CLaRE-ty Amid Chaos: Quantifying Representational Entanglement to Predict Ripple Effects in LLM Editing
arXiv:2603.19297v1 Announce Type: new Abstract: The static knowledge representations of large language models (LLMs) inevitably become outdated or incorrect over time. While model-editing techniques offer …
Manit Baser, Alperen Yildiz, Dinil Mon Divakaran, Mohan Gurusamy
7 views
