{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.4851953222194576,
  "global_step": 3900,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 2.6399999999999995e-05,
      "loss": 12.4264,
      "step": 50
    },
    {
      "epoch": 0.01,
      "learning_rate": 5.6399999999999995e-05,
      "loss": 7.1545,
      "step": 100
    },
    {
      "epoch": 0.01,
      "eval_cer": 0.990295846526883,
      "eval_loss": 4.435502529144287,
      "eval_runtime": 2212.4333,
      "eval_samples_per_second": 21.798,
      "eval_steps_per_second": 1.363,
      "eval_wer": 0.9803344576185385,
      "step": 100
    },
    {
      "epoch": 0.02,
      "learning_rate": 8.639999999999999e-05,
      "loss": 3.2621,
      "step": 150
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001164,
      "loss": 2.9431,
      "step": 200
    },
    {
      "epoch": 0.02,
      "eval_cer": 0.990295846526883,
      "eval_loss": 3.032039165496826,
      "eval_runtime": 2093.0103,
      "eval_samples_per_second": 23.041,
      "eval_steps_per_second": 1.441,
      "eval_wer": 0.9803344576185385,
      "step": 200
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00014639999999999998,
      "loss": 2.9236,
      "step": 250
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00017639999999999998,
      "loss": 2.9188,
      "step": 300
    },
    {
      "epoch": 0.04,
      "eval_cer": 0.990295846526883,
      "eval_loss": 2.976602077484131,
      "eval_runtime": 2082.4254,
      "eval_samples_per_second": 23.159,
      "eval_steps_per_second": 1.448,
      "eval_wer": 0.9803344576185385,
      "step": 300
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00020639999999999998,
      "loss": 2.8775,
      "step": 350
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0002364,
      "loss": 2.7809,
      "step": 400
    },
    {
      "epoch": 0.05,
      "eval_cer": 0.9696764691640024,
      "eval_loss": 2.7085533142089844,
      "eval_runtime": 2082.5137,
      "eval_samples_per_second": 23.158,
      "eval_steps_per_second": 1.448,
      "eval_wer": 0.9589848218162464,
      "step": 400
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00026639999999999997,
      "loss": 2.5919,
      "step": 450
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0002964,
      "loss": 2.3631,
      "step": 500
    },
    {
      "epoch": 0.06,
      "eval_cer": 0.862518823322377,
      "eval_loss": 1.9892584085464478,
      "eval_runtime": 2074.1857,
      "eval_samples_per_second": 23.251,
      "eval_steps_per_second": 1.454,
      "eval_wer": 0.9145354269893812,
      "step": 500
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0002994410095705937,
      "loss": 1.5803,
      "step": 550
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0002988057931735411,
      "loss": 0.9821,
      "step": 600
    },
    {
      "epoch": 0.07,
      "eval_cer": 0.1523255882620419,
      "eval_loss": 0.5998358726501465,
      "eval_runtime": 2075.5254,
      "eval_samples_per_second": 23.236,
      "eval_steps_per_second": 1.453,
      "eval_wer": 0.3576434163952664,
      "step": 600
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0002981705767764885,
      "loss": 0.7758,
      "step": 650
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0002975353603794359,
      "loss": 0.6585,
      "step": 700
    },
    {
      "epoch": 0.09,
      "eval_cer": 0.10982663055738294,
      "eval_loss": 0.4177716076374054,
      "eval_runtime": 2075.6462,
      "eval_samples_per_second": 23.234,
      "eval_steps_per_second": 1.453,
      "eval_wer": 0.25202041187922813,
      "step": 700
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0002969001439823833,
      "loss": 0.6077,
      "step": 750
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0002962649275853307,
      "loss": 0.5657,
      "step": 800
    },
    {
      "epoch": 0.1,
      "eval_cer": 0.09587676230587619,
      "eval_loss": 0.36425185203552246,
      "eval_runtime": 2094.5001,
      "eval_samples_per_second": 23.025,
      "eval_steps_per_second": 1.439,
      "eval_wer": 0.21289942193910186,
      "step": 800
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00029562971118827813,
      "loss": 0.5295,
      "step": 850
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0002949944947912255,
      "loss": 0.4998,
      "step": 900
    },
    {
      "epoch": 0.11,
      "eval_cer": 0.09022533657576749,
      "eval_loss": 0.3022037148475647,
      "eval_runtime": 2108.0878,
      "eval_samples_per_second": 22.877,
      "eval_steps_per_second": 1.43,
      "eval_wer": 0.20626376037718358,
      "step": 900
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00029435927839417295,
      "loss": 0.4887,
      "step": 950
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0002937240619971203,
      "loss": 0.4902,
      "step": 1000
    },
    {
      "epoch": 0.12,
      "eval_cer": 0.0833139440810026,
      "eval_loss": 0.2917855978012085,
      "eval_runtime": 2122.6451,
      "eval_samples_per_second": 22.72,
      "eval_steps_per_second": 1.42,
      "eval_wer": 0.1899083786482299,
      "step": 1000
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0002930888456000677,
      "loss": 0.4776,
      "step": 1050
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00029245362920301513,
      "loss": 0.468,
      "step": 1100
    },
    {
      "epoch": 0.14,
      "eval_cer": 0.08389881054318646,
      "eval_loss": 0.2726401686668396,
      "eval_runtime": 2129.6342,
      "eval_samples_per_second": 22.645,
      "eval_steps_per_second": 1.416,
      "eval_wer": 0.1888892520157431,
      "step": 1100
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00029181841280596254,
      "loss": 0.44,
      "step": 1150
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00029118319640890995,
      "loss": 0.4332,
      "step": 1200
    },
    {
      "epoch": 0.15,
      "eval_cer": 0.07579453612837772,
      "eval_loss": 0.2764703929424286,
      "eval_runtime": 2387.2613,
      "eval_samples_per_second": 20.201,
      "eval_steps_per_second": 1.263,
      "eval_wer": 0.1726946095004212,
      "step": 1200
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00029054798001185736,
      "loss": 0.4363,
      "step": 1250
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00028991276361480477,
      "loss": 0.4178,
      "step": 1300
    },
    {
      "epoch": 0.16,
      "eval_cer": 0.07956404299571626,
      "eval_loss": 0.2525430917739868,
      "eval_runtime": 2685.8951,
      "eval_samples_per_second": 17.955,
      "eval_steps_per_second": 1.123,
      "eval_wer": 0.18290121608219176,
      "step": 1300
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0002892775472177521,
      "loss": 0.4072,
      "step": 1350
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0002886423308206996,
      "loss": 0.4171,
      "step": 1400
    },
    {
      "epoch": 0.17,
      "eval_cer": 0.07121410576965095,
      "eval_loss": 0.24207398295402527,
      "eval_runtime": 2387.6421,
      "eval_samples_per_second": 20.198,
      "eval_steps_per_second": 1.263,
      "eval_wer": 0.16114939876197457,
      "step": 1400
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00028800711442364694,
      "loss": 0.4074,
      "step": 1450
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00028737189802659435,
      "loss": 0.4049,
      "step": 1500
    },
    {
      "epoch": 0.19,
      "eval_cer": 0.07119875184394742,
      "eval_loss": 0.23904423415660858,
      "eval_runtime": 2682.696,
      "eval_samples_per_second": 17.977,
      "eval_steps_per_second": 1.124,
      "eval_wer": 0.1614502011908041,
      "step": 1500
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00028673668162954176,
      "loss": 0.3901,
      "step": 1550
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0002861014652324892,
      "loss": 0.3963,
      "step": 1600
    },
    {
      "epoch": 0.2,
      "eval_cer": 0.07037200199837249,
      "eval_loss": 0.22700898349285126,
      "eval_runtime": 2633.1938,
      "eval_samples_per_second": 18.315,
      "eval_steps_per_second": 1.145,
      "eval_wer": 0.16208515443303412,
      "step": 1600
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0002854662488354366,
      "loss": 0.3813,
      "step": 1650
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.000284831032438384,
      "loss": 0.3924,
      "step": 1700
    },
    {
      "epoch": 0.21,
      "eval_cer": 0.07119875184394742,
      "eval_loss": 0.2278161644935608,
      "eval_runtime": 2675.9933,
      "eval_samples_per_second": 18.022,
      "eval_steps_per_second": 1.127,
      "eval_wer": 0.16218386565136397,
      "step": 1700
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0002841958160413314,
      "loss": 0.3938,
      "step": 1750
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00028356059964427876,
      "loss": 0.3942,
      "step": 1800
    },
    {
      "epoch": 0.22,
      "eval_cer": 0.06874165130289871,
      "eval_loss": 0.22094863653182983,
      "eval_runtime": 2685.0595,
      "eval_samples_per_second": 17.961,
      "eval_steps_per_second": 1.123,
      "eval_wer": 0.1562058342331839,
      "step": 1800
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0002829253832472262,
      "loss": 0.3674,
      "step": 1850
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0002822901668501736,
      "loss": 0.3745,
      "step": 1900
    },
    {
      "epoch": 0.24,
      "eval_cer": 0.0659505438242377,
      "eval_loss": 0.21558238565921783,
      "eval_runtime": 2439.0418,
      "eval_samples_per_second": 19.773,
      "eval_steps_per_second": 1.236,
      "eval_wer": 0.1506740042005625,
      "step": 1900
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.000281654950453121,
      "loss": 0.3711,
      "step": 1950
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0002810197340560684,
      "loss": 0.3598,
      "step": 2000
    },
    {
      "epoch": 0.25,
      "eval_cer": 0.06898353468628977,
      "eval_loss": 0.221131831407547,
      "eval_runtime": 2663.9985,
      "eval_samples_per_second": 18.103,
      "eval_steps_per_second": 1.132,
      "eval_wer": 0.15953867178719996,
      "step": 2000
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0002803845176590158,
      "loss": 0.3756,
      "step": 2050
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0002797493012619632,
      "loss": 0.3653,
      "step": 2100
    },
    {
      "epoch": 0.26,
      "eval_cer": 0.06659281034712863,
      "eval_loss": 0.2138465791940689,
      "eval_runtime": 2389.088,
      "eval_samples_per_second": 20.186,
      "eval_steps_per_second": 1.262,
      "eval_wer": 0.1510541757846708,
      "step": 2100
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00027911408486491063,
      "loss": 0.3751,
      "step": 2150
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00027847886846785804,
      "loss": 0.3576,
      "step": 2200
    },
    {
      "epoch": 0.27,
      "eval_cer": 0.06642438959287295,
      "eval_loss": 0.2111993283033371,
      "eval_runtime": 2676.0178,
      "eval_samples_per_second": 18.022,
      "eval_steps_per_second": 1.127,
      "eval_wer": 0.1520546273217979,
      "step": 2200
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00027784365207080545,
      "loss": 0.353,
      "step": 2250
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00027720843567375286,
      "loss": 0.3569,
      "step": 2300
    },
    {
      "epoch": 0.29,
      "eval_cer": 0.06564322909592542,
      "eval_loss": 0.21856307983398438,
      "eval_runtime": 2642.3302,
      "eval_samples_per_second": 18.251,
      "eval_steps_per_second": 1.141,
      "eval_wer": 0.14776135629213985,
      "step": 2300
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00027657321927670027,
      "loss": 0.371,
      "step": 2350
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0002759380028796476,
      "loss": 0.3473,
      "step": 2400
    },
    {
      "epoch": 0.3,
      "eval_cer": 0.0664718686554331,
      "eval_loss": 0.2098075896501541,
      "eval_runtime": 2685.1293,
      "eval_samples_per_second": 17.96,
      "eval_steps_per_second": 1.123,
      "eval_wer": 0.15094546005096968,
      "step": 2400
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0002753027864825951,
      "loss": 0.3574,
      "step": 2450
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00027466757008554244,
      "loss": 0.3735,
      "step": 2500
    },
    {
      "epoch": 0.31,
      "eval_cer": 0.06522796445920521,
      "eval_loss": 0.20497867465019226,
      "eval_runtime": 2686.3997,
      "eval_samples_per_second": 17.952,
      "eval_steps_per_second": 1.122,
      "eval_wer": 0.1481148491685914,
      "step": 2500
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00027403235368848985,
      "loss": 0.3617,
      "step": 2550
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00027339713729143726,
      "loss": 0.3523,
      "step": 2600
    },
    {
      "epoch": 0.32,
      "eval_cer": 0.06236906349320707,
      "eval_loss": 0.20248664915561676,
      "eval_runtime": 2418.5335,
      "eval_samples_per_second": 19.94,
      "eval_steps_per_second": 1.247,
      "eval_wer": 0.14132578503764698,
      "step": 2600
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0002727619208943847,
      "loss": 0.3466,
      "step": 2650
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0002721267044973321,
      "loss": 0.3443,
      "step": 2700
    },
    {
      "epoch": 0.34,
      "eval_cer": 0.0631530585610537,
      "eval_loss": 0.19711410999298096,
      "eval_runtime": 2678.8408,
      "eval_samples_per_second": 18.003,
      "eval_steps_per_second": 1.125,
      "eval_wer": 0.14343206900714522,
      "step": 2700
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0002714914881002795,
      "loss": 0.3353,
      "step": 2750
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0002708562717032269,
      "loss": 0.3271,
      "step": 2800
    },
    {
      "epoch": 0.35,
      "eval_cer": 0.06349556921136332,
      "eval_loss": 0.1974596232175827,
      "eval_runtime": 2413.2013,
      "eval_samples_per_second": 19.984,
      "eval_steps_per_second": 1.249,
      "eval_wer": 0.14444185809195217,
      "step": 2800
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00027022105530617426,
      "loss": 0.3411,
      "step": 2850
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0002695858389091217,
      "loss": 0.347,
      "step": 2900
    },
    {
      "epoch": 0.36,
      "eval_cer": 0.06194080707319925,
      "eval_loss": 0.19428882002830505,
      "eval_runtime": 2689.3153,
      "eval_samples_per_second": 17.932,
      "eval_steps_per_second": 1.121,
      "eval_wer": 0.14129443755615034,
      "step": 2900
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0002689506225120691,
      "loss": 0.3414,
      "step": 2950
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0002683154061150165,
      "loss": 0.3291,
      "step": 3000
    },
    {
      "epoch": 0.37,
      "eval_cer": 0.06220513080954164,
      "eval_loss": 0.19397367537021637,
      "eval_runtime": 2678.8937,
      "eval_samples_per_second": 18.002,
      "eval_steps_per_second": 1.125,
      "eval_wer": 0.14132445110226416,
      "step": 3000
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0002676801897179639,
      "loss": 0.3472,
      "step": 3050
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0002670449733209113,
      "loss": 0.3386,
      "step": 3100
    },
    {
      "epoch": 0.39,
      "eval_cer": 0.06227859343867701,
      "eval_loss": 0.18810917437076569,
      "eval_runtime": 2096.7687,
      "eval_samples_per_second": 23.0,
      "eval_steps_per_second": 1.438,
      "eval_wer": 0.143275331599662,
      "step": 3100
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0002664097569238587,
      "loss": 0.3561,
      "step": 3150
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00026577454052680613,
      "loss": 0.3346,
      "step": 3200
    },
    {
      "epoch": 0.4,
      "eval_cer": 0.062040017054668245,
      "eval_loss": 0.19153311848640442,
      "eval_runtime": 2107.1655,
      "eval_samples_per_second": 22.887,
      "eval_steps_per_second": 1.431,
      "eval_wer": 0.141297105426916,
      "step": 3200
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00026513932412975354,
      "loss": 0.3362,
      "step": 3250
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0002645041077327009,
      "loss": 0.3286,
      "step": 3300
    },
    {
      "epoch": 0.41,
      "eval_cer": 0.06275858077759366,
      "eval_loss": 0.18800656497478485,
      "eval_runtime": 2108.9053,
      "eval_samples_per_second": 22.868,
      "eval_steps_per_second": 1.43,
      "eval_wer": 0.14638073317090447,
      "step": 3300
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00026386889133564836,
      "loss": 0.3088,
      "step": 3350
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0002632336749385957,
      "loss": 0.3211,
      "step": 3400
    },
    {
      "epoch": 0.42,
      "eval_cer": 0.06097185625418542,
      "eval_loss": 0.1875077337026596,
      "eval_runtime": 2192.9622,
      "eval_samples_per_second": 21.991,
      "eval_steps_per_second": 1.375,
      "eval_wer": 0.1381210052803832,
      "step": 3400
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0002625984585415431,
      "loss": 0.3185,
      "step": 3450
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00026196324214449053,
      "loss": 0.3272,
      "step": 3500
    },
    {
      "epoch": 0.44,
      "eval_cer": 0.060900047124741195,
      "eval_loss": 0.18167421221733093,
      "eval_runtime": 2110.7367,
      "eval_samples_per_second": 22.848,
      "eval_steps_per_second": 1.428,
      "eval_wer": 0.14127176065464214,
      "step": 3500
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00026132802574743794,
      "loss": 0.321,
      "step": 3550
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00026069280935038535,
      "loss": 0.3246,
      "step": 3600
    },
    {
      "epoch": 0.45,
      "eval_cer": 0.05953118559471069,
      "eval_loss": 0.18253999948501587,
      "eval_runtime": 2122.2554,
      "eval_samples_per_second": 22.724,
      "eval_steps_per_second": 1.421,
      "eval_wer": 0.13588532957874988,
      "step": 3600
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00026005759295333276,
      "loss": 0.3266,
      "step": 3650
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0002594223765562802,
      "loss": 0.3239,
      "step": 3700
    },
    {
      "epoch": 0.46,
      "eval_cer": 0.06006573842343529,
      "eval_loss": 0.18680287897586823,
      "eval_runtime": 2677.6172,
      "eval_samples_per_second": 18.011,
      "eval_steps_per_second": 1.126,
      "eval_wer": 0.13703584884644604,
      "step": 3700
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00025878716015922753,
      "loss": 0.3272,
      "step": 3750
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.000258151943762175,
      "loss": 0.32,
      "step": 3800
    },
    {
      "epoch": 0.47,
      "eval_cer": 0.05884947129347375,
      "eval_loss": 0.18254593014717102,
      "eval_runtime": 2588.5152,
      "eval_samples_per_second": 18.631,
      "eval_steps_per_second": 1.165,
      "eval_wer": 0.13350625582346165,
      "step": 3800
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00025751672736512235,
      "loss": 0.3148,
      "step": 3850
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00025688151096806976,
      "loss": 0.3185,
      "step": 3900
    },
    {
      "epoch": 0.49,
      "eval_cer": 0.06071367408812445,
      "eval_loss": 0.1804201900959015,
      "eval_runtime": 2677.674,
      "eval_samples_per_second": 18.01,
      "eval_steps_per_second": 1.126,
      "eval_wer": 0.14048273787569457,
      "step": 3900
    }
  ],
  "max_steps": 24114,
  "num_train_epochs": 3,
  "total_flos": 3.282673713375086e+19,
  "trial_name": null,
  "trial_params": null
}