Skip to content

Comment on OpenAI o1 Results on ARC-AGI-Pubparent

Comments

So basically it's a kind of overfitting with pattern matching features? This doesn't undermine the power of LLMs but it is great to study their limitations.

AboutSource Built by g1lg1l

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