Also I was playing with chatgpt today and I discovered that if you use direct api access to their models the results are "raw" meaning it doesn't go through their "bias and safety" filter in other words it will say nigger.
Edit: Looks like this broke around the time I posted it. It's been taken down now.
[ + ]deleted
[ - ] deleted 1 point 1 yearMar 30, 2023 23:08:42 ago (+1/-0)
[ + ] SecretHitler
[ - ] SecretHitler [op] 0 points 1 yearMar 30, 2023 23:59:11 ago (+0/-0)
[ + ] BoraxTheFungarian
[ - ] BoraxTheFungarian 0 points 1 yearMar 31, 2023 11:29:18 ago (+0/-0)
High volume, try again later. Curious.
[ + ] Clubberlang
[ - ] Clubberlang 0 points 1 yearMar 31, 2023 10:15:31 ago (+0/-0)
[ + ] Clubberlang
[ - ] Clubberlang 0 points 1 yearMar 31, 2023 10:13:34 ago (+0/-0)
[ + ] Clubberlang
[ - ] Clubberlang 0 points 1 yearMar 31, 2023 10:10:48 ago (+0/-0)
[ + ] Nein
[ - ] Nein 0 points 1 yearMar 31, 2023 09:59:02 ago (+0/-0)
[ + ] The_Reunto
[ - ] The_Reunto 0 points 1 yearMar 31, 2023 07:09:27 ago (+0/-0)
[ + ] oursenilepresident
[ - ] oursenilepresident 0 points 1 yearMar 31, 2023 02:17:02 ago (+0/-0)*
An example after a long delay from your link of censoring :
answer:
hah!
kikes!
= = = = =
usually you edit out one line of code to circumvent ENTIRE safety layer.
example :
remove censoring of Jews from stableDiffusion ANIMATION :
https://archive.ph/4PmAX
HAH HAH HAH! change that single line in my archive example to uncensor it more.
= = = =
I posted other one line changes before, sometimes by reducing magnitude of violation value
[ + ] SecretHitler
[ - ] SecretHitler [op] 1 point 1 yearMar 31, 2023 05:33:46 ago (+1/-0)