Skip to content

Comment on Hacker News, Sans AIparent

Comments

It's not an LLM.

Why use big program when regex do trick?

You can get the best of both worlds with a small embedding model

Show me an embedding model that runs as fast as an optimized regex engine and I'll buy you a beer.

Probably not as fast as a simple regex but static embedding models can get stupid fast e.g https://www.flowercomputer.com/news/fast-static-embedding/

Master of efficiency

AboutSource Built by g1lg1l

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