ComMer: a Framework for Compressing and Merging User Data for Personalization

TL;DR

ComMer is a novel framework that efficiently personalizes large language models by compressing user data into compact representations, enabling better performance in constrained inference scenarios for skill learning tasks.

Contribution

The paper introduces ComMer, a new method for compressing and merging user data to personalize LLMs without fine-tuning or extensive retraining.

Findings

Outperforms existing methods in skill learning tasks under inference constraints.

Shows limitations in knowledge-intensive tasks due to information loss.

Provides insights into trade-offs in multi-document compression for personalization.

Abstract

Large Language Models (LLMs) excel at a wide range of tasks, but adapting them to new data, particularly for personalized applications, poses significant challenges due to resource and computational constraints. Existing methods either rely on exposing fresh data to the model through the prompt, which is limited by context size and computationally expensive at inference time, or fine-tuning, which incurs substantial training and update costs. In this paper, we introduce ComMer - Compress and Merge - a novel framework that efficiently personalizes LLMs by compressing users' documents into compact representations, which are then merged and fed into a frozen LLM. We evaluate ComMer on two types of personalization tasks - personalized skill learning, using the tweet paraphrasing dataset and the personalized news headline generation dataset from the LaMP benchmark, and knowledge-intensive, using the PerLTQA dataset. Our experiments demonstrate that in constrained inference budget scenarios ComMer achieves superior quality in skill learning tasks, while highlighting limitations in knowledge-intensive settings due to the loss of detailed information. These results offer insights into trade-offs and potential optimizations in multi-document compression for personalization.