all_results.json

{
    "epoch": 1.971563981042654,
    "eval_logits/chosen": 105.5069808959961,
    "eval_logits/rejected": 99.0961685180664,
    "eval_logps/chosen": -374.1521911621094,
    "eval_logps/rejected": -382.3683166503906,
    "eval_loss": 3.341306686401367,
    "eval_rewards/accuracies": 0.6666666865348816,
    "eval_rewards/chosen": -0.5241988301277161,
    "eval_rewards/margins": 0.49066951870918274,
    "eval_rewards/rejected": -1.0148683786392212,
    "eval_runtime": 119.4596,
    "eval_samples": 750,
    "eval_samples_per_second": 6.278,
    "eval_steps_per_second": 0.201,
    "total_flos": 0.0,
    "train_loss": 2.9736096469255595,
    "train_runtime": 2166.7366,
    "train_samples": 6750,
    "train_samples_per_second": 6.231,
    "train_steps_per_second": 0.048
}