Search papers, labs, and topics across Lattice.
1
0
3
Current multi-modal LLMs struggle with the messy, real-world visual data captured by wearable devices, achieving only 24-52% accuracy on the new WearVQA benchmark.
