Search papers, labs, and topics across Lattice.
1
0
3
Forget end-to-end video understanding: RieMind shows that explicitly grounding LLMs in 3D scene graphs unlocks a 16% jump in spatial reasoning, suggesting structured representations are the key.
