marcov/hans_promptsource

该数据集包含多个特征，如前提、假设、标签等，标签分为‘entailment’和‘non-entailment’两类。此外，数据集还包括解析前提、解析假设、二进制解析前提、二进制解析假设、启发式方法、子案例、模板、模板名称、渲染输入和渲染输出等特征。数据集分为训练集和验证集，分别包含300,000个样本，总大小为481,714,110字节。

This dataset includes features such as premise, hypothesis, and label, with labels categorized into entailment and non-entailment. It also contains features for parsing premise and hypothesis, as well as features related to templates and rendering. The dataset is divided into a training set and a validation set, each containing 300000 samples.