Yunsheng Li is a Member of Technical Staff at Microsoft AI. His research interests include large language models (post-training, reinforcement learning, and agents for SWE and security) and multi-modal large language models. His representative works include MAI-Code-1, Phi-4-mini, Phi-3-vision, MicroNet and BDL.

Projects

• 2025-present: MAI-Code-1 and MAI-Cyber-1, the in-house models at Microsoft AI for SWE and cybersecurity tasks.
• 2024-2025: Phi-4-mini and Phi-4-mini-reasoning, small language models with multi-modal and reasoning capabilities.
• 2023-2024: Phi-3-vision and Phi-3.5-vision, small multi-modal language models.
• 2022-2023: The background removal models used in Windows Paint and Photos and Microsoft Designer.
• 2015-2021: Ph.D. at University of California, San Diego, on efficient neural network architecture design and domain adaptation.

Publications

1. Improving Code Localization with Repository Memory
   Boshi Wang, Weijian Xu, Yunsheng Li, Mei Gao, Yujia Xie, Huan Sun, Dongdong Chen
   International Conference on Learning Representations (ICLR), 2026
   [Code]
2. RubricRL: Simple Generalizable Rewards for Text-to-Image Generation
   Xuelu Feng, Yunsheng Li, Ziyu Wan, Zixuan Gao, Junsong Yuan, Dongdong Chen, Chunming Qiao
   [arXiv]
3. Benchmarking Large and Small MLLMs
   Xuelu Feng, Yunsheng Li, Dongdong Chen, Mei Gao, Mengchen Liu, Junsong Yuan, Chunming Qiao
   Machine Vision and Applications (MVA), 2025
4. Pluralistic Salient Object Detection
   Xuelu Feng, Yunsheng Li, Dongdong Chen, Chunming Qiao, Junsong Yuan, Lu Yuan, Gang Hua
   IEEE Transactions on Image Processing (TIP), 2025
5. Phi-4-Mini-Reasoning: Exploring the Limits of Small Reasoning Language Models in Math
   Haoran Xu, Baolin Peng, Hany Awadalla, Dongdong Chen, Yen-Chun Chen, Mei Gao, Young Jin Kim, Yunsheng Li, Liliang Ren, Yelong Shen, Shuohang Wang, Weijian Xu, Jianfeng Gao, Weizhu Chen
   [arXiv]
6. Show and Segment: Universal Medical Image Segmentation via In-Context Learning
   Yunhe Gao, Di Liu, Zhuowei Li, Yunsheng Li, Dongdong Chen, Mu Zhou, Dimitris N. Metaxas
   Conference on Computer Vision and Pattern Recognition (CVPR), 2025
   [arXiv]
7. Olympus: A Universal Task Router for Computer Vision Tasks
   Yuanze Lin, Yunsheng Li, Dongdong Chen, Weijian Xu, Ronald Clark, Philip H. S. Torr
   Conference on Computer Vision and Pattern Recognition (CVPR), 2025 (Highlight)
   [arXiv]
8. Phi-4-Mini Technical Report: Compact yet Powerful Multimodal Language Models via Mixture-of-LoRAs
   Microsoft Phi Team
   [arXiv] [HuggingFace]
9. ToolBridge: An Open-Source Dataset to Equip LLMs with External Tool Capabilities
   Zhenchao Jin, Mengchen Liu, Dongdong Chen, Lingting Zhu, Yunsheng Li, Lequan Yu
   [arXiv]
10. SynChart: Synthesizing Charts from Language Models
    Mengchen Liu, Qixiu Li, Dongdong Chen, Dong Chen, Jianmin Bao, Yunsheng Li
    [arXiv]
11. Rethinking Visual Prompting for Multimodal Large Language Models with External Knowledge
    Yuanze Lin, Yunsheng Li, Dongdong Chen, Weijian Xu, Ronald Clark, Philip Torr, Lu Yuan
    [arXiv]
12. Fully Authentic Visual Question Answering Dataset from Online Communities
    Chongyan Chen, Mengchen Liu, Noel Codella, Yunsheng Li, Lu Yuan, Danna Gurari
    European Conference on Computer Vision (ECCV), 2024
    [arXiv] [Website]
13. Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone
    Microsoft Phi Team
    [arXiv] [HuggingFace]
14. SCHEME: Scalable Channel Mixer for Vision Transformers
    Deepak Sridhar, Yunsheng Li, Nuno Vasconcelos
    [arXiv]
15. Dense Network Expansion for Class Incremental Learning
    Zhiyuan Hu, Yunsheng Li, Jiancheng Lyu, Dashan Gao, Nuno Vasconcelos
    Conference on Computer Vision and Pattern Recognition (CVPR), 2023
    [Paper]
16. Should All Proposals Be Treated Equally in Object Detection?
    Yunsheng Li, Yinpeng Chen, Xiyang Dai, Dongdong Chen, Mengchen Liu, Pei Yu, Ying Jin, Lu Yuan, Zicheng Liu, Nuno Vasconcelos
    European Conference on Computer Vision (ECCV), 2022
    [arXiv] [Code]
17. MicroNet: Improving Image Recognition with Extremely Low FLOPs
    Yunsheng Li, Yinpeng Chen, Xiyang Dai, Dongdong Chen, Mengchen Liu, Lu Yuan, Zicheng Liu, Lei Zhang, Nuno Vasconcelos
    International Conference on Computer Vision (ICCV), 2021
    [arXiv] [Website] [Code]
18. Dynamic Transfer for Multi-Source Domain Adaptation
    Yunsheng Li, Lu Yuan, Yinpeng Chen, Pei Wang, Nuno Vasconcelos
    Conference on Computer Vision and Pattern Recognition (CVPR), 2021
    [arXiv] [Website] [Code]
19. Revisiting Dynamic Convolution via Matrix Decomposition
    Yunsheng Li, Yinpeng Chen, Xiyang Dai, Mengchen Liu, Dongdong Chen, Ye Yu, Lu Yuan, Zicheng Liu, Mei Chen, Nuno Vasconcelos
    International Conference on Learning Representations (ICLR), 2021
    [arXiv] [Website] [Code]
20. Deep Hashing with Hash-Consistent Large Margin Proxy Embeddings
    Pedro Morgado, Yunsheng Li, Jose Costa Pereira, Mohammad Saberian, Nuno Vasconcelos
    International Journal of Computer Vision (IJCV), 2021
    [arXiv]
21. Explainable Object-Induced Action Decision for Autonomous Vehicles
    Yiran Xu, Xiaoyin Yang, Lihang Gong, Hsuan-Chu Lin, Tz-Ying Wu, Yunsheng Li, Nuno Vasconcelos
    Conference on Computer Vision and Pattern Recognition (CVPR), 2020
    [arXiv] [Website]
22. Semantic Fisher Scores for Task Transfer: Using Objects to Classify Scenes
    Mandar Dixit, Yunsheng Li, Nuno Vasconcelos
    IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), 2020
    [arXiv]
23. Bidirectional Learning for Domain Adaptation of Semantic Segmentation
    Yunsheng Li, Lu Yuan, Nuno Vasconcelos
    Conference on Computer Vision and Pattern Recognition (CVPR), 2019
    [arXiv] [Website] [Code]
24. Efficient Multi-Domain Learning by Covariance Normalization
    Yunsheng Li, Nuno Vasconcelos
    Conference on Computer Vision and Pattern Recognition (CVPR), 2019
    [Paper] [Website]
25. Deep Scene Image Classification with the MFAFVNet
    Yunsheng Li, Mandar Dixit, Nuno Vasconcelos
    International Conference on Computer Vision (ICCV), 2017
    [Paper] [Website]

Patents

• Dynamic Matrix Convolution with Channel Fusion, US 12,223,412 B2, granted 2025.
• Runtime Ranking of Object Detection, US 2025/0252731 A1, published 2025.
• Pluralistic Salient Object Detection, EP 4 597 442 A1, published 2025.

Professional Activities

• Conference reviewer: CVPR, ICCV, ECCV, ICLR, et al.
• Journal reviewer: T-PAMI, T-MM, IJCV, et al.