Saved in:
Bibliographic Details
Main Authors: Rinberg, Roy, Bhalla, Usha, Shilov, Igor, Calmon, Flavio P., Gandikota, Rohit
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2512.04144
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917123936223232
author Rinberg, Roy
Bhalla, Usha
Shilov, Igor
Calmon, Flavio P.
Gandikota, Rohit
author_facet Rinberg, Roy
Bhalla, Usha
Shilov, Igor
Calmon, Flavio P.
Gandikota, Rohit
contents Targeted interventions on language models, such as unlearning, debiasing, or model editing, are a central method for refining model behavior and keeping knowledge up to date. While these interventions aim to modify specific information within models (e.g., removing virology content), their effects often propagate to related but unintended areas (e.g., allergies); these side-effects are commonly referred to as the ripple effect. In this work, we present RippleBench-Maker, an automatic tool for generating Q&A datasets that allow for the measurement of ripple effects in any model-editing task. RippleBench-Maker builds on a Wikipedia-based RAG pipeline (WikiRAG) to generate multiple-choice questions at varying semantic distances from the target concept (e.g., the knowledge being unlearned). Using this framework, we construct RippleBench-Bio, a benchmark derived from the WMDP (Weapons of Mass Destruction Paper) dataset, a common unlearning benchmark. We evaluate eight state-of-the-art unlearning methods and find that all exhibit non-trivial accuracy drops on topics increasingly distant from the unlearned knowledge, each with distinct propagation profiles. To support ongoing research, we release our codebase for on-the-fly ripple evaluation, along with the benchmark, RippleBench-Bio.
format Preprint
id arxiv_https___arxiv_org_abs_2512_04144
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle RippleBench: Capturing Ripple Effects Using Existing Knowledge Repositories
Rinberg, Roy
Bhalla, Usha
Shilov, Igor
Calmon, Flavio P.
Gandikota, Rohit
Artificial Intelligence
Targeted interventions on language models, such as unlearning, debiasing, or model editing, are a central method for refining model behavior and keeping knowledge up to date. While these interventions aim to modify specific information within models (e.g., removing virology content), their effects often propagate to related but unintended areas (e.g., allergies); these side-effects are commonly referred to as the ripple effect. In this work, we present RippleBench-Maker, an automatic tool for generating Q&A datasets that allow for the measurement of ripple effects in any model-editing task. RippleBench-Maker builds on a Wikipedia-based RAG pipeline (WikiRAG) to generate multiple-choice questions at varying semantic distances from the target concept (e.g., the knowledge being unlearned). Using this framework, we construct RippleBench-Bio, a benchmark derived from the WMDP (Weapons of Mass Destruction Paper) dataset, a common unlearning benchmark. We evaluate eight state-of-the-art unlearning methods and find that all exhibit non-trivial accuracy drops on topics increasingly distant from the unlearned knowledge, each with distinct propagation profiles. To support ongoing research, we release our codebase for on-the-fly ripple evaluation, along with the benchmark, RippleBench-Bio.
title RippleBench: Capturing Ripple Effects Using Existing Knowledge Repositories
topic Artificial Intelligence
url https://arxiv.org/abs/2512.04144