Search papers, labs, and topics across Lattice.
University of California
1
0
2
Models trained on CoVA-SFT outperform existing multimodal reasoning baselines by over 2x, revealing the potential of structured visual-textual integration in AI.
