eval_results.json

{
    "epoch": 1.9994743758212876,
    "eval_chosen_logps": -132.98548889160156,
    "eval_chosen_rewards": -0.4551008343696594,
    "eval_log_diff_policy": 6.635798931121826,
    "eval_logits": -2.239074468612671,
    "eval_logp_accuracy": 0.5631991028785706,
    "eval_loss": 44.87443923950195,
    "eval_objective": 44.855743408203125,
    "eval_rejected_logps": -139.6212921142578,
    "eval_rejected_rewards": -0.518205463886261,
    "eval_reward_accuracy": 0.5883668661117554,
    "eval_runtime": 489.9987,
    "eval_samples": 10722,
    "eval_samples_per_second": 21.882,
    "eval_steps_per_second": 0.912
}