HALLUCANA: Fixing LLM Hallucination with A Canary Lookahead

Tianyi Li; Erenay Dayanik; Shubhi Tyagi; Andrea Pierleoni

NAACL 2025findings1 citations

HALLUCANA: Fixing LLM Hallucination with A Canary Lookahead

Tianyi Li, Erenay Dayanik, Shubhi Tyagi, Andrea Pierleoni

Abstract

In this paper, we present HALLUCANA, a canary lookahead to detect and correct factual hallucinations of Large Language Models (LLMs) in long-form generation. HALLUCANA detects and intervenes as soon as traces of hallucination emerge, during and even before generation. To support timely detection, we exploit the internal factuality representation in the LLM hidden space, where we investigate various proxies to the LLMs’ factuality self-assessment, and discuss its relation to the models’ context familiarity from their pre-training. On biography generation, our method improves generation quality by up to 2.5x, while consuming over 6 times less compute.

BibTeX

@inproceedings{li-etal-2025-hallucana,
    title = "{HALLUCANA}: Fixing {LLM} Hallucination with A Canary Lookahead",
    author = "Li, Tianyi  and
      Dayanik, Erenay  and
      Tyagi, Shubhi  and
      Pierleoni, Andrea",
    editor = "Chiruzzo, Luis  and
      Ritter, Alan  and
      Wang, Lu",
    booktitle = "Findings of the Association for Computational Linguistics: NAACL 2025",
    month = apr,
    year = "2025",
    address = "Albuquerque, New Mexico",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2025.findings-naacl.12/",
    pages = "213--230",
    ISBN = "979-8-89176-195-7"
}