Large language models (LLMs) can improve their accuracy on tasks by refining and revising their output based on feedback. The SCREWS framework enables exploration in this space by providing modules for sampling, conditional resampling, and selection, leading to improved reasoning strategies for various tasks.

https://arxiv.org/abs//2309.13075

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

PODCASTS:

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

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