all_results.json

{
    "epoch": 2.8115942028985508,
    "eval_logits/chosen": 0.28515625,
    "eval_logits/rejected": 0.3828125,
    "eval_logps/chosen": -172.0,
    "eval_logps/rejected": -184.0,
    "eval_loss": 0.620069146156311,
    "eval_rewards/accuracies": 1.0,
    "eval_rewards/chosen": 0.01470947265625,
    "eval_rewards/margins": 0.1650390625,
    "eval_rewards/rejected": -0.150390625,
    "eval_runtime": 5.7268,
    "eval_samples_per_second": 42.606,
    "eval_steps_per_second": 1.397
}