@misc{abadji2026lagunam1xs2technicalreport,
      title={Laguna M.1/XS.2 Technical Report}, 
      author={Julien Abadji and Marah Abdin and Connor Adams and Eric Alcaide and Mustafa Altun and Michele Artoni and Junze Bao and Uday Barar and Vassilis Bekiaris and Arkadii Bessonov and Benjamin Bütikofer and Jonathan Chang and Yen-Chun Chen and Dmitry Chernenkov and Yang Chi and Filippos Christianos and Fenia Christopoulou and Razvan-Andrei Ciocoiu and Tzachi Cohen and Yohann Coppel and Dmitrii Emelianenko and Brandon Fergerson and Brian Fitzgerald and Matthias Gallé and Alex Golonzovskyi and George Grigorev and Yiyang Hao and Christian Hensel and Jan Huenermann and Ye Ji and Sarthak Joshi and Eiso Kant and Kabir Khandpur and Seonghyeon Kim and Vladimir Kirichenko and Umut Kocasarac and Ilya Kochik and Ivan Komarov and Chaerin Kong and Anurag Koul and François-Joseph Lacroix and Sergei Laktionov and Waren Long and Quentin Malartic and Vadim Markovtsev and Afonso Marques and Robert McHardy and Carlos Mocholí and Dmitry Monakhov and Adam Morris and Martin Muller and Christian Mürtz and Robin Nabel and Thien Nguyen and Rok Novosel and Szymon Ozog and Aalhad Patankar and Aleksei Petrov and Alexandre Piché and Arthur Pignet and Teodor Poncu and Phil Potter and Alexander Rakowski and Pierre-Yves Ritschard and Jay Roberts and Joe Rowell and Piotr Sarna and Pierre-André Savalle and Uladzislau Sazanovich and Nikita Shapovalov and Arsenii Shevchenko and Mikhail Shilkov and Andrei Sokol and Mohamed Soliman and Jack Stephenson and Victor Storchan and Dragos-Constantin Tantaru and Artem Tyurin and Adrian Wälchli and Pengming Wang and Jianxiao Yang and Renat Zayashnikov and Alexander Zelenka Martin and Nikolay Zinov and Caroline Bercier and José Caldeira and Margarida Garcia and Tom George and Kabeer Gharzai and Glenn Hitchcock and Carson Klingenberg and Ivo Pinto and Varun Randery and Noah Smith and Arina Sugako and Jason Warner},
      year={2026},
      eprint={2605.27605},
      archivePrefix={arXiv},
      primaryClass={cs.AI},
      url={https://arxiv.org/abs/2605.27605}, 
}

@inproceedings{gema-etal-2025-done,
    title = "Are We Done with {MMLU}?",
    author = "Gema, Aryo Pradipta  and
      Leang, Joshua Ong Jun  and
      Hong, Giwon  and
      Devoto, Alessio  and
      Mancino, Alberto Carlo Maria  and
      Saxena, Rohit  and
      He, Xuanli  and
      Zhao, Yu  and
      Du, Xiaotang  and
      Ghasemi Madani, Mohammad Reza  and
      Barale, Claire  and
      McHardy, Robert  and
      Harris, Joshua  and
      Kaddour, Jean  and
      Van Krieken, Emile  and
      Minervini, Pasquale",
    editor = "Chiruzzo, Luis  and
      Ritter, Alan  and
      Wang, Lu",
    booktitle = "Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers)",
    month = apr,
    year = "2025",
    address = "Albuquerque, New Mexico",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2025.naacl-long.262/",
    pages = "5069--5096",
    ISBN = "979-8-89176-189-6",
    abstract = "Maybe not. We identify and analyse errors in the popular Massive Multitask Language Understanding (MMLU) benchmark. Even though MMLU is widely adopted, our analysis demonstrates numerous ground truth errors that obscure the true capabilities of LLMs. For example, we find that 57{\%} of the analysed questions in the Virology subset contain errors. To address this issue, we introduce a comprehensive framework for identifying dataset errors using a novel error annotation protocol. Then, we create MMLU-Redux, which is a subset of 5,700 manually re-annotated questions across all 57 MMLU subjects. Using MMLU-Redux, we demonstrate significant discrepancies with the model performance metrics that were originally reported. Our results strongly advocate for revising MMLU`s error-ridden questions to enhance its future utility and reliability as a benchmark. Therefore, we open up MMLU-Redux for additional annotation."
}

@article{ramirez2024anatomy,
      title={Anatomy of Industrial Scale Multilingual ASR},
      author={McCann Ramirez, Francis and Chkhetiani, Luka and Ehrenberg, Andrew and McHardy, Robert and Botros, Rami and Khare, Yash and Vanzo, Andrea and Peyash, Taufiquzzaman and Oexle, Gabriel and Liang, Michael and Sklyar, Ilya and Fakhan, Enver and Etefy, Ahmed and McCrystal, Daniel and Flamini, Sam and Donato, Domenic and Yoshioka, Takuya},
      year={2024},
      journal={arXiv preprint arXiv:2404.09841},
}

@inproceedings{delaunay-etal-2023-lightmhc,
    title = "LightMHC: A Light Model for pMHC Structure Prediction with Graph Neural Networks",
    author = "Delaunay, Antoine P.  and
      Fu, Yunguan  and
      Gorbushin, Nikolai and
      McHardy, Robert and
      Djermani, Bachir A. and
      Copoiu, Liviu and
      Rooney, Michael and
      Lang, Maren and
      Tovchigrechko, Andrey and
      Şahin, Uğur and
      Beguir, Karim and
      Lopez Carranza, Nicolas
      ",
    booktitle = "Machine Learning in Structural Biology Workshop at the 37th Conference on Neural Information Processing Systems (NeurIPS)",
    month = dec,
    year = "2023",
    address = "New Orleans, Louisiana",
    publisher = "",
    url = "https://www.mlsb.io/",
    doi = "10.1101/2023.11.21.568015",
    pages = "",
}

@article{kaddour2023challenges,
  title={Challenges and Applications of Large Language Models},
  author={Kaddour, Jean and Harris, Joshua and Mozes, Maximilian and Bradley, Herbie and Raileanu, Roberta and McHardy, Robert},
  journal={arXiv preprint arXiv:2307.10169},
  year={2023}
}

@inproceedings{delaunay-etal-2022-gnn,
    title = "Peptide-MHC Structure Prediction With Mixed Residue and Atom Graph Neural Network",
    author = "Delaunay, Antoine P.  and
      Fu, Yunguan  and
      Bégué, Alberto and
      McHardy, Robert and
      Djermani, Bachir A. and
      Rooney, Michael and
      Tovchigrechko, Andrey and
      Copoiu, Liviu and
      Skwark, Marcin J. and
      Lopez Carranza, Nicolas and
      Lang, Maren and
      Beguir, Karim and
      Şahin, Uğur
      ",
    booktitle = "Machine Learning in Structural Biology Workshop at the 36th Conference on Neural Information Processing Systems (NeurIPS)",
    month = dec,
    year = "2022",
    address = "New Orleans, Louisiana",
    publisher = "",
    url = "https://www.mlsb.io/",
    doi = "10.1101/2022.11.23.517618",
    pages = "",
}

@inproceedings{mchardy-etal-2019-adversarial,
    title = "Adversarial Training for Satire Detection: Controlling for Confounding Variables",
    author = "McHardy, Robert  and
      Adel, Heike  and
      Klinger, Roman",
    booktitle = "Proceedings of the 2019 Conference of the North {A}merican Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers)",
    month = jun,
    year = "2019",
    address = "Minneapolis, Minnesota",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/N19-1069",
    doi = "10.18653/v1/N19-1069",
    pages = "660--665",
    abstract = "The automatic detection of satire vs. regular news is relevant for downstream applications (for instance, knowledge base population) and to improve the understanding of linguistic characteristics of satire. Recent approaches build upon corpora which have been labeled automatically based on article sources. We hypothesize that this encourages the models to learn characteristics for different publication sources (e.g., {``}The Onion{''} vs. {``}The Guardian{''}) rather than characteristics of satire, leading to poor generalization performance to unseen publication sources. We therefore propose a novel model for satire detection with an adversarial component to control for the confounding variable of publication source. On a large novel data set collected from German news (which we make available to the research community), we observe comparable satire classification performance and, as desired, a considerable drop in publication classification performance with adversarial training. Our analysis shows that the adversarial component is crucial for the model to learn to pay attention to linguistic properties of satire.",
}