Quantile Regression for Distributional Reward Models in RLHF

TL;DR

This paper introduces Quantile Reward Models (QRMs) for reinforcement learning from human feedback, which learn a distribution over rewards to better capture human preferences and improve policy safety.

Contribution

The paper presents a novel distributional reward modeling approach using quantile regression, enhancing preference representation and downstream RL applications.

Findings

QRM outperforms traditional models on RewardBench.

Distributional estimates enable risk-aware RL.

QRM reduces negative responses in LLM policies.

Abstract

Reinforcement learning from human feedback (RLHF) has become a key method for aligning large language models (LLMs) with human preferences through the use of reward models. However, traditional reward models typically generate point estimates, which oversimplify the diversity and complexity of human values and preferences. In this paper, we introduce Quantile Reward Models (QRMs), a novel approach to reward modeling that learns a distribution over rewards instead of a single scalar value. Our method uses quantile regression to estimate a full, potentially multimodal distribution over preferences, providing a more powerful and nuanced representation of preferences. This distributional approach can better capture the diversity of human values, addresses label noise, and accommodates conflicting preferences by modeling them as distinct modes in the distribution. Our experimental results show that QRM outperforms comparable traditional point-estimate models on RewardBench. Furthermore, we demonstrate that the additional information provided by the distributional estimates can be utilized in downstream applications, such as risk-aware reinforcement learning, resulting in LLM policies that generate fewer extremely negative responses. Our code and model are released at https://github.com/Nicolinho/QRM.