Search papers, labs, and topics across Lattice.
NII LLMC
1
0
3
4
Training VLMs on Jagle, the largest Japanese multimodal dataset, not only crushes existing models on Japanese tasks, but *also* boosts English performance when combined with English data.
