LGCNS/KorQuAD_2.0

KorQuAD 2.0是一个由93,651个问题-答案对组成的韩语机器阅读理解数据集。它要求在整篇Wikipedia文章（而不仅仅是1-2段）中寻找答案，能够有效地搜索非常长的文档。此外，它还需要理解包含表格和列表的HTML结构。答案单元可能不仅限于单词或子跨度，还可能是段落、表格或列表等不同长度和形态的内容。

KorQuAD 2.0 is a Korean Machine Reading Comprehension dataset composed of 93,651 question-answer pairs. It requires finding answers in the entire Wikipedia article (not just 1-2 paragraphs) and must be able to efficiently search through very long documents. Additionally, it demands understanding of HTML structures that include tables and lists. The answer units can be more than just words or sub-spans, they can also be paragraphs, tables, or lists of various lengths and forms.