Fine tune and lora
#12
by
andreclaudino
- opened
It's a great work, but now the model is not able to create uncensored content, that includes creating videos from NSFW images. Making different localizations like adapting backgrounds and police uniforms for different countries would be great.
What is the best approach for that? How to fine tune the model or create a LoRA that allows achieving targets like these? How much video time is needed, what are the hardware requirements?
Thanks.