About Me 👋
Hi, I’m Ying Li, a postdoctoral researcher at Westlake University working on Efficient AI. My goal is to make LLMs, MLLMs, and generative models — and the agentic systems built on them — more efficient and effective: faster to run, lighter to deploy, and better at the tasks that matter.
My research spans inference acceleration (speculative & dynamic decoding), model compression (pruning, efficient MoE), efficient generative models (visual autoregressive generation), agents & world models, and AI for Science.
Research Interests 🔍
“Being able to compress well is closely related to intelligence.” — Marcus Hutter (2005), Hutter Prize
“The success of machine learning algorithms generally depends on data representation.” — Bengio, Courville & Vincent (2013), Representation Learning: A Review and New Perspectives
- Efficient AI — speculative & dynamic decoding, model compression (pruning / efficient MoE), inference acceleration.
- Generative Models — visual autoregressive generation, fast / parallel decoding, efficient architecture conversion.
- Agents & World Models — efficient inference & planning for agents and world models.
- AI for Science — data-efficient modeling for biomedicine & molecular design.
News 💬
- [Jul, 2026] ✈️ Attending ICML 2026 in Seoul — presenting two posters and attending the ByteDance ICML banquet.
- [Jun, 2026] 🎉 I had two papers provisionally accepted by ECCV 2026: my first-author LISA (locality-informed speculative decoding) and EVAR (edge VAR via principled pruning) — congrats to Zefang Wang.
- [Jun, 2026] ✈️ Attending CVPR 2026.
- [Jun, 2026] 🏆 I was honored with the CVPR 2026 Compute Transparency Champion award.
- [May, 2026] 📄 Released preprint RankE — end-to-end post-training for discrete text-to-image generation with decoder co-evolution — congrats to Siyong Jian.
- [May, 2026] 🏆 I was recognized as a Gold Reviewer at ICML 2026.
- [May, 2026] 🎉 I had two first-author papers accepted by ICML 2026: Prism-MoE and ARC-Decode.
- [Feb, 2026] 🎉 Parallel Jacobi Decoding is accepted by CVPR 2026 — congrats to Boya Liao.
- [Dec, 2025] ✈️ Attending NeurIPS 2025 — presenting FreqExit at the poster session.
- [Sep, 2025] 🎉 I had my first-author paper FreqExit accepted by NeurIPS 2025.
- [Jan, 2025] 🎉 I had my co-first-author paper Intelligent Design of Lipid Nanoparticles for Enhanced Gene Therapeutics published in Molecular Pharmaceutics and selected as Editor's Choice (<1%) — congrats to Yichen Yuan.
- [Jan, 2025] 🎉 Joined Westlake University, Encode Lab, as a Postdoctoral Researcher.
- [Aug, 2024] 🎉 I had my first-author paper LEMTL accepted by BIBM 2024.
- [Dec, 2022] 💼 Joined Zhejiang Lab, working on Intelligent Computing and AI for Science.
- [Dec, 2022] 🎓 Received my Ph.D. from Shanghai Jiao Tong University.
- [Jun, 2017] 🎓 Received my M.Eng. from Dalian University of Technology.
Experiences 📝
| Apr 2025 – Present | Research Intern, Alibaba (Taotian) |
| Jan 2025 – Present |
Postdoctoral Researcher, Westlake University (Encode Lab) Advisor: Prof. Huan Wang |
| Dec 2022 – 2024 |
Researcher, Zhejiang Lab Intelligent Computing / AI for Science |
| Sep 2017 – Dec 2022 |
Ph.D., Shanghai Jiao Tong University Advisor: Prof. Xuewu Cao |
| M.Eng., Jun 2017 | Dalian University of Technology |
Publications 📚
(Full list on Google Scholar. Selected recent papers below. ✉️ = corresponding author, 🌟 = co-first author.)
[Arxiv] RankE: End-to-End Post-Training for Discrete Text-to-Image Generation with Decoder Co-Evolution
Siyong Jian, Siyuan Li, Luyuan Zhang, Zedong Wang, Xin Jin, Ying Li, Cheng Tan, Huan Wang✉️
Preprint, 2026
[Paper]
[Code]
[Paper] EVAR: Edge Visual Autoregressive Models via Principled Pruning
Zefang Wang, Ying Li, Yanyu Li, Mingluo Su, Simin Xu, Guanzhong Tian, Huan Wang✉️
ECCV, 2026 (provisionally accepted)
[Paper] LISA: Locality-Informed Speculative Decoding for Accelerating Autoregressive Image Generation
Ying Li, Siyong Jian, Zhaode Wang, Zhiwen Chen, Chengfei Lv, Huan Wang✉️
ECCV, 2026 (provisionally accepted)
[Paper] ARC-Decode: Accelerated Decoding with Risk-Bounded Acceptance
Ying Li, Zhaode Wang, Zhiwen Chen, Chengfei Lv, Huan Wang✉️
ICML, 2026
[Paper]
[Code]
[Website]
[Paper] Prism-MoE: Efficient Dense-to-MoE Conversion for Visual Autoregressive Generation
Ying Li, Zefang Wang, Zhaode Wang, Zhiwen Chen, Chengfei Lv, Huan Wang✉️
ICML, 2026
[Paper]
[Code]
[Website]
[Paper] Parallel Jacobi Decoding for Fast Autoregressive Image Generation
Boya Liao, Ying Li, Siyong Jian, Huan Wang✉️
CVPR, 2026
[Project]
[Code]
[Paper] FreqExit: Enabling Early-Exit Inference for Visual Autoregressive Models via Frequency-Aware Guidance
Ying Li, Chengfei Lv, Huan Wang✉️
NeurIPS, 2025
[Paper]
[Code]
[Journal] Intelligent Design of Lipid Nanoparticles for Enhanced Gene Therapeutics
Yichen Yuan🌟, Ying Li🌟, Guo Li🌟, Liqun Lei, Xingxu Huang, Ming Li, Yuan Yao✉️
Molecular Pharmaceutics, 2025 (Editor's Choice, <1%)
[Paper]
[Paper] LEMTL: Enhancing the pharmacokinetic predictions of multitask learning with existing pharmacokinetics
Ying Li, Xiao Deng, Jingsong Lv, Hongyang Chen, Yao Yang✉️
BIBM, 2024
[Paper]
Academic Service 🤝
- Conference Reviewer: ICML (2026), NeurIPS (2026)