RSS

2PuNCheeZ

Karma: 12

[Question] Do mul­ti­modal LLMs (like 4o) use OCR un­der the hood to read dense text in images?

2PuNCheeZ15 Jun 2025 11:20 UTC
4 points
1 comment1 min readLW link

François Chol­let on the limi­ta­tions of LLMs in reasoning

2PuNCheeZ30 Jul 2024 20:04 UTC
1 point
1 comment2 min readLW link
(x.com)