arXiv Machine Learning By Ananth Eswar, Pratinav Seth, Utsav Avaiya, Vinay Kumar Sankarapu

Faithfulness to Refusal: A Causal Audit of Neuron Selectors

Read the original on arXiv Machine Learning →

arXiv:2607. 05355v1 Announce Type: cross Abstract: Attribution scores increasingly identify which neuron rows of a language model matter for applications such as pruning, interpretability, and editing for safety, yet whether they identify causally important rows is rarely tested directly.

Summary generated by The Flow from the publisher's feed. The full article lives at arXiv Machine Learning.

arXiv AI
Jun 9

When Behavioral Safety Evaluation Fails: A Representation-Level Perspective

arXiv:2606. 08044v1 Announce Type: cross Abstract: Large Language Model (LLM) safety has often been evaluated at the behavior level, which provides limited evidence of internal robustness, as these evaluations target outputs rather than representation-level vulnerability under intervention.

By Enyi Jiang, Anders Gj{\o}lbye, Yibo Jacky Zhang, Sanmi Koyejo
arXiv Machine Learning
Jun 15

A Low-Rank Subspace Analysis of LLM Interventions

arXiv:2606. 14388v1 Announce Type: new Abstract: Interventions designed to modify a particular behavior in LLMs, such as refusal or sycophancy, often produce unintended changes in other behaviors.

By Angira Sharma, Christian Schroeder de Witt, Philip Torr, Anisoara Calinescu, Jialin Yu