đź‘‹ About Me
I am a Second-year M.S. student in Computer Technology at Tsinghua University. I am fortunate to be supervised by Prof. Chun Yuan in the CVML group. Before that, I obtained my Honors Bachelor’s degree in Computer Science and Technology from the Yingcai Honors College at the University of Electronic Science and Technology of China in 2024.
My research interests include LLM/Agent post-training (SFT/OPD/RL), Agent Harness design, multimodal understanding, and model distillation & self-training.
✨ News
- Jul. 2026: One paper on an omnimodal progressive reasoning agent for segmentation was submitted to AAAI-2027.
- Jul. 2026: 🎉 One paper on temporal forgery localization (MG-RWKV) was accepted by ECCV-2026.
- Jun. 2026: One paper on desktop GUI agent benchmark (DeskCraft) has been submitted to EMNLP-2026.
📚 Publications
* denotes equal contribution
-
MG-RWKV: Multi-Grained Context-Aware RWKV for Temporal Forgery Localization
THU-AEuropean Conference on Computer Vision (ECCV), 2026. -
DeskCraft: Benchmarking Desktop Agents on Professional Workflows and Human-in-the-Loop Collaboration
THU-AUnder review at Conference on Empirical Methods in Natural Language Processing (EMNLP), 2026. [Project Page] -
FCL-COD: Weakly Supervised Camouflaged Object Detection with Frequency-aware and Contrastive Learning
CCF-AIEEE/CVF Conf. on Computer Vision and Pattern Recognition (CVPR), 2026. -
AdaMem: Adaptive User-Centric Memory for Long-Horizon Dialogue Agents
THU-AUnder review at Conference on Empirical Methods in Natural Language Processing (EMNLP), 2026. -
IteRPrimE: Zero-shot Referring Image Segmentation with Iterative Grad-CAM Refinement and Primary Word Emphasis
CCF-AAAAI Conference on Artificial Intelligence (AAAI), 2025. -
Natural-Language Agent Harnesses
CCF-AUnder review at Conference on Neural Information Processing Systems (NeurIPS), 2026. -
Cognitive Kernel-Pro: A Framework for Deep Research Agents and Agent Foundation Models Training
Technical Report, 2026. -
SAHSR: Semantic Alignment and Hard-Sample Retraining Framework for Cross-Modal Person Re-Identification
CCF-BIEEE International Conference on Multimedia and Expo (ICME), 2025 (Oral).
🚀 Projects
-
Alibaba Cloud Data+AI Global Challenge — Research Agent Track
🏆 Rank 4 / 1028 (Top 0.39%) · [Competition Page]
Built an end-to-end Research Agent for complex multi-hop reasoning questions. -
ZotLink: MCP Server for Zotero Connector
[Project Page] ·
Production-ready MCP server for Zotero to save open preprints with rich metadata and smart PDF attachments.
🏆 Selected Honors and Awards
- National Scholarship, 2022
- China “Internet+” Innovation and Entrepreneurship Competition, National Silver Award, 2022
- IEEEXtreme 16.0 Programming Competition, Global Rank 118/2992, 2022
- Lanqiao Cup National Competition, Third Prize, 2023
- Mathematical Contest in Modeling (MCM), Honorable Mention, 2023
- UESTC Outstanding Student Scholarship, 2021, 2022, 2023
- UESTC Honor’s Degree of Bachelor of Engineering, 2024