Equivalent Expressions Dataset
收藏arXiv2025-09-30 收录
下载链接：
https://mlpgroup.github.io/expemb
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含了大约460万的数学等价表达式对，这些表达式对是通过使用SymPy从已知的一系列简单数学表达式中生成的。该数据集包括带有不同运算符的一元表达式，并排除了在解析时产生NaN结果的例子。这些表达式的序列长度平均为16.18，标准差为4.29。数据规模达到4,662,232个输入-输出对，包含2,744,809个独特的表达式。任务是对数学表达式生成向量表示（嵌入）。
This dataset contains approximately 4.6 million pairs of mathematically equivalent expressions, which were generated from a collection of known simple mathematical expressions via SymPy. It includes unary expressions with diverse operators, and excludes any examples that yield NaN results during parsing. The average sequence length of these expressions is 16.18, with a standard deviation of 4.29. In total, the dataset consists of 4,662,232 input-output pairs and encompasses 2,744,809 unique mathematical expressions. The task is to generate vector representations (embeddings) for mathematical expressions.
提供机构：
MLP Group
