3 ms·
Back when GPT-3 was first announced I got kind of scared, and decided to download the then-current Kiwix ZIM archives of Wikipedia, Stack Overflow, Wikihow, Wik
by LeoPanthera 4y ago
Back when GPT-3 was first announced I got kind of scared, and decided to download the then-current Kiwix ZIM archives of Wikipedia, Stack Overflow, Wikihow, Wikisource, and a number of other similar sites.
I'm kind of glad that I did, and intend to keep these versions "forever", as examples of pre-LLM human-generated content.
- dreamyfigment 4y agoAny chance you can upload those versions to archive.org?
- LeoPanthera 4y agoThat's a good idea. If they're not already there I will do so. Edit: The Internet Archive already has a reasonably comprehensive ZIM archive, just filter by year for 2019 or earlier: https://archive.org/details/zimarchive?sort=-week&and[]=year%3A%222019%22 https://archive.org/details/zimarchive?sort=-week&and[]=year...
- sitkack 4y agoWe are already seeing this with sites that pump as many prompts through SD and spam the internet with junk images. Future systems will at least have to have quality discriminators when training on these images.