Combining Domain and Alignment Vectors Provides Better Knowledge-Safety Trade-offs in LLMs
Megh Thakkar, Quentin Fournier, Matthew Riemer, Pin-Yu Chen, Amal Zouaq, Payel Das, Sarath Chandar. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers). 2025.
Authors
- Sarath Chandar
- Matthew Riemer
- Megh Thakkar (ORCID: https://orcid.org/0009-0004-8224-622X)
- Pin‐Yu Chen (ORCID: https://orcid.org/0000-0003-1039-8369)
- Payel Das (ORCID: https://orcid.org/0000-0002-7288-0516)
- Amal Zouaq (ORCID: https://orcid.org/0000-0002-4791-0752)
- Quentin Fournier
Publication Details
- Published
- 2025-01-01
- DOI
- https://doi.org/10.18653/v1/2025.acl-short.22
- Primary Topic
- Digital Rights Management and Security
- Type
- article
- Field-Weighted Citation Impact
- 0.00
Funders
- Canadian Institute for Advanced Research
- Alliance de recherche numérique du Canada
- Samsung
- Natural Sciences and Engineering Research Council of Canada