I am a second-year master student at Department of Electronic Engineering and Information Science, University of Science and Technology of China (USTC). My research interest is in trustworthy AI, particularly theory of generative models including diffusion models and large language models, and their applications in optimizing algorithms.
HRP: High-Rank Preheating for Superior LoRA Initialization.arXiv preprint arXiv:2502.07739. (2025).
TL;DR: We theoretically illustrate the importance of LoRA initialization and propose a method for superior LoRA initialization.
Yuzhu Chen, Fengxiang He, Shi Fu, Xinmei Tian, and Dacheng Tao.Adaptive Time-Stepping Schedules for Diffusion Models.Conference on Uncertainty in Artificial Intelligence (UAI), 2024.
TL;DR: We propose adaptive time-stepping schedules for diffusion models by minimizing a series of convergence bounds.
Shi Fu, Yuzhu Chen, Yingjie Wang, and Dacheng Tao.On Championing Foundation Models: From Explainability to Interpretability.arXiv preprint arXiv:2410.11444 (2024).
TL;DR: We survey interpretability analysis of foundation models, including inference capability and training dynamics to their ethical implications.