Representation-Guided Parameter-Efficient LLM Unlearning

TL;DR

This paper introduces REGLU, a novel representation-guided unlearning method for LLMs that improves the forget-retain balance by leveraging geometric properties of representation spaces.

Contribution

REGLU uses representation-guided initialization and orthogonal regularization to enhance unlearning precision and model utility, surpassing existing methods.

Findings

REGLU outperforms state-of-the-art baselines on TOFU and WMDP benchmarks.

REGLU achieves better unlearning quality with higher model utility.

The method effectively disentangles forget and retain information in LLMs.

Abstract

Large Language Models (LLMs) often memorize sensitive or harmful information, necessitating effective machine unlearning techniques. While existing parameter-efficient unlearning methods have shown promise, they still struggle with the forget-retain trade-off. This can be attributed to their reliance on parameter importance metrics to identify parameters that are important exclusively for the forget set, which is fundamentally limited by the superposition phenomenon. Due to the polysemantic nature of LLM parameters, such an importance metric may struggle to disentangle parameters associated with the forget and retain sets. In this work, we propose Representation-Guided Low-rank Unlearning (REGLU), a novel approach that leverages the geometric properties of representation spaces to achieve robust and precise unlearning. First, we develop a representation-guided initialization for LoRA that identifies the optimal subspace for selective forgetting. Second, we introduce a regularization loss that constrains the outputs of the LoRA update to lie in the orthogonal complement of the retain set's representation subspace, thereby minimizing interference with the model's performance on the retain set. We evaluate REGLU on the TOFU and WMDP benchmarks across multiple models. Our results demonstrate that REGLU consistently outperforms state-of-the-art baselines, achieving superior unlearning quality while maintaining higher model utility.