PAWS语序对抗问题匹配数据集

数据集里包含了释义对和非释义对，即识别一对句子是否具有相同的释义（含义），特点是具有高度重叠词汇，重点考察模型对句法结构的理解能力。

This dataset comprises both paraphrase pairs and non-paraphrase pairs. Its core task is to determine whether a given pair of sentences are mutual paraphrases (i.e., convey identical semantic meanings). Characterized by highly overlapping vocabulary between each sentence pair, this dataset primarily evaluates models' ability to understand syntactic structures.