trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.9976019184652278,
  "global_step": 208,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.96,
      "learning_rate": 4.9500000000000004e-05,
      "loss": 14.7256,
      "step": 100
    },
    {
      "epoch": 1.92,
      "learning_rate": 9.95e-05,
      "loss": 4.5709,
      "step": 200
    },
    {
      "epoch": 2.0,
      "step": 208,
      "total_flos": 9.397835469173249e+17,
      "train_loss": 9.413687669313871,
      "train_runtime": 890.0468,
      "train_samples_per_second": 7.496,
      "train_steps_per_second": 0.234
    }
  ],
  "max_steps": 208,
  "num_train_epochs": 2,
  "total_flos": 9.397835469173249e+17,
  "trial_name": null,
  "trial_params": null
}