trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 24.984615384615385,
  "global_step": 800,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 12.49,
      "learning_rate": 0.00023999999999999998,
      "loss": 5.2627,
      "step": 400
    },
    {
      "epoch": 12.49,
      "eval_loss": 2.4135375022888184,
      "eval_runtime": 64.4301,
      "eval_samples_per_second": 7.853,
      "eval_wer": 1.0,
      "step": 400
    },
    {
      "epoch": 24.98,
      "learning_rate": 0.00010434782608695651,
      "loss": 0.7954,
      "step": 800
    },
    {
      "epoch": 24.98,
      "eval_loss": 1.134879469871521,
      "eval_runtime": 63.6881,
      "eval_samples_per_second": 7.945,
      "eval_wer": 0.6729081334113517,
      "step": 800
    }
  ],
  "max_steps": 960,
  "num_train_epochs": 30,
  "total_flos": 3.0450267331053005e+18,
  "trial_name": null,
  "trial_params": null
}