trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 8.658008658008658,
  "eval_steps": 500,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 2.16,
      "learning_rate": 7.835497835497836e-06,
      "loss": 0.0,
      "step": 500
    },
    {
      "epoch": 4.33,
      "learning_rate": 5.670995670995671e-06,
      "loss": 0.0,
      "step": 1000
    },
    {
      "epoch": 6.49,
      "learning_rate": 3.506493506493507e-06,
      "loss": 0.0,
      "step": 1500
    },
    {
      "epoch": 8.66,
      "learning_rate": 1.3419913419913422e-06,
      "loss": 0.0,
      "step": 2000
    }
  ],
  "logging_steps": 500,
  "max_steps": 2310,
  "num_train_epochs": 10,
  "save_steps": 500,
  "total_flos": 2.4948275527365427e+17,
  "trial_name": null,
  "trial_params": null
}