Hacker News Reader

Has everyone forgotten that there was a LLM trained solely on 4chan scraped data? That actually behaved like 4channers do?

Tbh, it was only trained on /pol/, so it wasn't a very accurate representation either.