hanane/attributionBench_augmented_unshuffled_posandgenonly

这是一个包含问题、断言及其相关信息的文本数据集，适用于自然语言处理任务，如断言验证、信息检索等。数据集中的每个样本都包括问题、断言、原始断言字符串、响应、引用、网页引用、归因标签、源数据集ID、NLI分数、对齐分数、示例类型、原始标签、标签切换、错误类型、选中的内容与角色、拒绝的内容与角色、消息内容与角色、选中的分数和拒绝的分数等信息。数据集分为训练集和验证集，可以用于模型训练和评估。

This is a text dataset containing questions, claims, and related information, suitable for natural language processing tasks such as claim verification, information retrieval, etc. Each sample in the dataset includes information such as questions, claims, original claim strings, responses, references, webpage references, attribution labels, source dataset IDs, NLI scores, alignment scores, example types, original labels, label switches, error types, chosen content and roles, rejected content and roles, message content and roles, score chosen, and score rejected. The dataset is split into training and validation sets, which can be used for model training and evaluation.