Search papers, labs, and topics across Lattice.
1
0
2
Native multimodal models outperform structured vision-language pipelines in Change VQA, challenging assumptions about the benefits of model size.
