all_results.json

{
    "epoch": 0.9981298423724285,
    "eval_logits/chosen": 0.409419447183609,
    "eval_logits/rejected": 0.41578754782676697,
    "eval_logps/chosen": -239.2622528076172,
    "eval_logps/rejected": -259.08453369140625,
    "eval_loss": 0.6081043481826782,
    "eval_rewards/accuracies": 0.6129032373428345,
    "eval_rewards/chosen": -1.5096983909606934,
    "eval_rewards/margins": 0.21358628571033478,
    "eval_rewards/rejected": -1.7232848405838013,
    "eval_runtime": 177.9012,
    "eval_samples_per_second": 11.023,
    "eval_steps_per_second": 0.174
}