Skip to content

Dataset · manu

Project Gutenberg books

manu/project_gutenberg

Over 70k free books in multiple languages for text generation.

What's inside

The dataset is collected from the Project Gutenberg library, featuring over 70k free e-books. Each example corresponds to a single book and contains a service header and footer between the tags *** Start of *** and *** End of ***. Language splits are available, including French, English, Chinese, Portuguese, Polish, and Dutch; data can be streamed.

What you can do with it

  • Fine-tuning language models on long literary texts
  • Training LoRA for book-style generation
  • Evaluating generation quality on literary texts
Tasks
Text generation
Languages
French, English, Chinese, Portuguese, Polish, Dutch, Russian, Swedish and 3 more
Size
10K–100K rows
Downloads this month
155.2K
Likes
77

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("manu/project_gutenberg")