Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
Affiliation:
4
0
9
Diffusion models can learn new objectives by stochastically routing attention within their existing architecture, achieving 4x faster RL training without sacrificing generation quality.
Spatial memory via frustum slab routing slashes video world model inference time by over 4x without sacrificing generation quality.
System One models can achieve direct, low-latency decision-making by treating answers as structured choices rather than serialized text, with Jeva demonstrating robust null detection and permutation invariance.
HarnessEval-W transforms world model evaluation from mere scoring to a transparent reasoning process that mirrors human judgment.