Chupei Wang

Researcher on LLM memory & interpretability

Bachelor, University of Virginia, Physics Department

About

With a foundation in physics and philosophy—including a two‑year gap during college at the University of Chicago Divinity School (Eastern philosophy of religion)—Chupei explores where logic and mind meet their limits, probing how the edges of science and the humanities intersect. He is driven by a curiosity about where cognitive architectures—biological and artificial—break down, and what these failures teach us about intelligence. After graduation, he had a startup experience in China. Currently seeking lab and research opportunities.

Seeking RA / PhD — details

I’m seeking RA positions (onsite/remote) and future PhD opportunities to study the intersection of human cognition and algorithms. After a B.S. in Physics (University of Virginia), I self‑studied philosophy of science (admitted but not enrolled to a philosophy program in the University of Bristol), and during college I took a two‑year gap at the University of Chicago Divinity School to study the philosophy of religion (mainly Eastern). Those experiences inspired me to independently initiate and fund a PI‑for‑LLMs project—recently accepted to the ICML 2025 Workshop on Long Context Foundation Models.

Email me for RA inquiry / collaboration. Seeking RA & PhD opportunities

Group / Collaborators

We are an interdisciplinary group probing the boundaries between human and machine intelligence.

  • Jiaqiu Vince Sun* — PhD Candidate, NYU Center for Neuroscience. Former professional architect turned neuroscientist; integrates spatial design, cognitive neuroscience, and philosophy of mind to study how memory emerges and diverges in brains and artificial systems; focuses on higher‑level functions such as self‑monitoring and control. 📫 vince.sun@nyu.edu
  • Chupei Wang* — Bachelor, University of Virginia (Physics). Physics + philosophy background; UChicago Divinity School (Eastern philosophy of religion, gap during college). Studies failure modes at the intersection of cognitive architectures and algorithms. 📫 cw4bb@virginia.edu
  • Sam Zheyang Zheng — Guest researcher at the Center for Computational Neuroscience (CCN) and currently obtaining his Ph.D. at New York University (NYU), co‑advised by Alex Williams and Gyorgy Buzsaki (NYU Langone). Previously earned a B.A. in Mathematics (minor in Philosophy) at the University of Chicago. Works closely with experimentalists to develop statistical tools for neuroscientific questions, with a focus on the multi‑scale and flexible nature of the hippocampal neural code.

View more

Why approach LLMs from a cognitive‑science angle?

After earning my undergraduate degree in physics, I spent the past two years self‑studying philosophy of science (admitted but not enrolled to a philosophy program in the University of Bristol). During college, I also took a two‑year gap at the University of Chicago Divinity School to study the philosophy of religion (mainly Eastern). I was interested in how humans think and formulate theories from physics to philosophy. All those experiences inspired me to independently initiate and fund this PI‑for‑LLMs research—recently accepted to the ICML 2025 workshop. This recognition boosted my confidence and made me eager to learn more and to find a lab to continue thinking and observing.

I like making cross‑disciplinary comparisons and standing and observing at the intersections between the human and the natural world—from the mind and cognitive mechanisms to physics and algorithms. In particular, I am interested in uncovering where the structural features of knowledge within a field shape both its unique insights and its blind spots. The first cognitive‑with‑LLM project also grew from a gap‑year startup PM experience: with human engineers, spec clarification was natural; with LLMs, conversations felt different—fluent code but gaps in context‑aware reasoning. In Jan 2025 I began probing working memory; a proactive interference (PI) paradigm exposes this bottleneck cleanly.

Projects

Proactive Interference (PI) in LLMs — Cog4LLM

Interference-limited retrieval, measured minimally; simple mind‑flow with optional details.

ICML 2025 Workshop — Accepted Ongoing Boundary cases Population codes

Open Demo Details / Progress

Contributions (CW / JVS)
  • Equal contribution. Co‑initiated cognitive‑inspired WM stress tests.
  • CW (Chupei Wang): Framed LLM retrieval errors as interference; led iterative trial‑and‑error testing; designed Experiment 1; pioneered prompt‑hacking interventions.
  • JVS (Jiaqiu Vince Sun): Implemented/automated pipelines; data/analysis; co‑engineered Experiment 2; expanded to Experiment 3 (value‑length); developed NL interventions and error analyses.

Paper: arXiv · Code: GitHub

Interpretability: Population Codes → Circuits → Boundary Cases

Taking cues from population coding and inhibition/gating motifs in the brain, I use circuit analysis to ask whether similar population-like representations and control signals exist in LLMs—and where they break. Early evidence suggests limits that appear in models but not in human data. I focus on boundary cases: stress tests that expose failure surfaces, then map them back to candidate mechanisms (ranking, selection, inhibition) to turn cognitive hypotheses into falsifiable circuit predictions.

Ongoing Boundary cases Circuit analysis

Ongoing — progress posted weekly. This week: 2025-08-10.

Open project

Creative — Music Supervision (The Three‑Body)

Supervised music for early experimental animation series on Bilibili; coordinated NYU composers; oversaw post‑production.

Music Animation

Watch clip

更多 / More
  • 协调纽约大学作曲系成员,完成《三体》第一代实验动画(版权方实验系列)的配乐沟通与监督;系列在 B 站累计播放约 1.2 亿+。
  • 自第二季第 3 话起担任音乐监制;示例片段署名约 24:50 处(音乐制作人员名单:监制)。
  • 职责:以文字与示例沟通作曲意图与改进意见,统筹后期音乐制作。

Humanities & Philosophy — Zhuangzi and Tiantai

Seminar work at UChicago Divinity School with Prof. Brook Ziporyn, on classical Chinese (Zhuangzi; Tiantai). Minimal note with reference links.

Humanities

Open project

Contact

Email: cw4bb@virginia.edu

Download CV