RetNet is proposed as a foundation architecture for large language models, achieving training parallelism, low-cost inference, and good performance. It supports three computation paradigms and shows favorable scaling results, making it a strong successor to Transformer.

https://arxiv.org/abs//2307.08621

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