Track new results across pure and applied mathematics.
Search papers, labs, and topics across Lattice.
University of Oxford, Independent Researcher
4
0
9
LLM agents can now dynamically design their own post-training recipe spaces, leading to a 2.6x performance boost on complex multi-domain tasks.
Conversational avatars no longer require latency-heavy cascades: generating speech and full-body gesture end-to-end from shared LLM hidden states runs 5.4× faster than two-pass pipelines while matching cascaded teacher fidelity.
GameWAM reveals that a unified approach to world dynamics and action generation can significantly enhance performance in video game agents, achieving more with less.