Modotte/MathX-20M

Modotte数据集是MathX系列的一部分，专为基于指令的模型调优和微调而设计的高质量数学推理数据集。该数据集从高质量的公共来源中精选，并通过封闭和开源模型的合成数据增强。它提供了2000万经过精心策划的逐步思考数据示例，涵盖从基础算术到高级微积分的广泛数学领域。数据集经过多阶段过滤和验证过程，确保答案的正确性和详细数学思维。此外，数据集还包括问题复杂性分布、覆盖的数学领域、使用案例、数据集格式、质量保证和性能指标等详细信息。

The Modotte dataset is part of the MathX lineup, designed for instruction-based model tuning and fine-tuning. It is a high-quality mathematical reasoning dataset curated from public sources and enhanced with synthetic data from both closed and open-source models. The dataset offers 20 million highly curated examples of step-by-step thinking data, covering a wide range of mathematical domains from basic arithmetic to advanced calculus. It undergoes multi-stage filtering and verification processes to ensure answer correctness and detailed mathematical reasoning. The dataset also includes information on problem complexity distribution, covered mathematical domains, use cases, dataset format, quality assurance, and performance metrics.