TY  - RPRT
TI  - Are All Prompt Components Value-Neutral? Understanding the Heterogeneous Adversarial Robustness of Dissected Prompt in Large Language Models
AU  - Yujia Zheng
AU  - Tianhao Li
AU  - Haotian Huang
AU  - Tianyu Zeng
AU  - Jingyu Lu
AU  - Chuangxin Chu
AU  - Yuekai Huang
AU  - Ziyou Jiang
AU  - Qian Xiong
AU  - Yuyao Ge
AU  - Mingyang Li
PY  - 2025
UR  - https://arxiv.org/abs/2508.01554
ID  - 2508.01554
ER  -