Search papers, labs, and topics across Lattice.
Zhejiang University
1
0
3
Even the best vision-language models struggle with predictive spatial reasoning, scoring less than 60% on a new benchmark designed to test anticipating unseen outcomes.
