Functional Homotopy: Smoothing Discrete Optimization via Continuous Parameters for LLM Jailbreak Attacks
Abstract
Optimization methods are widely employed in deep learning to address and mitigate undesired model responses. While gradient-based techniques have proven effective for image models, their application to language models is hindered by the discrete nature of the input space. This study introduces a novel optimization approach, termed the *functional homotopy* method, which leverages the functional duality between model training and input generation. By constructing a series of easy-to-hard optimization problems, we iteratively solve these using principles derived from established homotopy methods. We apply this approach to jailbreak attack synthesis for large language models (LLMs), achieving a 20%-30% improvement in success rate over existing methods in circumventing established safe open-source models such as Llama-2 and Llama-3.
Cite
Text
Wang et al. "Functional Homotopy: Smoothing Discrete Optimization via Continuous Parameters for LLM Jailbreak Attacks." International Conference on Learning Representations, 2025.Markdown
[Wang et al. "Functional Homotopy: Smoothing Discrete Optimization via Continuous Parameters for LLM Jailbreak Attacks." International Conference on Learning Representations, 2025.](https://mlanthology.org/iclr/2025/wang2025iclr-functional/)BibTeX
@inproceedings{wang2025iclr-functional,
title = {{Functional Homotopy: Smoothing Discrete Optimization via Continuous Parameters for LLM Jailbreak Attacks}},
author = {Wang, Zi and Anshumaan, Divyam and Hooda, Ashish and Chen, Yudong and Jha, Somesh},
booktitle = {International Conference on Learning Representations},
year = {2025},
url = {https://mlanthology.org/iclr/2025/wang2025iclr-functional/}
}