Brain-Inspired Two-Stage Approach: Enhancing Mathematical Reasoning by Imitating Human Thought Processes

TL;DR

A novel approach, named Brain, to imitate human thought processes to enhance mathematical reasoning abilities, using the Frontal Lobe Model to generate plans, and then employing the Parietal Lobe Model to generate code and execute to obtain answers.

Abstract

Although large language models demonstrate emergent abilities in solving math word problems, there is a challenging task in complex multi-step mathematical reasoning tasks. To improve model performance on mathematical reasoning tasks, previous work has conducted supervised fine-tuning on open-source models by improving the quality and quantity of data. In this paper, we propose a novel approach, named Brain, to imitate human thought processes to enhance mathematical reasoning abilities, using the Frontal Lobe Model to generate plans, and then employing the Parietal Lobe Model to generate code and execute to obtain answers. First, we achieve SOTA performance in comparison with Code LLaMA 7B based models through this method. Secondly, we find that plans can be explicitly extracted from natural language, code, or formal language. Our code and data are publicly available at https://github.com/cyzhh/Brain.

Figures (5)

• Figure 1: Brain, using a combined approach of the Frontal Lobe Model and the Parietal Lobe Model to simulate the human problem-solving thought process.
• Figure 2: The overview of our proposed method, Brain.
• Figure 3: The few-shot prompt $C$for GSM8K high-quality plan creating. The black color text is the requirement $R$. The teal and blue color text is the one of example pairs, which contains input $\overline{x_1}$ including example question and program and the output $\overline{y_1}$ including example plan.
• Figure 4: The few-shot prompt $C^{'}$ for GSM8K high-quality score dataset creating. The black color text is the requirement $R$. The teal and blue color text is the one of example pairs, which contains input $\overline{x_1}$ including example question and example plan $\overline{z_1}$ and the output $\overline{y_1}$ including example reasons and score.
• Figure 5: The few-shot prompt $C^{"}$ for GSM8K high-quality plan creating. The black color text is the requirement $R$. The teal and blue color text is the one of example pairs, which contains input $\overline{x_1}$ including example question and solution and the output $\overline{y_1}$ including example plan.