mlfoundations-dev/oh_v1.3_alpaca_x8

该数据集包含对话数据，每个对话包含来源和内容两个字段，均为字符串类型。此外，每个数据项还包含一个字符串类型的shard_id。数据集提供了一个训练集分割，包含超过113万个例子，总数据量约为1.57GB，下载大小约为866MB。数据文件的路径在配置中指定。

The dataset contains conversation data, where each conversation is identified by the from field indicating the initiator and the value field indicating the conversation content. The dataset also includes a shard_id field for identifying data shards. The dataset is split into a training set with 1,132,996 samples, totaling 1,570,543,647.0 bytes. The download size of the dataset is 866,322,450 bytes.