Metraipolis › archive › dc_5df70ebb5993
Fine-Tuning a 3B-Parameter LLM on a Smartphone: Characterizing Sustained Training — arXiv:2610.06325
- Source
- arXiv
- Original
- https://arxiv.org/abs/2610.06325
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-06T05:27:36.373Z
- SHA-256 of the archived text
73fb9d6f1b6c7428591ba38fb75bc0326d30e9851a6c331eadb76337965c4725- Stable URL of this page
https://metraipolis.com/documento/dc_5df70ebb5993
How to cite
Fine-Tuning a 3B-Parameter LLM on a Smartphone: Characterizing Sustained Training — arXiv:2610.06325 — arXiv, https://arxiv.org/abs/2610.06325 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_5df70ebb5993 (Metraipolis archive dc_5df70ebb5993, sha256 73fb9d6f1b6c7428591ba38fb75bc0326d30e9851a6c331eadb76337965c4725)
BibTeX
@misc{dc5df70ebb5993,
title = {Fine-Tuning a 3B-Parameter LLM on a Smartphone: Characterizing Sustained Training — arXiv:2610.06325},
howpublished = {arXiv},
url = {https://arxiv.org/abs/2610.06325},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_5df70ebb5993; sha256 73fb9d6f1b6c7428591ba38fb75bc0326d30e9851a6c331eadb76337965c4725},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv, https://arxiv.org/abs/2610.06325. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Multi-billion-parameter LLMs now run on phones for inference, and training them on the device would personalize them without user data leaving the phone. Prior work has measured individual training steps of such models on phones, but not complete training runs, and not whether adapters trained on the device improve personalization. We present the first systematic characterization of a multi-billion-parameter LLM fine-tuned on a mobile device, covering memory, per-step time, thermal behavior, and energy. An iPhone 17 Pro can fine-tune a 3B-parameter LLM to a typical user within one battery charge, and the resulting adapters improve personalization as much as adapters trained on a server. Sustained training throttles the phone to about half its initial throughput, and none of the pausing or burst schedules we tested recovers it. Nearly all of each training step is spent in the frozen base model, most of it in the backward pass, which nine of the ten other runtimes we audited do not accelerate. Apple's MLX had a kernel for it that was never dispatched and was incorrect, and our repair, now merged upstream, trains an adapter 1.47x faster on a third less energy. On-device fine-tuning is feasible on current phones, and making it efficient requires runtimes and operating systems to treat training as a first-class workload. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno). Categorie arXiv: Distributed, Parallel, and Cluster Computing (cs.DC) ; Artificial Intelligence (cs.AI); Performance (cs.PF). Pagina del lavoro: https://arxiv.org/abs/2610.06325 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-06T05:24:06.145Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).