Enabling On-Device Large Language Model Personalization with Self-Supervised Data Selection and Synthesis

arXiv:2311.12275v4 Announce Type: replace
Abstract: After a large language model (LLM) is deployed on edge devices, it is desirable for these devices to learn from user-generated conversation data to generate user-specific and personalized responses in real-time. However, user-generated data usually contains sensitive and private information, and uploading such data to the cloud for annotation is not preferred if not prohibited. While it is possible to obtain annotation locally by directly asking users to provide preferred responses, such annotations have to …

abstract arxiv conversation cs.cl data devices edge edge devices enabling generate generated however information language language model large language large language model learn llm personalization personalized real-time responses synthesis type