Search papers, labs, and topics across Lattice.
1
0
3
2
Even the best MLLMs still fall far short of human performance in collaborative spatial reasoning, revealing a critical gap in their ability to build and maintain shared mental models through dialogue.
