Remote

The following papers were accepted but the authors are unable to present in person.

Spotlight Posters

1.
Mechanistic Interpretability as Statistical Estimation: A Variance Analysis of EAP-IG Spotlight
Maxime Méloux, François Portet, Maxime Peyrard
2.
False Sense of Security: Why Probing-based Malicious Input Detection Fails to Generalize Spotlight
Cheng Wang, Zeming Wei, Qin Liu, Wenxuan Zhou, Muhao Chen
3.
Towards Understanding the Nature of Attention with Low-Rank Sparse Decomposition Spotlight
Zhengfu He, Junxuan Wang, Rui Lin, Xuyang Ge, Wentao Shu, Qiong Tang, Junping Zhang, Xipeng Qiu
4.
Probing by Analogy: Decomposing Probes into Activations for Better Interpretability and Inter-Model Generalization Spotlight
Patrick Leask, Noura Al Moubayed
5.
Towards a Mechanistic Understanding of Robustness in Finetuned Reasoning Models Spotlight
Aashiq Muhamed, Xuandong Zhao, Mona T. Diab, Virginia Smith, Dawn Song

Posters

Sorted alphabetically by first author.

1.
Robustly Improving LLM Fairness in Realistic Settings via Interpretability Adam Karvonen, Samuel Marks
2.
Universal Neurons in GPT-2: Emergence, Persistence, and Functional Impact Advey Nandan, Cheng-Ting Chou, Amrit Kurakula, Cole Blondin, Kevin Zhu, Vasu Sharma, Sean O'Brien
3.
Reverse Engineering a Stateful Reasoning Circuit Akshit Kumar, Dipti Sharma, Parameswari Krishnamurthy
4.
Don't Believe the Belief Hype! Alessandro Corona Mendozza
5.
Dual Mechanisms of Value Expression: Decomposing Intrinsic and Prompted Values in Language Models Jongwook Han, Jongwon Lim, InJin Kong, Yohan Jo
6.
Adaptive Task Vectors for Large Language Models Joonseong Kang, Soojeong Lee, Sumin Park, Subeen Park, Taero Kim, Jihee Kim, Ryunyi LEE, Kyungwoo Song
7.
Symbolic vs. Continuous Features in Transformers: A Digital Communication System's Explanation Kan Deng
8.
PII Jailbreaking in LLMs via Activation Steering Reveals Personal Information Leakage Krishna Kanth Nakka, Dmitrii Usynin, Xue Jiang, Xuebing Zhou
9.
Learned Structure in Cartridges: Keys as Shareable Routers in Self-Studied Representations Mauri Diaz
10.
Localizing Reasoning Training-Induced Changes in Large Language Models Max Klabunde, Florian Lemmerich
11.
Three Desiderata for Faithfulness in Machine Learning Explanations: The Case for Causal Abstraction Mette Friis Andersen, Maria Heuss, Ana Lucic