trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.24964633435965716,
  "global_step": 1500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.08,
      "learning_rate": 0.00028506,
      "loss": 0.8206,
      "step": 500
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00027005999999999997,
      "loss": 0.7132,
      "step": 1000
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00025506,
      "loss": 0.6744,
      "step": 1500
    }
  ],
  "max_steps": 10000,
  "num_train_epochs": 2,
  "total_flos": 8209887942985728.0,
  "trial_name": null,
  "trial_params": null
}