Search papers, labs, and topics across Lattice.
3
0
7
OTRO reduces tokenizer latency to just 4.5% overhead while effectively safeguarding against side-channel attacks in LLMs.
Selective layer looping in masked diffusion language models unlocks surprising gains in both training efficiency and reasoning performance, beating naive depth scaling.
Forget hand-labeled data: a new agent-based system automatically generates artifact-annotated datasets, enabling VLMs and diffusion models to better understand and fix visual flaws.
