Search papers, labs, and topics across Lattice.
3
0
6
18
Forget vague goals – robots learn to *really* listen when you tell them *how* to do things, boosting real-world manipulation success by 25%.
Forget hand-crafted benchmarks: CUA-Gym's auto-generated training data lets computer-use agents crush existing open-source models on real-world tasks.
Despite impressive OCR performance on existing benchmarks, today's best LMMs still struggle with the messy realities of enterprise document processing.
