Search papers, labs, and topics across Lattice.
1
0
3
2
LLMs and VLMs can encode viewpoint, but they still hallucinate observations when reasoning about viewpoint rotation, revealing a critical gap in their spatial intelligence.
