sanchit-gandhi/cosmopedia_web_textbooks_logprobs

---
dataset_info:
features:
- name: prompt
dtype: string
- name: text_token_length
dtype: int64
- name: text
dtype: string
- name: seed_data
dtype: string
- name: format
dtype: string
- name: audience
dtype: string
- name: logprobs
dtype: float64
splits:
- name: train
num_bytes: 29924148869
num_examples: 5000000
download_size: 16394553341
dataset_size: 29924148869
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
---

The dataset includes multiple features such as prompt, text_token_length, text, seed_data, format, audience, and logprobs. The data types range from string to int64. The dataset is primarily for training, containing 5 million examples. The download size is 16394553341 bytes, and the actual size is 29924148869 bytes.

原始信息汇总

数据集概述

数据集特征

• prompt：字符串类型
• text_token_length：整数类型（int64）
• text：字符串类型
• seed_data：字符串类型
• format：字符串类型
• audience：字符串类型
• logprobs：浮点数类型（float64）

数据集拆分

• train：
  • 数据量：29,924,148,869 字节
  • 示例数量：5,000,000

数据集大小

• 下载大小：16,394,553,341 字节
• 数据集大小：29,924,148,869 字节

配置

• config_name: default
• data_files:
  • split: train
  • path: data/train-*