marcov/crows_pairs_promptsource

该数据集包含多个特征，如id、sent_more、sent_less等，其中stereo_antistereo和bias_type是分类标签，annotations是一个嵌套的分类标签序列。数据集还包含匿名作者和注释者的信息，以及模板和渲染的输入输出。数据集分为一个测试集，包含10556个样本。

The dataset includes multiple features such as id, sent_more, sent_less, where stereo_antistereo and bias_type are categorical labels, and annotations is a nested sequence of categorical labels. The dataset also contains information about anonymous writers and annotators, as well as templates and rendered input-output pairs. The dataset is divided into a test set with 10556 samples.