BEATS: Optimizing LLM Mathematical Capabilities with BackVerify and Adaptive Disambiguate based Efficient Tree Search
This addresses the challenge of enhancing mathematical reasoning in LLMs for applications requiring logical rigor, though it appears incremental as it builds on existing search-based methods.
The paper tackles the problem of improving LLMs' mathematical problem-solving abilities, which are often suboptimal and computationally expensive, by proposing BEATS, a method that uses iterative prompting, back-verification, and pruning tree search, resulting in a score increase from 36.94 to 61.52 on the MATH benchmark, outperforming GPT4's 42.5.
Large Language Models (LLMs) have exhibited exceptional performance across a broad range of tasks and domains. However, they still encounter difficulties in solving mathematical problems due to the rigorous and logical nature of mathematics. Previous studies have employed techniques such as supervised fine-tuning (SFT), prompt engineering, and search-based methods to improve the mathematical problem-solving abilities of LLMs. Despite these efforts, their performance remains suboptimal and demands substantial computational resources. To address this issue, we propose a novel approach, BEATS, to enhance mathematical problem-solving abilities. Our method leverages newly designed prompts that guide the model to iteratively rewrite, advance by one step, and generate answers based on previous steps. Additionally, we introduce a new back-verification technique that uses LLMs to validate the correctness of the generated answers. Furthermore, we employ a pruning tree search to optimize search time while achieving strong performance. Notably, our method improves Qwen2-7b-Instruct's score from 36.94 to 61.52, outperforming GPT4's 42.5 on the MATH benchmark.