Junyi Wu 吴俊逸

I am a junior undergraduate student (2023.9 - Present) at the Departmen of EE., Shanghai Jiao Tong University (SJTU). I worked closely with Prof. Guohao Dai and Prof. Yulun Zhang. My research interests lie in LLM/DiT/dLLM models acceleration and compression through techniques such as quantization, cache mechanism and specific algorithmic methods.

Please feel free to reach out via email or WeChat if you would like to connect further.

News

Sep 02, 2025	Our paper BalanceGS is accepted by ASP-DAC 2026.
Jun 26, 2025	Our paper QuantCache is accepted by ICCV 2025.
Mar 11, 2025	1 paper is accepted by ISCA 2025.

Publications

^† Equal Contribution, ^* Corresponding Author(s)

ISCA

SpecEE: Accelerating Large Language Model Inference with Speculative Early Exiting

Jiaming Xu , Jiayi Pan , Yongkang Zhou , Siming Chen , Jinhao Li , Yaoxiu Lian , Junyi Wu , and Guohao Dai^*

In International Symposium on Computer Architecture, 2025

PDF Code
ICCV

QuantCache: Adaptive Importance-Guided Quantization with Hierarchical Latent and Layer Caching for Video Generation

Junyi Wu^† , Zhiteng Li^† , Zheng Hui , Yulun Zhang^* , Linghe Kong , and Xiaokang Yang

In International Conference on Computer Vision, 2025

PDF Code
ASP-DAC

BalanceGS: Adaptive Workload Optimization for High-Fidelity 3D Gaussian Splatting on GPU

Junyi Wu^† , Jiaming Xu^† , Jinhao Li , Yongkang Zhou , Jiayi Pan , Xingyang Li , and Guohao Dai^*

In 31st Asia and South Pacific Design Automation Conference, 2026

PDF
arxiv

DVD-Quant: Data-free Video Diffusion Transformers Quantization

Zhiteng Li^† , Hanxuan Li^† , Junyi Wu , Kai Liu , Linghe Kong^* , Guihai Chen , Yulun Zhang^* , and Xiaokang Yang

arXiv preprint arXiv:2505.18663, 2025

Under Review

PDF Code