Skip to content

Comment on Focus and Context and LLMs

Comments

This is definitely the right problem to focus on. I think the answer is a different LLM structure that has unlimited context. The transformer with causal masks for training block got us here but they are now limiting us in many massive ways.

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.