Publications

2026

1. Tensor Product Representation Probes Reveal Shared Structure Across Linear Directions

   Andrew Lee , Fernanda Viegas, and Martin Wattenberg

   Workshop on Connecting Low-rank Representations in AI (@ ICML) Mechanistic Interpretability Workshop (@ ICML) 2026
2. Geometry-Adaptive Explainer for Faithful Dictionary-Based Interpretability under Distribution Shift

   Sungjun Lim, Heedong Kim, Andrew Lee , and Kyungwoo Song

   Mechanistic Interpretability Workshop (@ ICML) 2026
3. Agentic Reinforcement Learning for Search Misaligns Instruction-Tuning

   Yushi Yang, Shreyansh Padarha, Sarah Ball, Andrew Lee , and Adam Mahdi

   Workshop on Agents in the Wild: Safety, Security, and Beyond (@ ICML)\\Failure Modes of Agentic AI (@ ICML) 2026
4. Decomposing Query-Key Feature Interactions Using Contrastive Covariances

   Andrew Lee , Yonatan Belinkov, Fernanda Viegas, and Martin Wattenberg

   ICML 2026
5. From Isolation to Entanglement: When Do Interpretability Methods Identify and Disentangle Known Concepts?

   Aaron Mueller, Andrew Lee , Shruti Joshi, Ekdeep Singh Lubana, Dhanya Sridhar, and Patrik Reizinger

   ACL 2026
6. Valence–Arousal Subspace in LLMs: Circular Emotion Geometry and Multi-Behavioral Control

   Lihao Sun, Lewen Yan, Xiaoya Lu, Andrew Lee , Jie Zhang, and Jing Shao

   Mechanistic Interpretability Workshop (@ ICML) 2026
7. Into the Rabbit Hull: From Task-Relevant Concepts in DINO to Minkowski Geometry

   Thomas Fel, Binxu Wang, Michael A. Lepori, Matthew Kowal, Andrew Lee , Randall Balestriero, Sonia Joseph, Ekdeep S. Lubana, Talia Konkle, Demba Ba, and Martin Wattenberg

   ICLR 2026

2025

1. Shared Global and Local Geometry of Language Model Embeddings

   Andrew Lee , Melanie Weber, Fernanda Viegas, and Martin Wattenberg

   COLM 2025 - Outstanding Paper Award
2. ICLR: In-Context Learning of Representations

   *Core Francisco Park, *Andrew Lee , *Ekdeep Singh Lubana, *Yongyi Yang, Maya Okawa, Kento Nishi, Martin Wattenberg, and Hidenori Tanaka

   ICLR 2025
3. Better World Models Can Lead to Better Post-Training Performance

   Prakhar Gupta, Henry Conklin, Sarah-Jane Leslie, and Andrew Lee

   Mechanistic Interpretability @ NeurIPS 2025 - Spotlight
4. Why Can't Transformers Learn Multiplication? Reverse-Engineering Reveals Long-Range Dependency Pitfalls

   Xiaoyan Bai, Itamar Pres, Yuntian Deng, Chenhao Tan, Stuart Shieber, Fernanda Viegas, Martin Wattenberg, and Andrew Lee

   Preprint 2025
5. How Does DPO Reduce Toxicity? A Mechanistic Neuron-Level Analysis

   Yushi Yang, Filip Sondej, Harry Mayne, Andrew Lee , and Adam Mahdi

   EMNLP 2025
6. Eeyore: Realistic Depression Simulation via Expert-in-the-Loop Supervised and Preference Optimization

   Siyang Liu, Bianca Brie, Wenda Li, Laura Biester, Andrew Lee , James Pennebaker, and Rada Mihalcea

   Findings of ACL 2025
7. Agentic Reinforcement Learning for Search is Unsafe

   Yushi Yang, Shreyansh Padarha, Andrew Lee , and Adam Mahdi

   Preprint 2025

2024

1. A mechanistic understanding of alignment algorithms: A case study on dpo and toxicity

   Andrew Lee , Xiaoyan Bai, Itamar Pres, Martin Wattenberg, Jonathan K Kummerfeld, and Rada Mihalcea

   ICML 2024 - Oral (Top 1.5% of submissions)
2. Emergence of Hidden Capabilities: Exploring Learning Dynamics in Concept Space

   Core Francisco Park, Maya Okawa, Andrew Lee , Ekdeep Singh Lubana, and Hidenori Tanaka

   NeurIPS 2024 - Spotlight

2023

1. Emergent linear representations in world models of self-supervised sequence models

   *Neel Nanda, *Andrew Lee , and Martin Wattenberg

   BlackboxNLP (EMNLP) 2023 - Honorable Mention, Best Paper
2. Empathy Identification Systems are not Accurately Accounting for Context

   Andrew Lee , Jonathan Kummerfeld, Larry An, and Rada Mihalcea

   EACL 2023
3. A PhD Student's Perspective on Research in NLP in the Era of Very Large Language Models

   Oana Ignat, Zhijing Jin, Artem Abzaliev, Laura Biester, Santiago Castro, Naihao Deng, Xinyi Gao, Aylin Gunal, Jacky He, Ashkan Kazemi, and others

   2023
4. Some things are more CRINGE than others: Preference Optimization with the Pairwise Cringe Loss

   Jing Xu, Andrew Lee , Sainbayar Sukhbaatar, and Jason Weston

   Preprint 2023

2022

1. Augmenting Task-Oriented Dialogue Systems with Relation Extraction

   Andrew Lee , Zhenguo Chen, Kevin Leach, and Jonathan K. Kummerfeld

   AAAI 2022 DSTC10 Workshop
2. Improving Chess Commentaries by Combining Language Models with Symbolic Reasoning Engines

   Andrew Lee , David Wu, Emily Dinan, and Mike Lewis

   Preprint 2022

2021

1. Micromodels for Efficient, Explainable, and Reusable Systems: A Case Study on Mental Health

   Andrew Lee , Jonathan Kummerfeld, Lawrence An, and Rada Mihalcea

   Findings of EMNLP 2021

   [Code]

2019

1. An Evaluation Dataset for Intent Classification and Out-of-Scope Prediction

   Stefan Larson, Anish Mahendran, Joseph J Peper, Christopher Clarke, Andrew Lee , Parker Hill, Jonathan K Kummerfeld, Kevin Leach, Michael A Laurenzano, Lingjia Tang, and Jason Mars

   EMNLP 2019

   [Data]
2. Outlier Detection for Improved Data Quality and Diversity in Dialog Systems

   Stefan Larson, Anish Mahendran, Andrew Lee , Jonathan K Kummerfeld, Parker Hill, Michael A Laurenzano, Johann Hauswald, Lingjia Tang, and Jason Mars

   NAACL 2019