all_results.json

{
    "epoch": 2.0,
    "eval_logits/chosen": -3.0438718795776367,
    "eval_logits/rejected": -3.0380868911743164,
    "eval_logps/chosen": -64.8961181640625,
    "eval_logps/rejected": -71.05843353271484,
    "eval_loss": 0.6855542659759521,
    "eval_rewards/accuracies": 0.5954925417900085,
    "eval_rewards/chosen": -0.06184223294258118,
    "eval_rewards/margins": 0.016940835863351822,
    "eval_rewards/rejected": -0.0787830725312233,
    "eval_runtime": 381.8015,
    "eval_samples": 4304,
    "eval_samples_per_second": 11.273,
    "eval_steps_per_second": 1.409,
    "total_flos": 0.0,
    "train_loss": 0.6830481951767456,
    "train_runtime": 69306.1943,
    "train_samples": 92858,
    "train_samples_per_second": 2.68,
    "train_steps_per_second": 0.084
}