Search papers, labs, and topics across Lattice.
2
0
6
MLLMs still struggle to reliably predict the long-term consequences of actions in egocentric videos, even with structured scene annotations.
Ditch the deep thought: this new agentic search framework slashes reasoning steps by 70% while boosting accuracy by prioritizing parallel evidence gathering.
