Search papers, labs, and topics across Lattice.
1
0
3
2
Image-generation models can outperform text-output VLMs in spatial tasks when evaluated through visual answers, revealing a critical shift in how we assess spatial intelligence in AI.
