Selected work

Research

I work on efficient inference for diffusion and audio language models, as well as robotic ultrasound. Selected projects are grouped below by topic.

01 Diffusion LLM inference

Sampling, prompting, and verification methods for faster generation.

02 Audio LLM efficiency

Long-context memory and practical KV-cache management.

03 Robotic ultrasound

Navigation and probe guidance for medical robotics.

01

Diffusion language models

Diffusion LLM inference

Work on sampling, prompting, and self-speculative decoding for faster diffusion language-model inference.

ICLR 2026 · Accepted · CCF-A

SlowFast Sampling

Switches between exploratory and accelerated stages using token certainty, convergence, and position; 15.63× speedup on LLaDA.

ACL 2026 · Submitted

Thinking inside the Mask

Integrates intermediate reasoning into the mask-denoising process through in-place prompting.

ACL 2026 · Submitted

Self-Speculative Decoding

Uses self-drafting and parallel verification to accelerate diffusion language-model inference.

02

Audio and multimodal LLMs

Audio LLM memory

Work on preserving important acoustic and linguistic context while reducing memory use.

ACM MM 2026 · Accepted · First author

AudioKV

Uses semantic-acoustic alignment and Spectral Score Smoothing to retain important KV-cache entries; 60% lower memory overhead.

ACL 2026 · Submitted

AudioMarathon

A benchmark for long-context audio understanding and efficiency.

03

Embodied medical intelligence

Robotic ultrasound

Work on autonomous navigation and real-time probe guidance for ultrasonography.

ICRA 2026 · Accepted · CCF-B · Equal contribution

UltraHiT

A hierarchical Transformer for generalizable internal-carotid-artery robotic ultrasonography.

MICCAI 2026 · Accepted · CCF-B

VA-Adapter

A parameter-efficient vision-action adapter for real-time echocardiography probe guidance.