Shawn PresserThe Pile: An 800GB Dataset of Diverse Text for Language ModelingGPT-NeoX (GitHub)EleutherAI (GitHub)All names