Search papers, labs, and topics across Lattice.
6
0
10
7
HarnessEval-W transforms world model evaluation from mere scoring to a transparent reasoning process that mirrors human judgment.
DSWAM bridges the gap between coarse user commands and fine-grained robot actions, outperforming traditional models in real-world task execution.
BiPACE transforms credit assignment in LLM training, boosting validation success rates by over 6% without the need for critics or extra rollouts.
Existing video world models struggle with long-term memory retention, and MBench exposes their critical limitations while providing a structured path for future improvements.
LLMs waste up to 88% of their chain-of-thought tokens generating *after* they've already internally determined the answer.
Ditch the linear CFG gains: Sliding Mode Control offers provably stable and semantically richer diffusion guidance, especially when you crank up the guidance scale.
