{"version":"1","article":{"id":"phone-npu-lora-training-study-2026","slug":"phone-npu-lora-training-study-2026","language":"en","url":"https://haberpaper.com/en/news/phone-npu-lora-training-study-2026","api_url":"https://haberpaper.com/api/v1/articles/phone-npu-lora-training-study-2026?lang=en","title":"First-Order LoRA Training with Phone NPUs","summary":"The preprint examines whether smartphone neural processing units can help fine-tune language models locally with low-rank adaptation. It focuses on the repeated operations involving frozen model weights that remain costly even when only a small number of parameters are trained.","category":"business","region":"world","topics":["ai","semiconductors"],"published_at":"2026-10-20T19:09:32+03:00","modified_at":"2026-10-20T19:09:32+03:00","publication_basis":"publication-record","event_period":"2026-10-17","event_period_basis":"event-or-reference-period","sources":[{"name":"arXiv","url":"https://arxiv.org/abs/2610.24816v1","publishedAt":"2026-10-17","checkedAt":"2026-10-20"}],"translations":[{"language":"tr","url":"https://haberpaper.com/tr/news/phone-npu-lora-training-study-2026","api_url":"https://haberpaper.com/api/v1/articles/phone-npu-lora-training-study-2026?lang=tr"},{"language":"en","url":"https://haberpaper.com/en/news/phone-npu-lora-training-study-2026","api_url":"https://haberpaper.com/api/v1/articles/phone-npu-lora-training-study-2026?lang=en"},{"language":"ar","url":"https://haberpaper.com/ar/news/phone-npu-lora-training-study-2026","api_url":"https://haberpaper.com/api/v1/articles/phone-npu-lora-training-study-2026?lang=ar"},{"language":"fr","url":"https://haberpaper.com/fr/news/phone-npu-lora-training-study-2026","api_url":"https://haberpaper.com/api/v1/articles/phone-npu-lora-training-study-2026?lang=fr"},{"language":"es","url":"https://haberpaper.com/es/news/phone-npu-lora-training-study-2026","api_url":"https://haberpaper.com/api/v1/articles/phone-npu-lora-training-study-2026?lang=es"}],"data_types":[],"data":null,"data_compilations":[],"data_downloads":[],"content":{"paragraphs":["The preprint examines whether smartphone neural processing units can help fine-tune language models locally with low-rank adaptation. It focuses on the repeated operations involving frozen model weights that remain costly even when only a small number of parameters are trained.","The motivation is to keep personal data on-device during adaptation. The work is a systems research study; its feasibility claims should be read within the phones, models and tests reported by the authors."],"key_points":[],"sections":[]},"text":"The preprint examines whether smartphone neural processing units can help fine-tune language models locally with low-rank adaptation. It focuses on the repeated operations involving frozen model weights that remain costly even when only a small number of parameters are trained.\n\nThe preprint examines whether smartphone neural processing units can help fine-tune language models locally with low-rank adaptation. It focuses on the repeated operations involving frozen model weights that remain costly even when only a small number of parameters are trained.\n\nThe motivation is to keep personal data on-device during adaptation. The work is a systems research study; its feasibility claims should be read within the phones, models and tests reported by the authors.","corrections":[],"editorial_illustration":{"kind":"symbolic-editorial-illustration","documentary":false,"scope":"section","asset_id":"business","url":"https://haberpaper.com/assets/illustrations/business-1536.webp","width":1536,"height":1024,"alt":"Symbolic industrial workshop with machinery, a warehouse and a crane.","language":"en"},"section_illustration":{"kind":"symbolic-editorial-illustration","documentary":false,"scope":"section","asset_id":"business","url":"https://haberpaper.com/assets/illustrations/business-1536.webp","width":1536,"height":1024,"alt":"Symbolic industrial workshop with machinery, a warehouse and a crane.","language":"en"}}}