News for hackers. Stuff that matters.

A comment on The Pile is a 825 GiB diverse, open source language modelling data set. Parent comment

Re:The Pile is a 825 GiB diverse, open source language model... by whimsicalism on Friday March 08, 2024 @10:36AM (#39642004)
I think it is almost certain that GPT was trained on full books
[ Parent | on HN ]

There are no comments on this comment yet.