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": 3.874887488748875e-05,
      "loss": 3.6766,
      "step": 500
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.7497749774977498e-05,
      "loss": 2.2606,
      "step": 1000
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.6246624662466247e-05,
      "loss": 2.0216,
      "step": 1500
    },
    {
      "epoch": 1.8,
      "learning_rate": 4.995499549954996e-06,
      "loss": 1.9092,
      "step": 2000
    },
    {
      "epoch": 2.0,
      "step": 2222,
      "train_runtime": 2399.7982,
      "train_samples_per_second": 0.926
    }
  ],
  "max_steps": 2222,
  "num_train_epochs": 2,
  "total_flos": 234317752276746240,
  "trial_name": null,
  "trial_params": null
}