{ "epoch": 3.0, "eval_logits/chosen": 3.015625, "eval_logits/rejected": 4.75, "eval_logps/chosen": -1784.0, "eval_logps/rejected": -1928.0, "eval_loss": 1.839203119277954, "eval_rewards/accuracies": 0.595703125, "eval_rewards/chosen": -14.6875, "eval_rewards/margins": 1.71875, "eval_rewards/rejected": -16.375, "eval_runtime": 46.9056, "eval_samples": 2000, "eval_samples_per_second": 42.639, "eval_steps_per_second": 0.682 }