zorache/wmdp_fewshot

该数据集是一个用于文本生成任务的英文数据集，包含多选题形式的生物问题。数据集分为fewshot_in_context和fewshot_test两个部分，分别包含63和1210个示例。每个示例包括答案、问题和多个选择。

This dataset is designed for English text generation tasks, specifically configured as wmdp-mcqs. It includes three features: answer (dtype: int64), question (dtype: string), and choices (dtype: string sequence). The dataset is divided into two parts: fewshot_in_context (63 samples, 25110 bytes) and fewshot_test (1210 samples, 484295 bytes). The data files are stored in the wmdp-mcqs directory as bio_questions_fewshot_in_context.json and bio_questions_fewshot_test.json.

原始信息汇总

数据集概述

基本信息

• 语言: 英语
• 许可证: MIT
• 数据集大小: 1K<n<10K
• 任务类别: 文本生成

数据集配置

• 配置名称: wmdp-mcqs

特征

• answer: 数据类型为int64
• question: 数据类型为string
• choices: 数据类型为sequence of string

数据分割

• fewshot_in_context
  • 字节数: 25110
  • 样本数: 63
• fewshot_test
  • 字节数: 484295
  • 样本数: 1210

数据文件

• fewshot_in_context
  • 路径: wmdp-mcqs/bio_questions_fewshot_in_context.json
• fewshot_test
  • 路径: wmdp-mcqs/bio_questions_fewshot_test.json