Gusarich/math-expressions-1m

该数据集是一个包含1,000,000个数学表达式的全面集合，包括有效和无效的表达式。数据集旨在用于各种机器学习和自然语言处理任务，如计算、数学推理和验证。数据集的特征包括涉及基本算术操作和各种数学函数的表达式、有效和无效表达式的比例、每个表达式的评估结果、数字的范围和类型、以及表达式的深度。

The Math Expressions 1M dataset is a comprehensive collection of 1,000,000 mathematical expressions, including both valid and invalid expressions. This dataset is intended for use in various machine learning and natural language processing tasks, such as calculations, mathematical reasoning, and validation. The dataset consists of expressions generated using basic arithmetic operations and various mathematical functions, varying in complexity with both simple and nested structures, and are labeled with their evaluation results. Valid expressions have their numerical results, while invalid expressions are marked as INVALID. The numbers used in the expressions range from 0.0001 to 1,000,000,000, including both positive and negative values, generated with logarithmic distribution, some are integers, while others are floating-point numbers with up to four decimal places. The expressions have varying depths, with a maximum depth of 3 levels, including both simple and more complex, nested structures.