all_results.json

{
    "epoch": 3.0,
    "eval_logits/chosen": -0.8499579429626465,
    "eval_logits/rejected": -0.8532128930091858,
    "eval_logps/chosen": -167.44473266601562,
    "eval_logps/rejected": -209.41464233398438,
    "eval_loss": 2.6467697620391846,
    "eval_rewards/accuracies": 0.6425532102584839,
    "eval_rewards/chosen": 14.379803657531738,
    "eval_rewards/margins": 1.6944859027862549,
    "eval_rewards/rejected": 12.685317993164062,
    "eval_runtime": 1054.998,
    "eval_samples_per_second": 0.668,
    "eval_steps_per_second": 0.668,
    "train_loss": 0.5555986505650652,
    "train_runtime": 48641.0087,
    "train_samples_per_second": 0.296,
    "train_steps_per_second": 0.037
}