Work-in-Progress: On-device Retrieval Augmented Generation with Knowledge Graphs for Personalized Large Language Models

On-device large language models (LLMs) have promising benefits, including personalization, enhanced user privacy, and offline operations. We present our work-in-progress approach to on-device LLMs leveraging retrieval augmented generation (RAG) based on personalized on-device knowledge graphs (KGs). We aim to enhance the personalized experience of LLMs while preserving privacy by keeping sensitive data locally.