allenai-sera-unified-31600

该数据集包含31,600个训练样本，总大小约为6.79GB。每个样本包含以下字段：1) 'conversations'字段（列表结构），内含对话内容('content'字符串字段)和发言角色('role'字符串字段)；2) 'instance_id'字符串字段；3) 'source'字符串字段。数据集仅提供训练集划分，未说明具体应用场景或数据来源。下载压缩包大小约为1.93GB，解压后数据集大小为6.79GB。

This dataset contains 31,600 training samples with an overall size of approximately 6.79 GB. Each sample includes the following fields: 1) the "conversations" field, a list structure that contains conversation content (stored in the string field "content") and the speaker's role (stored in the string field "role"); 2) the "instance_id" string field; 3) the "source" string field. Only the training split is provided for this dataset, with no specific application scenarios or data sources specified. The compressed download package has a size of approximately 1.93 GB, while the uncompressed dataset occupies 6.79 GB of storage.

原始信息汇总

数据集概述

基本信息

• 数据集名称: allenai-sera-unified-31600
• 发布者: laion
• 数据集地址: https://huggingface.co/datasets/laion/allenai-sera-unified-31600

数据规模

• 训练集样本数量: 31,600
• 训练集大小: 6,676,608,567.242575 字节
• 下载大小: 1,905,620,636 字节
• 数据集总大小: 6,676,608,567.242575 字节

数据结构

特征字段

1. conversations
   • 类型: 列表
   • 结构:
     • content: 字符串类型
     • role: 字符串类型
2. instance_id: 字符串类型
3. source: 字符串类型

数据划分

• 划分名称: train
• 数据文件路径: data/train-*

配置信息

• 配置名称: default
• 数据文件关联:
  • 划分: train
  • 路径: data/train-*