CLIPer: Tailoring Diverse User Preference via Classifier-Guided Inference-Time Personalization

TL;DR

CLIPer is a lightweight, classifier-guided method for dynamically personalizing language model outputs at inference time without extensive fine-tuning.

Contribution

It introduces a novel inference-time personalization technique that uses classifiers to steer LLMs according to user preferences efficiently.

Findings

Effective in delivering personalized language generation.

Scalable and introduces negligible computational overhead.

Works across single and multi-dimensional preferences.

Abstract

Personalized LLMs can significantly enhance user experiences by tailoring responses to preferences such as helpfulness, conciseness, and humor. However, fine-tuning models to address all possible combinations of user preferences is computationally expensive and impractical. In this paper, we introduce \textbf{CLIPer}(\textbf{Cl}assifier-guided \textbf{I}nference-time \textbf{Per}sonalization), a lightweight personalization approach that leverages a classifier model to steer LLM generation dynamically to different user preferences at inference time. Our method eliminates the need for extensive fine-tuning, inducing negligible additional computational overhead while enabling more controllable and nuanced personalization across single and multi-dimensional preferences. Comprehensive empirical analyses demonstrate the scalability and effectiveness of our approach in delivering personalized language generation.