OpenUnlearning: Accelerating LLM Unlearning via Unified Benchmarking of Methods and Metrics

TL;DR

OpenUnlearning tackles fragmentation in LLM unlearning by providing a unified benchmarking framework that integrates multiple unlearning methods, metrics, and datasets, plus a meta-evaluation of evaluation metrics. It standardizes evaluation across 3 major benchmarks (TOFU, MUSE, WMDP) and offers 450+ ground-truth checkpoints to stress-test metrics, enabling principled cross-method comparisons. The paper reports that 8 unlearning methods were benchmarked using 10 metrics with SimNPO achieving top performance, while acknowledging limitations of current ranking schemes. OpenUnlearning is open-sourced under the MIT license and designed as a living framework to accelerate rigorous, reproducible progress in LLM unlearning, with community-driven extensions.

Abstract

Robust unlearning is crucial for safely deploying large language models (LLMs) in environments where data privacy, model safety, and regulatory compliance must be ensured. Yet the task is inherently challenging, partly due to difficulties in reliably measuring whether unlearning has truly occurred. Moreover, fragmentation in current methodologies and inconsistent evaluation metrics hinder comparative analysis and reproducibility. To unify and accelerate research efforts, we introduce OpenUnlearning, a standardized and extensible framework designed explicitly for benchmarking both LLM unlearning methods and metrics. OpenUnlearning integrates 13 unlearning algorithms and 16 diverse evaluations across 3 leading benchmarks (TOFU, MUSE, and WMDP) and also enables analyses of forgetting behaviors across 450+ checkpoints we publicly release. Leveraging OpenUnlearning, we propose a novel meta-evaluation benchmark focused specifically on assessing the faithfulness and robustness of evaluation metrics themselves. We also benchmark diverse unlearning methods and provide a comparative analysis against an extensive evaluation suite. Overall, we establish a clear, community-driven pathway toward rigorous development in LLM unlearning research.

Figures (1)

• Figure 1: OpenUnlearning is an extensible library for benchmarking LLM unlearning methods and metrics. It provides a unified framework for implementing unlearning methods, unlearning metrics, and stress-testing tools to verify unlearning robustness. This figure illustrates the unlearning pipeline in terms of implementation‑level components.