Align-then-Unlearn: Embedding Alignment for LLM Unlearning

TL;DR

This paper introduces Align-then-Unlearn, a novel embedding space approach for effectively removing specific knowledge from large language models while preserving overall utility.

Contribution

It proposes a new framework that unlearns targeted information by aligning semantic embeddings, improving robustness over token-level methods.

Findings

Effective removal of targeted knowledge demonstrated

Minimal impact on overall model performance

Embedding-based unlearning outperforms token-level approaches

Abstract

As large language models (LLMs) are trained on massive datasets, they have raised significant privacy and ethical concerns due to their potential to inadvertently retain sensitive information. Unlearning seeks to selectively remove specific data from trained models, such as personal information or copyrighted content. Current approaches targeting specific output sequences at the token level often fail to achieve complete forgetting and remain susceptible to prompt rephrasing. We propose Align-then-Unlearn, a novel framework that performs unlearning in the semantic embedding space rather than directly on output tokens. Align-then-Unlearn first augments the LLM with an embedding prediction module trained to anticipate future context representations. Unlearning is then achieved by fine-tuning the model to minimize the similarity between these predicted embeddings and a target embedding that represents the concept to be removed. Initial results show that Align-then-Unlearn effectively removes targeted knowledge with minimal degradation in overall model utility. These findings suggest that embedding-based unlearning offers a promising and robust approach to removing conceptual knowledge. Our code is available at https://github.com/ExplainableML/align-then-unlearn.