Towards On-Device Personalization: Cloud-device Collaborative Data Augmentation for Efficient On-device Language Model

ACL ARR 2025 February Submission4684 Authors

Abstract: With the advancement of large language models (LLMs), significant progress has been made in various Natural Language Processing (NLP) tasks. However, most existing LLMs still face two key challenges that hinder their broader applications: (1) their responses exhibit universal characteristics and lack personalization tailored to specific users, and (2) they are highly dependent on cloud infrastructure due to intensive computational requirements, leading to response delays and user privacy concerns. Recent research has primarily focused on either developing cloud-based personalized LLMs or exploring the on-device deployment of general LLMs. However, few studies have addressed both limitations by investigating personalized on-device LMs. To bridge this gap, this paper introduces CDCDA-PLM, a framework for deploying a personalized LLM on user devices with the assistance of a powerful cloud-based LLM while satisfying personalized user privacy requirements. Specifically, to overcome the data sparsity of on-device personal data, users have the flexibility to selectively share personal data with the server-side LLM to generate more synthetic personal data. By combining this synthetic data with locally stored user data, we fine-tune the personalized parameter-efficient fine-tuning (PEFT) modules of the small on-device model to capture user personas effectively. Our experiments demonstrate the effectiveness of CDCDA-PLM across six tasks in a widely used personalization benchmark.

Paper Type: Long

Research Area: Information Retrieval and Text Mining

Research Area Keywords: Large Language Model, Personalization, On-device LLM

Contribution Types: Approaches to low-resource settings

Languages Studied: English

Submission Number: 4684