My name is Qiyuan Chen (陈启源) and I am currently a second-year Ph.D. student at the College of Computer Science and Technology, Zhejiang University, supervised by Prof. Jian Wu. Before that, I got my B.Sc. degree from the School of Mathematics and Statistics, Central China Normal University, under the guidance of Prof. Bo Li and Dr. Haitong Yang.

My current research interests primarily include: Machine Learning (ML) and Natural Language Processing (NLP). Specifically,

Semi-Supervised Learning and Long-Tailed Learning
Tabular Data Prediction and Synthesis
Preference Optimization in LLMs/VLMs

Feel free to contact me if your research lies within these or related areas!

And I work really closely with Jiahe Chen and Huawei Fan.

In my spare time, I love doing full-stack development, usually with FastAPI and Next.js. I also frequently participate in hackathons to build fun and interesting projects. Feel free to check out my work on GitHub!

I also enjoy digital devices (mechanical keyboards, NAS, etc.), badminton, running and chess in my spare time.

🔥 News

2026.04, One paper on Vision-Language Reward Modeling is accepted by ACL 2026 Main. See you in San Diego!
2026.03, Poco has reached over 1k stars on GitHub! Welcome to use it and provide suggestions.
2026.02, One paper on Long-Tailed Semi-Supervised Learning is accepted by ICLR 2026. Congratulations to Qian Shao!
2025.08, One paper on Synthetic Preference Data is accepted by EMNLP 2025 Main. See you in Soochow!

📝 Publications

(*: Equal contribution; $\dagger$: Corresponding author(s))

2026

Learning What Matters: Dynamic Dimension Selection and Aggregation for Interpretable Vision-Language Reward Modeling [NLP]; Qiyuan Chen, Hongsen Huang, Jiahe Chen, Qian Shao, Jintai Chen, Hongxia Xu, Renjie Hua, Ren Chuan, Jian Wu; ACL; 2026.
CoLA: Co-Calibrated Logit Adjustment for Long-Tailed Semi-Supervised Learning [ML]; Qian Shao, Qiyuan Chen*, Jiahe Chen, Zepeng Li, Qianqian Tang, Hongxia Xu, Jian Wu; ICLR; 2026. [CCF A; TH-CPL A]
DREAM: Distribution-aware Re-sampling with Equiangular Alignment Mechanism for Long-Tailed Semi-Supervised Learning [ML]; Qian Shao, Jiahe Chen, Qiyuan Chen*, Qianqian Tang, Jintai Chen, Hongxia Xu, Jiangrui Kang, Jian Wu; ICME; 2026. [CCF B]

2025

Icon $^2$: Aligning Large Language Models Using Self-Synthetic Preference Data via Inherent Regulation [NLP]; Qiyuan Chen, Hongsen Huang, Qian Shao, Jiahe Chen, Jintai Chen, Hongxia Xu, Renjie Hua, Ren Chuan, Jian Wu; EMNLP; 2025. [CCF B; TH-CPL A]
Alleviating Hallucination in Large Vision-Language Models with Active Retrieval Augmentation [NLP]; Xiaoye Qu, Qiyuan Chen*, Wei Wei, Jiashuo Sun, Daizong Liu, Yu Cheng; ACM TOMM; 2025. [CCF B; TH-CPL B]
AI Approaches for Predicting Progression to Acute Coronary Syndrome among Stable Coronary Heart Disease Patients [ML]; Haozhong Ma, Hexiang Bai, Jiahuan Yan, Qiyuan Chen, Zihan Ma, Shuo Tong, Yuxuan Zhan, Ruijia Wu, Hongxia Xu, Jian Wu; npj Cardiovascular Health; 2025.
CC-GSEO-Bench: A Content-Centric Benchmark for Measuring Source Influence in Generative Search Engines [NLP]; Qiyuan Chen, Jiahe Chen, Hongsen Huang, Qian Shao, Jintai Chen, Renjie Hua, Hongxia Xu, Ruijia Wu, Ren Chuan, Jian Wu; arXiv; 2025.
MM-DADM: Multimodal Drug-Aware Diffusion Model for Virtual Clinical Trials; [ML]; Qian Shao, Bang Du, Zepeng Li, Qiyuan Chen, Jiahe Chen, Hongxia Xu, Jimeng Sun, Jian Wu, Jintai Chen; arXiv; 2025.
Curing Semantic Drift: A Dynamic Approach to Grounding Generation in Large Vision-Language Models [NLP]; Jiahe Chen, Jiaying He, Qiyuan Chen, Qian Shao, Jiahe Ying, Hongxia Xu, Jintai Chen, Jianwei Zheng, Jian Wu; arXiv; 2025.

2024

Enhancing Semi-Supervised Learning via Representative and Diverse Sample Selection [ML]; Qian Shao, Jiangrui Kang, Qiyuan Chen*, Zepeng Li, Hongxia Xu, Yiwen Cao, Jiajuan Liang, Jian Wu; NIPS; 2024. [CCF A; TH-CPL A]
Cross-Table Pretraining towards a Universal Function Space for Heterogeneous Tabular Data [ML]; Jintai Chen, Zhen Lin, Qiyuan Chen, Jimeng Sun; arXiv; 2024.
Take Your Steps: Hierarchically Efficient Pulmonary Disease Screening via CT Compression [ML]; Qian Shao, Kai Zhang, Bang Du, Zepeng Li, Yixuan Wu, Qiyuan Chen, Jian Wu, Jintai Chen; KDD-AIDSH; 2024. [KDD Workshop]
Can a Deep Learning Model be a Sure Bet for Tabular Prediction? [ML]; Jintai Chen, Jiahuan Yan, Qiyuan Chen, Danny Chen, Jian Wu, Jimeng Sun; KDD; 2024. [CCF A; TH-CPL A]
Mind’s Mirror: Distilling Self-Evaluation Capability and Comprehensive Thinking from Large Language Models [NLP]; Weize Liu, Guocong Li, Kai Zhang, Bang Du, Qiyuan Chen, Xuming Hu, Hongxia Xu, Jintai Chen, Jian Wu; NAACL; 2024. [CCF B; TH-CPL B]

💻 Projects

1. Luotuo: An Instruction-following Chinese Language model, LoRA tuning on LLaMA

This project has already received over 3.6k stars.

This project contains multiple sub-projects. For more sub-projects, please view the project homepage.

2. Poco: Your Pocket Coworker

This project has already received over 1.3k stars.

A more beautiful and easier-to-use alternative to OpenClaw, featuring a nicer Web UI, built-in IM support, and a sandboxed runtime. Powered by a Claude Code–based agent.

Competition Awards:

AI Hackathon Tour, Hangzhou Station — 3rd Place
AI Hackathon Tour, Zhejiang University Station — 1st Place
AI Hackathon Tour, National Finals — 3rd Place

3. Token Arena

1,000+ total users with 100+ daily active users

An open-source AI usage tracking platform that automatically collects token consumption data from local AI coding CLIs, with a clean Web dashboard for analytics.

4. WeDoctor Medical LLM Post-Training

Achieved 1st place on MedBench leaderboard

A medical domain LLM project focusing on post-training techniques:

SFT: Implemented Self-Instruct and Magpie instruction generation pipelines to automatically construct instruction-tuning pairs from medical documents
RL: Developed reward model training, DPO, CoT data generation, and RLVR approaches
VLM: Applied segmentation, detection, classification models and external knowledge bases for hallucination correction to generate SFT and DPO preference data

5. ZJU-Beamer-Template

A LaTeX Beamer template designed for creating presentation slides with Zhejiang University’s official visual identity and branding.

6. Agentify

A vibe workflow platform that enables agents to generate reusable agentic workflows from SOPs or natural language, with auto-generated UIs for seamless interaction.

🏅 Honors and Awards

Academic

2025.09, Outstanding Graduate Student
2024.05, Outstanding Graduates (Undergraduate)
2023.11, Principal’s Scholarship (Undergraduate) (ONLY 10 people in the whole school each year)
2021.11, China Undergraduate Mathematical Contest in Model (CUMCM), First Prize, under the guidance of Prof. Bo Li
2019.11, The 19th Awarding Program For Future Scientists (The largest pre-college scientific research event in China), Second Prize

Hackathon

2026.05, Douyin AI InnovAI Hackathon Tour, Hangzhou Station — 2nd Place
2026.05, Douyin AI InnovAI Hackathon Tour, Hangzhou Station — Most Popular Award
2026.03, AI Hackathon Tour, National Finals — 3rd Place
2026.03, AI Hackathon Tour, Zhejiang University Station — 1st Place
2026.01, AI Hackathon Tour, Hangzhou Station — 3rd Place

📝 Selected Blogs

RL

Agent

📖 Educations

2024.09 - 2029.06 (expected), Ph.D. in Artificial Intelligence, College of Computer Science and Technology, Zhejiang University. Supervised by Prof. Jian Wu and Dr. Jintai Chen.
2020.09 - 2024.06, B.Sc. in Statistics, School of Mathematics and Statistics, Central China Normal University. Advised by Prof. Bo Li and Dr. Haitong Yang. Thanks to Prof. Bo Li, I was able to spend four unforgettable and wonderful years there.

🔎 Reviews

Review for Conferences: ACL, EMNLP, NIPS, AAAI
Review for Journals: JBHI

This homepage is visited times