{
  "best_metric": 0.1725786179304123,
  "best_model_checkpoint": "./checkpoint-4000",
  "epoch": 70.17543859649123,
  "global_step": 4000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.75,
      "learning_rate": 1.26275e-05,
      "loss": 6.0202,
      "step": 100
    },
    {
      "epoch": 3.51,
      "learning_rate": 2.5002499999999997e-05,
      "loss": 2.9203,
      "step": 200
    },
    {
      "epoch": 5.26,
      "learning_rate": 3.73775e-05,
      "loss": 1.7768,
      "step": 300
    },
    {
      "epoch": 7.02,
      "learning_rate": 4.9752499999999995e-05,
      "loss": 1.331,
      "step": 400
    },
    {
      "epoch": 7.02,
      "eval_cer": 0.08315629742033384,
      "eval_loss": 0.32805994153022766,
      "eval_runtime": 15.7786,
      "eval_samples_per_second": 21.231,
      "eval_steps_per_second": 0.38,
      "eval_wer": 0.46174863387978143,
      "step": 400
    },
    {
      "epoch": 8.77,
      "learning_rate": 5e-05,
      "loss": 1.165,
      "step": 500
    },
    {
      "epoch": 10.53,
      "learning_rate": 5e-05,
      "loss": 1.0755,
      "step": 600
    },
    {
      "epoch": 12.28,
      "learning_rate": 5e-05,
      "loss": 1.0359,
      "step": 700
    },
    {
      "epoch": 14.04,
      "learning_rate": 5e-05,
      "loss": 0.9968,
      "step": 800
    },
    {
      "epoch": 14.04,
      "eval_cer": 0.05275670207384927,
      "eval_loss": 0.21139617264270782,
      "eval_runtime": 15.2468,
      "eval_samples_per_second": 21.972,
      "eval_steps_per_second": 0.394,
      "eval_wer": 0.3114754098360656,
      "step": 800
    },
    {
      "epoch": 15.79,
      "learning_rate": 5e-05,
      "loss": 0.9746,
      "step": 900
    },
    {
      "epoch": 17.54,
      "learning_rate": 5e-05,
      "loss": 0.9629,
      "step": 1000
    },
    {
      "epoch": 19.3,
      "learning_rate": 5e-05,
      "loss": 0.9639,
      "step": 1100
    },
    {
      "epoch": 21.05,
      "learning_rate": 5e-05,
      "loss": 0.9367,
      "step": 1200
    },
    {
      "epoch": 21.05,
      "eval_cer": 0.048052604957005564,
      "eval_loss": 0.19284144043922424,
      "eval_runtime": 15.2894,
      "eval_samples_per_second": 21.911,
      "eval_steps_per_second": 0.392,
      "eval_wer": 0.27712724434035907,
      "step": 1200
    },
    {
      "epoch": 22.81,
      "learning_rate": 5e-05,
      "loss": 0.9265,
      "step": 1300
    },
    {
      "epoch": 24.56,
      "learning_rate": 5e-05,
      "loss": 0.9093,
      "step": 1400
    },
    {
      "epoch": 26.32,
      "learning_rate": 5e-05,
      "loss": 0.9126,
      "step": 1500
    },
    {
      "epoch": 28.07,
      "learning_rate": 5e-05,
      "loss": 0.9066,
      "step": 1600
    },
    {
      "epoch": 28.07,
      "eval_cer": 0.047698533131006575,
      "eval_loss": 0.19045932590961456,
      "eval_runtime": 15.3651,
      "eval_samples_per_second": 21.803,
      "eval_steps_per_second": 0.39,
      "eval_wer": 0.2728337236533958,
      "step": 1600
    },
    {
      "epoch": 29.82,
      "learning_rate": 5e-05,
      "loss": 0.8954,
      "step": 1700
    },
    {
      "epoch": 31.58,
      "learning_rate": 5e-05,
      "loss": 0.8807,
      "step": 1800
    },
    {
      "epoch": 33.33,
      "learning_rate": 5e-05,
      "loss": 0.8753,
      "step": 1900
    },
    {
      "epoch": 35.09,
      "learning_rate": 5e-05,
      "loss": 0.869,
      "step": 2000
    },
    {
      "epoch": 35.09,
      "eval_cer": 0.04486595852301467,
      "eval_loss": 0.181670144200325,
      "eval_runtime": 15.5069,
      "eval_samples_per_second": 21.603,
      "eval_steps_per_second": 0.387,
      "eval_wer": 0.25644028103044497,
      "step": 2000
    },
    {
      "epoch": 36.84,
      "learning_rate": 4.76725e-05,
      "loss": 0.8637,
      "step": 2100
    },
    {
      "epoch": 38.6,
      "learning_rate": 4.5297500000000005e-05,
      "loss": 0.8557,
      "step": 2200
    },
    {
      "epoch": 40.35,
      "learning_rate": 4.29225e-05,
      "loss": 0.8537,
      "step": 2300
    },
    {
      "epoch": 42.11,
      "learning_rate": 4.0547500000000004e-05,
      "loss": 0.8319,
      "step": 2400
    },
    {
      "epoch": 42.11,
      "eval_cer": 0.04334850784016186,
      "eval_loss": 0.18102943897247314,
      "eval_runtime": 15.5131,
      "eval_samples_per_second": 21.595,
      "eval_steps_per_second": 0.387,
      "eval_wer": 0.24902419984387197,
      "step": 2400
    },
    {
      "epoch": 43.86,
      "learning_rate": 3.81725e-05,
      "loss": 0.837,
      "step": 2500
    },
    {
      "epoch": 45.61,
      "learning_rate": 3.57975e-05,
      "loss": 0.8098,
      "step": 2600
    },
    {
      "epoch": 47.37,
      "learning_rate": 3.34225e-05,
      "loss": 0.8212,
      "step": 2700
    },
    {
      "epoch": 49.12,
      "learning_rate": 3.1047500000000004e-05,
      "loss": 0.8113,
      "step": 2800
    },
    {
      "epoch": 49.12,
      "eval_cer": 0.042134547293879616,
      "eval_loss": 0.17691758275032043,
      "eval_runtime": 15.2685,
      "eval_samples_per_second": 21.941,
      "eval_steps_per_second": 0.393,
      "eval_wer": 0.24043715846994534,
      "step": 2800
    },
    {
      "epoch": 50.88,
      "learning_rate": 2.8672500000000004e-05,
      "loss": 0.7928,
      "step": 2900
    },
    {
      "epoch": 52.63,
      "learning_rate": 2.6297500000000004e-05,
      "loss": 0.7871,
      "step": 3000
    },
    {
      "epoch": 54.39,
      "learning_rate": 2.3922500000000003e-05,
      "loss": 0.777,
      "step": 3100
    },
    {
      "epoch": 56.14,
      "learning_rate": 2.1571249999999998e-05,
      "loss": 0.7624,
      "step": 3200
    },
    {
      "epoch": 56.14,
      "eval_cer": 0.04319676277187658,
      "eval_loss": 0.1837695837020874,
      "eval_runtime": 15.5608,
      "eval_samples_per_second": 21.528,
      "eval_steps_per_second": 0.386,
      "eval_wer": 0.24434035909445745,
      "step": 3200
    },
    {
      "epoch": 57.89,
      "learning_rate": 1.9196249999999998e-05,
      "loss": 0.7624,
      "step": 3300
    },
    {
      "epoch": 59.65,
      "learning_rate": 1.682125e-05,
      "loss": 0.7517,
      "step": 3400
    },
    {
      "epoch": 61.4,
      "learning_rate": 1.444625e-05,
      "loss": 0.7417,
      "step": 3500
    },
    {
      "epoch": 63.16,
      "learning_rate": 1.207125e-05,
      "loss": 0.7328,
      "step": 3600
    },
    {
      "epoch": 63.16,
      "eval_cer": 0.041476985331310064,
      "eval_loss": 0.17556767165660858,
      "eval_runtime": 15.151,
      "eval_samples_per_second": 22.111,
      "eval_steps_per_second": 0.396,
      "eval_wer": 0.2302888368462139,
      "step": 3600
    },
    {
      "epoch": 64.91,
      "learning_rate": 9.69625e-06,
      "loss": 0.7334,
      "step": 3700
    },
    {
      "epoch": 66.67,
      "learning_rate": 7.321250000000002e-06,
      "loss": 0.7261,
      "step": 3800
    },
    {
      "epoch": 68.42,
      "learning_rate": 4.946250000000002e-06,
      "loss": 0.7268,
      "step": 3900
    },
    {
      "epoch": 70.18,
      "learning_rate": 2.5712500000000027e-06,
      "loss": 0.7209,
      "step": 4000
    },
    {
      "epoch": 70.18,
      "eval_cer": 0.04046535154274153,
      "eval_loss": 0.1725786179304123,
      "eval_runtime": 14.417,
      "eval_samples_per_second": 23.236,
      "eval_steps_per_second": 0.416,
      "eval_wer": 0.2259953161592506,
      "step": 4000
    },
    {
      "epoch": 70.18,
      "step": 4000,
      "total_flos": 3.2826266564131357e+20,
      "train_loss": 0.07268305778503419,
      "train_runtime": 4050.6325,
      "train_samples_per_second": 126.4,
      "train_steps_per_second": 0.988
    }
  ],
  "max_steps": 4000,
  "num_train_epochs": 71,
  "total_flos": 3.2826266564131357e+20,
  "trial_name": null,
  "trial_params": null
}