HuRTE
收藏资源简介:
HuRTE数据集是匈牙利语版本的Recognizing Textual Entailment (RTE)数据集,属于匈牙利语言理解评估基准工具包(HuLU)的一部分。该数据集通过翻译和重新注释GLUE基准中的RTE数据集实例创建。数据集包含4,504个实例,每个实例包含一个(有时是多句的)前提和一个单句假设,任务是判断前提是否蕴含假设,属于二分类任务。数据集分为训练集、验证集和测试集,分别包含2,131、242和2,131个实例。测试集不包含标签,评估模型需联系作者或访问HuLU网站。数据格式为JSON,包含id、premise、hypothesis和label四个键。
The HuRTE dataset is the Hungarian version of the Recognizing Textual Entailment (RTE) dataset, which forms part of the Hungarian Language Understanding Evaluation Benchmark Toolkit (HuLU). This dataset was developed by translating and re-annotating the RTE dataset instances from the GLUE benchmark. The dataset contains 4,504 instances in total, each consisting of a (sometimes multi-sentence) premise and a single-sentence hypothesis. The task of this dataset is to determine whether the premise entails the hypothesis, which is a binary classification task. The dataset is split into training, validation and test sets, which hold 2,131, 242 and 2,131 instances respectively. The test set does not include ground-truth labels, and researchers intending to evaluate their models should contact the dataset authors or visit the HuLU official website. The data is formatted as JSON, with four core keys: id, premise, hypothesis and label.




