trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 30.0,
  "global_step": 1020,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 14.71,
      "learning_rate": 5.098039215686274e-06,
      "loss": 0.0049,
      "step": 500
    },
    {
      "epoch": 29.41,
      "learning_rate": 1.9607843137254904e-07,
      "loss": 0.0012,
      "step": 1000
    },
    {
      "epoch": 30.0,
      "step": 1020,
      "total_flos": 1529333009280000.0,
      "train_loss": 0.002993474760695986,
      "train_runtime": 193.0701,
      "train_samples_per_second": 124.307,
      "train_steps_per_second": 5.283
    }
  ],
  "max_steps": 1020,
  "num_train_epochs": 30,
  "total_flos": 1529333009280000.0,
  "trial_name": null,
  "trial_params": null
}