Respuestas al mensaje #4,139

Eh, I mean none of the stable diffusion models were so I do think it's a little surprising.

I think this is actually a mistake they made while training it. "Convert the training dataset into lower case so the model will be case...