Llamaverse-3.1-8B-Instruct -Abliteration request
#819
by
Lonmine
- opened
The model already can do NSFW prompts for image generation however it refuses to use the knowledge of the other models it was merged with for RP, even jailbreaking don't seem to suffice, or the knowledge could have been lost in the merges, i don't know how abliteration actually works, but maybe that would fix the model.
https://huggingface.co/sethuiyer/Llamaverse-3.1-8B-Instruct
I could do an uncensored version of it but I would not use the abliteration technique as I'm a person of quality and so if I uncensor I want to do it properly by training it on an uncensored dataset despite it being much more expensive computationally.