{"slug": "team-dactyl-at-pan-2026-bayesian-data-mixing-and-empirical-x-risk-minimization", "title": "Team DACTYL at PAN 2026: Bayesian Data Mixing and Empirical X-risk Minimization for AI-text Detection", "summary": "Team DACTYL's MCGrad model achieved a mean score of 0.974 on the PAN 2026 test set, ranking second on the leaderboard for AI-generated text detection. The researchers fine-tuned BERT-tiny models with Bayesian classification heads to select training data across three datasets, addressing out-of-distribution performance issues. Their DeBERTa-V3-large and ModernBERT-large models are released on Hugging Face.", "body_md": "arXiv:2607.17382v1 Announce Type: new\nAbstract: Existing research shows that AI-generated text detection classifiers achieve strong in-distribution (ID) performance but do not maintain the same performance on out-of-distribution (OOD) texts, suggesting overfitting to dataset-specific features. However, combining different training datasets doesn't always improve performance and, in some cases, can even encourage shortcut learning. To address this issue, we fine-tune BERT-tiny models with Bayesian classification heads to select texts across three different datasets to use as a consolidated training set. We trained three different classifiers: fine-tuned DeBERTa-V3-large and ModernBERT-large classifiers via empirical X-risk minimization, and an MCGrad model that calibrates the predictions from the ModernBERT-large classifier. The DeBERTa-V3-large-large classifier achieves a mean score of 0.882 on the PAN 2026 test set across five metrics: AUROC, $F_1$, C@1, Brier score, and $F_{0.5u}$. ModernBERT-large achieves a score of 0.96 while MCGrad achieves the best score of the three with a mean score of 0.974, ranking second on the leaderboard. Our results highlight that careful dataset curation can lead to strong OOD performance. We release our ModernBERT-large and DeBERTa-V3-large models at https://huggingface.co/collections/ShantanuT01/panclef-2026 .", "url": "https://wpnews.pro/news/team-dactyl-at-pan-2026-bayesian-data-mixing-and-empirical-x-risk-minimization", "canonical_source": "https://www.machinebrief.com/news/team-dactyl-at-pan-2026-bayesian-data-mixing-and-empirical-x-pvdi", "published_at": "2026-07-21 04:00:00+00:00", "updated_at": "2026-07-21 05:34:29.569602+00:00", "lang": "en", "topics": ["artificial-intelligence", "machine-learning", "natural-language-processing"], "entities": ["Team DACTYL", "PAN 2026", "BERT-tiny", "DeBERTa-V3-large", "ModernBERT-large", "MCGrad", "Hugging Face"], "alternates": {"html": "https://wpnews.pro/news/team-dactyl-at-pan-2026-bayesian-data-mixing-and-empirical-x-risk-minimization", "markdown": "https://wpnews.pro/news/team-dactyl-at-pan-2026-bayesian-data-mixing-and-empirical-x-risk-minimization.md", "text": "https://wpnews.pro/news/team-dactyl-at-pan-2026-bayesian-data-mixing-and-empirical-x-risk-minimization.txt", "jsonld": "https://wpnews.pro/news/team-dactyl-at-pan-2026-bayesian-data-mixing-and-empirical-x-risk-minimization.jsonld"}}