Unlearning in- vs. out-of-distribution data in LLMs under gradient-based method
This work addresses the challenge of effectively removing data influence in LLMs, which is incremental as it builds on existing gradient-based methods to analyze data properties like memorization and difficulty.
The paper tackles the problem of evaluating and understanding machine unlearning in large language models, particularly for in-distribution vs. out-of-distribution data, finding that unlearning out-of-distribution examples requires more steps but offers a better trade-off, while in-distribution examples lead to rapid performance decay.
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.