@inproceedings{wang-etal-2025-beyond-prompt,
    title = "Beyond Prompt Engineering: Robust Behavior Control in {LLM}s via Steering Target Atoms",
    author = "Wang, Mengru  and
      Xu, Ziwen  and
      Mao, Shengyu  and
      Deng, Shumin  and
      Tu, Zhaopeng  and
      Chen, Huajun  and
      Zhang, Ningyu",
    editor = "Che, Wanxiang  and
      Nabende, Joyce  and
      Shutova, Ekaterina  and
      Pilehvar, Mohammad Taher",
    booktitle = "Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
    month = jul,
    year = "2025",
    address = "Vienna, Austria",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingestion-acl-25/2025.acl-long.1139/",
    pages = "23381--23399",
    ISBN = "979-8-89176-251-0",
    abstract = "Precise control over language model generation is vital for ensuring both safety and reliability. Although prompt engineering and steering are commonly used to intervene in model behaviors, the vast number of parameters in models often results in highly intertwined internal representations. This interdependency can limit control precision and sometimes lead to unintended side effects. Recent research has explored the use of sparse autoencoders (SAE) to disentangle knowledge in high-dimensional spaces for steering.However, these applications have been limited to toy tasks owing to the nontrivial issue of locating ``atomic knowledge components''. In this paper, we propose Steering Target Atoms (STA), a novel method that isolates and manipulates disentangled knowledge components to enhance safety. Comprehensive experiments demonstrate the effectiveness of our approach. Further analysis reveals that steering exhibits superior robustness and flexibility, particularly in adversarial scenarios. We also apply the steering strategy to the large reasoning model, confirming its effectiveness in precise reasoning control."
}