trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 10.0,
  "global_step": 2030,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 2.46,
      "learning_rate": 3.768472906403941e-05,
      "loss": 6.5048,
      "step": 500
    },
    {
      "epoch": 4.93,
      "learning_rate": 2.5369458128078822e-05,
      "loss": 3.552,
      "step": 1000
    },
    {
      "epoch": 7.39,
      "learning_rate": 1.3054187192118228e-05,
      "loss": 3.34,
      "step": 1500
    },
    {
      "epoch": 9.85,
      "learning_rate": 7.389162561576355e-07,
      "loss": 3.238,
      "step": 2000
    },
    {
      "epoch": 10.0,
      "step": 2030,
      "total_flos": 6865519834275840.0,
      "train_loss": 4.144926990076826,
      "train_runtime": 1387.4174,
      "train_samples_per_second": 14.603,
      "train_steps_per_second": 1.463
    }
  ],
  "max_steps": 2030,
  "num_train_epochs": 10,
  "total_flos": 6865519834275840.0,
  "trial_name": null,
  "trial_params": null
}