trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.9990393852065322,
  "global_step": 1040,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.92,
      "learning_rate": 1.0576923076923078e-05,
      "loss": 0.1836,
      "step": 500
    },
    {
      "epoch": 3.85,
      "learning_rate": 9.615384615384617e-07,
      "loss": 0.0586,
      "step": 1000
    },
    {
      "epoch": 4.0,
      "step": 1040,
      "total_flos": 190203697799424.0,
      "train_loss": 0.11814142912626266,
      "train_runtime": 189.0089,
      "train_samples_per_second": 176.161,
      "train_steps_per_second": 5.502
    }
  ],
  "max_steps": 1040,
  "num_train_epochs": 4,
  "total_flos": 190203697799424.0,
  "trial_name": null,
  "trial_params": null
}