Adnan KhanAdnanProduct & AI
San Francisco Bay Area
Get in touch

Source note2 min read

LACUNA: A Testbed for Evaluating Localization Precision for LLM Unlearning

Source

arXiv

arxiv.org/abs/2607.02513v1

Read the source

Source

Plain-Language Brief

Researchers introduce a new testbed called LACUNA that allows for the evaluation of unlearning methods at the parameter level, rather than just the output level. They find that current state-of-the-art (SOTA) unlearning methods are not precise enough and are susceptible to resurfacing attacks. However, when localization is successful, a simple gradient-based unlearning method achieves strong erasure and robustness.

What the Document Says

  • LACUNA is a testbed introduced for evaluating the precision of unlearning methods at the parameter level.
  • The testbed injects personally identifiable information (PII) into predefined parameters of 1B and 7B OLMo-based models via masked continual pretraining.
  • Current SOTA unlearning methods are evaluated using LACUNA, and it is found that they are highly imprecise and susceptible to resurfacing attacks.
  • When localization is successful, a simple gradient-based unlearning method achieves strong erasure and robustness.

Key Concepts Mentioned

  • LACUNA
  • unlearning
  • parameter-level evaluation
  • PII
  • OLMo-based models
  • resurfacing attacks

Claims Made by the Document

  • LACUNA is the first unlearning testbed with ground-truth parameter-level localization.
  • Current SOTA unlearning methods are imprecise and susceptible to resurfacing attacks.
  • When localization is successful, a simple gradient-based unlearning method achieves strong erasure and robustness.

Evidence Used

  • The text states that LACUNA injects PII into predefined parameters of 1B and 7B OLMo-based models and allows for direct evaluation of whether unlearning targets the weights responsible for knowledge storage.
  • The text mentions that current SOTA unlearning methods are benchmarked using LACUNA, and it is found that they are highly imprecise and susceptible to resurfacing attacks.
  • The text states that when localization is successful, a simple gradient-based unlearning method achieves strong erasure and robustness.

Methods / Approach

UNKNOWN

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

UNKNOWN

Open Questions

UNKNOWN

Notes for Future Q&A

May include interpretation. Labeled where it does.

UNKNOWN

Raw Abstract / Excerpt

LLMs memorize sensitive training data, including personally identifiable information (PII), creating a pressing need for reliable post hoc removal methods. Unlearning has emerged as a promising solution, with state-of-the-art(SOTA) methods often following a localize-first, unlearn-second paradigm that targets specific model parameters. However, existing benchmarks evaluate unlearning solely at the output level, leaving open the question of whether unlearning truly erases knowledge from a model's parameters or merely obfuscates it, a concern reinforced by the success of resurfacing attacks. To bridge this gap, we introduce LACUNA: the first unlearning testbed with ground-truth parameter-level localization. LACUNA injects PII of synthetic individuals into predefined parameters of 1B and 7B OLMo-based models via masked continual pretraining, enabling direct evaluation of whether unlearning targets the weights responsible for knowledge storage. We use LACUNA to benchmark current SOTA unlearning methods and find that, despite strong output-level performance, existing methods are highly imprecise and susceptible to resurfacing attacks. We further show that when localization is successful, even a simple gradient-based unlearning method achieves strong erasure and robustness to resurfacing attacks, highlighting the importance of precise unlearning. We release LACUNA to complement behavioral evaluations and drive further advances in robust, localization-based unlearning.