HUOZIIME: An On-Device LLM-enhanced Input Method for Deep Personalization

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Shan, Baocai, Xu, Yuzhuang, Che, Wanxiang
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910132560986112
author Shan, Baocai
Xu, Yuzhuang
Che, Wanxiang
author_facet Shan, Baocai
Xu, Yuzhuang
Che, Wanxiang
contents Mobile input method editors (IMEs) are the primary interface for text input, yet they remain constrained to manual typing and struggle to produce personalized text. While lightweight large language models (LLMs) make on-device auxiliary generation feasible, enabling deeply personalized, privacy-preserving, and real-time generative IMEs poses fundamental challenges.To this end, we present HUOZIIME, a personalized on-device IME powered by LLM. We endow HUOZIIME with initial human-like prediction ability by post-training a base LLM on synthesized personalization data. Notably, a hierarchical memory mechanism is designed to continually capture and leverage user-specific input history. Furthermore, we perform systemic optimizations tailored to on-device LLMbased IME deployment, ensuring efficient and responsive operation under mobile constraints.Experiments demonstrate efficient on-device execution and high-fidelity memory-driven personalization. Code and package are available at https://github.com/Shan-HIT/HuoziIME.
format Preprint
id arxiv_https___arxiv_org_abs_2604_14159
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle HUOZIIME: An On-Device LLM-enhanced Input Method for Deep Personalization
Shan, Baocai
Xu, Yuzhuang
Che, Wanxiang
Computation and Language
Artificial Intelligence
Mobile input method editors (IMEs) are the primary interface for text input, yet they remain constrained to manual typing and struggle to produce personalized text. While lightweight large language models (LLMs) make on-device auxiliary generation feasible, enabling deeply personalized, privacy-preserving, and real-time generative IMEs poses fundamental challenges.To this end, we present HUOZIIME, a personalized on-device IME powered by LLM. We endow HUOZIIME with initial human-like prediction ability by post-training a base LLM on synthesized personalization data. Notably, a hierarchical memory mechanism is designed to continually capture and leverage user-specific input history. Furthermore, we perform systemic optimizations tailored to on-device LLMbased IME deployment, ensuring efficient and responsive operation under mobile constraints.Experiments demonstrate efficient on-device execution and high-fidelity memory-driven personalization. Code and package are available at https://github.com/Shan-HIT/HuoziIME.
title HUOZIIME: An On-Device LLM-enhanced Input Method for Deep Personalization
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2604.14159