Search papers, labs, and topics across Lattice.
Queen Mary University of London
1
0
3
4
Current video Q&A benchmarks can be fooled by textual regularities, failing to actually ground reasoning in the video's physical reality.
