# MS MARCO Training Dataset This dataset consists of 4 separate datasets, each using the MS MARCO Queries and passages: * `triplets`: This subset contains triplets of query-id, positive-id, negative
MS MARCO训练数据集，包含三个子数据集：triplets、bert-ensemble-mse和bert-ensemble-margin-mse。triplets为包含查询ID、正例ID和负例ID的三元组数据集；bert-ensemble-mse为包含查询ID、段落ID和分数的元组数据集；bert-ensemble-margin-mse为包含查询ID、正例ID、负例ID和分数的四元组数据集。同
