Work in Progress
Weight of 0.9 is good for Anything 3.0.
This is trained not extracted. People have asked about this. I don't think it's as good as my Owler Model. I'm not experienced with making style Loras. I've tried extracting a Lora from my model and It doesn't look at all the same.
My model is a merge of two separate checkpoints I trained. I don't know if I will be able to a Lora that can produce similar results.
I'll try to make an updated version that is better. Go easy on me.
Description
FAQ
Comments (9)
Damn, my man is both a hard and a fast working man, thank you :)
Nice Lora and thanks for sharing? May I ask, what are you using for training? I do use Kohya but, I wonder how you have a file size like 1mb.
I also use Kohya. Lower the network Rank to 8 or below. I don't know what this Lora was set to.
Thank you
Was your class name "art style"? Do you use regularisation images?
@sekoseko I didn't use any regularization images. Maybe I should.
@Shurik I asked about regularisation images because you have great loras of people. Do you use reg images for them?
@sekoseko I think I did for Vayntrub and Skye Blue. I actually forgot I was doing that...
Thank you for all of your answers.