Process supervision is shown to significantly outperform outcome supervision in training language models to solve complex reasoning problems, as demonstrated on the MATH dataset. Active learning is also shown to improve the efficacy of process supervision.

https://arxiv.org/abs//2305.20050

YouTube: https://www.youtube.com/@ArxivPapers

TikTok: https://www.tiktok.com/@arxiv_papers

Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1692476016

Spotify: https://podcasters.spotify.com/pod/show/arxiv-papers