About

"Efficiency isn't about doing less — it's about understanding more."

Hi! My name is Junhan Zhu (诸俊涵). I am a 3rd year undergraduate student at Westlake University, majoring in Electrical and Information Engineering. My research interests include Computer Vision, Generative AI, and Efficient AI, especially in the area of model compression and acceleration.

I will share my research and personal thoughts on my blog, "Lonely Zone". Check it out here for my latest updates and thoughts! [Lonely Zone]

I am deeply grateful for the guidance and support I have received from my mentors, senior students, and friends.

I am always open to collaborations and discussions. Please feel free to contact me at zhujunhan@westlake.edu.cn.

You can attach my CV here: CV_JunhanZhu.pdf

I am actively seeking 27 Fall PhD opportunities!

News

Jan, 2026

🎉 [ICLR 2026] OBS-Diff has been accepted to ICLR 2026! 🇧🇷 A huge thanks to my supervisor, Prof. Huan Wang, and all co-authors. Let's talk about sparsity and efficiency in Brazil! NEW

Jan, 2026

✈️ [Exchange] Started my exchange semester at Nanyang Technological University (NTU) in Singapore!

Jan, 2026

🤗 [Release] OBS-Diff models are now available on HuggingFace! Achieving training-free, one-shot structured pruning for SDXL (10-30%) and SD3.5-Large (15-30%).

Dec, 2025

🚀 [Blog] My personal blog, "Lonely Zone", is launched! Check it out here: [Lonely Zone]!

Oct, 2025

🏆 [Award] Honored to receive the Outstanding Bachelor's Student award from Westlake University.

Oct, 2025

📑 [Paper] Our work, OBS-Diff, is released as a preprint. It is my first paper as a leading author! [arXiv] [code]

May, 2025

🔧 [Project] Released SparAlloc, a modular benchmark and toolkit for decoupled sparsity allocation in LLM pruning. [code]

Mar, 2025

🔬 [Visit] Started visiting at ENCODE Lab at Westlake University, supervised by Prof. Huan Wang.

Jul, 2023

🎉 [Admission] Admitted to Westlake University, skipping the final year of high school!

Education

Nanyang Technological University
Exchange Student
Jan. 2026 - Jun. 2026
Exchange student at Nanyang Technological University.

Research Interests: Computer Vision, Efficient AI, World Model

Westlake University
B.Eng in Electronic and Information Engineering
Jul. 2023 - Present (Expected 2027)
Admitted by skipping the final year of high school.

GPA: 4.04 / 4.3 Rank: 2 / 23 📄 Transcript

Research Interests: Computer Vision, Generative AI, Efficient AI

Experiences

Jan, 2026 - Present

Visiting Student @ MMLab@NTU

Supervisor: Associate Prof. Ziwei Liu

Mar, 2025 - Nov, 2025

Visiting Student @ ENCODE Lab

Supervisor: Assistant Prof. Huan Wang

Publication and Projects

I have just started my reserch journey, with research interests in the intersection of Computer Vision and Efficient AI.
You can check the summary of my research activities in this slides.
Here are some of my publications and projects:

OBS-Diff: Accurate Pruning For Diffusion Models in One-Shot

Junhan Zhu, Hesong Wang, Mingluo Su, Zefang Wang, Huan Wang^*

ICLR 2026

Project / Code / arXiv / 🤗 SDXL / 🤗 SD3.5-Large

Propose the first training-free, one-shot pruning framework for diffusion models, demonstrating broad applicability across diverse architectures and pruning granularities.

Cross-Resolution Diffusion Models via Network Pruning

Jiaxuan Ren^†, Junhan Zhu^†, Huan Wang^*

(^† Equal contribution, ^* Corresponding author)

Under Review

CR-Diff repurposes network pruning to enhance generalizability by removing "adverse weights" that cause degradation at non-default resolutions.

SparAlloc: A Simple and Modular Framework for Decoupled Sparsity Allocation in Layerwise Pruning for LLM

Junhan Zhu

Open Source Project

Code

A modular benchmark toolkit for decoupled sparsity allocation in LLM pruning. Enables fair comparison and flexible combination of pruning algorithms.

About

News

Education

Experiences

Publication and Projects

Selected Honors & Awards

Academic Services