@inproceedings{nguyen-etal-2025-distributional,
    title = "Distributional Surgery for Language Model Activations",
    author = "Nguyen, Bao  and
      Nguyen, Binh  and
      Nguyen, Duy  and
      Nguyen, Viet Anh",
    editor = "Christodoulopoulos, Christos  and
      Chakraborty, Tanmoy  and
      Rose, Carolyn  and
      Peng, Violet",
    booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2025",
    month = nov,
    year = "2025",
    address = "Suzhou, China",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/author-page-yu-wang-polytechnic/2025.findings-emnlp.435/",
    doi = "10.18653/v1/2025.findings-emnlp.435",
    pages = "8192--8212",
    ISBN = "979-8-89176-335-7",
    abstract = "Language models, while capable of generating remarkably coherent and seemingly accurate text, can occasionally produce undesirable content including harmful or toxic outputs. In this paper, we present a new two-stage approach to detect and mitigate undesirable content generations by rectifying activations. First, we train an ensemble of layerwise classifiers to detect undesirable content using activations by minimizing a smooth surrogate of the risk-aware score. Then, for detected undesirable contents, we propose layerwise distributional steering policies that transform the attention heads. These policies are computed through principled semidefinite programming aims to minimally perturb the attention distribution while probabilistically guaranteeing the effectiveness of the editions. Empirical evaluations across multiple language models and datasets show that our method outperforms baselines in reducing the generation of undesirable output."
}