copenlu/reality-check-on-context-utilisation

该数据集用于评估模型对上下文的利用，并测试自动化的声明验证方法。数据集包含来自CounterFact、ConflictQA和DRUID的样本，并附有Llama和Pythia模型的预测结果。数据集分为两种版本：few-shot和zero-shot，每种版本都有对应的数据文件和特征。数据集的详细信息可以在相关研究论文中找到。

This dataset is used for evaluating model context usage and testing automated claim verification methods. It includes samples from CounterFact, ConflictQA, and DRUID with predictions from Llama and Pythia models. The dataset comes in two versions: few-shot and zero-shot, each with corresponding data files and features. More details about the dataset can be found in the related research paper.