Search papers, labs, and topics across Lattice.
1
0
3
VLMs can't count blocks because they lack a view-consistent spatial interface, but decomposing scenes into orthographic projections fixes it.
