Search papers, labs, and topics across Lattice.
4
0
5
7
LLM agents can now build lasting, reusable "competence" for specific knowledge sources, moving beyond simple retrieval to deep, adaptive understanding.
HEART achieves 84% task completion on real-world tasks, outperforming leading models while slashing API costs by up to 85%.
Latent reasoning can now be validated in real-time, with a novel method that boosts accuracy by over 16 points on critical benchmarks.
Ditch the fixed interface: DiffMAS unlocks surprisingly large gains in multi-agent reasoning by jointly optimizing latent communication, outperforming text-based and prior latent methods by a wide margin.
