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 archon1410 on Thursday March 07, 2024 @01:50PM (#39633059)
> The Pile is old news, check out more recent datasets like; https://huggingface.co/datasets/bigcode/the-stack-v2

— https://the-eye.eu/public/AI/pile/readme.txt

[ Parent | on HN ]
Re:The Pile is a 825 GiB diverse, open source language model... by natch on Thursday March 07, 2024 @02:55PM (#39633860)
Super odd message since the stack v2 seems to be exclusively code and The Pile is (mostly?) text.
[ Parent | on HN ]