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