@inproceedings{yang-etal-2026-calibration,
    title = "When Calibration Rankings Reverse: Accuracy-Controlled Evaluation for Fair Comparison of {LLM}s",
    author = "Yang, Zhichao  and
      Zhang, Caiqi  and
      Yang, Ruihan  and
      Li, Chengzu  and
      Collier, Nigel  and
      Yang, Deqing",
    editor = "Lee, Hung-Yi  and
      Dev, Sunipa  and
      Li, Jessy  and
      Elliott, Desmond",
    booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
    month = oct,
    year = "2026",
    address = "Budapest, Hungary",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2026.emnlp-main.1144/",
    pages = "26206--26231",
    ISBN = "979-8-89176-459-0"
}