@inproceedings{lior-etal-2025-reliableeval,
    title = "{R}eliable{E}val: A Recipe for Stochastic {LLM} Evaluation via Method of Moments",
    author = "Lior, Gili  and
      Habba, Eliya  and
      Levy, Shahar  and
      Caciularu, Avi  and
      Stanovsky, Gabriel",
    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.594/",
    doi = "10.18653/v1/2025.findings-emnlp.594",
    pages = "11146--11153",
    ISBN = "979-8-89176-335-7",
    abstract = "LLMs are highly sensitive to prompt phrasing, yet standard benchmarks typically report performance using a single prompt, raising concerns about the reliability of such evaluations. In this work, we argue for a stochastic method of moments evaluation over the space of meaning-preserving prompt perturbations. We introduce a formal definition of *reliable evaluation* that accounts for prompt sensitivity, and suggest ReliableEval - a method for estimating the number of prompt resamplings needed to obtain meaningful results. Using our framework, we stochastically evaluate five frontier LLMs and find that even top-performing models like GPT-4o and Claude-3.7-Sonnet exhibit substantial prompt sensitivity. Our approach is model-, task-, and metric-agnostic, offering a recipe for meaningful and robust LLM evaluation."
}