trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 70.0,
  "global_step": 770,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 27.27,
      "learning_rate": 0.0001495,
      "loss": 5.2793,
      "step": 300
    },
    {
      "epoch": 27.27,
      "eval_loss": 3.073740243911743,
      "eval_runtime": 13.7846,
      "eval_samples_per_second": 21.836,
      "eval_steps_per_second": 2.757,
      "eval_wer": 1.0,
      "step": 300
    },
    {
      "epoch": 54.55,
      "learning_rate": 0.00015833333333333332,
      "loss": 1.5348,
      "step": 600
    },
    {
      "epoch": 54.55,
      "eval_loss": 0.6312472820281982,
      "eval_runtime": 13.5036,
      "eval_samples_per_second": 22.29,
      "eval_steps_per_second": 2.814,
      "eval_wer": 0.6334310850439883,
      "step": 600
    },
    {
      "epoch": 70.0,
      "step": 770,
      "total_flos": 1.0714540362717635e+19,
      "train_loss": 2.7389903675426135,
      "train_runtime": 3757.1772,
      "train_samples_per_second": 13.116,
      "train_steps_per_second": 0.205
    }
  ],
  "max_steps": 770,
  "num_train_epochs": 70,
  "total_flos": 1.0714540362717635e+19,
  "trial_name": null,
  "trial_params": null
}