Search papers, labs, and topics across Lattice.
The Chinese University of Hong Kong, Shanghai AI Laboratory
1
0
3
Image editing can boost multimodal LLM reasoning by 5 points, but only if the editor is trained to understand the reasoning process itself.
