RT by @lugaricano: Translation: Auto-Regressive LLMs scaling is giving diminishing returns.
As I've said repeatedly, a new architecture will emerge for the next leap, perhaps along the lines of the Objective-Driven AI I've been proposing.
But deep learning will still be the foundation. No wall.
🐦🔗: https://nitter.cz/ylecun/status/1725411082855182681#m
[2023-11-17 07:10 UTC]