trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9999040030719017,
  "global_step": 1953,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.26,
      "learning_rate": 7.43983614951357e-06,
      "loss": 0.1187,
      "step": 500
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.8796722990271386e-06,
      "loss": 0.0011,
      "step": 1000
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.319508448540707e-06,
      "loss": 0.0006,
      "step": 1500
    },
    {
      "epoch": 1.0,
      "step": 1953,
      "train_runtime": 2937.1499,
      "train_samples_per_second": 0.665
    }
  ],
  "max_steps": 1953,
  "num_train_epochs": 1,
  "total_flos": 262568325479694336,
  "trial_name": null,
  "trial_params": null
}