@inproceedings{d-etal-2026-bitzkrieg,
    title = "Bitzkrieg at {S}em{E}val-2026 Task 13: Calibration-Aware Dual {C}ode{BERT} for Multilingual Machine-Generated Code Detection",
    author = "D., Thenmozhi  and
      S, Adithya  and
      Malisetty, Harshil  and
      P, Aadit  and
      R, Rohan",
    editor = "Kochmar, Ekaterina  and
      Ghosh, Debanjan  and
      North, Kai  and
      Komachi, Mamoru",
    booktitle = "Proceedings of the 20th {I}nternational {W}orkshop on {S}emantic {E}valuation (2026)",
    month = jul,
    year = "2026",
    address = "San Diego, California, USA",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingest-acl-workshops/2026.semeval-1.282/",
    pages = "2233--2237",
    ISBN = "979-8-89176-414-9",
    abstract = "We describe our submission to SemEval-2026 Task 13, addressing binary detection (Subtask A), generator attribution (Subtask B), and hybrid/adversarial authorship classification (Subtask C) of machine-generated code (MGC). For Subtask A, we fine-tune two CodeBERT models with complementary sampling strategies and apply percentile-based post-hoc calibration, improving Macro-F1 from 0.47 to 0.56 without additional training. For Subtask B, we combine TF-IDF n-grams, frozen CodeBERT embeddings, and language features with XGBoost, using synthetic augmentation and class weighting to handle an 11-class dataset skewed 88{\%} toward the human class, achieving Macro-F1 of 0.289. For Subtask C, we fine-tune a CodeBERT classifier for four-way authorship classification, achieving Macro-F1 of 0.49. Our results highlight the importance of probability calibration for binary detection and class balancing for multi-class attribution."
}