Prototypical Reward Network for Data-Efficient RLHF

TL;DR

This paper introduces Proto-RM, a prototypical network-based reward model that improves data efficiency and stability in RLHF for fine-tuning large language models with limited human feedback.

Contribution

Proto-RM leverages prototypical networks to enhance reward modeling in data-scarce settings, enabling more stable and accurate learning from fewer human feedback samples.

Findings

Proto-RM outperforms traditional reward models in low-data scenarios.

It achieves comparable or better results with significantly less human feedback.

Experimental results demonstrate improved LLM fine-tuning efficiency.

Abstract

The reward model for Reinforcement Learning from Human Feedback (RLHF) has proven effective in fine-tuning Large Language Models (LLMs). Notably, collecting human feedback for RLHF can be resource-intensive and lead to scalability issues for LLMs and complex tasks. Our proposed framework Proto-RM leverages prototypical networks to enhance reward models under limited human feedback. By enabling stable and reliable structural learning from fewer samples, Proto-RM significantly enhances LLMs' adaptability and accuracy in interpreting human preferences. Extensive experiments on various datasets demonstrate that Proto-RM significantly improves the performance of reward models and LLMs in human feedback tasks, achieving comparable and usually better results than traditional methods, while requiring significantly less data. in data-limited scenarios. This research offers a promising direction for enhancing the efficiency of reward models and optimizing the fine-tuning of language models under restricted feedback conditions.