{
    "epoch": 0.999244142101285,
    "eval_epsilon_dpo/beta": 0.009046695195138454,
    "eval_epsilon_dpo/beta_margin_grad_mean": -0.4353906810283661,
    "eval_epsilon_dpo/beta_margin_grad_std": 0.1097957193851471,
    "eval_epsilon_dpo/beta_margin_mean": 0.27432820200920105,
    "eval_epsilon_dpo/beta_margin_std": 0.46992063522338867,
    "eval_epsilon_dpo/loss_margin_mean": 30.660152435302734,
    "eval_kl/n_epsilon_steps": 0.2830105721950531,
    "eval_kl/p_epsilon_steps": 0.7161091566085815,
    "eval_logits/chosen": -0.20607627928256989,
    "eval_logits/rejected": -0.3409919738769531,
    "eval_logps/chosen": -133.02056884765625,
    "eval_logps/ref_chosen": -87.42715454101562,
    "eval_logps/ref_rejected": -104.23548889160156,
    "eval_logps/rejected": -180.4890594482422,
    "eval_loss": 0.5927965641021729,
    "eval_rewards/accuracies": 0.7235915660858154,
    "eval_rewards/chosen": -0.4138784408569336,
    "eval_rewards/margins": 0.27432820200920105,
    "eval_rewards/rejected": -0.6882066130638123,
    "eval_runtime": 47.5074,
    "eval_samples": 2303,
    "eval_samples_per_second": 48.477,
    "eval_steps_per_second": 1.516
}