Search papers, labs, and topics across Lattice.
Hong Kong University of Science and Technology
1
0
2
Tool-augmented models can achieve high accuracy without visual inputs, relying instead on structured text scaffolds to guide reasoning.
