LAiW
收藏资源简介:
LAiW是一个专为评估中文法律领域大型语言模型(LLMs)而设计的数据集,由四川大学等机构的研究人员创建。该数据集基于法律实践的逻辑,将法律能力分为三个层次:基本信息检索、法律基础推理和复杂法律应用,每个层次包含多个任务,以确保全面评估。数据集的建设结合了多个公开源数据集和少量专有数据,旨在通过自动化评估和人工评估,揭示当前法律LLMs在实际应用中的能力和局限性。
LAiW is a dataset specifically designed for evaluating large language models (LLMs) in the Chinese legal domain, created by researchers from Sichuan University and other institutions. Rooted in the logic of legal practice, this dataset classifies legal capabilities into three tiers: basic information retrieval, basic legal reasoning, and complex legal application. Each tier includes multiple tasks to ensure a comprehensive evaluation. Constructed by integrating multiple publicly available legal datasets and a small amount of proprietary data, this dataset aims to reveal the current capabilities and limitations of legal LLMs in real-world applications through both automated and manual evaluations.




