[umutisik]

And very likely this is coming to computer vision too with multi-modal GPT.

[thewataccount]

I've already had some success with Llava although existing image classifiers are still going to outperform it.