<ul><li>Driven by privacy protection laws and regulations, unlearning in Large Language Models (LLMs) is gaining attention.</li><li>Researchers propose a metric called Memory Removal Difficulty (MRD) to quantify sample-level unlearning difficulty.</li><li>The study explores the characteristics of hard-to-unlearn and easy-to-unlearn samples in LLM unlearning.</li><li>An MRD-based weighted sampling method is proposed to optimize existing unlearning algorithms, improving efficiency and effectiveness.</li></ul>

A Neuro-inspired Interpretation of Unlearning in Large Language Models through Sample-level Unlearning Difficulty

Discover more