trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 92,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0869565217391304,
      "grad_norm": 9.991344451904297,
      "learning_rate": 2.595756834225089e-06,
      "logits/chosen": 18.9567813873291,
      "logits/rejected": 21.65990447998047,
      "logps/chosen": -88.46855163574219,
      "logps/rejected": -192.53219604492188,
      "loss": 0.2648,
      "rewards/accuracies": 0.8654292225837708,
      "rewards/chosen": 1.6632243394851685,
      "rewards/margins": 2.6804497241973877,
      "rewards/rejected": -1.0172255039215088,
      "step": 50
    }
  ],
  "logging_steps": 50,
  "max_steps": 92,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}