snap-stanford/hotpotqa_four_agents_pipeline-preference_scorer

这是一个包含对话生成、问题重写、信息提取、提示生成和维基百科检索任务的数据集。数据集的特征包括上下文、选择的回应、被拒绝的回应、对应分数、附加信息和相关变量。数据集分为answer_generator、question_rewriter、info_extractor、hint_generator和wikipedia_retriever五个部分，分别用于不同的NLP任务。

This dataset includes tasks such as dialogue generation, question rewriting, information extraction, hint generation, and Wikipedia retrieval. The features of the dataset include context, chosen response, rejected response, corresponding scores, additional information, and related variables. The dataset is divided into five parts: answer_generator, question_rewriter, info_extractor, hint_generator, and wikipedia_retriever, each for different NLP tasks.