suggest weight: 0.5-0.6
推荐权重：0.5-0.6
倍率：0.5-0.6
ーーー
NO COMMERCIAL USE
Description
FAQ
Comments (4)
do you use kohya? If yes, can you share your setting paramater like network dimension, learning rate, etc? I'm kinda struggling getting a good result. Seeing your good model make me wonder what went wrong on my side
>do you use kohya?
Yes
>can you share your setting paramater like network dimension
Of course.
you can find parameters on the following page
https://seesaawiki.jp/nai_ch/d/Dreambooth-LoRA
written in Japanese but they attached images and parameters are there.
Also do you edit text files to describe the character you want to learn? I think that's a important point.
sorry for my bad English too.
@doyagao Thanks for sharing~! Yes i did create the text file, and im using tool like WD autotagger to create the tag and modify the tag result.
to be fair, my main source of my image data is screenshot image from anime. Maybe that can be a problem, idk.
Now i just need to understand what you just gave to me. Once again, thank you
@zetsubou hope you have a good result!
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.