trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "global_step": 1373,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.36,
      "learning_rate": 2.8437246963562755e-05,
      "loss": 1.1212,
      "step": 500
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.2242914979757085e-05,
      "loss": 0.5928,
      "step": 1000
    },
    {
      "epoch": 1.0,
      "step": 1373,
      "total_flos": 6.849180050433638e+16,
      "train_loss": 0.76789391414145,
      "train_runtime": 1113.5246,
      "train_samples_per_second": 78.913,
      "train_steps_per_second": 1.233
    }
  ],
  "max_steps": 1373,
  "num_train_epochs": 1,
  "total_flos": 6.849180050433638e+16,
  "trial_name": null,
  "trial_params": null
}