{
    "epoch": 1.0,
    "eval_logits/chosen": -2.808791399002075,
    "eval_logits/rejected": -2.8062777519226074,
    "eval_logps/chosen": -31.203502655029297,
    "eval_logps/rejected": -34.66593933105469,
    "eval_loss": 0.629190981388092,
    "eval_rewards/accuracies": 0.5166113376617432,
    "eval_rewards/chosen": 0.05526338145136833,
    "eval_rewards/margins": 0.03271109238266945,
    "eval_rewards/rejected": 0.022552290931344032,
    "eval_runtime": 112.7274,
    "eval_samples": 343,
    "eval_samples_per_second": 3.043,
    "eval_steps_per_second": 0.381,
    "train_loss": 1.2163990609057538,
    "train_runtime": 2720.9807,
    "train_samples": 3079,
    "train_samples_per_second": 1.132,
    "train_steps_per_second": 0.141
}