Fundamental Problems With Model Editing: How Should Rational Belief Revision Work in LLMs?

Hase, Peter, Hofweber, Thomas, Zhou, Xiang · arXiv · 2024