Siyu Chen

PhD student, Yale University

Joined

May 2021

Names

Siyu Chen (Preferred)

Emails

****@mails.tsinghua.edu.cn (Confirmed)

****@yale.edu (Confirmed)

Personal Links

Homepage

Career & Education History

PhD student

Yale University (yale.edu)

2022 – Present

Undergrad student

Tsinghua University, Tsinghua University (tsinghua.edu.cn)

2018 – 2022

Advisors, Relations & Conflicts

No relations added

Expertise

Stats

2022 – Present

2021 – Present

Publications

How Transformers Learn Causal Structures In-Context: Explainable Mechanism Meets Theoretical Guarantee
Jianzhe Wei, Siyu Chen, Jianliang He, Zhuoran Yang
- ICLR 2026 Poster
- Readers: Everyone
Learning to Incentivize on the Fly: Leader-Follower Games with Policy Recommendation
Weizheng Zhang, Siyu Chen, Zhuoran Yang
- Submitted to ICLR 2026
- Readers: Everyone
On the Mechanism and Dynamics of Modular Addition: Fourier Features, Lottery Ticket, and Grokking
Jianliang He, Leda Wang, Siyu Chen, Zhuoran Yang
- Submitted to ICLR 2026
- Readers: Everyone
Unlocking Out-of-Distribution Generalization in Transformers via Latent Space Reasoning
Awni Altabaa, Siyu Chen, John Lafferty, Zhuoran Yang
- Submitted to ICLR 2026
- Readers: Everyone
Taming Polysemanticity in LLMs: Theory-Grounded Feature Recovery via Sparse Autoencoders
Siyu Chen, Heejune Sheen, Xuyuan Xiong, Tianhao Wang, Zhuoran Yang
- ICLR 2026 Poster
- Readers: Everyone
Mechanistic Interpretability of In-Context Learning Generalization through Structured Task Curriculum
Chenxiang Yan, Jianliang He, Siyu Chen, Xintian Pan, Zhuoran Yang
- Submitted to ICLR 2026
- Readers: Everyone
Interpreting Multi-Layer Transformers for In-Context Linear Regression with Varying Covariance
Changqi Wang, Jianliang He, Siyu Chen, Xintian Pan, Xiangyu Yue, Zhuoran Yang
- Submitted to ICLR 2026
- Readers: Everyone
Active Advantage-Aligned Online Reinforcement Learning with Offline Data
Xuefeng Liu, Hung T. C. Le, Siyu Chen, Rick Stevens, Zhuoran Yang, Matthew Walter, Yuxin Chen
- EXAIT@ICML 2025 Poster
- Readers: Everyone
An Optimized Franz-Parisi Criterion and its Equivalence with SQ Lower Bounds
Siyu Chen, Theodor Misiakiewicz, Ilias Zadik, Peiyuan Zhang
- NeurIPS 2025 oral
- Readers: Everyone
In-Context Linear Regression Demystified: Training Dynamics and Mechanistic Interpretability of Multi-Head Softmax Attention
Jianliang He, Xintian Pan, Siyu Chen, Zhuoran Yang
- ICML 2025 poster
- Readers: Everyone

View all 20 publications

Co-Authors

View all 37 co-authors