sileod/missing-item-prediction

gapfinder数据集包含两种配置：对比配置（contrastive）和默认配置（default）。每个配置都分为训练集、验证集和测试集。数据集的特征包括标签y、项目集合items_set、提示prompt和项目数量N_items。对比配置的训练集包含4400个示例，大小为63796428字节；验证集和测试集分别包含3300个示例，大小分别为39152469字节。默认配置的训练集大小为64093400字节，验证集和测试集大小为39375100字节。数据集的总下载大小和存储大小也在文件中给出。

The gapfinder dataset comes with two configurations: contrastive and default. Each configuration is split into training, validation, and test sets. The features of the dataset include y (string type), items_set (string type), prompt (string type), and N_items (integer type). The contrastive configurations training set contains 4400 examples, totaling 63796428 bytes; the validation and test sets each contain 3300 examples, totaling 39152469 bytes. The default configurations training set is 64093400 bytes, while the validation and test sets are 39375100 bytes each. The total download size and storage size of the dataset are also provided in the file.