No, it does not work like that, it actually can process the image itself there i... | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

weird-eye-issue 70 days ago | parent | context | favorite | on: I used Claude Code to get a second opinion on my M...

No, it does not work like that, it actually can process the image itself there is not an intermediate image to text step

voidUpdate 70 days ago [–]

How does a Large Language Model process images then?

jappgar 70 days ago | [–]

It can only deal in tokens, so you're essentially right that it creates a textual description before describing it back to you. This process is obviously incredibly lossy and details are easily missed

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact