eval_results.json

{
    "epoch": 0.9969104016477858,
    "eval_logits/chosen": -1.0279620885849,
    "eval_logits/rejected": -1.1372919082641602,
    "eval_logps/chosen": -335.9989929199219,
    "eval_logps/rejected": -419.17950439453125,
    "eval_loss": 0.6490800976753235,
    "eval_rewards/accuracies": 0.5625,
    "eval_rewards/chosen": -0.18141408264636993,
    "eval_rewards/margins": 0.044088974595069885,
    "eval_rewards/rejected": -0.22550305724143982,
    "eval_runtime": 7.6515,
    "eval_samples": 1000,
    "eval_samples_per_second": 130.694,
    "eval_steps_per_second": 1.046
}