mlfoundations-dev/oh_v1.3_metamath_x2

该数据集包含对话内容（conversations）和分片ID（shard_id）两个主要特征。对话内容进一步细分为来源（from）和值（value）两个字段。数据集分为训练集（train），包含913,084个例子，占用1,502,847,241字节。下载大小为824,099,227字节，数据集总大小为1,502,847,241字节。默认配置下的数据文件路径为data/train-*。

The dataset includes two main features: conversations and shard_id. Conversations are further divided into from and value fields. The dataset is split into a training set (train) containing 913,084 examples, occupying 1,502,847,241 bytes. The download size is 824,099,227 bytes, and the total dataset size is 1,502,847,241 bytes. The default configuration specifies the data file path as data/train-*.