Towards Robust Evaluation of Unlearning in LLMs via Data Transformations

TL;DR

This paper investigates the robustness of machine unlearning techniques in large language models by examining how data transformations affect their ability to forget specific information, emphasizing the importance of diverse data formats for reliable evaluation.

Contribution

It introduces a method to evaluate unlearning robustness in LLMs through data transformations, highlighting the need for diverse data formats in unlearning assessment.

Findings

Data transformations can impact the effectiveness of unlearning in LLMs.

Diverse data formats are essential for reliable evaluation of unlearning.

Current unlearning techniques may not fully prevent information recall after data format changes.

Abstract

Large Language Models (LLMs) have shown to be a great success in a wide range of applications ranging from regular NLP-based use cases to AI agents. LLMs have been trained on a vast corpus of texts from various sources; despite the best efforts during the data pre-processing stage while training the LLMs, they may pick some undesirable information such as personally identifiable information (PII). Consequently, in recent times research in the area of Machine Unlearning (MUL) has become active, the main idea is to force LLMs to forget (unlearn) certain information (e.g., PII) without suffering from performance loss on regular tasks. In this work, we examine the robustness of the existing MUL techniques for their ability to enable leakage-proof forgetting in LLMs. In particular, we examine the effect of data transformation on forgetting, i.e., is an unlearned LLM able to recall forgotten information if there is a change in the format of the input? Our findings on the TOFU dataset highlight the necessity of using diverse data formats to quantify unlearning in LLMs more reliably.