Search papers, labs, and topics across Lattice.
8
0
12
2
The effectiveness of on-policy distillation hinges not on the size of the teacher model, but on the quality of the guiding signal, revealing critical pathologies that can undermine exploration.
JITOMA slashes active graph size and latency, ensuring robots can efficiently navigate long-horizon tasks without perceptual overload.
DW-FM reveals that aligning generative model training with decision-sensitive outcomes can drastically reduce decision regret in stochastic optimization tasks.
Action-Grounded Representation Alignment (AGRA) transforms how robots interpret visual data, enabling them to focus on crucial interaction regions and significantly enhancing manipulation performance.
State-dependent action constraints can be effectively managed in DRL, leading to near-optimal solutions in complex queueing networks.
Capturing structured relationships in images can boost open-vocabulary object detection performance by over 10% on novel categories.
LMs can now selectively abstain from answering with provable guarantees, thanks to a new method that uses representation geometry to better gauge when they're out of their depth.
Google's AI Overviews favor Google-owned content and penalize sites blocking its AI crawler, raising serious questions about fairness and bias in the emerging generative search landscape.
