←back to thread

Karpathy on DeepSeek-OCR paper: Are pixels better inputs to LLMs than text?

(twitter.com)
237 points JnBrymn | 1 comments | 21 Oct 25 17:43 UTC | HN request time: 0.312s | source
1. cnxhk ◴[22 Oct 25 23:54 UTC] No.45676635[source]▶
The paper is quite interesting but efficiency on OCR tasks does not mean it could be plugged into a general llm directly without performance loss. If you train a tokenizer only on OCR text you might be able to get better compression already.