Yuqiao Tan

Yuqiao Tan (谭宇乔)

First-year master student at CASIA, supervised by Shizhu He. My research focuses on LLM Reasoning, LLM Interpretability, Reinforcement Learning, and Personalized Agent. I'm willing to any discussion, and I'm open to internship opportunities and collaborations. If you find my work interesting, feel free to contact me!

tanyuqiao2025@ia.ac.cn

GitHub  /  Google Scholar

News

Publications

— 2025 —

Bottom-up Policy Optimization: Your Language Model Policy Secretly Contains Internal Policies  [paper]  [code]
Yuqiao Tan*, Minzheng Wang*, Shizhu He, et al.
Preprint

The Zero-Step Thinking: An Empirical Study of Mode Selection as Harder Early Exit in Reasoning Models  [paper]  [code]
Yuqiao Tan, Shizhu He, Kang Liu, Jun Zhao
NeurIPS 2025 Efficient Reasoning Workshop

Neural Incompatibility: The Unbridgeable Gap of Cross-Scale Parametric Knowledge Transfer in LLMs  [paper]  [code]
Yuqiao Tan, Shizhu He, Kang Liu, Jun Zhao
ACL 2025

Dynamic Parametric Retrieval Augmented Generation for Test-time Knowledge Enhancement  [paper]  [code]
Yuqiao Tan, Shizhu He, Huanxuan Liao, Jun Zhao, Kang Liu
Preprint

MuAP: Multi-step Adaptive Prompt Learning for Vision-Language Model with Missing Modality  [paper]  [code]
Ruiting Dai*, Yuqiao Tan*, et al.
ICMR 2025

— 2024 —

G-SAP: Graph-based Structure-Aware Prompt Learning over Heterogeneous Knowledge  [paper]
Ruiting Dai*, Yuqiao Tan*, et al.
ICMR 2024

Projects

Education

Internships

Honors

Invited Talks

Services