Enhancing Problem-Solving Abilities with Reinforcement Learning-Augmented Large Language Models
Kai Xi, Xiaowei Bi, Zheng Xu, Fu Lei, Zheze Yang · 2024
The problem-solving abilities of large language models (LLMs) have significantly advanced in various fields, yet their potential for tackling complex tasks remains underexplored. This paper introduces a novel framework that enhances LLM performance by incorporating reinforcement learning (RL). By combining a policy model, which generates candidate solutions, and a reward model, which evaluates their accuracy, our approach creates a feedback loop that systematically improves the model's solution-generation capabilities. Tested across challenging datasets, our framework achieves superior performance compared to traditional models, highlighting the potential of LLMs augmented with RL to solve complex tasks more effectively. This research provides a robust foundation for advancing automated problem-solving systems.