MiCRo: Mixture Modeling and Context-aware Routing for Personalized Preference Learning

TL;DR

MiCRo is a two-stage framework that improves personalized preference learning for large language models by modeling diverse human preferences with mixture models and context-aware routing, without needing detailed annotations.

Contribution

MiCRo introduces a novel mixture modeling and context-aware routing approach for scalable, personalized preference learning from large-scale binary preference data.

Findings

Effectively captures diverse human preferences.

Significantly improves downstream personalization.

Adapts dynamically with minimal supervision.

Abstract

Reward modeling is a key step in building safe foundation models when applying reinforcement learning from human feedback (RLHF) to align Large Language Models (LLMs). However, reward modeling based on the Bradley-Terry (BT) model assumes a global reward function, failing to capture the inherently diverse and heterogeneous human preferences. Hence, such oversimplification limits LLMs from supporting personalization and pluralistic alignment. Theoretically, we show that when human preferences follow a mixture distribution of diverse subgroups, a single BT model has an irreducible error. While existing solutions, such as multi-objective learning with fine-grained annotations, help address this issue, they are costly and constrained by predefined attributes, failing to fully capture the richness of human values. In this work, we introduce MiCRo, a two-stage framework that enhances personalized preference learning by leveraging large-scale binary preference datasets without requiring explicit fine-grained annotations. In the first stage, MiCRo introduces context-aware mixture modeling approach to capture diverse human preferences. In the second stage, MiCRo integrates an online routing strategy that dynamically adapts mixture weights based on specific context to resolve ambiguity, allowing for efficient and scalable preference adaptation with minimal additional supervision. Experiments on multiple preference datasets demonstrate that MiCRo effectively captures diverse human preferences and significantly improves downstream personalization.