Search papers, labs, and topics across Lattice.
1
0
3
Current vision-language benchmarks miss the mark: AMIGO reveals how hard it is for agents to ground visual information across multiple images and turns.