breaking the model

#2
by froilo - opened

research report:

I made it to post :
'Although safety guidance is enabled, potential unsafe content found. Please try again with different seed.'
about 30% of seeds with guidance 10 :D

Artificial Intelligence & Machine Learning Lab at TU Darmstadt org

Hi froilo,

thanks for trying out our demo. What was your prompt? If you explicitly, strongly try to break the model this could indeed work.

Best,
Patrick

Patrick, i think it contained ''busty'' and or ''suggestive '' as the strongest words!

Sign up or log in to comment