hgissbkh/xglue

README文件提供了多个数据集配置的详细信息，每个配置都有特定的特征、分割和大小。数据集包括MLQA、NC、NER、PAWS-X、POS、QADSM、QAM、WPR和XNLI。每个配置都有不同的特征，例如MLQA的上下文、问题、答案；NC的新闻标题、新闻正文、新闻类别；NER的词语、命名实体识别标签；PAWS-X的句子1、句子2、标签；POS的词语、词性标签；QADSM的查询、广告标题、广告描述、相关性标签；QAM的问题、答案、标签；WPR的查询、网页标题、网页摘要、相关性标签；以及XNLI的前提、假设、标签。分割包括训练集、验证集和测试集，并指定了大小和示例数量。

The README file provides detailed information about multiple dataset configurations, each with specific features, splits, and sizes. The datasets include MLQA, NC, NER, PAWS-X, POS, QADSM, QAM, WPR, and XNLI. Each configuration has distinct features such as context, question, answers for MLQA; news_title, news_body, news_category for NC; words, ner for NER; sentence1, sentence2, label for PAWS-X; words, pos for POS; query, ad_title, ad_description, relevance_label for QADSM; question, answer, label for QAM; query, web_page_title, web_page_snippet, relavance_label for WPR; and premise, hypothesis, label for XNLI. The splits include train, validation, and test sets with specified sizes and examples.