hanane/attributionBench_augmentedselectErrortypesmixedalltrainbalanced

该数据集是一个文本数据集，包含了问题、断言、原始断言字符串、响应、参考列表、引用链接、网页参考（此处为null）、归属标签、源数据集、唯一标识符、NLI得分、对齐得分、样例类型、原始标签、标签切换、错误类型、选中的内容及其角色、拒绝的内容及其角色、消息及其角色、选中的得分和拒绝的得分。数据集分为训练集和验证集，可用于文本分类、信息抽取等自然语言处理任务。

The dataset is a text dataset that includes fields such as questions, claims, raw claim strings, responses, reference lists, citation links, webpage references (which is null), attribution labels, source datasets, unique identifiers, NLI scores, alignment scores, example types, original labels, label switches, error types, chosen content and its role, rejected content and its role, messages and their roles, score for chosen, and score for rejected. The dataset is split into training and development sets and can be used for tasks such as text classification and information extraction in natural language processing.