README.md

metadata

dataset_info:
  features:
    - name: dataType
      dtype: string
    - name: title
      dtype: string
    - name: content
      dtype: string
    - name: uniqueKey
      dtype: string
    - name: titleUkey
      dtype: string
    - name: id
      dtype: int64
  splits:
    - name: train
      num_bytes: 58043923125
      num_examples: 16905023
  download_size: 25662051889
  dataset_size: 58043923125

Dataset Card for "pretrain_zh"

Tigerbot pretrain数据的中文部分。

包含(未压缩前) 书籍zh-books 12G, 新闻zh-news 25G, 百科类zh-wiki 19G

更多语料请关注开源模型及持续更新 https://github.com/TigerResearch/TigerBot

Usage

import datasets

ds_sft = datasets.load_dataset('TigerResearch/pretrain_zh')