Full-length answer generation dataset
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/kolk/ConfnetPointerGenBaseline
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是一个包含259,788个样本的大型口语提问、事实性答案以及相对应的完整文本答案的数据集。这些样本选自SQuAD和HarvestingQA，并额外加入了来自NewsQA和Freebase的评价样本。该数据集的规模属于大规模，其任务是生成完整的答案。
This is a large-scale dataset containing 259,788 samples of spoken questions, factual answers, and their corresponding full-text answers. These samples are selected from SQuAD and HarvestingQA, with additional evaluation samples sourced from NewsQA and Freebase. As a large-scale dataset, its designated task is to generate complete answers.
