MathClean: A Benchmark for Synthetic Mathematical Data Cleaning
This work addresses the need for effective cleaning methods in synthetic math data to improve reasoning abilities in LLMs, but it is incremental as it focuses on benchmarking rather than proposing a new cleaning method.
The paper tackles the problem of inaccuracies in synthetic mathematical data for training large language models by introducing the MathClean benchmark, which consists of 4,000 questions and answers with error annotations, and shows that state-of-the-art models like GPT-o1 and DeepSeek-R1 perform poorly on it.
With the rapid development of large language models (LLMs), the quality of training data has become crucial. Among the various types of training data, mathematical data plays a key role in enabling LLMs to acquire strong reasoning abilities. While high-quality open-source data is important, it is often insufficient for pre-training, necessitating the addition of synthetic math problems. However, synthetic math questions and answers can introduce inaccuracies, which may degrade both the training data and web data. Therefore, an effective method for cleaning synthetic math data is essential. In this paper, we propose the MathClean benchmark to evaluate the effectiveness of math data cleaning models. The MathClean benchmark consists of 2,000 correct questions and 2,000 erroneous questions with additional 2,000 correct and erroneous answers sourced from augmented data based on GSM8K and MATH. Moreover, we also annotate error types for each question or answer, since it can assess whether models can correctly identify the error categories for future improvements. Finally, we present comprehensive evaluations using state-of-the-art (SOTA) models. Our results demonstrate that even strong models like GPT-o1 and DeepSeek-R1 perform poorly on this benchmark, highlighting the utility of MathClean. Our code and data is available at https://github.com/YuYingLi0/MathClean.