trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "global_step": 2222,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.45,
      "learning_rate": 7.74977497749775e-05,
      "loss": 3.0294,
      "step": 500
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.4995499549954995e-05,
      "loss": 1.7905,
      "step": 1000
    },
    {
      "epoch": 1.35,
      "learning_rate": 3.2493249324932494e-05,
      "loss": 1.5222,
      "step": 1500
    },
    {
      "epoch": 1.8,
      "learning_rate": 9.990999099909991e-06,
      "loss": 1.3883,
      "step": 2000
    },
    {
      "epoch": 2.0,
      "step": 2222,
      "train_runtime": 2105.7691,
      "train_samples_per_second": 1.055
    }
  ],
  "max_steps": 2222,
  "num_train_epochs": 2,
  "total_flos": 161621779478999040,
  "trial_name": null,
  "trial_params": null
}