withpi/msmarco-mnrl-hard-triplets-grouped-80s-20m-joined_embedding_tokenized_8k_5_embedding

该数据集包含多个字符串类型的查询语句（query_1至query_4），对应的正面和负面段落（positive_passage和negative_passage），以及它们的哈希值。还包括一个整型类别（category）和查询计数（query_count）。此外，还有输入ID和注意力掩码的序列信息。数据集分为训练集和测试集，提供了各自的样本数量和文件大小。未提供数据集的具体用途或内容描述。

The dataset includes multiple string-type query statements (query_1 to query_4), corresponding positive and negative passages (positive_passage and negative_passage), and their hash values. It also contains an integer category (category) and a query count (query_count). In addition, there are sequence information for input IDs and attention masks. The dataset is split into training and test sets, with the number of samples and file sizes provided for each. No specific description of the datasets purpose or content is given.