trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.06091370558375635,
  "eval_steps": 500,
  "global_step": 3,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 0.0001,
      "loss": 1.5897,
      "step": 1
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0002,
      "loss": 1.7194,
      "step": 2
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0,
      "loss": 1.51,
      "step": 3
    },
    {
      "epoch": 0.06,
      "step": 3,
      "total_flos": 519041285554176.0,
      "train_loss": 1.6063518921534221,
      "train_runtime": 71.9618,
      "train_samples_per_second": 0.167,
      "train_steps_per_second": 0.042
    }
  ],
  "logging_steps": 1,
  "max_steps": 3,
  "num_train_epochs": 1,
  "save_steps": 500,
  "total_flos": 519041285554176.0,
  "trial_name": null,
  "trial_params": null
}