Search papers, labs, and topics across Lattice.
1
0
3
Video-LLMs can spot a shape-shifting object, but still fail at basic physics, revealing a critical gap in predictive world modeling that HOCA-Bench exposes.