Li, Puyin Li, Puyin

Li, Puyin

李浦银

PhD Student

Stanford University

About Me

Hello! I am a first-year PhD student in Computer Science at Stanford University, currently rotating with Tobias Gerstenberg and Thomas Icard. Previously, I was a master's student in Symbolic Systems at Stanford, advised by Thomas Icard, and a research assistant at the Stanford Translational AI (STAI) Lab, supervised by Ehsan Adeli.

I am broadly interested in evaluating, improving, and interpreting foundation models' reasoning abilities as cognitive agents. This interest spans topics including causality, visual understanding and reasoning, mechanistic interpretability, formal methods, etc. My aim is to create powerful, faithful, and interpretable AI systems that can robustly interact with the noisy (physical) real world. I believe in cross-disciplinary collaboration in AI.

Prior to Stanford, I studied logic and argumentation with Dov Gabbay and Beishui Liao (廖备水) at ZLAIRE. I am also a Morningside Culture China Scholar at Zhejiang University, which is a life-changing program initiated by a beloved professor Shengchun Zhou (周生春).

I enjoy music and tennis. I play the piano, guitar, and bass, and I’m in a band called The Wishes. I also sing alto in the Stanford Sing++ a cappella group.

Education

  • Ph.D. in Computer Science, 2026- School of Engineering, Stanford University
  • M.S. in Symbolic Systems, 2023–2026 School of Humanities and Sciences, Stanford University
  • B.A. in Philosophy (Minor in Physics), 2018–2023 Chu Kochen Honors College, Zhejiang University
  • Non-Degree Programs Visiting Student in Philosophy of Science, Mansfield College, University of Oxford, 2020-2021 Visiting Student in Management Sciences & Economy, Guanghua School of Management, Peking University, 2021-2022 Tsinghua Logic Summer School 2023; ESSLLI 2024; NASSLLI 2025.

Recent News

  • [09/2026] I started my PhD in Computer Science at Stanford University 🌲.
  • [07/2026] The QuantiPhy Challenge is accepted as an official NeurIPS 2026 competition! Submissions are open until Oct 23, 2026 — come join us 🏆.
  • [04/2026] QuantiPhy full set is now available on HuggingFace🤗.
  • [03/2026] QuantiPhy is accepted by CVPR 2026

Selected Publications

QuantiPhy
QuantiPhy: A Quantitative Benchmark Evaluating Physical Reasoning Abilities of Vision-Language Models
CVPR, 2026
Li Puyin*, Tiange Xiang*, Ella Mao*, Shirley Wei, Xinye Chen, Adnan Masood, Li Fei-fei†, Ehsan Adeli†
Apple
Bucketing the Good Apples: A Method for Diagnosing and Improving Causal Abstraction
Preprint, 2026
Li Puyin*, Jiyuan Tan*, Ahmad Jabbar, Thomas Icard†, Atticus Geiger†

Contact

CoDa, Stanford University