@inproceedings{huber-etal-2026-cosy,
    title = "{C}o{S}y: Conversational Synthesis for Grounded Question Answering",
    author = "Huber, Patrick  and
      Einolghozati, Arash  and
      Conway, Rylan  and
      Narang, Kanika  and
      Smith, Matt  and
      Nayyar, Waqar  and
      Sagar, Adithya  and
      Aly, Ahmed A  and
      Shrivastava, Akshat",
    editor = "Mille, Simon  and
      Gehrmann, Sebastian  and
      Schmidtov{\'a}, Patr{\'i}cia  and
      Du{\v{s}}ek, Ond{\v{r}}ej  and
      Fadaee, Marzieh  and
      Lo, Kyle  and
      Santus, Enrico  and
      Stanovsky, Gabriel",
    booktitle = "Proceedings of the Fifth Workshop on Generation, Evaluation and Metrics ({GEM})",
    month = jul,
    year = "2026",
    address = "San Diego, California, USA",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingest-acl-workshops/2026.gem-main.2/",
    pages = "1--10",
    ISBN = "979-8-89176-423-1",
    abstract = "High-quality, large-scale conversational datasets are scarce, making it difficult to train on-device language models ({\textasciitilde}1B parameters) as effective assistants. We introduce CoSy (Conversational Synthesis), a novel framework for generating diverse, steerable, multi-turn conversations at scale. CoSY combines three key mechanisms: (1) conversational graphs that ensure natural dialogue flow, (2) turn-based prompt augmentations for diversity, and (3) explicit linguistic phenomena for coherence. We evaluate CoSy on conversational grounded reasoning tasks (i.e. answering questions based on contextual information), a core on-device use case.Our on-device sized models trained on CoSy-synthesized data achieve competitive performance with human-annotated baselines and outperform instruction-tuned models of up to 70B parameters in zero-shot settings."
}