Search papers, labs, and topics across Lattice.
1
0
3
MLLMs can achieve 4x faster inference without sacrificing accuracy by intelligently focusing on only the image regions relevant to the query.