```So how do we get GPT4V to “see” what’s in the box? We give it back the previous result’s HTML, together with a note to “fill in” the white box with this provided HTML.
And it works.```, Why is this feasible? It looks really incredible. how does chatGPT-4v know which object the arrow point at?it's just a blank white rectangle && original html code? I don't understand, could you explain this? thanks in advance.
I made a test, using an arrow to point at an area(used a rectangle to mark this area), then gpt-4v generated the new page with a wrong area modification.
I think your current method is not very solid. @linearuncle
any development for llama2?
```So how do we get GPT4V to “see” what’s in the box? We give it back the previous result’s HTML, together with a note to “fill in” the white box with this provided HTML.
And it works.```, Why is this feasible? It looks really incredible. how does chatGPT-4v know which object the arrow point at?it's just a blank white rectangle && original html code? I don't understand, could you explain this? thanks in advance.
I made a test, using an arrow to point at an area(used a rectangle to mark this area), then gpt-4v generated the new page with a wrong area modification.
I think your current method is not very solid. @linearuncle
Game... Changer... 🏆
Interesting.
Hey buddy! can we talk?
http://meetings.jux.io/meetings/amashiah
You have exceeded your available quota, please check
your plan and billing details, giving error
Love this Steve amazing work.