davanstrien/ledgar-top10-cascade-2tier

该数据集是一个由LLM（大型语言模型）标注的数据集，名为davanstrien/ledgar-top10-cascade-2tier，通过classify-and-augment工具生成。数据集配置使用了两个层级的模型：第一层是LiquidAI/LFM2.5-1.2B-Instruct，第二层是Qwen/Qwen3-4B-Instruct-2507。数据集包含200条输入行和200条输出行，标注了10个标签，包括Governing Laws、Notices、Counterparts等。标签分布表展示了各标签的数量统计，Cascade audit部分则详细记录了模型处理过程中的各项指标和层级间的一致性情况。

This dataset is an LLM-annotated dataset named davanstrien/ledgar-top10-cascade-2tier, produced by the classify-and-augment tool. The configuration uses a two-tier cascade of models: Tier 1 is LiquidAI/LFM2.5-1.2B-Instruct, and Tier 2 is Qwen/Qwen3-4B-Instruct-2507. The dataset consists of 200 input rows and 200 output rows, annotated with 10 labels including Governing Laws, Notices, Counterparts, etc. The label distribution table shows the count statistics for each label, and the Cascade audit section details various metrics during model processing and inter-tier agreement.