@inproceedings{sheng-etal-2025-regularized,
    title = "Regularized Contrastive Decoding with Hard Negative Samples for {LLM} Hallucination Mitigation",
    author = "Sheng, Haonan  and
      Hu, Dou  and
      Wei, Lingwei  and
      Zhou, Wei  and
      Hu, Songlin",
    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.322/",
    doi = "10.18653/v1/2025.findings-emnlp.322",
    pages = "6061--6073",
    ISBN = "979-8-89176-335-7",
    abstract = "Large language models are prone to generate hallucinations, which can undermine their reliability in high-stakes applications. Some works on LLM hallucination mitigation use the model{'}s internal signals to contrast different output during inference stage. However, these works often focus on simple forms of hallucinations, and struggle to effectively mitigate hallucinations. To address the issue, this paper exploits hard negative samples to construct a factually weaker model for improving contrastive decoding. We propose a new inference-time method, Regularized Contrastive Decoding (RCD), to capture correct hallucination signals for mitigating hallucinations in LLMs. RCD learns more diverse hallucination patterns via adversarial-aware fine-tuning and mitigates hallucinations via contrastive decoding. Experiments on four hallucination benchmarks demonstrate that our method achieves better LLM hallucination mitigation performance. Further analysis shows RCD generalizes well across different model sizes, task formats, perturbation methods and training data sizes."
}