trainer_state.json

{
  "best_metric": 0.09600851684808731,
  "best_model_checkpoint": "safety_instruct_detect/checkpoint-686",
  "epoch": 2.0,
  "global_step": 686,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_accuracy": 0.9537414965986395,
      "eval_loss": 0.14035563170909882,
      "eval_runtime": 35.4642,
      "eval_samples_per_second": 145.076,
      "eval_steps_per_second": 4.145,
      "step": 343
    },
    {
      "epoch": 1.46,
      "learning_rate": 5.451895043731778e-06,
      "loss": 0.2205,
      "step": 500
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9706511175898931,
      "eval_loss": 0.09600851684808731,
      "eval_runtime": 35.524,
      "eval_samples_per_second": 144.832,
      "eval_steps_per_second": 4.138,
      "step": 686
    }
  ],
  "max_steps": 686,
  "num_train_epochs": 2,
  "total_flos": 1291799489699544.0,
  "trial_name": null,
  "trial_params": null
}