Search papers, labs, and topics across Lattice.
1
0
3
5
Unified multimodal models can ace visual understanding and generation tasks, yet still fail to maintain basic semantic consistency between them.
