MoF: Preference-Aware Mixture Modeling for Black-Box LLM Personalization

2026-10-06

AI Summary

This study addresses the challenges of linear parameter growth with increasing users and the difficulty of adapting to new users in black-box large model personalization. To this end, this work proposes a Mixture of Preference Surfaces framework. By eliminating user-specific scoring heads, the method models individual preferences as combinations of shared latent surfaces. Furthermore, it designs a history-conditioned dynamic routing network based on a Mixture-of-Experts architecture to achieve scalable personalization with minimal parameters. Experimental results demonstrate that the proposed framework significantly enhances both personalization performance and parameter efficiency across diverse tasks, while exhibiting superior zero-shot generalization capabilities for unseen users.

Abstract

Proprietary Large Language Models (LLMs) have demonstrated remarkable capabilities across a wide range of tasks, yet aligning their outputs with diverse user preferences remains challenging. Existing personalization approaches for black-box LLMs often rely on user-specific scoring heads, causing the number of personalized parameters to grow linearly with the number of users and requiring additional adaptation for unseen users. To address these limitations, we propose Mixture-of-Facets (MoF), a scalable personalization framework for black-box LLMs that models user preferences as compositions of shared latent preference facets rather than dedicated user-specific parameters. MoF performs personalization through history-conditioned routing over shared facet heads, enabling personalization for users unseen during training without additional parameter updates. Across diverse personalization tasks, MoF delivers stronger personalization performance while maintaining a more scalable and parameter-efficient design than prior approaches. Additional analysis indicates strong generalization to unseen users.

Problem

• Black-Box LLM
• Personalization
• Scalability
• Parameter Efficiency
• Unseen Users

Innovation

• Black-Box LLM Personalization
• Mixture-of-Facets
• Preference Modeling
• Parameter Efficiency
• Zero-shot Generalization