REVIEW 5 cited by
Mechanistic Unlearning: Robust Knowledge Unlearning and Editing via Mechanistic Localization
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
Methods for knowledge editing and unlearning in large language models seek to edit or remove undesirable knowledge or capabilities without compromising general language modeling performance. This work investigates how mechanistic interpretability -- which, in part, aims to identify model components (circuits) associated to specific interpretable mechanisms that make up a model capability -- can improve the precision and effectiveness of editing and unlearning. We find a stark difference in unlearning and edit robustness when training components localized by different methods. We highlight an important distinction between methods that localize components based primarily on preserving outputs, and those finding high level mechanisms with predictable intermediate states. In particular, localizing edits/unlearning to components associated with the lookup-table mechanism for factual recall 1) leads to more robust edits/unlearning across different input/output formats, and 2) resists attempts to relearn the unwanted information, while also reducing unintended side effects compared to baselines, on both a sports facts dataset and the CounterFact dataset across multiple models. We also find that certain localized edits disrupt the latent knowledge in the model more than any other baselines, making unlearning more robust to various attacks.
Forward citations
Cited by 5 Pith papers
-
RippleBench: Capturing Ripple Effects Using Existing Knowledge Repositories
RippleBench automatically generates questions at increasing semantic distance from unlearned topics and shows all eight tested unlearning methods degrade accuracy that recovers only slowly with distance.
-
DUSK: Do Not Unlearn Shared Knowledge
DUSK benchmarks machine unlearning under overlapping forget and retain documents, showing existing methods remove surface text but fail to preserve shared knowledge while erasing unique content.
-
SoK: Machine Unlearning for Large Language Models
A new taxonomy for LLM unlearning distinguishes removal-intended from suppression-intended methods, and argues that gradient ascent methods functionally behave like suppression.
-
Editing as Unlearning: Are Knowledge Editing Methods Strong Baselines for Large Language Model Unlearning?
WISE and AlphaEdit, two knowledge editing methods, are competitive unlearning baselines when unlearning is framed as editing a model's answer into a refusal.
-
Step-by-Step Reasoning Attack: Revealing 'Erased' Knowledge in Large Language Models
Step-by-step reasoning prompts can recover purportedly erased facts from unlearned LLMs, but the paper's quantitative evidence is internally inconsistent.
Discussion (0). Continue with ORCID to comment.