Yuan Zhang Ph.D. Student Center on Frontiers of Computing Studies, Peking University Email: zhangyuan [at] alumni.pku.edu.cn

Biography

His major research interests lie within efficient machine learning algorithms, like

• Knowledge Distillation (KD)
• Efficient Large Models (MLLM)

Education

Peking University (PKU), Beijing, China PhD Student at Center on Frontiers of Computing Studies. Sep. 2024 – Jun. 2028 Research: Efficiency for Multimodal Large Language Model (MLLM) Research: Knowledge Distillation

Peking University (PKU), Beijing, China Master Degree. Sep. 2020 – Jun. 2023 Rank: 3/110 National Scholarship of China

Hohai University (HHU), Nanjing, China Bachelor Degree. Sep. 2016 – Jun. 2020 Rank: 1/112 National Scholarship of China

Publications

• CCF-AUnveiling the Tapestry of Consistency in Large Vision-Language Models
  Yuan Zhang, Fei Xiao, Tao Huang, Chunkai Fan, H Dong, J Li, J Wang, Kuan Cheng, Shanghang Zhang, Haoyuan Guo
  Advances in Neural Information Processing Systems (NeurIPS), 2024.

  PDF Bib Code Video

• CCF-AFreeKD: Knowledge Distillation via Semantic Frequency Prompt
  Yuan Zhang, Tao Huang, Jiaming Liu, Tao Jiang, Kuan Cheng, Shanghang Zhang
  IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2024.

  PDF Bib Code

• CCF-AAvatar Knowledge Distillation: Self-ensemble Teacher Paradigm with Uncertainty
  Yuan Zhang, Weihua Chen*, Yichen Lu*, Tao Huang, Xiuyu Sun, Jian Cao
  Proceedings of the 31th ACM International Conference on Multimedia (ACM MM), 2023.

  PDF Bib Code Video

• CCF-AMasked Distillation with Receptive Tokens
  Tao Huang*, Yuan Zhang*, Shan You, Fei Wang, Chen Qian, Jian Cao, Chang Xu
  International Conference on Learning Representations (ICLR), 2023.

  PDF Bib Code Video

• CCF-AKnowledge Diffusion for Distillation
  Tao Huang, Yuan Zhang, Mingkai Zheng, Shan You, Fei Wang, Chen Qian, Chang Xu
  Advances in Neural Information Processing Systems (NeurIPS), 2023.

  PDF Bib Code

• CCF-ACloud-Device Collaborative Learning for Multimodal Large Language Models
  Guanqun Wang*, Jiaming Liu*, Chenxuan Li*, Yuan Zhang, Ray Zhang, Yijiang Liu, Shanghang Zhang
  IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2024.

  PDF Bib

• CCF-BA free lunch from ViT: Adaptive Attention Multi-scale Fusion Transformer for Fine-grained Visual Recognition
  Yuan Zhang, Jian Cao, Ling Zhang, Xiangcheng Liu, Feng Ling, Weiqian Chen
  IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2022.

  PDF Bib Video

• CCF-CRazor SNN: Efficient Spiking Neural Network with Temporal Embeddings
  Yuan Zhang, Jian Cao, Ling Zhang, Jue Chen, Wenyu Sun, Yuan Wang
  32nd International Conference on Artificial Neural Networks (ICANN), 2023. [Oral]

  PDF Bib Code Video

Manuscripts

• arXivDAMO-YOLO: A Report on Real-Time Object Detection Design
  Xianzhe Xu*, Yiqi Jiang*, Weihua Chen*, Yilun Huang*, Yuan Zhang*, Xiuyu Sun
  arXiv preprint arXiv:2211.15444 (2022).

  PDF Bib Code Video

• arXivSparseVLM: Visual Token Sparsification for Efficient Vision-Language Model Inference
  Yuan Zhang*, ChunKai Fan*, Junpeng Ma*, Wenzhao Zheng, Tao Huang, Kuan Cheng, Denis Gudovskiy, Tomoyuki Okuno, Yohei Nakata, Kurt Keutzer, Shanghang Zhang
  arXiv preprint arXiv:2410.04417 (2024).

  PDF Bib Code

Research Experience

Berkeley Artificial Intelligence Research Lab (BAIR), California, USA Research Intern (Advisor: Wenzhao Zheng). Sep. 2024 – Now Vision Foundation Model. Efficiency in LVLMs.

ByteDance Seed Foundation Model, Beijing, China Researcher (Advisor: Haoyuan Guo). Jul. 2023 – Jul. 2024 Vision Foundation Model. Consistency in LVLMs.

Alibaba DAMO Academy, Beijing, China Research Intern (Advisor: Weihua Chen). Jun. 2022 – Feb. 2023 Research on knowledge distillation (KD) and propose AvatarKD. Distill for DAMO-YOLO.

SenseTime Research, Beijing, China Research Intern (Advisor: Tao Huang). Mar. 2022 – Jun. 2022 Research on knowledge distillation (KD) and propose MasKD. Develop on MMRazor.

Academic Services

• CVPR 2024 | ECCV 2024 | ACM MM 2024 | NeurIPS 2024
• ICLR 2025 | AISTATS 2025 | CVPR 2025

Open Source Experience

• YOLOv5-Multibackbone-Compression
  
  Duty: Owner.

  Oct. 2021 – Mar. 2022
  Description: YOLOv5 Series Multi-backbone(TPH-YOLOv5, Ghostnet, ShuffleNetv2, Mobilenetv3Small, EfficientNetLite, PP-LCNet, SwinTransformer YOLO), Module(CBAM, DCN), Pruning (EagleEye, Network Slimming), Quantization (MQBench) and Deployment (TensorRT, ncnn) Compression Tool Box.
  
• DAMO-YOLO
  
  Duty: Distill and Deployment.

  Jun. 2022 – Sep. 2022
  Description: First make distillation great again on all size models of YOLO series (1.0 AP+), especially on the small size.
  

Awards and Honor

• Excellent Graduate of Peking University
  
  Top 10% in school.

  June 2023
• National Scholarship of China
  
  Master period (¥ 20,000).

  Nov. 2021
• National Scholarship of China
  
  Undergraduate period (¥ 8,000).

  Nov. 2018

Visitors