trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 100.0,
  "global_step": 8200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 3.66,
      "learning_rate": 2.0859999999999997e-05,
      "loss": 6.9294,
      "step": 300
    },
    {
      "epoch": 6.1,
      "eval_loss": 2.971196413040161,
      "eval_runtime": 36.7267,
      "eval_samples_per_second": 32.483,
      "eval_steps_per_second": 1.035,
      "eval_wer": 1.0,
      "step": 500
    },
    {
      "epoch": 7.32,
      "learning_rate": 4.1859999999999996e-05,
      "loss": 3.05,
      "step": 600
    },
    {
      "epoch": 10.98,
      "learning_rate": 6.285999999999999e-05,
      "loss": 2.8305,
      "step": 900
    },
    {
      "epoch": 12.2,
      "eval_loss": 1.707324504852295,
      "eval_runtime": 37.8146,
      "eval_samples_per_second": 31.549,
      "eval_steps_per_second": 1.005,
      "eval_wer": 0.9478846435368175,
      "step": 1000
    },
    {
      "epoch": 14.63,
      "learning_rate": 6.8075e-05,
      "loss": 2.0414,
      "step": 1200
    },
    {
      "epoch": 18.29,
      "learning_rate": 6.515833333333332e-05,
      "loss": 1.4795,
      "step": 1500
    },
    {
      "epoch": 18.29,
      "eval_loss": 0.5756120681762695,
      "eval_runtime": 37.4871,
      "eval_samples_per_second": 31.824,
      "eval_steps_per_second": 1.014,
      "eval_wer": 0.6397306397306397,
      "step": 1500
    },
    {
      "epoch": 21.95,
      "learning_rate": 6.224166666666666e-05,
      "loss": 1.3433,
      "step": 1800
    },
    {
      "epoch": 24.39,
      "eval_loss": 0.4967685639858246,
      "eval_runtime": 37.9583,
      "eval_samples_per_second": 31.429,
      "eval_steps_per_second": 1.001,
      "eval_wer": 0.5423803249890207,
      "step": 2000
    },
    {
      "epoch": 25.61,
      "learning_rate": 5.9325e-05,
      "loss": 1.2567,
      "step": 2100
    },
    {
      "epoch": 29.27,
      "learning_rate": 5.6408333333333327e-05,
      "loss": 1.1766,
      "step": 2400
    },
    {
      "epoch": 30.49,
      "eval_loss": 0.4184603989124298,
      "eval_runtime": 37.222,
      "eval_samples_per_second": 32.051,
      "eval_steps_per_second": 1.021,
      "eval_wer": 0.4743083003952569,
      "step": 2500
    },
    {
      "epoch": 32.93,
      "learning_rate": 5.349166666666666e-05,
      "loss": 1.0943,
      "step": 2700
    },
    {
      "epoch": 36.59,
      "learning_rate": 5.0574999999999996e-05,
      "loss": 1.0017,
      "step": 3000
    },
    {
      "epoch": 36.59,
      "eval_loss": 0.3302731513977051,
      "eval_runtime": 36.8764,
      "eval_samples_per_second": 32.351,
      "eval_steps_per_second": 1.03,
      "eval_wer": 0.35778070560679254,
      "step": 3000
    },
    {
      "epoch": 40.24,
      "learning_rate": 4.7658333333333324e-05,
      "loss": 0.9358,
      "step": 3300
    },
    {
      "epoch": 42.68,
      "eval_loss": 0.30026641488075256,
      "eval_runtime": 37.9381,
      "eval_samples_per_second": 31.446,
      "eval_steps_per_second": 1.002,
      "eval_wer": 0.3050797833406529,
      "step": 3500
    },
    {
      "epoch": 43.9,
      "learning_rate": 4.4741666666666665e-05,
      "loss": 0.8798,
      "step": 3600
    },
    {
      "epoch": 47.56,
      "learning_rate": 4.183472222222222e-05,
      "loss": 0.8358,
      "step": 3900
    },
    {
      "epoch": 48.78,
      "eval_loss": 0.30453845858573914,
      "eval_runtime": 36.3987,
      "eval_samples_per_second": 32.776,
      "eval_steps_per_second": 1.044,
      "eval_wer": 0.2883911579563753,
      "step": 4000
    },
    {
      "epoch": 51.22,
      "learning_rate": 3.8918055555555554e-05,
      "loss": 0.8059,
      "step": 4200
    },
    {
      "epoch": 54.88,
      "learning_rate": 3.600138888888889e-05,
      "loss": 0.7647,
      "step": 4500
    },
    {
      "epoch": 54.88,
      "eval_loss": 0.2865545451641083,
      "eval_runtime": 36.6315,
      "eval_samples_per_second": 32.568,
      "eval_steps_per_second": 1.037,
      "eval_wer": 0.2677499634021373,
      "step": 4500
    },
    {
      "epoch": 58.54,
      "learning_rate": 3.3084722222222216e-05,
      "loss": 0.7482,
      "step": 4800
    },
    {
      "epoch": 60.98,
      "eval_loss": 0.2829342782497406,
      "eval_runtime": 36.4673,
      "eval_samples_per_second": 32.714,
      "eval_steps_per_second": 1.042,
      "eval_wer": 0.2585273020055629,
      "step": 5000
    },
    {
      "epoch": 62.2,
      "learning_rate": 3.016805555555555e-05,
      "loss": 0.7223,
      "step": 5100
    },
    {
      "epoch": 65.85,
      "learning_rate": 2.725138888888889e-05,
      "loss": 0.6943,
      "step": 5400
    },
    {
      "epoch": 67.07,
      "eval_loss": 0.2782347798347473,
      "eval_runtime": 37.212,
      "eval_samples_per_second": 32.06,
      "eval_steps_per_second": 1.021,
      "eval_wer": 0.24784072610159566,
      "step": 5500
    },
    {
      "epoch": 69.51,
      "learning_rate": 2.433472222222222e-05,
      "loss": 0.6758,
      "step": 5700
    },
    {
      "epoch": 73.17,
      "learning_rate": 2.1418055555555552e-05,
      "loss": 0.6586,
      "step": 6000
    },
    {
      "epoch": 73.17,
      "eval_loss": 0.2911244034767151,
      "eval_runtime": 36.7087,
      "eval_samples_per_second": 32.499,
      "eval_steps_per_second": 1.035,
      "eval_wer": 0.25369638413116674,
      "step": 6000
    },
    {
      "epoch": 76.83,
      "learning_rate": 1.850138888888889e-05,
      "loss": 0.6425,
      "step": 6300
    },
    {
      "epoch": 79.27,
      "eval_loss": 0.2817089259624481,
      "eval_runtime": 36.6478,
      "eval_samples_per_second": 32.553,
      "eval_steps_per_second": 1.037,
      "eval_wer": 0.24623042014346364,
      "step": 6500
    },
    {
      "epoch": 80.49,
      "learning_rate": 1.558472222222222e-05,
      "loss": 0.6313,
      "step": 6600
    },
    {
      "epoch": 84.15,
      "learning_rate": 1.2677777777777776e-05,
      "loss": 0.6067,
      "step": 6900
    },
    {
      "epoch": 85.37,
      "eval_loss": 0.29103317856788635,
      "eval_runtime": 37.7988,
      "eval_samples_per_second": 31.562,
      "eval_steps_per_second": 1.005,
      "eval_wer": 0.24359537403015663,
      "step": 7000
    },
    {
      "epoch": 87.8,
      "learning_rate": 9.761111111111111e-06,
      "loss": 0.6064,
      "step": 7200
    },
    {
      "epoch": 91.46,
      "learning_rate": 6.844444444444444e-06,
      "loss": 0.5974,
      "step": 7500
    },
    {
      "epoch": 91.46,
      "eval_loss": 0.28752732276916504,
      "eval_runtime": 38.6504,
      "eval_samples_per_second": 30.866,
      "eval_steps_per_second": 0.983,
      "eval_wer": 0.24300980822719953,
      "step": 7500
    },
    {
      "epoch": 95.12,
      "learning_rate": 3.927777777777777e-06,
      "loss": 0.5812,
      "step": 7800
    },
    {
      "epoch": 97.56,
      "eval_loss": 0.2852196991443634,
      "eval_runtime": 38.817,
      "eval_samples_per_second": 30.734,
      "eval_steps_per_second": 0.979,
      "eval_wer": 0.23964280486019615,
      "step": 8000
    },
    {
      "epoch": 98.78,
      "learning_rate": 1.011111111111111e-06,
      "loss": 0.5804,
      "step": 8100
    },
    {
      "epoch": 100.0,
      "step": 8200,
      "total_flos": 2.6094292924080824e+19,
      "train_loss": 1.2572689837944218,
      "train_runtime": 11083.5384,
      "train_samples_per_second": 23.512,
      "train_steps_per_second": 0.74
    }
  ],
  "max_steps": 8200,
  "num_train_epochs": 100,
  "total_flos": 2.6094292924080824e+19,
  "trial_name": null,
  "trial_params": null
}