ddrg/named_math_formulas_ft

基于71个著名数学恒等式的数学数据集。每个条目包括一个著名的数学恒等式名称（例如，勾股定理）和一个潜在的公式表示（例如，$a^2+b^2=c^2$ 或文本描述），以及一个标签，指示公式表示是否与名称所述的数学恒等式相同。数据集中真例的数量是假例的十分之一，使得数据集可以在每个epoch改变假例的情况下用于训练。

Mathematical dataset based on 71 famous mathematical identities. Each entry consists of a name of a well known mathematical identity (e.g., Pythagorean Theorem) and a potential representation (e.g., $a^2+b^2=c^2$ or a textual description), together with a label indicating whether the formula representation describes the same mathematical identity as the name. The dataset contains ten times more falsified versions than true ones, making it suitable for training with changing false examples every epoch.