Safety Arithmetic: A Framework for Test-time Safety Alignment of Language Models by Steering Parameters and Activations

Safety Arithmetic: A Framework for Test-time Safety Alignment of Language Models by Steering Parameters and Activations Rima Hazra author Sayan Layek author Somnath Banerjee author Soujanya Poria author 2024-11 text Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing Yaser Al-Onaizan editor Mohit Bansal editor Yun-Nung Chen editor Association for Computational Linguistics Miami, Florida, USA conference publication hazra-etal-2024-safety 10.18653/v1/2024.emnlp-main.1212 https://aclanthology.org/2024.emnlp-main.1212/ 2024-11 21759 21776