timaeus/pile-pubmed_abstracts

该数据集包含文本和元数据两个主要部分。文本部分为字符串类型，元数据部分包含一个名为pile_set_name的字符串字段。数据集包含一个训练分割，共有100,000个样本，总数据量为135,484,704字节。数据集的下载大小为75,713,626字节，与训练分割的大小相同。数据文件默认存储在data/train-*路径下。

The dataset consists of two main features: text and meta. The text feature is of string type, while the meta feature is a structure containing a string field named pile_set_name. The dataset includes a train split with 100,000 examples, totaling 135,484,704 bytes. The download size of the dataset is 75,713,626 bytes, matching the size of the train split. The default configuration specifies that the data files are located at data/train-*.