Salesforce/cos_e
收藏资源简介：
Commonsense Explanations (CoS-E) 数据集旨在训练语言模型，使其能够自动生成解释，这些解释可以在训练和推理过程中使用。数据集包含两个版本（v1.0和v1.11），每个版本都有训练集和验证集。数据集的字段包括id、question、choices、answer、abstractive_explanation和extractive_explanation。数据集的语言为英语，且为单语数据集。数据集的创建过程涉及众包，但具体的注释过程和注释者信息未提供。数据集的许可证信息未知。
The Commonsense Explanations (CoS-E) dataset is designed to train language models to automatically generate explanations that can be used during both training and inference stages. It has two versions, v1.0 and v1.11, each with a training split and a validation split. The dataset's fields include id, question, choices, answer, abstractive_explanation, and extractive_explanation. It is an English-only monolingual dataset. The dataset was created via crowdsourcing, but specific annotation procedures and annotator information are not provided. The license information of the dataset is unknown.
数据集概述
基本信息
- 数据集名称: Commonsense Explanations (CoS-E)
- 语言: 英语
- 许可: 未知
- 多语言性: 单语种
- 数据集大小: 10K<n<100K
- 源数据集: 扩展自commonsense_qa
- 任务类别: 问答
- 任务ID: 开放领域问答
- PapersWithCode ID: cos-e
- 美观名称: Commonsense Explanations
数据集配置
v1.0
- 特征:
id: 字符串question: 字符串choices: 字符串序列answer: 字符串abstractive_explanation: 字符串extractive_explanation: 字符串
- 分割:
train: 7610个样本, 2067971字节validation: 950个样本, 260669字节
- 下载大小: 1588340字节
- 数据集大小: 2328640字节
v1.11
- 特征:
id: 字符串question: 字符串choices: 字符串序列answer: 字符串abstractive_explanation: 字符串extractive_explanation: 字符串
- 分割:
train: 9741个样本, 2702777字节validation: 1221个样本, 329897字节
- 下载大小: 1947552字节
- 数据集大小: 3032674字节
数据集结构
数据实例
v1.0
- 下载大小: 4.30 MB
- 生成数据集大小: 2.34 MB
- 总磁盘使用量: 6.64 MB
- 示例: json { "abstractive_explanation": "this is open-ended", "answer": "b", "choices": ["a", "b", "c"], "extractive_explanation": "this is selected train", "id": "42", "question": "question goes here." }
v1.11
- 下载大小: 6.53 MB
- 生成数据集大小: 3.05 MB
- 总磁盘使用量: 9.58 MB
- 示例: json { "abstractive_explanation": "this is open-ended", "answer": "b", "choices": ["a", "b", "c"], "extractive_explanation": "this is selected train", "id": "42", "question": "question goes here." }
数据字段
v1.0
id: 字符串question: 字符串choices: 字符串列表answer: 字符串abstractive_explanation: 字符串extractive_explanation: 字符串
v1.11
id: 字符串question: 字符串choices: 字符串列表answer: 字符串abstractive_explanation: 字符串extractive_explanation: 字符串
数据分割
| 配置名称 | 训练集样本数 | 验证集样本数 |
|---|---|---|
| v1.0 | 7610 | 950 |
| v1.11 | 9741 | 1221 |
