
    {
      "epoch": 0.05402887750349324,
      "grad_norm": 3.1459827423095703,
      "learning_rate": 1.3725263157894737e-05,
      "loss": 5.4152,
      "step": 174
    },
    {
      "epoch": 0.054339388293743206,
      "grad_norm": 4.134361267089844,
      "learning_rate": 1.319736842105263e-05,
      "loss": 5.8967,
      "step": 175
    },
    {
      "epoch": 0.05464989908399317,
      "grad_norm": 8.840166091918945,
      "learning_rate": 1.2669473684210527e-05,
      "loss": 5.2314,
      "step": 176
    },
    {
      "epoch": 0.05496040987424313,
      "grad_norm": 3.1707940101623535,
      "learning_rate": 1.214157894736842e-05,
      "loss": 5.8824,
      "step": 177
    },
    {
      "epoch": 0.055270920664493094,
      "grad_norm": 2.967905044555664,
      "learning_rate": 1.1613684210526315e-05,
      "loss": 5.9145,
      "step": 178
    },
    {
      "epoch": 0.05558143145474305,
      "grad_norm": 4.332707405090332,
      "learning_rate": 1.108578947368421e-05,
      "loss": 5.7882,
      "step": 179
    },
    {
      "epoch": 0.05589194224499301,
      "grad_norm": 3.2459423542022705,
      "learning_rate": 1.0557894736842104e-05,
      "loss": 5.0901,
      "step": 180
    },
    {
      "epoch": 0.056202453035242975,
      "grad_norm": 2.94150710105896,
      "learning_rate": 1.003e-05,
      "loss": 5.3414,
      "step": 181
    },
    {
      "epoch": 0.05651296382549294,
      "grad_norm": 5.878596305847168,
      "learning_rate": 9.502105263157896e-06,
      "loss": 4.6183,
      "step": 182
    },
    {
      "epoch": 0.0568234746157429,
      "grad_norm": 4.656161308288574,
      "learning_rate": 8.974210526315789e-06,
      "loss": 5.2575,
      "step": 183
    },
    {
      "epoch": 0.057133985405992856,
      "grad_norm": 3.8339030742645264,
      "learning_rate": 8.446315789473684e-06,
      "loss": 5.2496,
      "step": 184
    },
    {
      "epoch": 0.05744449619624282,
      "grad_norm": 4.35939884185791,
      "learning_rate": 7.918421052631579e-06,
      "loss": 5.0508,
      "step": 185
    },
    {
      "epoch": 0.05775500698649278,
      "grad_norm": 4.872880935668945,
      "learning_rate": 7.3905263157894725e-06,
      "loss": 5.2661,
      "step": 186
    },
    {
      "epoch": 0.058065517776742744,
      "grad_norm": 3.8898978233337402,
      "learning_rate": 6.862631578947368e-06,
      "loss": 5.0692,
      "step": 187
    },
    {
      "epoch": 0.058376028566992706,
      "grad_norm": 9.882028579711914,
      "learning_rate": 6.3347368421052634e-06,
      "loss": 5.5958,
      "step": 188
    },
    {
      "epoch": 0.05868653935724266,
      "grad_norm": 6.079442024230957,
      "learning_rate": 5.806842105263158e-06,
      "loss": 5.6492,
      "step": 189
    },
    {
      "epoch": 0.058997050147492625,
      "grad_norm": 5.305960655212402,
      "learning_rate": 5.278947368421052e-06,
      "loss": 5.9404,
      "step": 190
    },
    {
      "epoch": 0.05930756093774259,
      "grad_norm": 9.125515937805176,
      "learning_rate": 4.751052631578948e-06,
      "loss": 5.1715,
      "step": 191
    },
    {
      "epoch": 0.05961807172799255,
      "grad_norm": 13.075247764587402,
      "learning_rate": 4.223157894736842e-06,
      "loss": 5.723,
      "step": 192
    },
    {
      "epoch": 0.059928582518242506,
      "grad_norm": 11.328529357910156,
      "learning_rate": 3.6952631578947362e-06,
      "loss": 5.354,
      "step": 193
    },
    {
      "epoch": 0.06023909330849247,
      "grad_norm": 7.932187080383301,
      "learning_rate": 3.1673684210526317e-06,
      "loss": 4.9868,
      "step": 194
    },
    {
      "epoch": 0.06054960409874243,
      "grad_norm": 4.846871852874756,
      "learning_rate": 2.639473684210526e-06,
      "loss": 5.9433,
      "step": 195
    },
    {
      "epoch": 0.06086011488899239,
      "grad_norm": 7.914638519287109,
      "learning_rate": 2.111578947368421e-06,
      "loss": 5.3406,
      "step": 196
    },
    {
      "epoch": 0.061170625679242356,
      "grad_norm": 11.115379333496094,
      "learning_rate": 1.5836842105263159e-06,
      "loss": 4.7257,
      "step": 197
    },
    {
      "epoch": 0.06148113646949231,
      "grad_norm": 12.03275203704834,
      "learning_rate": 1.0557894736842105e-06,
      "loss": 3.4278,
      "step": 198
    },
    {
      "epoch": 0.061791647259742274,
      "grad_norm": 8.894533157348633,
      "learning_rate": 5.278947368421053e-07,
      "loss": 3.2262,
      "step": 199
    },
    {
      "epoch": 0.06210215804999224,
      "grad_norm": 5.871228218078613,
      "learning_rate": 0.0,
      "loss": 4.6102,
      "step": 200
    },
    {
      "epoch": 0.06210215804999224,
      "eval_loss": 2.710070848464966,
      "eval_runtime": 174.2701,
      "eval_samples_per_second": 7.781,
      "eval_steps_per_second": 1.945,
      "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": 8.78275028016169e+16,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}