trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 10.0,
  "global_step": 2290,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 2.18,
      "learning_rate": 3.9082969432314415e-05,
      "loss": 6.5963,
      "step": 500
    },
    {
      "epoch": 4.37,
      "learning_rate": 2.816593886462882e-05,
      "loss": 3.1634,
      "step": 1000
    },
    {
      "epoch": 6.55,
      "learning_rate": 1.7248908296943234e-05,
      "loss": 1.3969,
      "step": 1500
    },
    {
      "epoch": 8.73,
      "learning_rate": 6.3318777292576415e-06,
      "loss": 0.8354,
      "step": 2000
    },
    {
      "epoch": 10.0,
      "step": 2290,
      "total_flos": 1.0781642018832384e+16,
      "train_loss": 2.697846571743228,
      "train_runtime": 1376.5937,
      "train_samples_per_second": 16.613,
      "train_steps_per_second": 1.664
    }
  ],
  "max_steps": 2290,
  "num_train_epochs": 10,
  "total_flos": 1.0781642018832384e+16,
  "trial_name": null,
  "trial_params": null
}