trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 18.46153846153846,
  "global_step": 1200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 3.08,
      "learning_rate": 0.00011999999999999999,
      "loss": 5.8244,
      "step": 200
    },
    {
      "epoch": 3.08,
      "eval_loss": 3.064483404159546,
      "eval_runtime": 32.6688,
      "eval_samples_per_second": 7.071,
      "eval_wer": 0.9981722641078364,
      "step": 200
    },
    {
      "epoch": 6.15,
      "learning_rate": 0.00023999999999999998,
      "loss": 3.0089,
      "step": 400
    },
    {
      "epoch": 6.15,
      "eval_loss": 2.9369325637817383,
      "eval_runtime": 32.3179,
      "eval_samples_per_second": 7.148,
      "eval_wer": 0.9981722641078364,
      "step": 400
    },
    {
      "epoch": 9.23,
      "learning_rate": 0.0002625,
      "loss": 2.2569,
      "step": 600
    },
    {
      "epoch": 9.23,
      "eval_loss": 1.1511069536209106,
      "eval_runtime": 32.4921,
      "eval_samples_per_second": 7.109,
      "eval_wer": 0.7900388393877085,
      "step": 600
    },
    {
      "epoch": 12.31,
      "learning_rate": 0.00018749999999999998,
      "loss": 0.9673,
      "step": 800
    },
    {
      "epoch": 12.31,
      "eval_loss": 0.8750982284545898,
      "eval_runtime": 32.4693,
      "eval_samples_per_second": 7.114,
      "eval_wer": 0.6248572081334247,
      "step": 800
    },
    {
      "epoch": 15.38,
      "learning_rate": 0.0001125,
      "loss": 0.6495,
      "step": 1000
    },
    {
      "epoch": 15.38,
      "eval_loss": 0.8149241805076599,
      "eval_runtime": 32.8858,
      "eval_samples_per_second": 7.024,
      "eval_wer": 0.5716244002741604,
      "step": 1000
    },
    {
      "epoch": 18.46,
      "learning_rate": 3.75e-05,
      "loss": 0.5093,
      "step": 1200
    },
    {
      "epoch": 18.46,
      "eval_loss": 0.8335053324699402,
      "eval_runtime": 32.889,
      "eval_samples_per_second": 7.024,
      "eval_wer": 0.5522047064199224,
      "step": 1200
    }
  ],
  "max_steps": 1300,
  "num_train_epochs": 20,
  "total_flos": 7.999569757266722e+18,
  "trial_name": null,
  "trial_params": null
}