Let's Reinforce Step by Step
This addresses the problem of unreliable reasoning in language models for mathematics, though it is incremental as it builds on existing RLHF methods.
The paper investigated using Reinforcement Learning from Human Feedback (RLHF) with two reward schemes (outcome-supervised and process-supervised) to improve language model reasoning on mathematical tasks, finding that process-supervised methods enhanced accuracy on simple problems (GSM8K) but reduced performance on complex ones (MATH).
While recent advances have boosted LM proficiency in linguistic benchmarks, LMs consistently struggle to reason correctly on complex tasks like mathematics. We turn to Reinforcement Learning from Human Feedback (RLHF) as a method with which to shape model reasoning processes. In particular, we explore two reward schemes, outcome-supervised reward models (ORMs) and process-supervised reward models (PRMs), to optimize for logical reasoning. Our results show that the fine-grained reward provided by PRM-based methods enhances accuracy on simple mathematical reasoning (GSM8K) while, unexpectedly, reducing performance in complex tasks (MATH). Furthermore, we show the critical role reward aggregation functions play in model performance. Providing promising avenues for future research, our study underscores the need for further exploration into fine-grained reward modeling for more reliable language models.