Are Dense Rewards Worth Their Tokens in GRPO?

(arxiv.org)

1 points | by vikashjohn2505 4 hours ago ago

2 comments

  • Benard-dev 3 hours ago ago

    Do dense rewards in GRPO actually save compute if they need more token scoring per rollout?

  • vikashjohn2505 4 hours ago ago

    [flagged]