@inproceedings{zhang-etal-2025-verifiable,
    title = "Verifiable by Design: Aligning Language Models to Quote from Pre-Training Data",
    author = "Zhang, Jingyu  and
      Marone, Marc  and
      Li, Tianjian  and
      Van Durme, Benjamin  and
      Khashabi, Daniel",
    editor = "Chiruzzo, Luis  and
      Ritter, Alan  and
      Wang, Lu",
    booktitle = "Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers)",
    month = apr,
    year = "2025",
    address = "Albuquerque, New Mexico",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/automate-pypi-publishing/2025.naacl-long.191/",
    doi = "10.18653/v1/2025.naacl-long.191",
    pages = "3748--3768",
    ISBN = "979-8-89176-189-6",
    abstract = "To trust the fluent generations of large language models (LLMs), humans must be able to \textit{verify} their correctness against trusted, external sources. Recent efforts, such as providing citations via retrieved documents or post-hoc provenance, enhance verifiability but provide no guarantees on their correctness. To address these limitations, we tackle the verifiability goal with a different philosophy: {\_}trivializing the verification process by developing models that quote verbatim statements from trusted sources in their pre-training data.{\_}We propose Quote-Tuning, which demonstrates the feasibility of aligning models to quote. The core of Quote-Tuning is a fast membership inference function that efficiently verifies text against trusted corpora. We leverage this tool to design a reward function to quantify quotes in model responses, and curate datasets for preference learning. Experiments show that Quote-Tuning significantly increases verbatim quotes from high-quality documents by up to 130{\%} relative to base models while maintaining response quality. Quote-Tuning is applicable in different tasks, generalizes to out-of-domain data and diverse model families, and provides additional benefits to truthfulness. Our method not only serves as a hassle-free method to increase quoting but also opens up avenues for improving LLM trustworthiness through better verifiability."
}