mlfoundations-dev/oh_v1.3_unnatural_instructions_x8

该数据集包含对话数据，每个对话包含来源（from）和内容（value）两个字段，均为字符串类型。此外，每个对话还包含一个分片ID（shard_id），也是字符串类型。数据集包含一个训练集，大小为1648072847.0字节，包含1191727个示例。数据集的下载大小为895381126字节，总大小为1648072847.0字节。数据集的配置文件指定了默认配置，数据文件路径为data/train-*。

The dataset contains conversation data, with each conversation including two fields: from and value, both of which are string types. Additionally, each conversation includes a shard ID (shard_id), also a string type. The dataset contains a training set with a size of 1648072847.0 bytes, comprising 1191727 examples. The download size of the dataset is 895381126 bytes, and the total size is 1648072847.0 bytes. The datasets configuration file specifies the default configuration, with data file paths as data/train-*.