pile_of_Law

我们收集了大量的法律的和行政数据。这些数据的用途有两方面：（1）汇总体现不同数据过滤规范和法律的标准的法律的和行政数据源;（2）收集一个数据集，可用于未来的法律领域语言模型预训练，这是诉诸司法举措的一个关键方向。因此，对数据源进行策划以告知：（1）法律的分析、知识或理解;（2）论点形成;（3）隐私过滤标准。像法典和法律这样的来源倾向于提供信息（1）。成绩单和法庭文件往往提供信息（2）。意见倾向于告知（1）和（3）。

We have collected a large volume of legal and administrative data. This data serves two purposes: (1) aggregating legal and administrative data sources that embody different data filtering norms and legal standards; (2) compiling a dataset applicable for pre-training future legal domain language models—a key direction for judicial initiatives. Therefore, the data sources are curated to inform three aspects: (1) legal analysis, knowledge, or understanding; (2) argumentation formation; (3) privacy filtering standards. Sources like codes and laws tend to support aspect (1); transcripts and court documents often contribute to aspect (2); while opinions typically inform both (1) and (3).