SOUL: Unlocking the Power of Second-Order Optimization for LLM Unlearning

TL;DR

This paper introduces SOUL, a second-order optimization framework for LLM unlearning, demonstrating that second-order methods outperform first-order approaches in removing undesired data influences while maintaining model utility.

Contribution

The paper reveals the importance of optimizer choice in LLM unlearning and develops a novel second-order unlearning method that improves effectiveness over traditional first-order techniques.

Findings

SOUL outperforms first-order methods across various tasks.

Second-order optimization enhances unlearning effectiveness.

SOUL is broadly applicable to different models and metrics.

Abstract

Large Language Models (LLMs) have highlighted the necessity of effective unlearning mechanisms to comply with data regulations and ethical AI practices. LLM unlearning aims at removing undesired data influences and associated model capabilities without compromising utility beyond the scope of unlearning. While interest in studying LLM unlearning is growing, the impact of the optimizer choice for LLM unlearning remains unexplored. In this work, we shed light on the significance of optimizer selection in LLM unlearning for the first time, establishing a clear connection between second-order optimization and influence unlearning (a classical approach using influence functions to update the model for data influence removal). This insight propels us to develop a second-order optimization-based LLM unlearning framework, termed Second-Order UnLearning (SOUL), which extends the static, one-shot model update using influence unlearning to a dynamic, iterative unlearning process. Our extensive experiments show that SOUL consistently outperforms conventional first-order methods across various unlearning tasks, models, and metrics, indicating that second-order optimization offers an effective and broadly applicable solution for LLM unlearning. Codes are available at https://github.com/OPTML-Group/SOUL.