1 paper
Stella Biderman, Kieran Bicheno, Leo Gao
This datasheet describes the Pile, a 825 GiB dataset of human-authored text compiled by EleutherAI for use in large-scale language modeling. The Pile is comprised of 22 different t…