Hao Mark Chen

AdaBlock-dLLM: Semantic-Aware Diffusion LLM Inference via Adaptive Block Size

Rethinking Optimal Verification Granularity for Compute-Efficient Test-Time Scaling

Progressive Mixed-Precision Decoding for Efficient LLM Inference

Hardware-Aware Parallel Prompt Decoding for Memory-Efficient Acceleration of LLM Inference