eval_results.json

{
    "epoch": 0.9999333733093477,
    "eval_logits/chosen": -1.3400534391403198,
    "eval_logits/rejected": -1.3050051927566528,
    "eval_logps/chosen": -281.1357727050781,
    "eval_logps/rejected": -299.5000915527344,
    "eval_loss": 0.6583243012428284,
    "eval_rewards/accuracies": 0.5955284833908081,
    "eval_rewards/chosen": -1.2959448099136353,
    "eval_rewards/margins": 0.1955154538154602,
    "eval_rewards/rejected": -1.4914603233337402,
    "eval_runtime": 246.7004,
    "eval_samples": 1962,
    "eval_samples_per_second": 7.953,
    "eval_steps_per_second": 0.997
}