Abstract
We propose RESTA to perform LLM realignment towards safety, which gets compromised due to downstream task fine-tuning. RESTA stands for REstoring Safety through Task Arithmetic. At its core, it involves a simple arithmetic addition of a safety vector to the weights of the compromised model. We demonstrate the effectiveness of RESTA in both parameter-efficient and full fine-tuning, covering a wide range of downstream tasks, including instruction following in Chinese, English, and Hindi, as well as problem-solving capabilities in Code and Math. We also showcase the generalizability of RESTA on three existing safety evaluation benchmarks and a multilingual benchmark dataset proposed as a part of this work, consisting of 550 harmful questions covering 11 categories, each with 5 sub-categories of harm. Overall, RESTA decreases the harmfulness of the compromised model from 18.6% to 5.1% and from 9.2% to 1.5% in parameter-efficient and full fine-tuning, respectively, while maintaining most of the model’s performance on the task. We release the source codes at: https://github.com/declare-lab/resta.- Anthology ID:
- 2024.acl-long.762
- Volume:
- Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
- Month:
- August
- Year:
- 2024
- Address:
- Bangkok, Thailand
- Editors:
- Lun-Wei Ku, Andre Martins, Vivek Srikumar
- Venue:
- ACL
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 14138–14149
- Language:
- URL:
- https://aclanthology.org/2024.acl-long.762
- DOI:
- Cite (ACL):
- Rishabh Bhardwaj, Duc Anh Do, and Soujanya Poria. 2024. Language Models are Homer Simpson! Safety Re-Alignment of Fine-tuned Language Models through Task Arithmetic. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 14138–14149, Bangkok, Thailand. Association for Computational Linguistics.
- Cite (Informal):
- Language Models are Homer Simpson! Safety Re-Alignment of Fine-tuned Language Models through Task Arithmetic (Bhardwaj et al., ACL 2024)
- PDF:
- https://preview.aclanthology.org/nschneid-patch-4/2024.acl-long.762.pdf