trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 2127,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.71,
      "learning_rate": 3.824635637047485e-05,
      "loss": 2.8218,
      "step": 500
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.6492712740949695e-05,
      "loss": 1.9118,
      "step": 1000
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.4739069111424541e-05,
      "loss": 1.5585,
      "step": 1500
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.9854254818993887e-06,
      "loss": 1.2761,
      "step": 2000
    },
    {
      "epoch": 3.0,
      "step": 2127,
      "total_flos": 608939462688768.0,
      "train_loss": 1.8521533960684062,
      "train_runtime": 370.0616,
      "train_samples_per_second": 22.975,
      "train_steps_per_second": 5.748
    }
  ],
  "max_steps": 2127,
  "num_train_epochs": 3,
  "total_flos": 608939462688768.0,
  "trial_name": null,
  "trial_params": null
}