You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We present a novel video-based visual-spatial intelligence benchmark (VSI-Bench) of over 5,000 question-answer pairs, and find that MLLMs exhibit competitive - though subhuman - visual-spatial intelligence.
The text was updated successfully, but these errors were encountered:
https://vision-x-nyu.github.io/thinking-in-space.github.io/
The text was updated successfully, but these errors were encountered: