Want to create an interactive transcript for this episode?
Podcast: Arxiv Papers
Episode: [QA] Rewarding the Unlikely: Lifting GRPO Beyond Distribution Sharpening
Description: This paper critiques GRPO's bias in training language models for theorem proving and introduces the unlikeliness reward to enhance performance and sample diversity, achieving competitive results.https://arxiv.org/abs//2506.02355YouTube: https://www.youtube.com/@ArxivPapersTikTok: https://www.tiktok.com/@arxiv_papersApple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1692476016Spotify: https://podcasters.spotify.com/pod/show/arxiv-papers