GPT4V evaluation prompt

BradyFU / Awesome-Multimodal-Large-Language-Models

:sparkles::sparkles:Latest Advances on Multimodal Large Language Models

10.88k stars 721 forks source link

GPT4V evaluation prompt #131

Closed LengSicong closed 4 months ago

LengSicong commented 4 months ago

Hi authors,

Congrats on this great work! However, the results we have reproduced from GPT4V are quite different from those reported in the leaderboard. May I ask you to kindly provide the prompt that you are using to evaluate GPT4V?

BradyFU commented 4 months ago

Hi Sicong, thanks for your attention. We only use the same prompt in the MME. GPT4V has many versions on different times.