3 ms·

Have you tried abliterated models? I'm curious if the current de-censorship methods are effective in that area / at that level.