Search papers, labs, and topics across Lattice.
1
0
2
MMCS achieves superior visual grounding with just 50K samples, outperforming traditional models trained on 600K image-text pairs.
