numerata/diffs-sample

这是一个包含训练集和测试集的数据集，其中每个样本具有input_ids、attention_mask和labels三个特征。input_ids是int32类型的列表，attention_mask是int8类型的列表，labels是int64类型的列表。训练集包含大约1.38亿个示例，大小为309,574,315,134字节；测试集包含13,844个示例，大小为30,962,235字节。整个数据集的大小为309,605,277,369字节，下载大小为102,145,724,638字节。

This dataset consists of training and test sets, with each sample having three features: input_ids, attention_mask, and labels. input_ids is a list of int32, attention_mask is a list of int8, and labels is a list of int64. The training set contains about 138,418,521 examples, with a size of 309,574,315,134 bytes; the test set contains 13,844 examples, with a size of 30,962,235 bytes. The entire dataset is 309,605,277,369 bytes in size, and the download size is 102,145,724,638 bytes.