Huihan Li

I am a PhD candidate in Computer Science at the University of Southern California, advised by Xiang Ren. During my PhD, I was awarded the Amazon ML Fellowship. Previously, I completed an M.S.E. in Computer Science at Princeton University, where I worked with Danqi Chen and was named a Siebel Scholar, and a B.A. in Computer Science and Cognitive & Linguistic Sciences at Wellesley College.

My research develops methods to diagnose and improve language-model reasoning, spanning reasoning diagnostics, post-training, agent routing and evaluation, and understanding how pretraining data shapes model behavior.

Along the way, I have released open-source tools and datasets: diagnostic tools for long reasoning traces (ARBOR, STIM), pipelines that trace generations back to the pretraining documents that shaped them (CULTURE-GEN, MEMOed), and evaluation resources for long-tail knowledge (LINK) and conversational QA (EvalConvQA).

I am seeking full-time Research Scientist and Applied Scientist roles starting in 2027. I expect to graduate by May 2027.

Selected Publications

* indicates equal contribution.

1. ARBOR: Locating Reasoning Failures in Long Chains of Thought through Reasoning State Tracking
   Huihan Li, Siyuan Wang, Naga Vamsi Ramana Dinavahi, Xiaohui Liu, Xiang Ren. arXiv preprint 2026.
   [Paper] [Code]
2. Diagnosing Memorization in Chain-of-Thought Reasoning, One Token at a Time
   Huihan Li*, You Chen*, Siyuan Wang, Yixin He, Ninareh Mehrabi, Rahul Gupta, Xiang Ren. EMNLP 2025.
   [Paper] [Code]
3. Attributing Culture-Conditioned Generations to Pretraining Corpora
   Huihan Li*, Arnav Goel*, Keyu He, Xiang Ren. ICLR 2025.
   [Paper] [Code]
4. CULTURE-GEN: Revealing Global Cultural Perception in Language Models through Natural Language Prompting
   Huihan Li, Liwei Jiang, Jena D. Hwang, Hyunwoo Kim, Sebastin Santy, Taylor Sorensen, Bill Yuchen Lin, Nouha Dziri, Xiang Ren, Yejin Choi. COLM 2024.
   [Paper] [Code]
5. Ditch the Gold Standard: Re-evaluating Conversational Question Answering
   Huihan Li*, Tianyu Gao*, Manan Goenka, Danqi Chen. ACL 2022. Outstanding Paper Award
   [Paper] [Code]

Full publication list →

Industry Research

• Adobe, Adobe Experience Platform — Applied Scientist Engineering Intern, 2026, with Yunyao Li
  Led a research and productionization project on reliable enterprise agent orchestration, building production-grounded evaluation for improving routing reliability against a growing capability catalog, nondeterministic agent trajectories, and sparse enterprise data.
• Meta, GenAI — Research Intern, 2025, with Liang Tan
  Studied how decoding-temperature schedules affect diversity and accuracy in GRPO post-training for mathematical reasoning.