The paper introduces BTLM-3B-8K, a new state-of-the-art 3 billion parameter language model that outperforms existing models and provides excellent long context performance. It is compact, requiring less memory and compute, making it accessible for mobile and edge devices.

https://arxiv.org/abs//2309.11568

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