kevin017/tokenized_bioS_inverse_QA_three_large_padding

该数据集包含了三个特征字段：input_ids、attention_mask和answers_tokenized。input_ids和attention_mask是整数序列，分别使用int32和int8数据类型。answers_tokenized是一个结构化的字段，它本身包含两个序列：attention_mask和input_ids，这两个序列都使用int64数据类型。数据集分为训练集和测试集，每个集合都包含34789个样本。数据集的总下载大小为2.49MB，完整大小为约174MB。提供了默认配置，指定了训练和测试数据文件的路径。

The dataset includes three feature fields: input_ids, attention_mask, and answers_tokenized. input_ids and attention_mask are integer sequences using int32 and int8 data types, respectively. answers_tokenized is a structured field that contains two sequences: attention_mask and input_ids, both using int64 data types. The dataset is split into a training set and a test set, each containing 34,789 samples. The total download size of the dataset is 2.49MB, and the full size is approximately 174MB. A default configuration is provided, specifying the paths to the training and test data files.