Unlearning in- vs. out-of-distribution data in LLMs under gradient-based method

TL;DR

This paper introduces a metric to evaluate unlearning in large language models, analyzing the trade-offs between unlearning quality and model performance for in- and out-of-distribution data, highlighting different challenges in each case.

Contribution

It formalizes a new metric for unlearning quality in generative models and systematically assesses the trade-offs and challenges in unlearning in- versus out-of-distribution data.

Findings

Unlearning out-of-distribution data requires more steps but offers better trade-offs.

Unlearning in-distribution data causes rapid performance decay.

Memorization and difficulty influence unlearning effectiveness.

Abstract

Machine unlearning aims to solve the problem of removing the influence of selected training examples from a learned model. Despite the increasing attention to this problem, it remains an open research question how to evaluate unlearning in large language models (LLMs), and what are the critical properties of the data to be unlearned that affect the quality and efficiency of unlearning. This work formalizes a metric to evaluate unlearning quality in generative models, and uses it to assess the trade-offs between unlearning quality and performance. We demonstrate that unlearning out-of-distribution examples requires more unlearning steps but overall presents a better trade-off overall. For in-distribution examples, however, we observe a rapid decay in performance as unlearning progresses. We further evaluate how example's memorization and difficulty affect unlearning under a classical gradient ascent-based approach.