CLDec 31, 2020

The Pile: An 800GB Dataset of Diverse Text for Language Modeling

arXiv:2101.00027v12841 citations
Originality Incremental advance
AI Analysis

This dataset addresses the problem of limited training data diversity for large-scale language models, benefiting researchers and developers in the NLP community.

This paper introduces The Pile, an 825 GiB English text corpus comprising 22 diverse high-quality subsets, designed to enhance the training of large-scale language models. Models trained on The Pile show significant performance improvements over those trained on Raw CC and CC-100 across all components of The Pile, and also improve performance on downstream evaluations.

Recent work has demonstrated that increased training dataset diversity improves general cross-domain knowledge and downstream generalization capability for large-scale language models. With this in mind, we present \textit{the Pile}: an 825 GiB English text corpus targeted at training large-scale language models. The Pile is constructed from 22 diverse high-quality subsets -- both existing and newly constructed -- many of which derive from academic or professional sources. Our evaluation of the untuned performance of GPT-2 and GPT-3 on the Pile shows that these models struggle on many of its components, such as academic writing. Conversely, models trained on the Pile improve significantly over both Raw CC and CC-100 on all components of the Pile, while improving performance on downstream evaluations. Through an in-depth exploratory analysis, we document potentially concerning aspects of the data for prospective users. We make publicly available the code used in its construction.

Code Implementations22 repos

Data from Papers with Code (CC-BY-SA-4.0)

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes