ALaRM: Align Language Models via Hierarchical Rewards Modeling
This work addresses the challenge of aligning LLMs with human preferences, particularly in complex text generation tasks, representing an incremental advancement over current RLHF methods.
The paper tackles the problem of aligning large language models with human preferences by introducing ALaRM, a framework that models hierarchical rewards in RLHF to address inconsistency and sparsity in supervision signals, resulting in improved performance in long-form question answering and machine translation tasks over existing baselines.
We introduce ALaRM, the first framework modeling hierarchical rewards in reinforcement learning from human feedback (RLHF), which is designed to enhance the alignment of large language models (LLMs) with human preferences. The framework addresses the limitations of current alignment approaches, which often struggle with the inconsistency and sparsity of human supervision signals, by integrating holistic rewards with aspect-specific rewards. This integration enables more precise and consistent guidance of language models towards desired outcomes, particularly in complex and open text generation tasks. By employing a methodology that filters and combines multiple rewards based on their consistency, the framework provides a reliable mechanism for improving model alignment. We validate our approach through applications in long-form question answering and machine translation tasks, employing gpt-3.5-turbo for pairwise comparisons, and demonstrate improvements over existing baselines. Our work underscores the effectiveness of hierarchical rewards modeling in refining LLM training processes for better human preference alignment. We release our code at https://ALaRM-fdu.github.io.