Mingjia Shi

PhD student, University of Virginia

4 papers at NeurIPS 2025

OpenReview· Semantic Scholar· Google Scholar

Poster Session 2

2 papers

Wednesday, December 3, 2025 · 4:30 PM → 7:30 PM

Exhibit Hall C,D,E

Tackling Feature-Classifier Mismatch in Federated Learning via Prompt-Driven Feature Transformation

#2810 · Xinghao Wu, Xuefeng Liu, Jianwei Niu, Guogang Zhu, Mingjia Shi, Shaojie Tang, Jing Yuan

REPA Works Until It Doesn’t: Early-Stopped, Holistic Alignment Supercharges Diffusion Training

#3614 · Ziqiao Wang, Wangbo Zhao, Yuhao Zhou, Zekai Li, Zhiyuan Liang, Mingjia Shi, Xuanlei Zhao, Pengfei Zhou, Kaipeng Zhang, Zhangyang Wang, Kai Wang, Yang You

We propose HASTE, which combines holistic alignment (feature and attention) with early termination to accelerate diffusion transformer training by 28× while maintaining quality.

Poster Session 6

2 papers

Friday, December 5, 2025 · 4:30 PM → 7:30 PM

Exhibit Hall C,D,E

Drag-and-Drop LLMs: Zero-Shot Prompt-to-Weights

#2604 · Zhiyuan Liang, Dongwen Tang, Yuhao Zhou, Xuanlei Zhao, Mingjia Shi, Wangbo Zhao, Zekai Li, Peihao Wang, Konstantin Schürholt, Damian Borth, Michael M. Bronstein, Yang You, Zhangyang Wang, Kai Wang

Generate high-performing LoRA parameters from prompts that is unseen in training.

Pruning-Robust Mamba with Asymmetric Multi-Scale Scanning Paths

#3809 · Jindi Lv, Yuhao Zhou, Mingjia Shi, Zhiyuan Liang, Panpan Zhang, Xiaojiang Peng, Wangbo Zhao, Zheng Zhu, Jiancheng Lv, Qing Ye, Kai Wang