Turbo-AI/train-multi-negatives-v3

该数据集包含118,956个训练示例，总大小为1,105,237,553字节。每个示例包含id、text字段，以及两个列表relevant和not_relevant，这两个列表也分别包含id和text字段。数据集的结构设计用于处理文本相关性任务，其中relevant列表可能包含与主文本相关的文本片段，而not_relevant列表则包含不相关的文本片段。

This dataset contains 118,956 training examples with a total size of 1,105,237,553 bytes. Each example includes id, text fields, and two lists: relevant and not_relevant, which also contain id and text fields. The dataset is structured for text relevance tasks, where the relevant list may contain text segments related to the main text, and the not_relevant list contains unrelated text segments.