Skip to content

Dataset · euirim

Selected English Wikipedia articles in Markdown

euirim/goodwiki

From 10 to 100 thousand high-quality Wikipedia articles with 179 million tokens preserving markup for generation and summarization.

What's inside

Inside are gathered from 10 to 100 thousand English-language articles with a total volume of 179 million tokens, marked by Wikipedia editors as 'good' or 'featured' (snapshot from September 4, 2023). The texts are presented in GitHub-flavored Markdown format preserving lists, code blocks, formulas, and quotes, and are also provided with brief descriptions and categories.

What you can do with it

  • Training language models to generate structured texts in Markdown format
  • Training and evaluating summarization models on verified encyclopedic articles
Tasks
Text generation, Summarization
Languages
English
Size
10K–100K rows
Downloads this month
201
Likes
56

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("euirim/goodwiki")