Mingze Wang

PhD student, Peking University

Joined

April 2022

Names

Mingze Wang (Preferred)

Emails

****@stu.pku.edu.cn (Confirmed)

Personal Links

Homepage

Google Scholar

DBLP

Career & Education History

PhD student

Peking University (pku.edu.cn)

2021 – Present

Undergrad student

Zhejiang University (zju.edu.cn)

2017 – 2021

Advisors, Relations & Conflicts

PhD Advisor

Weinan E

Present

Expertise

machine learning

2021 – Present

optimization

2021 – Present

Publications

Accelerating LLM Pre-Training through Flat-Direction Dynamics Enhancement
Shuchen Zhu, Rizhen Hu, Mingze Wang, Mou Sun, Xue Wang, Kun Yuan, Zaiwen Wen
- arxiv
- Readers: Everyone
How Transformers Get Rich: Approximation and Dynamics Analysis
Mingze Wang, Ruoxi Yu, Weinan E, Lei Wu
- ICLR 2026 Conference Withdrawn Submission
- Readers: Everyone
Fast Catch-Up, Late Switching: Optimal Batch Size Scheduling via Functional Scaling Laws
Jinbo Wang, Binghui Li, Zhanpeng Zhou, Mingze Wang, Yuxuan Sun, Jiaqi Zhang, Xunliang Cai, Lei Wu
- ICLR 2026 Poster
- Readers: Everyone
GradPower: Powering Gradients for Faster Language Model Pre-Training
Mingze Wang, Jinbo Wang, Jiaqi Zhang, Peng Pei, Wei Wang, Xunliang Cai, Weinan E, Lei Wu
- Submitted to ICLR 2026
- Readers: Everyone
Low-probability Tokens Sustain Exploration in Reinforcement Learning with Verifiable Reward
Guanhua Huang, Tingqiang Xu, Mingze Wang, Qi Yi, Xue Gong, Siheng Li, Ruibin Xiong, Kejiao Li, Yuhao Jiang, Bo Zhou
- ICLR 2026 Conference Withdrawn Submission
- Readers: Everyone
On The Surprising Effectiveness of a Single Global Merging in Decentralized Learning
Tongtian Zhu, Tianyu Zhang, Mingze Wang, Zhanpeng Zhou, Can Wang
- ICLR 2026 Oral
- Readers: Everyone
On the Learning Dynamics of Two-layer Linear Networks with Label Noise SGD
Tongcheng Zhang, Zhanpeng Zhou, Mingze Wang, Andi Han, Wei Huang, Taiji Suzuki, Junchi Yan
- HiLD at ICML 2025 Poster
- Readers: Everyone
How Transformers Get Rich: Training Dynamics Analysis
Mingze Wang, Ruoxi Yu, Weinan E, Lei Wu
- HiLD at ICML 2025 Poster
- Readers: Everyone
On the Expressive Power of Mixture-of-Experts for Structured Complex Tasks
Mingze Wang, Weinan E
- NeurIPS 2025 spotlight
- Readers: Everyone
The Sharpness Disparity Principle in Transformers for Accelerating Language Model Pre-Training
Jinbo Wang, Mingze Wang, Zhanpeng Zhou, Junchi Yan, Weinan E, Lei Wu
- CoRR 2025
- Readers: Everyone

View all 46 publications

Co-Authors

View all 52 co-authors