About Me

I am a Ph.D. student at University of Maryland, College Park (UMD), advised by Prof. Ang Li. Before beginning my journey at UMD, I obtained my bachelor’s degree at the School of Cyber Science and Engineering in Sichuan University, where I worked under the guidance of Prof. Mingjie Tang. I am also glad to have the chance as a visiting student at University of Illinois, Urbana Champaign under the guidance of Prof. Huan Zhang.

My research interests focus on identifying and addressing challenges in large language models (LLMs) and vision-language models (VLMs), spanning reasoning, training, and model-level issues, while also exploring paradigms for learning, adaptation, and autonomous intelligence.

Interests
  • Large Language Model
  • Vision Language Model
  • Trustworthy Machine Learning
Education
  • Ph.D. Student, 2025 - Present

    University of Maryland, College Park

  • Visiting Student, 2024 - 2025

    University of Illinois, Urbana Champaign

  • B.E. in Computer Engineering, 2021 - 2025

    Sichuan University

Recent News

  • [2026/09/24] One paper is accepted to NeurIPS 2026.
  • [2025/02/28] Our paper BadVision is accepted to CVPR 2025.
  • [2025/02/27] One paper is accepted to ICLR 2025.
  • [2024/05/02] Our paper BadPart is accepted to ICML 2024.

Recent Publications

Quickly discover relevant content by filtering publications.
(2026). Spectral-Sphere-Constrained Hyper-Connections. In Conference on Neural Information Processing Systems (NeurIPS 2026).

PDF Cite Code

(2026). TrustGen: A Platform of Dynamic Benchmarking on the Trustworthiness of Generative Foundation Models. In International Conference on Learning Representations (ICLR 2026).

PDF Cite Project

(2025). Stealthy Backdoor Attack in Self-supervised Vision Encoders for Large Vision Language Models. In Conference on Computer Vision and Pattern Recognition (CVPR 2025).

PDF Cite Code

(2025). Interleaved Scene Graph for Interleaved Text-and-Image Generation Assessment. In International Conference on Learning Representations (ICLR 2025, Spotlight).

PDF Cite Project

(2024). BadPart: Unified Black-box Adversarial Patch Attacks against Pixel-wise Regression Tasks. In International Conference on Machine Learning (ICML 2024).

PDF Cite Code

Honors & Awards

  • Dean’s Fellowship, University of Maryland, 2025