photo

Email: lez023 at ucsd dot edu

Lei (Max) Zhang

Lei (Max) Zhang

I am a first-year (2024 - ) Ph.D. student at University of California, San Diego supervised by Prof. Julian McAuley and closely collaborate with Prof. Zhijian Liu. Previously, I completed my master degree at Zhejiang University in 2024 and bachelor degree at South China University of Technology in 2021.

I like building multi-modal models, focusing on multi-modal understanding, generation, and alignment, all with an emphasis on efficiency, data, and visual tokenizer.

News 🎉

  • Sep. 2024 - Start Ph.D. journey at University of California, San Diego.
  • Aug. 2023 - Fair Network Pruning gets accepted by ICCV 2023.
  • Mar. 2023 - Efficient Dataset Distillation gets accepted by CVPR 2023 as Highlight.

Publications

Full Publications: Google Scholar

  • LLaVA-MoD: Making LLaVA Tiny via MoE Knowledge Distillation

    Fangxun Shu*, Yue Liao*, Le Zhuo*, Chenning Xu*, Lei Zhang*, Guanghao Zhang*, Haonan Shi*,

    Long Chen, Tao Zhong, Wanggui He, Siming Fu, Haoyuan Li, Si Liu, Hongsheng Li
    Tech report, 2024

    arXiv
  • Audio-Visual LLM for Video Understanding

    Fangxun Shu*, Lei Zhang*, Hao Jiang, Cihang Xie
    Tech report, 2023

    arXiv
  • Filter & Align: Leveraging Human Knowledge to Curate Image-Text Data

    Lei Zhang, Fangxun Shu, Tianyang Liu, Sucheng Ren, Hao Jiang, Cihang Xie
    Tech report, 2024

    arXiv
  • Towards Fairness-aware Adversarial Network Pruning

    Lei Zhang, Zhibo Wang, Xiaowei Dong, Yunhe Feng, Xiaoyi Pang, Zhifei Zhang, Kui Ren
    ICCV, 2023

    camera-ready
  • Accelerating Dataset Distillation via Model Augmentation

    Lei Zhang, Jie Zhang, Bowen Lei, Subhabrata Mukherjee, Xiang Pan, Bo Zhao, Caiwen Ding, Li

    Yao, Dongkuan Xu
    CVPR, 2023, Highlight

    arXiv / camera-ready

Interests

Multi-modal undestanding, generation, and alignment on efficient training and inference, data recepie: quality vs. quantity, visual tokenizer, etc.

Experiences

Services

Reviewer for CVPR, ECCV, ICCV, ICLR, ICML, and AISTATS.