Project Gutenberg

A collection of over 75,000 free eBooks, primarily consisting of classic literature, useful for training models on well-structured text.

About Dataset

Project Gutenberg is a digital library of over 75,000 freely available books, primarily consisting of literary works whose copyrights have expired or that are released under permissive terms. The collection includes classic novels, plays, essays, poetry, and reference works, mostly in plain-text formats suitable for computational processing. The texts are carefully digitized and proofread by volunteers, resulting in relatively clean and well-structured documents. The dataset is predominantly English but also includes works in many other languages.

Purpose of Dataset

Project Gutenberg Is Widely Used For Training And Evaluating Language Models On Long-form, Well-structured Text. It Is Particularly Useful For Learning Narrative Flow, Grammar, Literary Style, And Long-context Reasoning. Researchers Also Use It To Study Text Generation, Summarization, And Authorship Attribution. Because The Texts Are Largely Public Domain, The Dataset Is Suitable For Open Research, Reproducible Experimentation, And Building Language Models That Benefit From Exposure To Classic And Formal Writing Styles.