marcov/sick_promptsource

该数据集包含多个字段，如id、sentence_A、sentence_B、label等，其中label字段有三个类别：entailment（蕴含）、neutral（中立）和contradiction（矛盾）。数据集分为train、validation和test三个部分，分别包含22195、2475和24530个样本。文件配置部分指定了每个分割对应的文件路径。

This is a dataset for natural language inference (NLI) tasks, containing sentence pairs (sentence_A and sentence_B) and their relationship labels (entailment, neutral, contradiction). The dataset also includes relatedness scores, entailment relations between sentence pairs, original sentences, sentence source datasets, template names, template content, rendered input, and rendered output. The dataset is divided into training, validation, and test sets, containing 22195, 2475, and 24530 samples respectively.