bergson-recall-9000
收藏资源简介:
该数据集包含两个独立配置:questions和statements。questions配置包含36,000个训练样本,每个样本具有以下字段:question(问题)、answer(答案)、text(相关文本)、field(领域类别)和identifier(唯一标识符)。statements配置包含324,000个训练样本,每个样本具有以下字段:answer(答案)、fact(事实陈述)、field(领域类别)、identifier(唯一标识符)、question(问题)和template(模板标识)。数据集以文本形式组织,所有字段为字符串或整型。从字段推断,该数据集可能设计用于问答系统、事实检索、文本生成或事实验证等自然语言处理任务,涉及多领域的问题-答案对和事实陈述。
The dataset contains two independent configurations: questions and statements. The questions configuration includes 36,000 training samples, each with the following fields: question (question), answer (answer), text (related text), field (domain category), and identifier (unique identifier). The statements configuration includes 324,000 training samples, each with the following fields: answer (answer), fact (fact statement), field (domain category), identifier (unique identifier), question (question), and template (template identifier). The dataset is organized in text format, with all fields being strings or integers. Based on the fields, it may be designed for natural language processing tasks such as question answering systems, fact retrieval, text generation, or fact verification, involving multi-domain question-answer pairs and fact statements.
数据集概述:Bergson-recall-9000
基本信息
- 来源:EleutherAI
- 地址:https://huggingface.co/datasets/EleutherAI/bergson-recall-9000
数据集配置
该数据集包含两个配置(config):
1. questions 配置
- 训练集样本数:36,000
- 训练集大小:约5.06 MB
- 下载大小:约1.27 MB
- 特征字段:
question(字符串):问题answer(字符串):答案text(字符串):文本内容field(字符串):领域identifier(整数):标识符
2. statements 配置
- 训练集样本数:324,000
- 训练集大小:约49.21 MB
- 下载大小:约5.15 MB
- 特征字段:
answer(字符串):答案fact(字符串):事实field(字符串):领域identifier(整数):标识符question(字符串):问题template(整数):模板编号
数据规模总结
- 总样本数:360,000(36,000 + 324,000)
- 总数据集大小:约54.26 MB
- 总下载大小:约6.41 MB
- 数据划分:仅包含训练集,无验证集和测试集



