Search papers, labs, and topics across Lattice.
University of California, Los Angeles
1
0
2
10
VLMs achieve superior reasoning performance when their thoughts are explicitly tied to visual evidence, outperforming larger models in some tasks.
