Search papers, labs, and topics across Lattice.
3
0
7
5
LEGO-RL boosts coding agent performance by up to 8.4% while ensuring robust training signals and execution reliability.
UBEP slashes All-to-All latency by over 52%, unlocking the full potential of Mixture-of-Experts models on superpod architectures.
User-defined response prefixes in LLMs are a major safety risk, enabling CoT attacks to achieve near-perfect success rates on some models.
