This paper investigates how transformer-based language models retrieve information from long contexts, identifying special attention heads called retrieval heads as crucial for this task.

https://arxiv.org/abs//2404.15574

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