uscode-qa-with-context
收藏资源简介:
该数据集名为“USCode Questions and answers with context”,是一个关于美国法典(US Code)的问答数据集。它由四个现有数据集合并而成:uscode_qac、legalbench:rule_qa、USCode-QAPairs-Finetuning 和 synthetic-legal。数据集包含 700 个样本,划分为训练集(600 个样本)和测试集(100 个样本)。每个样本包含三个核心字段:问题(question)、答案(answer)和上下文(context),所有字段均为文本字符串。其中,115 个样本提供了相关的上下文信息,而其余 585 个样本则不包含上下文。该数据集适用于法律领域的自然语言处理任务,如问答系统、信息检索或法律文本理解模型的训练与评估。
The dataset is named USCode Questions and answers with context and is a question-answering dataset related to the United States Code (US Code). It is formed by merging four existing datasets: uscode_qac, legalbench:rule_qa, USCode-QAPairs-Finetuning, and synthetic-legal. The dataset contains 700 samples, divided into a training set (600 samples) and a test set (100 samples). Each sample includes three core fields: question, answer, and context, all of which are text strings. Among these, 115 samples provide relevant context information, while the remaining 585 samples do not include context. This dataset is suitable for natural language processing tasks in the legal domain, such as question answering systems, information retrieval, or training and evaluation of models for legal text understanding.
- 数据集名称:USCode Questions and answers with context
- 数据集描述:该数据集合并了多个来源,包括
uscode_qac、legalbench:rule_qa、USCode-QAPairs-Finetuning和synthetic-legal,共包含115条带有上下文的数据和585条不带上下文的数据。 - 数据集大小:数据集总大小为1,286,180字节,下载大小为326,369字节。
- 数据集拆分:
- 训练集:600个样本,大小为1,102,440字节。
- 测试集:100个样本,大小为183,740字节。
- 数据特征:
question(字符串):问题。answer(字符串):答案。context(字符串):上下文。
- 配置:
- 配置名称:
default - 训练集数据文件:
data/train-* - 测试集数据文件:
data/test-*
- 配置名称:




