Search papers, labs, and topics across Lattice.
2
0
4
18
Performance gaps in multilingual medical evaluations reveal that proprietary models outperform open-source ones, but translation quality can swing results dramatically.
Current models struggle with long-range narrative integration and cultural reasoning, revealing critical gaps in their understanding of high-context media.
