DJ-Research/pubmed_inference

该数据集包含三个不同配置clf、ft和po，每个配置都适用于不同的NLP任务。clf配置适合分类任务，包含问题、答案和上下文信息。ft配置和po配置除了包含clf的字段外，还有额外的字段，可能适用于文本生成或句子选择等任务。数据集分为训练集和验证集。

The dataset consists of three different configurations: clf, ft, and po, each suitable for different NLP tasks. The clf configuration is for classification tasks and includes fields for questions, answers, and context. The ft and po configurations have additional fields beyond those in clf, which may be used for tasks like text generation or sentence selection. The dataset is split into training and validation sets.