Selected Publications
CForce: Boosting Parallel Decoding for dLLMs via Consistency Forcing
* Equal contribution.
LoPA: Scaling dLLM Inference via Lookahead Parallel Decoding
* Equal contribution.
Diffusion LLMs Can Do Faster-Than-AR Inference via Discrete Diffusion Forcing
* Equal contribution.
UniCMs: A Unified Consistency Model For Efficient Multimodal Generation and Understanding
* Equal contribution.
Multi-Block Diffusion Language Models
Technical Stack
