trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.986078886310905,
  "eval_steps": 500,
  "global_step": 214,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.86,
      "learning_rate": 3.2710280373831774e-06,
      "loss": 6.5469,
      "step": 200
    },
    {
      "epoch": 1.99,
      "step": 214,
      "total_flos": 2829332565196800.0,
      "train_loss": 6.479159524507612,
      "train_runtime": 865.9454,
      "train_samples_per_second": 1.991,
      "train_steps_per_second": 0.247
    }
  ],
  "logging_steps": 200,
  "max_steps": 214,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 2000,
  "total_flos": 2829332565196800.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}