{
    "epoch": 1.0,
    "eval_logits/chosen": -2.9391283988952637,
    "eval_logits/rejected": -2.2059857845306396,
    "eval_logps/chosen": -168.49281311035156,
    "eval_logps/rejected": -193.87171936035156,
    "eval_loss": 0.6908513903617859,
    "eval_rewards/accuracies": 0.5625,
    "eval_rewards/chosen": 0.00629099877551198,
    "eval_rewards/margins": 0.0006325023132376373,
    "eval_rewards/rejected": 0.0056584966368973255,
    "eval_runtime": 20.5327,
    "eval_samples": 231,
    "eval_samples_per_second": 11.25,
    "eval_steps_per_second": 0.39,
    "train_loss": 0.680875283606509,
    "train_runtime": 1044.3353,
    "train_samples": 6013,
    "train_samples_per_second": 5.758,
    "train_steps_per_second": 0.09
}