{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.537318712415989,
  "eval_steps": 1000,
  "global_step": 5000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.017686593562079942,
      "grad_norm": 32.10449981689453,
      "learning_rate": 4.6000000000000004e-07,
      "loss": 2.6364,
      "step": 25
    },
    {
      "epoch": 0.035373187124159884,
      "grad_norm": 20.653064727783203,
      "learning_rate": 9.600000000000001e-07,
      "loss": 1.5119,
      "step": 50
    },
    {
      "epoch": 0.05305978068623983,
      "grad_norm": 20.740116119384766,
      "learning_rate": 1.46e-06,
      "loss": 0.984,
      "step": 75
    },
    {
      "epoch": 0.07074637424831977,
      "grad_norm": 18.87639045715332,
      "learning_rate": 1.9600000000000003e-06,
      "loss": 0.9579,
      "step": 100
    },
    {
      "epoch": 0.08843296781039972,
      "grad_norm": 18.41526985168457,
      "learning_rate": 2.46e-06,
      "loss": 0.8896,
      "step": 125
    },
    {
      "epoch": 0.10611956137247966,
      "grad_norm": 17.025005340576172,
      "learning_rate": 2.96e-06,
      "loss": 0.8909,
      "step": 150
    },
    {
      "epoch": 0.1238061549345596,
      "grad_norm": 17.609508514404297,
      "learning_rate": 3.46e-06,
      "loss": 0.8571,
      "step": 175
    },
    {
      "epoch": 0.14149274849663954,
      "grad_norm": 18.9322452545166,
      "learning_rate": 3.96e-06,
      "loss": 0.906,
      "step": 200
    },
    {
      "epoch": 0.1591793420587195,
      "grad_norm": 20.501718521118164,
      "learning_rate": 4.4600000000000005e-06,
      "loss": 0.8095,
      "step": 225
    },
    {
      "epoch": 0.17686593562079944,
      "grad_norm": 19.373750686645508,
      "learning_rate": 4.960000000000001e-06,
      "loss": 0.8398,
      "step": 250
    },
    {
      "epoch": 0.19455252918287938,
      "grad_norm": 16.65863037109375,
      "learning_rate": 5.460000000000001e-06,
      "loss": 0.8129,
      "step": 275
    },
    {
      "epoch": 0.21223912274495932,
      "grad_norm": 20.23094940185547,
      "learning_rate": 5.9600000000000005e-06,
      "loss": 0.8192,
      "step": 300
    },
    {
      "epoch": 0.22992571630703926,
      "grad_norm": 19.06565284729004,
      "learning_rate": 6.460000000000001e-06,
      "loss": 0.849,
      "step": 325
    },
    {
      "epoch": 0.2476123098691192,
      "grad_norm": 21.71100425720215,
      "learning_rate": 6.96e-06,
      "loss": 0.7948,
      "step": 350
    },
    {
      "epoch": 0.26529890343119916,
      "grad_norm": 17.772262573242188,
      "learning_rate": 7.4600000000000006e-06,
      "loss": 0.7934,
      "step": 375
    },
    {
      "epoch": 0.2829854969932791,
      "grad_norm": 17.064373016357422,
      "learning_rate": 7.960000000000002e-06,
      "loss": 0.7988,
      "step": 400
    },
    {
      "epoch": 0.30067209055535904,
      "grad_norm": 16.245346069335938,
      "learning_rate": 8.46e-06,
      "loss": 0.8256,
      "step": 425
    },
    {
      "epoch": 0.318358684117439,
      "grad_norm": 15.647849082946777,
      "learning_rate": 8.96e-06,
      "loss": 0.7855,
      "step": 450
    },
    {
      "epoch": 0.3360452776795189,
      "grad_norm": 19.271818161010742,
      "learning_rate": 9.460000000000001e-06,
      "loss": 0.7986,
      "step": 475
    },
    {
      "epoch": 0.3537318712415989,
      "grad_norm": 15.884185791015625,
      "learning_rate": 9.960000000000001e-06,
      "loss": 0.8113,
      "step": 500
    },
    {
      "epoch": 0.3714184648036788,
      "grad_norm": 14.456477165222168,
      "learning_rate": 9.94888888888889e-06,
      "loss": 0.7269,
      "step": 525
    },
    {
      "epoch": 0.38910505836575876,
      "grad_norm": 15.575183868408203,
      "learning_rate": 9.893333333333334e-06,
      "loss": 0.8071,
      "step": 550
    },
    {
      "epoch": 0.40679165192783867,
      "grad_norm": 17.374601364135742,
      "learning_rate": 9.837777777777778e-06,
      "loss": 0.7626,
      "step": 575
    },
    {
      "epoch": 0.42447824548991864,
      "grad_norm": 16.48969268798828,
      "learning_rate": 9.782222222222222e-06,
      "loss": 0.7429,
      "step": 600
    },
    {
      "epoch": 0.4421648390519986,
      "grad_norm": 16.197696685791016,
      "learning_rate": 9.72888888888889e-06,
      "loss": 0.7456,
      "step": 625
    },
    {
      "epoch": 0.4598514326140785,
      "grad_norm": 15.233689308166504,
      "learning_rate": 9.673333333333334e-06,
      "loss": 0.7541,
      "step": 650
    },
    {
      "epoch": 0.4775380261761585,
      "grad_norm": 14.872835159301758,
      "learning_rate": 9.617777777777778e-06,
      "loss": 0.7168,
      "step": 675
    },
    {
      "epoch": 0.4952246197382384,
      "grad_norm": 16.02081298828125,
      "learning_rate": 9.562222222222223e-06,
      "loss": 0.7081,
      "step": 700
    },
    {
      "epoch": 0.5129112133003184,
      "grad_norm": 15.401288032531738,
      "learning_rate": 9.506666666666667e-06,
      "loss": 0.7298,
      "step": 725
    },
    {
      "epoch": 0.5305978068623983,
      "grad_norm": 15.33830738067627,
      "learning_rate": 9.451111111111112e-06,
      "loss": 0.6965,
      "step": 750
    },
    {
      "epoch": 0.5482844004244782,
      "grad_norm": 11.412097930908203,
      "learning_rate": 9.395555555555556e-06,
      "loss": 0.7183,
      "step": 775
    },
    {
      "epoch": 0.5659709939865581,
      "grad_norm": 13.304567337036133,
      "learning_rate": 9.340000000000002e-06,
      "loss": 0.6847,
      "step": 800
    },
    {
      "epoch": 0.5836575875486382,
      "grad_norm": 11.349915504455566,
      "learning_rate": 9.284444444444444e-06,
      "loss": 0.6513,
      "step": 825
    },
    {
      "epoch": 0.6013441811107181,
      "grad_norm": 14.461771011352539,
      "learning_rate": 9.22888888888889e-06,
      "loss": 0.6613,
      "step": 850
    },
    {
      "epoch": 0.619030774672798,
      "grad_norm": 24.84481430053711,
      "learning_rate": 9.173333333333334e-06,
      "loss": 0.6935,
      "step": 875
    },
    {
      "epoch": 0.636717368234878,
      "grad_norm": 14.775162696838379,
      "learning_rate": 9.117777777777778e-06,
      "loss": 0.6126,
      "step": 900
    },
    {
      "epoch": 0.6544039617969579,
      "grad_norm": 13.53603744506836,
      "learning_rate": 9.062222222222224e-06,
      "loss": 0.6542,
      "step": 925
    },
    {
      "epoch": 0.6720905553590378,
      "grad_norm": 12.807902336120605,
      "learning_rate": 9.006666666666666e-06,
      "loss": 0.6512,
      "step": 950
    },
    {
      "epoch": 0.6897771489211177,
      "grad_norm": 13.615946769714355,
      "learning_rate": 8.951111111111112e-06,
      "loss": 0.6428,
      "step": 975
    },
    {
      "epoch": 0.7074637424831978,
      "grad_norm": 16.28773307800293,
      "learning_rate": 8.895555555555556e-06,
      "loss": 0.6485,
      "step": 1000
    },
    {
      "epoch": 0.7074637424831978,
      "eval_loss": 0.3581026792526245,
      "eval_runtime": 4019.5457,
      "eval_samples_per_second": 2.678,
      "eval_steps_per_second": 0.167,
      "eval_wer": 0.22099663615739654,
      "step": 1000
    },
    {
      "epoch": 0.7251503360452777,
      "grad_norm": 12.776633262634277,
      "learning_rate": 8.84e-06,
      "loss": 0.6253,
      "step": 1025
    },
    {
      "epoch": 0.7428369296073576,
      "grad_norm": 13.185070991516113,
      "learning_rate": 8.784444444444446e-06,
      "loss": 0.599,
      "step": 1050
    },
    {
      "epoch": 0.7605235231694376,
      "grad_norm": 11.916515350341797,
      "learning_rate": 8.72888888888889e-06,
      "loss": 0.6254,
      "step": 1075
    },
    {
      "epoch": 0.7782101167315175,
      "grad_norm": 15.015724182128906,
      "learning_rate": 8.673333333333334e-06,
      "loss": 0.6457,
      "step": 1100
    },
    {
      "epoch": 0.7958967102935974,
      "grad_norm": 13.483101844787598,
      "learning_rate": 8.617777777777778e-06,
      "loss": 0.6214,
      "step": 1125
    },
    {
      "epoch": 0.8135833038556773,
      "grad_norm": 10.465691566467285,
      "learning_rate": 8.562222222222224e-06,
      "loss": 0.5961,
      "step": 1150
    },
    {
      "epoch": 0.8312698974177574,
      "grad_norm": 13.515427589416504,
      "learning_rate": 8.506666666666668e-06,
      "loss": 0.6166,
      "step": 1175
    },
    {
      "epoch": 0.8489564909798373,
      "grad_norm": 11.983545303344727,
      "learning_rate": 8.451111111111112e-06,
      "loss": 0.6157,
      "step": 1200
    },
    {
      "epoch": 0.8666430845419172,
      "grad_norm": 12.49000072479248,
      "learning_rate": 8.395555555555557e-06,
      "loss": 0.5511,
      "step": 1225
    },
    {
      "epoch": 0.8843296781039972,
      "grad_norm": 14.379878997802734,
      "learning_rate": 8.34e-06,
      "loss": 0.6323,
      "step": 1250
    },
    {
      "epoch": 0.9020162716660771,
      "grad_norm": 13.675074577331543,
      "learning_rate": 8.284444444444446e-06,
      "loss": 0.5503,
      "step": 1275
    },
    {
      "epoch": 0.919702865228157,
      "grad_norm": 14.126933097839355,
      "learning_rate": 8.22888888888889e-06,
      "loss": 0.5654,
      "step": 1300
    },
    {
      "epoch": 0.937389458790237,
      "grad_norm": 12.32470417022705,
      "learning_rate": 8.173333333333334e-06,
      "loss": 0.5526,
      "step": 1325
    },
    {
      "epoch": 0.955076052352317,
      "grad_norm": 15.61474323272705,
      "learning_rate": 8.11777777777778e-06,
      "loss": 0.5485,
      "step": 1350
    },
    {
      "epoch": 0.9727626459143969,
      "grad_norm": 11.217473983764648,
      "learning_rate": 8.062222222222222e-06,
      "loss": 0.6228,
      "step": 1375
    },
    {
      "epoch": 0.9904492394764768,
      "grad_norm": 13.000173568725586,
      "learning_rate": 8.006666666666667e-06,
      "loss": 0.5661,
      "step": 1400
    },
    {
      "epoch": 1.0081358330385568,
      "grad_norm": 10.341316223144531,
      "learning_rate": 7.951111111111111e-06,
      "loss": 0.474,
      "step": 1425
    },
    {
      "epoch": 1.0258224266006368,
      "grad_norm": 9.769425392150879,
      "learning_rate": 7.895555555555557e-06,
      "loss": 0.418,
      "step": 1450
    },
    {
      "epoch": 1.0435090201627166,
      "grad_norm": 10.41649341583252,
      "learning_rate": 7.840000000000001e-06,
      "loss": 0.3768,
      "step": 1475
    },
    {
      "epoch": 1.0611956137247966,
      "grad_norm": 7.566088676452637,
      "learning_rate": 7.784444444444445e-06,
      "loss": 0.388,
      "step": 1500
    },
    {
      "epoch": 1.0788822072868764,
      "grad_norm": 11.578902244567871,
      "learning_rate": 7.72888888888889e-06,
      "loss": 0.3349,
      "step": 1525
    },
    {
      "epoch": 1.0965688008489565,
      "grad_norm": 10.13636302947998,
      "learning_rate": 7.673333333333333e-06,
      "loss": 0.3656,
      "step": 1550
    },
    {
      "epoch": 1.1142553944110365,
      "grad_norm": 10.895235061645508,
      "learning_rate": 7.617777777777778e-06,
      "loss": 0.3707,
      "step": 1575
    },
    {
      "epoch": 1.1319419879731163,
      "grad_norm": 10.668679237365723,
      "learning_rate": 7.562222222222223e-06,
      "loss": 0.3588,
      "step": 1600
    },
    {
      "epoch": 1.1496285815351963,
      "grad_norm": 10.62954044342041,
      "learning_rate": 7.506666666666668e-06,
      "loss": 0.3737,
      "step": 1625
    },
    {
      "epoch": 1.1673151750972763,
      "grad_norm": 10.885923385620117,
      "learning_rate": 7.451111111111111e-06,
      "loss": 0.3484,
      "step": 1650
    },
    {
      "epoch": 1.1850017686593561,
      "grad_norm": 12.320752143859863,
      "learning_rate": 7.395555555555556e-06,
      "loss": 0.3775,
      "step": 1675
    },
    {
      "epoch": 1.2026883622214362,
      "grad_norm": 10.216126441955566,
      "learning_rate": 7.340000000000001e-06,
      "loss": 0.3369,
      "step": 1700
    },
    {
      "epoch": 1.2203749557835162,
      "grad_norm": 12.078413009643555,
      "learning_rate": 7.284444444444445e-06,
      "loss": 0.3669,
      "step": 1725
    },
    {
      "epoch": 1.238061549345596,
      "grad_norm": 9.149145126342773,
      "learning_rate": 7.22888888888889e-06,
      "loss": 0.3453,
      "step": 1750
    },
    {
      "epoch": 1.255748142907676,
      "grad_norm": 16.312984466552734,
      "learning_rate": 7.173333333333335e-06,
      "loss": 0.3405,
      "step": 1775
    },
    {
      "epoch": 1.2734347364697558,
      "grad_norm": 11.405570983886719,
      "learning_rate": 7.117777777777778e-06,
      "loss": 0.3665,
      "step": 1800
    },
    {
      "epoch": 1.2911213300318358,
      "grad_norm": 11.892732620239258,
      "learning_rate": 7.062222222222223e-06,
      "loss": 0.3561,
      "step": 1825
    },
    {
      "epoch": 1.3088079235939158,
      "grad_norm": 8.842497825622559,
      "learning_rate": 7.006666666666667e-06,
      "loss": 0.3218,
      "step": 1850
    },
    {
      "epoch": 1.3264945171559956,
      "grad_norm": 14.237404823303223,
      "learning_rate": 6.951111111111112e-06,
      "loss": 0.3334,
      "step": 1875
    },
    {
      "epoch": 1.3441811107180757,
      "grad_norm": 11.255355834960938,
      "learning_rate": 6.8955555555555565e-06,
      "loss": 0.3293,
      "step": 1900
    },
    {
      "epoch": 1.3618677042801557,
      "grad_norm": 10.295186042785645,
      "learning_rate": 6.8400000000000014e-06,
      "loss": 0.3085,
      "step": 1925
    },
    {
      "epoch": 1.3795542978422355,
      "grad_norm": 8.852261543273926,
      "learning_rate": 6.784444444444445e-06,
      "loss": 0.3423,
      "step": 1950
    },
    {
      "epoch": 1.3972408914043155,
      "grad_norm": 8.795380592346191,
      "learning_rate": 6.7288888888888895e-06,
      "loss": 0.3436,
      "step": 1975
    },
    {
      "epoch": 1.4149274849663955,
      "grad_norm": 8.382344245910645,
      "learning_rate": 6.6733333333333335e-06,
      "loss": 0.3362,
      "step": 2000
    },
    {
      "epoch": 1.4149274849663955,
      "eval_loss": 0.309411883354187,
      "eval_runtime": 4009.7652,
      "eval_samples_per_second": 2.684,
      "eval_steps_per_second": 0.168,
      "eval_wer": 0.1831142924196198,
      "step": 2000
    },
    {
      "epoch": 1.4326140785284753,
      "grad_norm": 9.291923522949219,
      "learning_rate": 6.617777777777778e-06,
      "loss": 0.3433,
      "step": 2025
    },
    {
      "epoch": 1.4503006720905554,
      "grad_norm": 8.35601806640625,
      "learning_rate": 6.562222222222223e-06,
      "loss": 0.3344,
      "step": 2050
    },
    {
      "epoch": 1.4679872656526354,
      "grad_norm": 8.72737979888916,
      "learning_rate": 6.5066666666666665e-06,
      "loss": 0.3333,
      "step": 2075
    },
    {
      "epoch": 1.4856738592147152,
      "grad_norm": 8.06108283996582,
      "learning_rate": 6.451111111111111e-06,
      "loss": 0.3291,
      "step": 2100
    },
    {
      "epoch": 1.5033604527767952,
      "grad_norm": 9.253275871276855,
      "learning_rate": 6.395555555555556e-06,
      "loss": 0.3341,
      "step": 2125
    },
    {
      "epoch": 1.5210470463388752,
      "grad_norm": 10.817941665649414,
      "learning_rate": 6.34e-06,
      "loss": 0.3549,
      "step": 2150
    },
    {
      "epoch": 1.538733639900955,
      "grad_norm": 10.792004585266113,
      "learning_rate": 6.284444444444445e-06,
      "loss": 0.3353,
      "step": 2175
    },
    {
      "epoch": 1.556420233463035,
      "grad_norm": 7.487943649291992,
      "learning_rate": 6.22888888888889e-06,
      "loss": 0.326,
      "step": 2200
    },
    {
      "epoch": 1.574106827025115,
      "grad_norm": 9.3944673538208,
      "learning_rate": 6.173333333333333e-06,
      "loss": 0.3214,
      "step": 2225
    },
    {
      "epoch": 1.5917934205871949,
      "grad_norm": 10.089088439941406,
      "learning_rate": 6.117777777777778e-06,
      "loss": 0.3066,
      "step": 2250
    },
    {
      "epoch": 1.6094800141492749,
      "grad_norm": 9.607513427734375,
      "learning_rate": 6.062222222222223e-06,
      "loss": 0.3492,
      "step": 2275
    },
    {
      "epoch": 1.627166607711355,
      "grad_norm": 9.252910614013672,
      "learning_rate": 6.006666666666667e-06,
      "loss": 0.3137,
      "step": 2300
    },
    {
      "epoch": 1.6448532012734347,
      "grad_norm": 11.22940444946289,
      "learning_rate": 5.951111111111112e-06,
      "loss": 0.3408,
      "step": 2325
    },
    {
      "epoch": 1.6625397948355147,
      "grad_norm": 9.891559600830078,
      "learning_rate": 5.895555555555557e-06,
      "loss": 0.306,
      "step": 2350
    },
    {
      "epoch": 1.6802263883975948,
      "grad_norm": 8.754281044006348,
      "learning_rate": 5.84e-06,
      "loss": 0.3356,
      "step": 2375
    },
    {
      "epoch": 1.6979129819596745,
      "grad_norm": 10.18521499633789,
      "learning_rate": 5.784444444444445e-06,
      "loss": 0.3167,
      "step": 2400
    },
    {
      "epoch": 1.7155995755217543,
      "grad_norm": 7.963505744934082,
      "learning_rate": 5.72888888888889e-06,
      "loss": 0.3229,
      "step": 2425
    },
    {
      "epoch": 1.7332861690838346,
      "grad_norm": 9.744851112365723,
      "learning_rate": 5.673333333333334e-06,
      "loss": 0.2961,
      "step": 2450
    },
    {
      "epoch": 1.7509727626459144,
      "grad_norm": 8.96956729888916,
      "learning_rate": 5.617777777777779e-06,
      "loss": 0.3399,
      "step": 2475
    },
    {
      "epoch": 1.7686593562079942,
      "grad_norm": 10.65976619720459,
      "learning_rate": 5.562222222222222e-06,
      "loss": 0.3205,
      "step": 2500
    },
    {
      "epoch": 1.7863459497700744,
      "grad_norm": 10.630120277404785,
      "learning_rate": 5.506666666666667e-06,
      "loss": 0.3188,
      "step": 2525
    },
    {
      "epoch": 1.8040325433321542,
      "grad_norm": 9.28016471862793,
      "learning_rate": 5.451111111111112e-06,
      "loss": 0.3033,
      "step": 2550
    },
    {
      "epoch": 1.821719136894234,
      "grad_norm": 9.342653274536133,
      "learning_rate": 5.3955555555555565e-06,
      "loss": 0.3299,
      "step": 2575
    },
    {
      "epoch": 1.839405730456314,
      "grad_norm": 9.810345649719238,
      "learning_rate": 5.3400000000000005e-06,
      "loss": 0.3141,
      "step": 2600
    },
    {
      "epoch": 1.857092324018394,
      "grad_norm": 9.378629684448242,
      "learning_rate": 5.2844444444444454e-06,
      "loss": 0.2861,
      "step": 2625
    },
    {
      "epoch": 1.8747789175804739,
      "grad_norm": 9.487212181091309,
      "learning_rate": 5.228888888888889e-06,
      "loss": 0.3064,
      "step": 2650
    },
    {
      "epoch": 1.892465511142554,
      "grad_norm": 10.776664733886719,
      "learning_rate": 5.1733333333333335e-06,
      "loss": 0.3091,
      "step": 2675
    },
    {
      "epoch": 1.910152104704634,
      "grad_norm": 9.766382217407227,
      "learning_rate": 5.117777777777778e-06,
      "loss": 0.2854,
      "step": 2700
    },
    {
      "epoch": 1.9278386982667137,
      "grad_norm": 10.020450592041016,
      "learning_rate": 5.062222222222222e-06,
      "loss": 0.3398,
      "step": 2725
    },
    {
      "epoch": 1.9455252918287937,
      "grad_norm": 10.089125633239746,
      "learning_rate": 5.006666666666667e-06,
      "loss": 0.3081,
      "step": 2750
    },
    {
      "epoch": 1.9632118853908738,
      "grad_norm": 8.752697944641113,
      "learning_rate": 4.951111111111111e-06,
      "loss": 0.3033,
      "step": 2775
    },
    {
      "epoch": 1.9808984789529536,
      "grad_norm": 9.102664947509766,
      "learning_rate": 4.895555555555556e-06,
      "loss": 0.3018,
      "step": 2800
    },
    {
      "epoch": 1.9985850725150336,
      "grad_norm": 8.133709907531738,
      "learning_rate": 4.84e-06,
      "loss": 0.3068,
      "step": 2825
    },
    {
      "epoch": 2.0162716660771136,
      "grad_norm": 7.3281779289245605,
      "learning_rate": 4.784444444444445e-06,
      "loss": 0.1808,
      "step": 2850
    },
    {
      "epoch": 2.0339582596391934,
      "grad_norm": 8.63786792755127,
      "learning_rate": 4.728888888888889e-06,
      "loss": 0.1844,
      "step": 2875
    },
    {
      "epoch": 2.0516448532012737,
      "grad_norm": 6.717278957366943,
      "learning_rate": 4.673333333333333e-06,
      "loss": 0.1663,
      "step": 2900
    },
    {
      "epoch": 2.0693314467633535,
      "grad_norm": 6.7627739906311035,
      "learning_rate": 4.617777777777778e-06,
      "loss": 0.1543,
      "step": 2925
    },
    {
      "epoch": 2.0870180403254333,
      "grad_norm": 8.074238777160645,
      "learning_rate": 4.562222222222222e-06,
      "loss": 0.1515,
      "step": 2950
    },
    {
      "epoch": 2.104704633887513,
      "grad_norm": 6.387521266937256,
      "learning_rate": 4.506666666666667e-06,
      "loss": 0.1572,
      "step": 2975
    },
    {
      "epoch": 2.1223912274495933,
      "grad_norm": 6.854986190795898,
      "learning_rate": 4.451111111111112e-06,
      "loss": 0.1504,
      "step": 3000
    },
    {
      "epoch": 2.1223912274495933,
      "eval_loss": 0.2956615090370178,
      "eval_runtime": 4074.1591,
      "eval_samples_per_second": 2.642,
      "eval_steps_per_second": 0.165,
      "eval_wer": 0.16988383008683408,
      "step": 3000
    },
    {
      "epoch": 2.140077821011673,
      "grad_norm": 7.486941814422607,
      "learning_rate": 4.395555555555556e-06,
      "loss": 0.139,
      "step": 3025
    },
    {
      "epoch": 2.157764414573753,
      "grad_norm": 10.396411895751953,
      "learning_rate": 4.34e-06,
      "loss": 0.166,
      "step": 3050
    },
    {
      "epoch": 2.175451008135833,
      "grad_norm": 7.398874282836914,
      "learning_rate": 4.284444444444445e-06,
      "loss": 0.1654,
      "step": 3075
    },
    {
      "epoch": 2.193137601697913,
      "grad_norm": 6.542088985443115,
      "learning_rate": 4.228888888888889e-06,
      "loss": 0.1559,
      "step": 3100
    },
    {
      "epoch": 2.2108241952599927,
      "grad_norm": 6.458364009857178,
      "learning_rate": 4.173333333333334e-06,
      "loss": 0.1493,
      "step": 3125
    },
    {
      "epoch": 2.228510788822073,
      "grad_norm": 7.253146648406982,
      "learning_rate": 4.117777777777779e-06,
      "loss": 0.1488,
      "step": 3150
    },
    {
      "epoch": 2.246197382384153,
      "grad_norm": 7.556519508361816,
      "learning_rate": 4.062222222222223e-06,
      "loss": 0.136,
      "step": 3175
    },
    {
      "epoch": 2.2638839759462326,
      "grad_norm": 6.949014186859131,
      "learning_rate": 4.006666666666667e-06,
      "loss": 0.1626,
      "step": 3200
    },
    {
      "epoch": 2.281570569508313,
      "grad_norm": 7.269302845001221,
      "learning_rate": 3.951111111111112e-06,
      "loss": 0.1635,
      "step": 3225
    },
    {
      "epoch": 2.2992571630703926,
      "grad_norm": 9.310769081115723,
      "learning_rate": 3.895555555555556e-06,
      "loss": 0.1593,
      "step": 3250
    },
    {
      "epoch": 2.3169437566324724,
      "grad_norm": 6.7611894607543945,
      "learning_rate": 3.8400000000000005e-06,
      "loss": 0.1331,
      "step": 3275
    },
    {
      "epoch": 2.3346303501945527,
      "grad_norm": 8.675508499145508,
      "learning_rate": 3.784444444444445e-06,
      "loss": 0.1607,
      "step": 3300
    },
    {
      "epoch": 2.3523169437566325,
      "grad_norm": 8.40794563293457,
      "learning_rate": 3.728888888888889e-06,
      "loss": 0.1566,
      "step": 3325
    },
    {
      "epoch": 2.3700035373187123,
      "grad_norm": 8.73206615447998,
      "learning_rate": 3.673333333333334e-06,
      "loss": 0.1499,
      "step": 3350
    },
    {
      "epoch": 2.3876901308807925,
      "grad_norm": 7.350528717041016,
      "learning_rate": 3.617777777777778e-06,
      "loss": 0.1645,
      "step": 3375
    },
    {
      "epoch": 2.4053767244428723,
      "grad_norm": 6.264787197113037,
      "learning_rate": 3.5622222222222224e-06,
      "loss": 0.1593,
      "step": 3400
    },
    {
      "epoch": 2.423063318004952,
      "grad_norm": 7.77794885635376,
      "learning_rate": 3.5066666666666673e-06,
      "loss": 0.1499,
      "step": 3425
    },
    {
      "epoch": 2.4407499115670324,
      "grad_norm": 6.835781574249268,
      "learning_rate": 3.4511111111111113e-06,
      "loss": 0.1475,
      "step": 3450
    },
    {
      "epoch": 2.458436505129112,
      "grad_norm": 7.863101482391357,
      "learning_rate": 3.3955555555555558e-06,
      "loss": 0.1495,
      "step": 3475
    },
    {
      "epoch": 2.476123098691192,
      "grad_norm": 5.20499849319458,
      "learning_rate": 3.3400000000000006e-06,
      "loss": 0.1486,
      "step": 3500
    },
    {
      "epoch": 2.493809692253272,
      "grad_norm": 7.780925750732422,
      "learning_rate": 3.2844444444444447e-06,
      "loss": 0.1324,
      "step": 3525
    },
    {
      "epoch": 2.511496285815352,
      "grad_norm": 8.029056549072266,
      "learning_rate": 3.228888888888889e-06,
      "loss": 0.1654,
      "step": 3550
    },
    {
      "epoch": 2.529182879377432,
      "grad_norm": 8.651548385620117,
      "learning_rate": 3.173333333333334e-06,
      "loss": 0.1409,
      "step": 3575
    },
    {
      "epoch": 2.5468694729395116,
      "grad_norm": 7.2054548263549805,
      "learning_rate": 3.117777777777778e-06,
      "loss": 0.1706,
      "step": 3600
    },
    {
      "epoch": 2.564556066501592,
      "grad_norm": 9.639912605285645,
      "learning_rate": 3.0622222222222225e-06,
      "loss": 0.1446,
      "step": 3625
    },
    {
      "epoch": 2.5822426600636716,
      "grad_norm": 8.069547653198242,
      "learning_rate": 3.0066666666666674e-06,
      "loss": 0.1492,
      "step": 3650
    },
    {
      "epoch": 2.599929253625752,
      "grad_norm": 7.179859638214111,
      "learning_rate": 2.9511111111111114e-06,
      "loss": 0.1401,
      "step": 3675
    },
    {
      "epoch": 2.6176158471878317,
      "grad_norm": 6.412818908691406,
      "learning_rate": 2.895555555555556e-06,
      "loss": 0.1364,
      "step": 3700
    },
    {
      "epoch": 2.6353024407499115,
      "grad_norm": 6.2440571784973145,
      "learning_rate": 2.84e-06,
      "loss": 0.143,
      "step": 3725
    },
    {
      "epoch": 2.6529890343119913,
      "grad_norm": 3.8863301277160645,
      "learning_rate": 2.784444444444445e-06,
      "loss": 0.1416,
      "step": 3750
    },
    {
      "epoch": 2.6706756278740715,
      "grad_norm": 7.729201793670654,
      "learning_rate": 2.7288888888888893e-06,
      "loss": 0.1407,
      "step": 3775
    },
    {
      "epoch": 2.6883622214361513,
      "grad_norm": 6.973865032196045,
      "learning_rate": 2.6733333333333333e-06,
      "loss": 0.1555,
      "step": 3800
    },
    {
      "epoch": 2.7060488149982316,
      "grad_norm": 7.154475212097168,
      "learning_rate": 2.617777777777778e-06,
      "loss": 0.1541,
      "step": 3825
    },
    {
      "epoch": 2.7237354085603114,
      "grad_norm": 8.38300609588623,
      "learning_rate": 2.5622222222222226e-06,
      "loss": 0.1345,
      "step": 3850
    },
    {
      "epoch": 2.741422002122391,
      "grad_norm": 7.105490684509277,
      "learning_rate": 2.5066666666666667e-06,
      "loss": 0.136,
      "step": 3875
    },
    {
      "epoch": 2.759108595684471,
      "grad_norm": 6.707284927368164,
      "learning_rate": 2.451111111111111e-06,
      "loss": 0.1454,
      "step": 3900
    },
    {
      "epoch": 2.776795189246551,
      "grad_norm": 4.504308223724365,
      "learning_rate": 2.3955555555555556e-06,
      "loss": 0.1383,
      "step": 3925
    },
    {
      "epoch": 2.794481782808631,
      "grad_norm": 8.98666763305664,
      "learning_rate": 2.3400000000000005e-06,
      "loss": 0.1247,
      "step": 3950
    },
    {
      "epoch": 2.8121683763707113,
      "grad_norm": 7.637745380401611,
      "learning_rate": 2.2844444444444445e-06,
      "loss": 0.1503,
      "step": 3975
    },
    {
      "epoch": 2.829854969932791,
      "grad_norm": 8.50109577178955,
      "learning_rate": 2.228888888888889e-06,
      "loss": 0.1558,
      "step": 4000
    },
    {
      "epoch": 2.829854969932791,
      "eval_loss": 0.2815879285335541,
      "eval_runtime": 3977.9422,
      "eval_samples_per_second": 2.706,
      "eval_steps_per_second": 0.169,
      "eval_wer": 0.16456426503950558,
      "step": 4000
    },
    {
      "epoch": 2.847541563494871,
      "grad_norm": 6.490965366363525,
      "learning_rate": 2.1733333333333334e-06,
      "loss": 0.146,
      "step": 4025
    },
    {
      "epoch": 2.8652281570569507,
      "grad_norm": 6.164106845855713,
      "learning_rate": 2.117777777777778e-06,
      "loss": 0.1306,
      "step": 4050
    },
    {
      "epoch": 2.882914750619031,
      "grad_norm": 6.784088611602783,
      "learning_rate": 2.0622222222222223e-06,
      "loss": 0.1362,
      "step": 4075
    },
    {
      "epoch": 2.9006013441811107,
      "grad_norm": 7.150921821594238,
      "learning_rate": 2.006666666666667e-06,
      "loss": 0.1483,
      "step": 4100
    },
    {
      "epoch": 2.9182879377431905,
      "grad_norm": 6.3694329261779785,
      "learning_rate": 1.9511111111111113e-06,
      "loss": 0.1545,
      "step": 4125
    },
    {
      "epoch": 2.9359745313052708,
      "grad_norm": 5.38779878616333,
      "learning_rate": 1.8955555555555557e-06,
      "loss": 0.1395,
      "step": 4150
    },
    {
      "epoch": 2.9536611248673506,
      "grad_norm": 7.62282657623291,
      "learning_rate": 1.8400000000000002e-06,
      "loss": 0.1436,
      "step": 4175
    },
    {
      "epoch": 2.9713477184294304,
      "grad_norm": 9.915560722351074,
      "learning_rate": 1.7844444444444444e-06,
      "loss": 0.1332,
      "step": 4200
    },
    {
      "epoch": 2.9890343119915106,
      "grad_norm": 6.760516166687012,
      "learning_rate": 1.728888888888889e-06,
      "loss": 0.1258,
      "step": 4225
    },
    {
      "epoch": 3.0067209055535904,
      "grad_norm": 3.125908136367798,
      "learning_rate": 1.6733333333333335e-06,
      "loss": 0.1119,
      "step": 4250
    },
    {
      "epoch": 3.02440749911567,
      "grad_norm": 4.832523345947266,
      "learning_rate": 1.6177777777777778e-06,
      "loss": 0.0726,
      "step": 4275
    },
    {
      "epoch": 3.0420940926777504,
      "grad_norm": 3.6243960857391357,
      "learning_rate": 1.5622222222222225e-06,
      "loss": 0.0672,
      "step": 4300
    },
    {
      "epoch": 3.0597806862398302,
      "grad_norm": 4.241108417510986,
      "learning_rate": 1.506666666666667e-06,
      "loss": 0.0654,
      "step": 4325
    },
    {
      "epoch": 3.07746727980191,
      "grad_norm": 4.0127644538879395,
      "learning_rate": 1.4511111111111112e-06,
      "loss": 0.0801,
      "step": 4350
    },
    {
      "epoch": 3.0951538733639903,
      "grad_norm": 4.87974739074707,
      "learning_rate": 1.3955555555555556e-06,
      "loss": 0.0599,
      "step": 4375
    },
    {
      "epoch": 3.11284046692607,
      "grad_norm": 5.309682846069336,
      "learning_rate": 1.34e-06,
      "loss": 0.0636,
      "step": 4400
    },
    {
      "epoch": 3.13052706048815,
      "grad_norm": 2.6454548835754395,
      "learning_rate": 1.2844444444444445e-06,
      "loss": 0.0539,
      "step": 4425
    },
    {
      "epoch": 3.14821365405023,
      "grad_norm": 3.234809637069702,
      "learning_rate": 1.228888888888889e-06,
      "loss": 0.0531,
      "step": 4450
    },
    {
      "epoch": 3.16590024761231,
      "grad_norm": 5.55891227722168,
      "learning_rate": 1.1733333333333335e-06,
      "loss": 0.0805,
      "step": 4475
    },
    {
      "epoch": 3.1835868411743897,
      "grad_norm": 5.280560493469238,
      "learning_rate": 1.117777777777778e-06,
      "loss": 0.0658,
      "step": 4500
    },
    {
      "epoch": 3.2012734347364695,
      "grad_norm": 4.731041431427002,
      "learning_rate": 1.0622222222222222e-06,
      "loss": 0.0653,
      "step": 4525
    },
    {
      "epoch": 3.2189600282985498,
      "grad_norm": 3.2889113426208496,
      "learning_rate": 1.0066666666666668e-06,
      "loss": 0.0628,
      "step": 4550
    },
    {
      "epoch": 3.2366466218606296,
      "grad_norm": 3.6610448360443115,
      "learning_rate": 9.511111111111111e-07,
      "loss": 0.0739,
      "step": 4575
    },
    {
      "epoch": 3.25433321542271,
      "grad_norm": 4.585997104644775,
      "learning_rate": 8.955555555555557e-07,
      "loss": 0.0733,
      "step": 4600
    },
    {
      "epoch": 3.2720198089847896,
      "grad_norm": 4.393402099609375,
      "learning_rate": 8.400000000000001e-07,
      "loss": 0.0589,
      "step": 4625
    },
    {
      "epoch": 3.2897064025468694,
      "grad_norm": 6.0053205490112305,
      "learning_rate": 7.844444444444445e-07,
      "loss": 0.0669,
      "step": 4650
    },
    {
      "epoch": 3.307392996108949,
      "grad_norm": 6.425204277038574,
      "learning_rate": 7.28888888888889e-07,
      "loss": 0.0633,
      "step": 4675
    },
    {
      "epoch": 3.3250795896710295,
      "grad_norm": 3.262568473815918,
      "learning_rate": 6.733333333333334e-07,
      "loss": 0.0601,
      "step": 4700
    },
    {
      "epoch": 3.3427661832331093,
      "grad_norm": 4.711056709289551,
      "learning_rate": 6.177777777777778e-07,
      "loss": 0.0694,
      "step": 4725
    },
    {
      "epoch": 3.360452776795189,
      "grad_norm": 3.0668253898620605,
      "learning_rate": 5.622222222222223e-07,
      "loss": 0.0453,
      "step": 4750
    },
    {
      "epoch": 3.3781393703572693,
      "grad_norm": 2.9075567722320557,
      "learning_rate": 5.066666666666667e-07,
      "loss": 0.0594,
      "step": 4775
    },
    {
      "epoch": 3.395825963919349,
      "grad_norm": 3.7801194190979004,
      "learning_rate": 4.511111111111111e-07,
      "loss": 0.0718,
      "step": 4800
    },
    {
      "epoch": 3.413512557481429,
      "grad_norm": 5.163565158843994,
      "learning_rate": 3.9555555555555557e-07,
      "loss": 0.058,
      "step": 4825
    },
    {
      "epoch": 3.431199151043509,
      "grad_norm": 4.530272960662842,
      "learning_rate": 3.4000000000000003e-07,
      "loss": 0.0533,
      "step": 4850
    },
    {
      "epoch": 3.448885744605589,
      "grad_norm": 3.2570483684539795,
      "learning_rate": 2.844444444444445e-07,
      "loss": 0.0603,
      "step": 4875
    },
    {
      "epoch": 3.4665723381676687,
      "grad_norm": 2.4571080207824707,
      "learning_rate": 2.2888888888888892e-07,
      "loss": 0.0597,
      "step": 4900
    },
    {
      "epoch": 3.484258931729749,
      "grad_norm": 3.9720897674560547,
      "learning_rate": 1.7333333333333335e-07,
      "loss": 0.0589,
      "step": 4925
    },
    {
      "epoch": 3.501945525291829,
      "grad_norm": 4.322543621063232,
      "learning_rate": 1.1777777777777778e-07,
      "loss": 0.0758,
      "step": 4950
    },
    {
      "epoch": 3.5196321188539086,
      "grad_norm": 6.084353446960449,
      "learning_rate": 6.222222222222223e-08,
      "loss": 0.0586,
      "step": 4975
    },
    {
      "epoch": 3.537318712415989,
      "grad_norm": 2.7334227561950684,
      "learning_rate": 6.666666666666667e-09,
      "loss": 0.0619,
      "step": 5000
    },
    {
      "epoch": 3.537318712415989,
      "eval_loss": 0.29267358779907227,
      "eval_runtime": 3980.6566,
      "eval_samples_per_second": 2.704,
      "eval_steps_per_second": 0.169,
      "eval_wer": 0.15772901509817727,
      "step": 5000
    },
    {
      "epoch": 3.537318712415989,
      "step": 5000,
      "total_flos": 2.7274615074127872e+20,
      "train_loss": 0.36250473141670225,
      "train_runtime": 65548.1959,
      "train_samples_per_second": 2.441,
      "train_steps_per_second": 0.076
    }
  ],
  "logging_steps": 25,
  "max_steps": 5000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.7274615074127872e+20,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}