The paper introduces InfAlign, an inference-aware alignment framework that optimizes language model performance during inference, outperforming existing methods by 8-12% on helpfulness and harmlessness benchmarks.

https://arxiv.org/abs//2412.19792

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