deu05232/promptriever-RQ1-diff_version

该数据集是一个用于信息检索或问答任务的数据集，包含查询（query）及其对应的正面和负面段落（passages）。正面段落（positive_passages）包括文档ID（docid）、文本（text）和标题（title），而负面段落（negative_passages）仅包含基本字段。此外，数据集还包含指令（instruction）相关字段，如only_instruction和has_instruction，以及新负面样本（new_negatives）用于增强训练。数据集仅包含训练分割（train），共有423,706个示例，总大小约4.03 GB。This dataset is designed for information retrieval or question-answering tasks, containing queries (query) along with corresponding positive and negative passages. Positive passages (positive_passages) include fields such as document ID (docid), explanation (null), followir_score (null), joint_id (null), text, and title, while negative passages (negative_passages) contain basic fields like docid, text, and title. Additionally, the dataset includes instruction-related fields such as only_instruction and has_instruction, as well as new negative samples (new_negatives) for enhanced training. The dataset consists solely of a training split (train) with 423,706 examples and a total size of approximately 4.03 GB.