Search papers, labs, and topics across Lattice.
4
0
8
6
State-of-the-art agents lose over 45% of their performance the moment analytical reasoning requires directional constraint handoffs between SQL databases and open-web search.
VLA-Precision, an efficient real-world online RL framework featuring the Asymmetric Co-Bootstrapping (ACoB) algorithm and the ACoB-Stream architecture, and ACoB-Stream, a closed-loop experience--policy architecture that establishes invariant-state decoupling and on-demand streaming as design principles, deliver up to 10.9% improvements in throughput and computational efficiency.
Semantically distinct memory objects in coding agents reveal that equal token budgets can mislead about context delivery and management costs.
Forget slow, multi-step action generation: CF-VLA's coarse-to-fine approach slashes latency by 75% while boosting real-robot success rates to a new high of 83%.
