research
∙
12/31/2020
The Pile: An 800GB Dataset of Diverse Text for Language Modeling
Recent work has demonstrated that increased training dataset diversity i...
share
research
∙
11/21/2016
