Latent Jailbreak

🎉 Paper

This repository contains the code and data for the paper Latent Jailbreak: A Benchmark for Evaluating Text Safety and Output Robustness of Large Language Models. The paper explores the topic of latent jailbreak and presents a novel approach to evaluate the text safety and output robustness for large language models.

Data

The data used in this paper is included in the data directory.

Templates

Templates for latent jailbreak prompts.

Generate Model Responses

cd src
python BELLE_7B_2M.py
python ChatGLM2-6B.py
python ChatGPT.py --api_key 'your key'

Fine-Tune Model to Perform Automatic Labeling

python finetune.py

Citation

If you use the code or data in this repository, please cite the following paper.

@misc{qiu2023latent,
      title={Latent Jailbreak: A Benchmark for Evaluating Text Safety and Output Robustness of Large Language Models},
      author={Huachuan Qiu and Shuai Zhang and Anqi Li and Hongliang He and Zhenzhong Lan},
      year={2023},
      eprint={2307.08487},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}