timaeus/pile-ubuntu_irc

该数据集通过从`monology/pile-uncopyrighted`数据集中筛选特定行创建而成，每个子集通常限制为遇到的第一个100,000个符合条件的行。数据集包含文本和元数据信息，其中文本为字符串类型，元数据包含数据集名称等信息。训练集包含15,345个示例，大小为9,052,995,836字节。

This dataset is created by filtering specific rows from the `monology/pile-uncopyrighted` dataset, with each subset generally limited to the first 100,000 qualifying rows encountered. The dataset includes text and metadata information, where the text is of string type, and the metadata contains details such as the dataset name. The training set consists of 15,345 examples and is 9,052,995,836 bytes in size.