trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 1527,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.98,
      "learning_rate": 3.362802881466929e-05,
      "loss": 3.7943,
      "step": 500
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.7256057629338573e-05,
      "loss": 2.9242,
      "step": 1000
    },
    {
      "epoch": 2.95,
      "learning_rate": 8.840864440078585e-07,
      "loss": 2.426,
      "step": 1500
    },
    {
      "epoch": 3.0,
      "step": 1527,
      "total_flos": 366012115943424.0,
      "train_loss": 3.0355011216159262,
      "train_runtime": 240.3711,
      "train_samples_per_second": 25.373,
      "train_steps_per_second": 6.353
    }
  ],
  "max_steps": 1527,
  "num_train_epochs": 3,
  "total_flos": 366012115943424.0,
  "trial_name": null,
  "trial_params": null
}