Portrait of Xun Wu

Microsoft Research

Xun Wu (吴浔)

Email: xunwu [at] microsoft.com

I am a researcher at Microsoft Research. I am broadly interested in building frontier foundation models, with a particular focus on post-training for large language models (e.g., OPD and RL), mixture-of-experts, and low-bit language models (e.g., BitNet).

I received my master's degree in Computer Software Engineering from Tsinghua University in 2024.

Publications (selected)

2026

  1. ICLR
    zhao2026geometric.png
    Geometric-Mean Policy Optimization
    Yuzhong Zhao, Yue Liu, Junpeng Liu, Jingye Chen, Xun Wu, Yaru Hao, and 6 more authors
    In International Conference on Learning Representations, 2026
  2. arXiv
    ye2026onpolicycontext.png
    On-Policy Context Distillation for Language Models
    Tianzhu Ye, Li Dong, Xun Wu, Shaohan Huang, and Furu Wei
    arXiv preprint arXiv:2602.12275, 2026

2025

  1. NeurIPS
    jiang2025think.png
    Think Only When You Need with Large Hybrid-Reasoning Models
    Lingjie Jiang*, Xun Wu*, Shaohan Huang, Qingxiu Dong, Zewen Chi, Li Dong, and 4 more authors
    In Advances in Neural Information Processing Systems, 2025
  2. arXiv
    wu2025bitnet.png
    BitNet Distillation
    Xun Wu, Shaohan Huang, Wenhui Wang, Ting Song, Li Dong, Yan Xia, and 1 more author
    arXiv preprint arXiv:2510.13998, 2025

2024

  1. ICLR
    wu2024mixture.png
    Mixture of LoRA Experts
    Xun Wu, Shaohan Huang, and Furu Wei
    In International Conference on Learning Representations, 2024

Awards

Competition Awards

  • 2nd Place Mobile Intelligent Photography & Imaging Workshop CVPR 2023
  • 3rd Place Mobile AI for Learned Smartphone ISP Challenge ECCV 2022
  • 1st Place New Trends in Image Restoration and Enhancement Workshop CVPR 2022
  • Provincial Second Prize, 33rd Chinese Physics Olympiad