{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 1500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "grad_norm": 119.98773193359375,
      "learning_rate": 6.000000000000001e-07,
      "loss": 2.5941,
      "step": 10
    },
    {
      "epoch": 0.04,
      "grad_norm": 68.91211700439453,
      "learning_rate": 1.2666666666666669e-06,
      "loss": 2.2812,
      "step": 20
    },
    {
      "epoch": 0.06,
      "grad_norm": 53.26000213623047,
      "learning_rate": 1.9333333333333336e-06,
      "loss": 1.2541,
      "step": 30
    },
    {
      "epoch": 0.08,
      "grad_norm": 34.61829376220703,
      "learning_rate": 2.6e-06,
      "loss": 0.7484,
      "step": 40
    },
    {
      "epoch": 0.1,
      "grad_norm": 32.14445495605469,
      "learning_rate": 3.266666666666667e-06,
      "loss": 0.5454,
      "step": 50
    },
    {
      "epoch": 0.12,
      "grad_norm": 38.480838775634766,
      "learning_rate": 3.9333333333333335e-06,
      "loss": 0.8012,
      "step": 60
    },
    {
      "epoch": 0.14,
      "grad_norm": 27.056198120117188,
      "learning_rate": 4.600000000000001e-06,
      "loss": 0.6157,
      "step": 70
    },
    {
      "epoch": 0.16,
      "grad_norm": 27.715463638305664,
      "learning_rate": 5.2666666666666665e-06,
      "loss": 0.6487,
      "step": 80
    },
    {
      "epoch": 0.18,
      "grad_norm": 26.66598129272461,
      "learning_rate": 5.933333333333335e-06,
      "loss": 0.4948,
      "step": 90
    },
    {
      "epoch": 0.2,
      "grad_norm": 29.367525100708008,
      "learning_rate": 6.600000000000001e-06,
      "loss": 0.5841,
      "step": 100
    },
    {
      "epoch": 0.22,
      "grad_norm": 27.86680793762207,
      "learning_rate": 7.266666666666668e-06,
      "loss": 0.6287,
      "step": 110
    },
    {
      "epoch": 0.24,
      "grad_norm": 22.585514068603516,
      "learning_rate": 7.933333333333334e-06,
      "loss": 0.7193,
      "step": 120
    },
    {
      "epoch": 0.26,
      "grad_norm": 32.08122253417969,
      "learning_rate": 8.6e-06,
      "loss": 0.6858,
      "step": 130
    },
    {
      "epoch": 0.28,
      "grad_norm": 20.128015518188477,
      "learning_rate": 9.266666666666667e-06,
      "loss": 0.5262,
      "step": 140
    },
    {
      "epoch": 0.3,
      "grad_norm": 25.974939346313477,
      "learning_rate": 9.933333333333334e-06,
      "loss": 0.4959,
      "step": 150
    },
    {
      "epoch": 0.32,
      "grad_norm": 29.289508819580078,
      "learning_rate": 9.998903417374228e-06,
      "loss": 0.8063,
      "step": 160
    },
    {
      "epoch": 0.34,
      "grad_norm": 24.139253616333008,
      "learning_rate": 9.995113378907791e-06,
      "loss": 0.7603,
      "step": 170
    },
    {
      "epoch": 0.36,
      "grad_norm": 35.83868408203125,
      "learning_rate": 9.988618398427495e-06,
      "loss": 0.54,
      "step": 180
    },
    {
      "epoch": 0.38,
      "grad_norm": 17.248802185058594,
      "learning_rate": 9.979421993079853e-06,
      "loss": 0.5126,
      "step": 190
    },
    {
      "epoch": 0.4,
      "grad_norm": 39.79250717163086,
      "learning_rate": 9.967529142880592e-06,
      "loss": 0.5866,
      "step": 200
    },
    {
      "epoch": 0.42,
      "grad_norm": 28.33167266845703,
      "learning_rate": 9.952946288017899e-06,
      "loss": 0.6714,
      "step": 210
    },
    {
      "epoch": 0.44,
      "grad_norm": 28.86366844177246,
      "learning_rate": 9.93568132536494e-06,
      "loss": 0.7777,
      "step": 220
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.633607387542725,
      "learning_rate": 9.915743604203563e-06,
      "loss": 0.4985,
      "step": 230
    },
    {
      "epoch": 0.48,
      "grad_norm": 27.641942977905273,
      "learning_rate": 9.893143921161501e-06,
      "loss": 0.6369,
      "step": 240
    },
    {
      "epoch": 0.5,
      "grad_norm": 33.35298156738281,
      "learning_rate": 9.867894514365802e-06,
      "loss": 0.522,
      "step": 250
    },
    {
      "epoch": 0.52,
      "grad_norm": 24.139148712158203,
      "learning_rate": 9.840009056815674e-06,
      "loss": 0.7008,
      "step": 260
    },
    {
      "epoch": 0.54,
      "grad_norm": 18.907367706298828,
      "learning_rate": 9.809502648978311e-06,
      "loss": 0.5756,
      "step": 270
    },
    {
      "epoch": 0.56,
      "grad_norm": 12.178987503051758,
      "learning_rate": 9.776391810611719e-06,
      "loss": 0.5105,
      "step": 280
    },
    {
      "epoch": 0.58,
      "grad_norm": 20.952960968017578,
      "learning_rate": 9.740694471818988e-06,
      "loss": 0.5805,
      "step": 290
    },
    {
      "epoch": 0.6,
      "grad_norm": 22.51838493347168,
      "learning_rate": 9.702429963338812e-06,
      "loss": 0.629,
      "step": 300
    },
    {
      "epoch": 0.62,
      "grad_norm": 24.876291275024414,
      "learning_rate": 9.661619006077562e-06,
      "loss": 0.4561,
      "step": 310
    },
    {
      "epoch": 0.64,
      "grad_norm": 20.83371353149414,
      "learning_rate": 9.618283699888543e-06,
      "loss": 0.5428,
      "step": 320
    },
    {
      "epoch": 0.66,
      "grad_norm": 18.498687744140625,
      "learning_rate": 9.572447511604536e-06,
      "loss": 0.7385,
      "step": 330
    },
    {
      "epoch": 0.68,
      "grad_norm": 23.148433685302734,
      "learning_rate": 9.524135262330098e-06,
      "loss": 0.688,
      "step": 340
    },
    {
      "epoch": 0.7,
      "grad_norm": 23.14569854736328,
      "learning_rate": 9.473373114000493e-06,
      "loss": 0.5262,
      "step": 350
    },
    {
      "epoch": 0.72,
      "grad_norm": 14.241995811462402,
      "learning_rate": 9.420188555214537e-06,
      "loss": 0.45,
      "step": 360
    },
    {
      "epoch": 0.74,
      "grad_norm": 12.99563217163086,
      "learning_rate": 9.364610386349048e-06,
      "loss": 0.6064,
      "step": 370
    },
    {
      "epoch": 0.76,
      "grad_norm": 20.02166175842285,
      "learning_rate": 9.306668703962927e-06,
      "loss": 0.7147,
      "step": 380
    },
    {
      "epoch": 0.78,
      "grad_norm": 24.161821365356445,
      "learning_rate": 9.246394884499334e-06,
      "loss": 0.4635,
      "step": 390
    },
    {
      "epoch": 0.8,
      "grad_norm": 22.88410758972168,
      "learning_rate": 9.18382156729481e-06,
      "loss": 0.5259,
      "step": 400
    },
    {
      "epoch": 0.82,
      "grad_norm": 22.740713119506836,
      "learning_rate": 9.118982636904476e-06,
      "loss": 0.552,
      "step": 410
    },
    {
      "epoch": 0.84,
      "grad_norm": 21.54591178894043,
      "learning_rate": 9.051913204752972e-06,
      "loss": 0.5106,
      "step": 420
    },
    {
      "epoch": 0.86,
      "grad_norm": 18.233713150024414,
      "learning_rate": 8.982649590120982e-06,
      "loss": 0.5603,
      "step": 430
    },
    {
      "epoch": 0.88,
      "grad_norm": 18.04786491394043,
      "learning_rate": 8.911229300477716e-06,
      "loss": 0.6289,
      "step": 440
    },
    {
      "epoch": 0.9,
      "grad_norm": 17.79781150817871,
      "learning_rate": 8.837691011169944e-06,
      "loss": 0.5141,
      "step": 450
    },
    {
      "epoch": 0.92,
      "grad_norm": 23.359535217285156,
      "learning_rate": 8.762074544478622e-06,
      "loss": 0.4825,
      "step": 460
    },
    {
      "epoch": 0.94,
      "grad_norm": 22.502107620239258,
      "learning_rate": 8.68442084805442e-06,
      "loss": 0.6841,
      "step": 470
    },
    {
      "epoch": 0.96,
      "grad_norm": 28.555416107177734,
      "learning_rate": 8.604771972743848e-06,
      "loss": 0.6332,
      "step": 480
    },
    {
      "epoch": 0.98,
      "grad_norm": 15.663265228271484,
      "learning_rate": 8.523171049817974e-06,
      "loss": 0.6694,
      "step": 490
    },
    {
      "epoch": 1.0,
      "grad_norm": 14.243117332458496,
      "learning_rate": 8.439662267616093e-06,
      "loss": 0.5578,
      "step": 500
    },
    {
      "epoch": 1.02,
      "grad_norm": 15.848461151123047,
      "learning_rate": 8.354290847616954e-06,
      "loss": 0.2886,
      "step": 510
    },
    {
      "epoch": 1.04,
      "grad_norm": 19.934585571289062,
      "learning_rate": 8.267103019950529e-06,
      "loss": 0.3467,
      "step": 520
    },
    {
      "epoch": 1.06,
      "grad_norm": 13.80959415435791,
      "learning_rate": 8.178145998363585e-06,
      "loss": 0.2702,
      "step": 530
    },
    {
      "epoch": 1.08,
      "grad_norm": 30.82528305053711,
      "learning_rate": 8.087467954652608e-06,
      "loss": 0.2224,
      "step": 540
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.0778735876083374,
      "learning_rate": 7.99511799257793e-06,
      "loss": 0.2643,
      "step": 550
    },
    {
      "epoch": 1.12,
      "grad_norm": 21.56842041015625,
      "learning_rate": 7.901146121273165e-06,
      "loss": 0.2383,
      "step": 560
    },
    {
      "epoch": 1.1400000000000001,
      "grad_norm": 17.69034767150879,
      "learning_rate": 7.80560322816439e-06,
      "loss": 0.2724,
      "step": 570
    },
    {
      "epoch": 1.16,
      "grad_norm": 14.281789779663086,
      "learning_rate": 7.7085410514137e-06,
      "loss": 0.3057,
      "step": 580
    },
    {
      "epoch": 1.18,
      "grad_norm": 13.571429252624512,
      "learning_rate": 7.610012151902091e-06,
      "loss": 0.347,
      "step": 590
    },
    {
      "epoch": 1.2,
      "grad_norm": 22.4501895904541,
      "learning_rate": 7.510069884766802e-06,
      "loss": 0.2657,
      "step": 600
    },
    {
      "epoch": 1.22,
      "grad_norm": 4.763045310974121,
      "learning_rate": 7.408768370508577e-06,
      "loss": 0.2981,
      "step": 610
    },
    {
      "epoch": 1.24,
      "grad_norm": 21.819042205810547,
      "learning_rate": 7.3061624656844544e-06,
      "loss": 0.3681,
      "step": 620
    },
    {
      "epoch": 1.26,
      "grad_norm": 15.343484878540039,
      "learning_rate": 7.2023077332019755e-06,
      "loss": 0.2492,
      "step": 630
    },
    {
      "epoch": 1.28,
      "grad_norm": 8.440631866455078,
      "learning_rate": 7.0972604122308865e-06,
      "loss": 0.2866,
      "step": 640
    },
    {
      "epoch": 1.3,
      "grad_norm": 20.592426300048828,
      "learning_rate": 6.991077387748643e-06,
      "loss": 0.2569,
      "step": 650
    },
    {
      "epoch": 1.32,
      "grad_norm": 4.904416561126709,
      "learning_rate": 6.883816159736187e-06,
      "loss": 0.2827,
      "step": 660
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.7817937135696411,
      "learning_rate": 6.775534812040686e-06,
      "loss": 0.2119,
      "step": 670
    },
    {
      "epoch": 1.3599999999999999,
      "grad_norm": 5.5259318351745605,
      "learning_rate": 6.666291980922122e-06,
      "loss": 0.2444,
      "step": 680
    },
    {
      "epoch": 1.38,
      "grad_norm": 11.531004905700684,
      "learning_rate": 6.556146823300701e-06,
      "loss": 0.2244,
      "step": 690
    },
    {
      "epoch": 1.4,
      "grad_norm": 12.882322311401367,
      "learning_rate": 6.445158984722358e-06,
      "loss": 0.2498,
      "step": 700
    },
    {
      "epoch": 1.42,
      "grad_norm": 25.625864028930664,
      "learning_rate": 6.3333885670596285e-06,
      "loss": 0.2757,
      "step": 710
    },
    {
      "epoch": 1.44,
      "grad_norm": 35.734169006347656,
      "learning_rate": 6.220896095965428e-06,
      "loss": 0.3523,
      "step": 720
    },
    {
      "epoch": 1.46,
      "grad_norm": 29.897642135620117,
      "learning_rate": 6.107742488097338e-06,
      "loss": 0.2827,
      "step": 730
    },
    {
      "epoch": 1.48,
      "grad_norm": 30.538116455078125,
      "learning_rate": 5.993989018130173e-06,
      "loss": 0.2875,
      "step": 740
    },
    {
      "epoch": 1.5,
      "grad_norm": 22.707246780395508,
      "learning_rate": 5.879697285574655e-06,
      "loss": 0.263,
      "step": 750
    },
    {
      "epoch": 1.52,
      "grad_norm": 20.372825622558594,
      "learning_rate": 5.764929181420191e-06,
      "loss": 0.2907,
      "step": 760
    },
    {
      "epoch": 1.54,
      "grad_norm": 16.188474655151367,
      "learning_rate": 5.649746854619814e-06,
      "loss": 0.3206,
      "step": 770
    },
    {
      "epoch": 1.56,
      "grad_norm": 14.655498504638672,
      "learning_rate": 5.5342126784354265e-06,
      "loss": 0.3373,
      "step": 780
    },
    {
      "epoch": 1.58,
      "grad_norm": 13.933366775512695,
      "learning_rate": 5.41838921666158e-06,
      "loss": 0.304,
      "step": 790
    },
    {
      "epoch": 1.6,
      "grad_norm": 18.03886604309082,
      "learning_rate": 5.3023391897460715e-06,
      "loss": 0.1914,
      "step": 800
    },
    {
      "epoch": 1.62,
      "grad_norm": 15.681951522827148,
      "learning_rate": 5.18612544082573e-06,
      "loss": 0.1852,
      "step": 810
    },
    {
      "epoch": 1.6400000000000001,
      "grad_norm": 23.784828186035156,
      "learning_rate": 5.069810901695727e-06,
      "loss": 0.3489,
      "step": 820
    },
    {
      "epoch": 1.6600000000000001,
      "grad_norm": 12.440536499023438,
      "learning_rate": 4.953458558730917e-06,
      "loss": 0.2083,
      "step": 830
    },
    {
      "epoch": 1.6800000000000002,
      "grad_norm": 21.0580997467041,
      "learning_rate": 4.837131418777595e-06,
      "loss": 0.277,
      "step": 840
    },
    {
      "epoch": 1.7,
      "grad_norm": 17.277027130126953,
      "learning_rate": 4.720892475034181e-06,
      "loss": 0.2547,
      "step": 850
    },
    {
      "epoch": 1.72,
      "grad_norm": 18.04334831237793,
      "learning_rate": 4.604804672939295e-06,
      "loss": 0.2526,
      "step": 860
    },
    {
      "epoch": 1.74,
      "grad_norm": 23.68144416809082,
      "learning_rate": 4.4889308760856826e-06,
      "loss": 0.3214,
      "step": 870
    },
    {
      "epoch": 1.76,
      "grad_norm": 26.326650619506836,
      "learning_rate": 4.373333832178478e-06,
      "loss": 0.2073,
      "step": 880
    },
    {
      "epoch": 1.78,
      "grad_norm": 9.934289932250977,
      "learning_rate": 4.258076139056217e-06,
      "loss": 0.2154,
      "step": 890
    },
    {
      "epoch": 1.8,
      "grad_norm": 18.508403778076172,
      "learning_rate": 4.143220210792993e-06,
      "loss": 0.1455,
      "step": 900
    },
    {
      "epoch": 1.8199999999999998,
      "grad_norm": 22.585491180419922,
      "learning_rate": 4.028828243900141e-06,
      "loss": 0.2328,
      "step": 910
    },
    {
      "epoch": 1.8399999999999999,
      "grad_norm": 0.5546496510505676,
      "learning_rate": 3.914962183645733e-06,
      "loss": 0.2251,
      "step": 920
    },
    {
      "epoch": 1.8599999999999999,
      "grad_norm": 23.677593231201172,
      "learning_rate": 3.8016836905101157e-06,
      "loss": 0.3064,
      "step": 930
    },
    {
      "epoch": 1.88,
      "grad_norm": 22.756183624267578,
      "learning_rate": 3.6890541067956775e-06,
      "loss": 0.2951,
      "step": 940
    },
    {
      "epoch": 1.9,
      "grad_norm": 13.436799049377441,
      "learning_rate": 3.577134423408906e-06,
      "loss": 0.1715,
      "step": 950
    },
    {
      "epoch": 1.92,
      "grad_norm": 17.541547775268555,
      "learning_rate": 3.465985246832739e-06,
      "loss": 0.1682,
      "step": 960
    },
    {
      "epoch": 1.94,
      "grad_norm": 16.890501022338867,
      "learning_rate": 3.355666766307084e-06,
      "loss": 0.2152,
      "step": 970
    },
    {
      "epoch": 1.96,
      "grad_norm": 62.4870719909668,
      "learning_rate": 3.246238721235283e-06,
      "loss": 0.3343,
      "step": 980
    },
    {
      "epoch": 1.98,
      "grad_norm": 18.51460838317871,
      "learning_rate": 3.137760368834169e-06,
      "loss": 0.2769,
      "step": 990
    },
    {
      "epoch": 2.0,
      "grad_norm": 5.725079536437988,
      "learning_rate": 3.030290452045245e-06,
      "loss": 0.1502,
      "step": 1000
    },
    {
      "epoch": 2.02,
      "grad_norm": 9.278656959533691,
      "learning_rate": 2.9238871677243354e-06,
      "loss": 0.129,
      "step": 1010
    },
    {
      "epoch": 2.04,
      "grad_norm": 9.533599853515625,
      "learning_rate": 2.818608135126967e-06,
      "loss": 0.0928,
      "step": 1020
    },
    {
      "epoch": 2.06,
      "grad_norm": 8.221805572509766,
      "learning_rate": 2.714510364706531e-06,
      "loss": 0.0733,
      "step": 1030
    },
    {
      "epoch": 2.08,
      "grad_norm": 7.671918869018555,
      "learning_rate": 2.611650227242102e-06,
      "loss": 0.0747,
      "step": 1040
    },
    {
      "epoch": 2.1,
      "grad_norm": 5.903024196624756,
      "learning_rate": 2.5100834233126827e-06,
      "loss": 0.0658,
      "step": 1050
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.4098687171936035,
      "learning_rate": 2.40986495313435e-06,
      "loss": 0.0679,
      "step": 1060
    },
    {
      "epoch": 2.14,
      "grad_norm": 23.2823543548584,
      "learning_rate": 2.3110490867766644e-06,
      "loss": 0.0864,
      "step": 1070
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.38023042678833,
      "learning_rate": 2.213689334774479e-06,
      "loss": 0.1165,
      "step": 1080
    },
    {
      "epoch": 2.18,
      "grad_norm": 12.243284225463867,
      "learning_rate": 2.1178384191510344e-06,
      "loss": 0.0852,
      "step": 1090
    },
    {
      "epoch": 2.2,
      "grad_norm": 3.8006153106689453,
      "learning_rate": 2.023548244868051e-06,
      "loss": 0.0525,
      "step": 1100
    },
    {
      "epoch": 2.22,
      "grad_norm": 3.944629430770874,
      "learning_rate": 1.9308698717182874e-06,
      "loss": 0.0554,
      "step": 1110
    },
    {
      "epoch": 2.24,
      "grad_norm": 8.810611724853516,
      "learning_rate": 1.8398534866757455e-06,
      "loss": 0.036,
      "step": 1120
    },
    {
      "epoch": 2.26,
      "grad_norm": 23.007856369018555,
      "learning_rate": 1.7505483767185583e-06,
      "loss": 0.091,
      "step": 1130
    },
    {
      "epoch": 2.2800000000000002,
      "grad_norm": 0.49094370007514954,
      "learning_rate": 1.6630029021392007e-06,
      "loss": 0.0309,
      "step": 1140
    },
    {
      "epoch": 2.3,
      "grad_norm": 6.82590913772583,
      "learning_rate": 1.5772644703565564e-06,
      "loss": 0.1021,
      "step": 1150
    },
    {
      "epoch": 2.32,
      "grad_norm": 6.427361488342285,
      "learning_rate": 1.4933795102439558e-06,
      "loss": 0.09,
      "step": 1160
    },
    {
      "epoch": 2.34,
      "grad_norm": 14.492709159851074,
      "learning_rate": 1.4113934469871166e-06,
      "loss": 0.0501,
      "step": 1170
    },
    {
      "epoch": 2.36,
      "grad_norm": 3.1820945739746094,
      "learning_rate": 1.3313506774856177e-06,
      "loss": 0.048,
      "step": 1180
    },
    {
      "epoch": 2.38,
      "grad_norm": 42.043678283691406,
      "learning_rate": 1.2532945463111856e-06,
      "loss": 0.1067,
      "step": 1190
    },
    {
      "epoch": 2.4,
      "grad_norm": 3.8690338134765625,
      "learning_rate": 1.1772673222358421e-06,
      "loss": 0.0715,
      "step": 1200
    },
    {
      "epoch": 2.42,
      "grad_norm": 13.964327812194824,
      "learning_rate": 1.1033101753426285e-06,
      "loss": 0.0588,
      "step": 1210
    },
    {
      "epoch": 2.44,
      "grad_norm": 13.137357711791992,
      "learning_rate": 1.0314631547312738e-06,
      "loss": 0.0898,
      "step": 1220
    },
    {
      "epoch": 2.46,
      "grad_norm": 22.27714729309082,
      "learning_rate": 9.617651668308914e-07,
      "loss": 0.0671,
      "step": 1230
    },
    {
      "epoch": 2.48,
      "grad_norm": 6.549908638000488,
      "learning_rate": 8.942539543314799e-07,
      "loss": 0.07,
      "step": 1240
    },
    {
      "epoch": 2.5,
      "grad_norm": 8.432822227478027,
      "learning_rate": 8.289660757455803e-07,
      "loss": 0.074,
      "step": 1250
    },
    {
      "epoch": 2.52,
      "grad_norm": 13.46225357055664,
      "learning_rate": 7.659368856111926e-07,
      "loss": 0.0816,
      "step": 1260
    },
    {
      "epoch": 2.54,
      "grad_norm": 19.169546127319336,
      "learning_rate": 7.052005153466779e-07,
      "loss": 0.0587,
      "step": 1270
    },
    {
      "epoch": 2.56,
      "grad_norm": 7.869392395019531,
      "learning_rate": 6.467898547679913e-07,
      "loss": 0.0479,
      "step": 1280
    },
    {
      "epoch": 2.58,
      "grad_norm": 17.20852279663086,
      "learning_rate": 5.9073653427826e-07,
      "loss": 0.0609,
      "step": 1290
    },
    {
      "epoch": 2.6,
      "grad_norm": 24.62643814086914,
      "learning_rate": 5.370709077393721e-07,
      "loss": 0.0915,
      "step": 1300
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.43287232518196106,
      "learning_rate": 4.858220360348187e-07,
      "loss": 0.0647,
      "step": 1310
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.1185322999954224,
      "learning_rate": 4.370176713327118e-07,
      "loss": 0.0321,
      "step": 1320
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.1170597076416016,
      "learning_rate": 3.90684242057498e-07,
      "loss": 0.0403,
      "step": 1330
    },
    {
      "epoch": 2.68,
      "grad_norm": 9.258137702941895,
      "learning_rate": 3.468468385785023e-07,
      "loss": 0.0792,
      "step": 1340
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.038108617067337036,
      "learning_rate": 3.055291996230492e-07,
      "loss": 0.0643,
      "step": 1350
    },
    {
      "epoch": 2.7199999999999998,
      "grad_norm": 17.6850643157959,
      "learning_rate": 2.6675369942151864e-07,
      "loss": 0.0585,
      "step": 1360
    },
    {
      "epoch": 2.74,
      "grad_norm": 17.302579879760742,
      "learning_rate": 2.3054133559131163e-07,
      "loss": 0.0713,
      "step": 1370
    },
    {
      "epoch": 2.76,
      "grad_norm": 9.864686965942383,
      "learning_rate": 1.9691171776626882e-07,
      "loss": 0.1073,
      "step": 1380
    },
    {
      "epoch": 2.7800000000000002,
      "grad_norm": 0.8560196161270142,
      "learning_rate": 1.6588305697770313e-07,
      "loss": 0.0749,
      "step": 1390
    },
    {
      "epoch": 2.8,
      "grad_norm": 16.406612396240234,
      "learning_rate": 1.374721557928116e-07,
      "loss": 0.1064,
      "step": 1400
    },
    {
      "epoch": 2.82,
      "grad_norm": 26.20368194580078,
      "learning_rate": 1.1169439921578485e-07,
      "loss": 0.082,
      "step": 1410
    },
    {
      "epoch": 2.84,
      "grad_norm": 8.685345649719238,
      "learning_rate": 8.856374635655696e-08,
      "loss": 0.0384,
      "step": 1420
    },
    {
      "epoch": 2.86,
      "grad_norm": 11.66822338104248,
      "learning_rate": 6.809272287169988e-08,
      "loss": 0.0821,
      "step": 1430
    },
    {
      "epoch": 2.88,
      "grad_norm": 10.853692054748535,
      "learning_rate": 5.029241418156139e-08,
      "loss": 0.0839,
      "step": 1440
    },
    {
      "epoch": 2.9,
      "grad_norm": 10.97079849243164,
      "learning_rate": 3.517245946731529e-08,
      "loss": 0.0439,
      "step": 1450
    },
    {
      "epoch": 2.92,
      "grad_norm": 5.0863142013549805,
      "learning_rate": 2.27410464511707e-08,
      "loss": 0.0673,
      "step": 1460
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.250962018966675,
      "learning_rate": 1.3004906962578723e-08,
      "loss": 0.0711,
      "step": 1470
    },
    {
      "epoch": 2.96,
      "grad_norm": 22.011507034301758,
      "learning_rate": 5.969313292830126e-09,
      "loss": 0.0375,
      "step": 1480
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.2536417245864868,
      "learning_rate": 1.638075340010814e-09,
      "loss": 0.0433,
      "step": 1490
    },
    {
      "epoch": 3.0,
      "grad_norm": 10.799844741821289,
      "learning_rate": 1.3538545881042198e-11,
      "loss": 0.0557,
      "step": 1500
    },
    {
      "epoch": 3.0,
      "step": 1500,
      "total_flos": 1813196615712768.0,
      "train_loss": 0.3410530858039856,
      "train_runtime": 433.1751,
      "train_samples_per_second": 6.926,
      "train_steps_per_second": 3.463
    }
  ],
  "logging_steps": 10,
  "max_steps": 1500,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": false,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1813196615712768.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}