ROSE Doesn’t Do That: Boosting the Safety of Instruction-Tuned Large Language Models with Reverse Prompt Contrastive Decoding

Qihuang Zhong, Liang Ding, Juhua Liu, Bo Du, Dacheng Tao


Abstract
With the development of instruction-tuned large language models (LLMs), improving the safety of LLMs has become more critical. However, the current approaches for aligning the LLMs output with expected safety usually require substantial training efforts, e.g., high-quality safety data and expensive computational resources, which are costly and inefficient. To this end, we present reverse prompt contrastive decoding (ROSE), a simple-yet-effective method to directly boost the safety of existing instruction-tuned LLMs without any additional training. The principle of ROSE is to improve the probability of desired safe output via suppressing the undesired output induced by the carefully-designed reverse prompts. Experiments on 6 safety and 2 general-purpose tasks show that, our ROSE not only brings consistent and significant safety improvements (up to +13.8% safety score) upon 5 types of instruction-tuned LLMs, but also benefits the general-purpose ability of LLMs. In-depth analyses explore the underlying mechanism of ROSE, and reveal when and where to use it.
Anthology ID:
2024.findings-acl.814
Volume:
Findings of the Association for Computational Linguistics: ACL 2024
Month:
August
Year:
2024
Address:
Bangkok, Thailand
Editors:
Lun-Wei Ku, Andre Martins, Vivek Srikumar
Venue:
Findings
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
13721–13736
Language:
URL:
https://aclanthology.org/2024.findings-acl.814
DOI:
10.18653/v1/2024.findings-acl.814
Bibkey:
Cite (ACL):
Qihuang Zhong, Liang Ding, Juhua Liu, Bo Du, and Dacheng Tao. 2024. ROSE Doesn’t Do That: Boosting the Safety of Instruction-Tuned Large Language Models with Reverse Prompt Contrastive Decoding. In Findings of the Association for Computational Linguistics: ACL 2024, pages 13721–13736, Bangkok, Thailand. Association for Computational Linguistics.
Cite (Informal):
ROSE Doesn’t Do That: Boosting the Safety of Instruction-Tuned Large Language Models with Reverse Prompt Contrastive Decoding (Zhong et al., Findings 2024)
Copy Citation:
PDF:
https://aclanthology.org/2024.findings-acl.814.pdf