The paper introduces Length Controlled Policy Optimization (LCPO) for training reasoning models, enabling controlled output length and improved performance, outperforming existing methods while allowing for efficient compute allocation.

https://arxiv.org/abs//2503.04697

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