👋 About Me
I am Qinghua Xing (邢清画), an undergraduate student at Nankai University (南开大学). I conducted a research internship at Media Computing Lab (媒体计算实验室) under the supervision of Prof. Xiang Li (李翔).
Recently, I have been admitted via recommended exemption to University of Science and Technology of China (USTC, 中国科学技术大学) to pursue a Master of Science degree, under the supervision of Prof. Zhiwei Xiong (熊志伟).
During my undergraduate studies, I mainly focus on AI4Research. Our papers have been accepted by CCF-A top-tier conferences, including ICLR 2026 (NAIPv2) and AAAI 2025 (NAIPv1). I am currently working on Agents, Multimodal Learning, and Knowledge Editing.
Interests : LLMs, Agents, Multimodal Learning, Knowledge Editing.
🔥 News
- 2026.01: 🎉🎉 Our paper NAIPv2 has been accepted by ICLR 2026!
- 2025.09: 🎉🎉 I am honored to be admitted to University of Science and Technology of China (USTC) to pursue my master’s degree under the supervision of Prof. Zhiwei Xiong!
- 2024.12: 🎉🎉 Our paper NAIPv1 has been accepted by AAAI 2025!
📝 Publications
🎓 Education
- 2026.09 - Present (Incoming): Master's Student, College of Computer Science and Artificial Intelligence, University of Science and Technology of China.
- 2022.09 - 2026.06: Undergraduate Student, College of Cryptology and Cyber Science, Nankai University.
💼 Internships
- 2026.03 - Present: Huawei 2012 Lab, Base Large Model Department, Beijing
- Working on agent-based models and large model fine-tuning tasks, contributing to research and development in the field of advanced AI systems.
🔭 Things I am currently working on 🔭
🧩 NAIPv1
From Words to Worth: Newborn Article Impact Prediction with LLM — accepted to AAAI 2025. This work proposes an LLM-based framework for predicting the future impact of newly published papers before citation signals accumulate.
🧩 NAIPv2
NAIPv2: Debiased Pairwise Learning for Efficient Paper Quality Estimation . We estimate paper quality efficiently with a debiased pairwise learning scheme, introduce the Review Tendency Signal (RTS) to calibrate reviewer bias, and build a large-scale dataset for training—achieving linear-time inference and strong cross-venue generalization. You can try this demo here: 👉 NAIPv2 Demo
🌐 MediChain-LLM-Agent
A medical-domain LLM agent built with LangChain and a JavaScript frontend, using a custom drug database to suggest medication plans from patient symptoms (research prototype; not for clinical use). See MediChain-LLM-Advisor for details. We are experimenting with multi-agent frameworks to improve reasoning reliability, tool orchestration, and end-to-end task success.
🌐 Kaifeng Woodblock Print Generation Platform
In collaboration with the Kaifeng municipal government, this AIGC-powered platform promotes Intangible Cultural Heritage by digitally preserving and revitalizing the Zhuxianzhen woodblock print style. It generates heritage-aligned artworks and offers educational modules for knowledge dissemination and cultural transmission. Open-source website🌱: coming soon.
My website
https://1Reminding.github.io 🚀 Continuously updating
💬 Miscellaneous
- Tech interests / Tools: MLLMs, Knowledge Editing (KE), Chain-of-Thought (CoT); familiar with multi-agent development.
- Hobbies / Fun: 🎨 Drawing · 📷 Photography · 🏸 Badminton · 💎 Jewelry design
- Availability / Collaboration: Open to collaboration!😄