6k steps but lora unet : 0.0005 , lora text encode 0.00003
so smaller steps , and batch size 1 .




If you enjoyed the model I provided, I would greatly appreciate your consideration in making a donation to support my college education.
https://www.patreon.com/user?u=87243198
this was more practice on training person with rtx 3080 with 10vram.
for the dataset I used 10 pictures from the show using vlc.
you can see the dataset here : https://imgur.com/a/GDy1lQH
I used a specific version of dreambooth to get it to work
https://github.com/d8ahazard/sd_dreambooth_extension/tree/c5cb58328c555ac27679422b1da940a9b19de6f2 Here, this is the commit when they first added LoRA.
I added the settings in the pictures of (Jan Levinson 3000 Steps)
6k steps but lora unet : 0.0005 , lora text encode 0.00003
so smaller steps , and batch size 1 .
1. 재게시된 모델의 권리는 원 제작자에게 있습니다.
2. 모델 원작자가 모델을 인증받으려면 공식 채널을 통해 SeaArt.AI 직원에게 문의하세요. 저희는 모든 창작자의 권리를 보호하기 위해 노력합니다. 인증하러 이동
