@inproceedings{tjuatja-neubig-2025-behaviorbox,
    title = "{B}ehavior{B}ox: Automated Discovery of Fine-Grained Performance Differences Between Language Models",
    author = "Tjuatja, Lindia  and
      Neubig, Graham",
    editor = "Che, Wanxiang  and
      Nabende, Joyce  and
      Shutova, Ekaterina  and
      Pilehvar, Mohammad Taher",
    booktitle = "Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
    month = jul,
    year = "2025",
    address = "Vienna, Austria",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingestion-acl-25/2025.acl-long.923/",
    pages = "18851--18873",
    ISBN = "979-8-89176-251-0"
}