{"ID":2855434,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2510.15001","arxiv_id":"2510.15001","title":"VaultGemma: A Differentially Private Gemma Model","abstract":"We introduce VaultGemma 1B, a 1 billion parameter model within the Gemma family, fully trained with differential privacy. Pretrained on the identical data mixture used for the Gemma 2 series, VaultGemma 1B represents a significant step forward in privacy-preserving large language models. We openly release this model to the community","short_abstract":"We introduce VaultGemma 1B, a 1 billion parameter model within the Gemma family, fully trained with differential privacy. Pretrained on the identical data mixture used for the Gemma 2 series, VaultGemma 1B represents a significant step forward in privacy-preserving large language models. We openly release this model to...","url_abs":"https://arxiv.org/abs/2510.15001","url_pdf":"https://arxiv.org/pdf/2510.15001v2","authors":"[\"Amer Sinha\",\"Thomas Mesnard\",\"Ryan McKenna\",\"Daogao Liu\",\"Christopher A. Choquette-Choo\",\"Yangsibo Huang\",\"Da Yu\",\"George Kaissis\",\"Zachary Charles\",\"Ruibo Liu\",\"Lynn Chua\",\"Pritish Kamath\",\"Pasin Manurangsi\",\"Steve He\",\"Chiyuan Zhang\",\"Badih Ghazi\",\"Borja De Balle Pigem\",\"Prem Eruvbetine\",\"Tris Warkentin\",\"Armand Joulin\",\"Ravi Kumar\"]","published":"2025-10-15T21:59:53Z","proceeding":"cs.CR","tasks":"[\"cs.CR\",\"cs.AI\"]","methods":"[\"Language Model\"]","has_code":false}