trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.802407221664995,
  "eval_steps": 500,
  "global_step": 800,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.2,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.0915,
      "step": 200
    },
    {
      "epoch": 0.4,
      "learning_rate": 2.5e-05,
      "loss": 0.0484,
      "step": 400
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.25e-05,
      "loss": 0.0426,
      "step": 600
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0,
      "loss": 0.0399,
      "step": 800
    }
  ],
  "logging_steps": 200,
  "max_steps": 800,
  "num_train_epochs": 1,
  "save_steps": 200,
  "total_flos": 0.0,
  "trial_name": null,
  "trial_params": null
}