shomez/blink-pretrain-wikipedia

该数据集包含多个特征，包括id、input、output和meta。output特征进一步包含answer和provenance，其中provenance又包含wikipedia_id和title。meta特征包含mention、left_context和right_context。数据集只有一个train分割，包含9000000个示例，文件大小为10211044686字节。

The dataset includes multiple features such as id, input, output, and meta. The output feature further contains answer and provenance, with provenance including wikipedia_id and title. The meta feature contains mention, left_context, and right_context. The dataset has only one train split, containing 9000000 examples, with a file size of 10211044686 bytes.