Search papers, labs, and topics across Lattice.
1
0
2
A model-free benchmark reveals that vision-language models often misread rather than misreason, exposing a performance gap that a supervised vision model can surpass.
