EMNLP

Does Localization Inform Unlearning? A Rigorous Examination of Local Parameter Attribution for Knowledge Unlearning in Language Models

定位信息是否有助于去学习？对语言模型中知识去学习的局部参数归因的严格检验

Hwiyeong Lee, Uiji Hwang, Hyelim Lim, Taeuk Kim

AI总结 本文通过实验验证局部参数更新对语言模型知识去学习的有效性，发现参数局部性并不必然指示有效知识去除。

Comments The 2025 Conference on Empirical Methods in Natural Language Processing (EMNLP 2025)