Brain-Inspired Two-Stage Approach: Enhancing Mathematical Reasoning by Imitating Human Thought Processes (2403.00800v1)
Abstract: Although LLMs 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.
- Palm 2 technical report.
- Training verifiers to solve math word problems.
- Towards a unified view of answer calibration for multi-step reasoning.
- G-llava: Solving geometric problem with multi-modal large language model.
- Tora: A tool-integrated reasoning agent for mathematical problem solving.
- Reasoning with language model is planning with world model.
- A closer look at the self-verification abilities of large language models in logical reasoning.
- Metagpt: Meta programming for a multi-agent collaborative framework.
- Large language models cannot self-correct reasoning yet.
- Decomposed prompting: A modular approach for solving complex tasks.
- Making large language models better reasoners with step-aware verifier.
- Let’s verify step by step.
- Deductive verification of chain-of-thought reasoning.
- Wizardmath.
- OpenAI. 2023. Gpt-4 technical report.
- Refiner: Reasoning feedback on intermediate representations.
- Direct preference optimization: Your language model is secretly a reward model.
- Mathematical discoveries from program search with large language models. Nature.
- Code llama: Open foundation models for code.
- Deepseekmath: Pushing the limits of mathematical reasoning in open language models.
- Adaplanner: Adaptive planning from feedback with language models.
- Mirac Suzgun and Adam Tauman Kalai. 2024. Meta-prompting: Enhancing language models with task-agnostic scaffolding.
- Llama: Open and efficient foundation language models.
- Solving olympiad geometry without human demonstrations. Nature, 625(7995):476–482.
- Mathcoder: Seamless code integration in llms for enhanced mathematical reasoning.
- Plan-and-solve prompting: Improving zero-shot chain-of-thought reasoning by large language models.
- Math-shepherd: A label-free step-by-step verifier for llms in mathematical reasoning.
- Fine-grained human feedback gives better rewards for language model training.
- Get an a in math: Progressive rectification prompting.
- Tree of thoughts: Deliberate problem solving with large language models.
- Outcome-supervised verifiers for planning in mathematical reasoning.
- Metamath: Bootstrap your own mathematical questions for large language models.
- Scaling relationship on learning mathematical reasoning with large language models.
- Mammoth: Building math generalist models through hybrid instruction tuning.
- Cumulative reasoning with large language models.
- Verify-and-edit: A knowledge-enhanced chain-of-thought framework. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 5823–5840, Toronto, Canada. Association for Computational Linguistics.
- Progressive-hint prompting improves reasoning in large language models.
- Solving math word problems via cooperative reasoning induced language models. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 4471–4485, Toronto, Canada. Association for Computational Linguistics.
- Large language models can learn rules.
- Yezeng Chen (5 papers)
- Zui Chen (14 papers)
- Yi Zhou (438 papers)