Intrinsic Test of Unlearning Using Parametric Knowledge Traces

TL;DR

This paper introduces a new internal evaluation method for unlearning in large language models, focusing on changes in parametric knowledge traces, and presents a benchmark dataset to assess unlearning effectiveness.

Contribution

It proposes a novel parameter-based evaluation approach for unlearning, including the ConceptVectors benchmark dataset, revealing limitations of existing behavioral assessments.

Findings

Existing unlearning methods minimally affect concept vectors

Ablating concept vectors removes associated knowledge effectively

Behavioral tests may not reflect true unlearning success

Abstract

The task of "unlearning" certain concepts in large language models (LLMs) has attracted immense attention recently, due to its importance in mitigating undesirable model behaviours, such as the generation of harmful, private, or incorrect information. Current protocols to evaluate unlearning methods largely rely on behavioral tests, without monitoring the presence of unlearned knowledge within the model's parameters. This residual knowledge can be adversarially exploited to recover the erased information post-unlearning. We argue that unlearning should also be evaluated internally, by considering changes in the parametric knowledge traces of the unlearned concepts. To this end, we propose a general evaluation methodology that leverages vocabulary projections to inspect concepts encoded in model parameters. We use this approach to localize "concept vectors" - parameter vectors that encode concrete concepts - and construct ConceptVectors, a benchmark dataset containing hundreds of common concepts and their parametric knowledge traces within two open-source LLMs. Evaluation on ConceptVectors shows that existing unlearning methods minimally impact concept vectors and mostly suppress them during inference, while directly ablating these vectors demonstrably removes the associated knowledge and significantly reduces the model's susceptibility to adversarial manipulation. Our results highlight limitations in behavioral-based unlearning evaluations and call for future work to include parameter-based evaluations. To support this, we release our code and benchmark at https://github.com/yihuaihong/ConceptVectors.