mib-bench/ravel

这是一个包含模板、提示、实体和属性等信息的NLP数据集，同时包含对抗性版本的字段，适用于文本生成或对抗性学习任务。数据集分为训练集、验证集和测试集，提供了丰富的示例用于模型训练和评估。

This is an NLP dataset containing information such as templates, prompts, entities, and attributes, along with counterfactual versions that include fields like Continent, Country, Language, etc., suitable for text generation or adversarial learning tasks. The dataset is divided into training, validation, and test sets, providing a wealth of examples for model training and evaluation.