trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.1616842105263158,
  "eval_steps": 500,
  "global_step": 3,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.05,
      "learning_rate": 1.9848077530122083e-05,
      "loss": 11.2151,
      "step": 1
    },
    {
      "epoch": 0.16,
      "eval_loss": 10.685684204101562,
      "eval_runtime": 8.5482,
      "eval_samples_per_second": 58.492,
      "eval_steps_per_second": 7.37,
      "step": 3
    },
    {
      "epoch": 0.16,
      "step": 3,
      "total_flos": 5.890428031624806e+16,
      "train_loss": 11.974308331807455,
      "train_runtime": 561.6093,
      "train_samples_per_second": 16.916,
      "train_steps_per_second": 0.032
    }
  ],
  "logging_steps": 5,
  "max_steps": 18,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "total_flos": 5.890428031624806e+16,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}