Hearsay determination task from LegalBench

该数据集用于进行传闻证据判断任务，它要求通过多步骤推理来确定证据是否根据《联邦证据规则》构成不可采纳的传闻。此外，该数据集结合了一个结构化提示框架，该框架融合了神经符号方法进行规则应用。该数据集的规模包括4个训练示例和95个测试示例，其任务是进行传闻判断。

This dataset is designed for the hearsay evidence judgment task, which requires multi-step reasoning to determine whether the presented evidence constitutes inadmissible hearsay under the Federal Rules of Evidence. Furthermore, this dataset incorporates a structured prompting framework that integrates neuro-symbolic methods for rule application. The dataset contains 4 training instances and 95 test instances, with its core task centered on hearsay judgment.