Software Engineering Courses (SECourses)•3y ago

i also made a slightly more realistic merge of the model for my dog.

LLEGACIES it has 12gb

Furkan Gözükara SECourses•10/8/23, 1:09 PM

with 12 gb you can do SDXL lora

Furkan Gözükara SECourses•10/8/23, 1:09 PM

best option for you

Furkan Gözükara SECourses•10/8/23, 1:09 PM

sd 1.5 dreambooth perhaps with optimization but quality i dont know need testing

SSandstormleader i also made a slightly more realistic merge of the model for my dog.

Furkan Gözükara SECourses•10/8/23, 1:09 PM

nice

Furkan Gözükara SECourses•10/8/23, 1:10 PM

can you show your settings? configuration

FFurkan Gözükara SECourses with 12 gb you can do SDXL lora

LEGACIES•10/8/23, 1:10 PM

do you think that i will get bettwe result with a 24gb gpu on runpod?

LLEGACIES do you think that i will get bettwe result with a 24gb gpu on runpod?

Furkan Gözükara SECourses•10/8/23, 1:10 PM

yes

Furkan Gözükara SECourses•10/8/23, 1:10 PM

100%

Furkan Gözükara SECourses•10/8/23, 1:10 PM

with dreambooth of SDXL

Furkan Gözükara SECourses•10/8/23, 1:11 PM

look at this quality : https://twitter.com/GozukaraFurkan/status/1710995764162216205

LEGACIES•10/8/23, 1:11 PM

i dont know how to use SDXL, thats why im training for sd 1.5

FFurkan Gözükara SECourses yes

LEGACIES•10/8/23, 1:12 PM

do you have a video how to train model with dreambooth for sdxl?

LLEGACIES do you have a video how to train model with dreambooth for sdxl?

Furkan Gözükara SECourses•10/8/23, 1:12 PM

i have for lora

Furkan Gözükara SECourses•10/8/23, 1:12 PM

you only need to load config

Furkan Gözükara SECourses•10/8/23, 1:12 PM

from patreon

Furkan Gözükara SECourses•10/8/23, 1:12 PM

and prepare dataset nothing else

Furkan Gözükara SECourses•10/8/23, 1:12 PM

https://youtu.be/-xEwaQ54DI4

YouTubeSECourses

How To Do SDXL LoRA Training On RunPod With Kohya SS GUI Trainer & ...

If you don't have a GPU, or have a strong GPU, or you are using Mac and your computer not supporting Stable Diffusion training, SDXL training, then this is the tutorial you are looking for. In this tutorial, we will use a cheap cloud GPU service provider RunPod to use both Stable Diffusion Web UI Automatic1111 and Stable Diffusion trainer Kohya ...

Furkan Gözükara SECourses•10/8/23, 1:12 PM

after watching this video you can load config and use it

Furkan Gözükara SECourses•10/8/23, 1:12 PM

i will also hopefully record quick video for patreon

Furkan Gözükara SECourses•10/8/23, 1:13 PM

currently fixing kaggle notebook

FFurkan Gözükara SECourses i have for lora

LEGACIES•10/8/23, 1:13 PM

is ti still possible to train model with dreambooth?

LLEGACIES is ti still possible to train model with dreambooth?

Furkan Gözükara SECourses•10/8/23, 1:13 PM

yes

Stefanoid9343•10/8/23, 1:13 PM

@Dr. Furkan Gözükara I’m using runpod what disk space should I use for training 100GB for both? persistence and the other one?

Furkan Gözükara SECourses•10/8/23, 1:13 PM

with lora

SStefanoid9343 @Dr. Furkan Gözükara I’m using runpod what disk space should I use for training ...

Furkan Gözükara SECourses•10/8/23, 1:14 PM

you need only volume disk

Furkan Gözükara SECourses•10/8/23, 1:14 PM

not the other one

FFurkan Gözükara SECourses yes

LEGACIES•10/8/23, 1:14 PM

can you please have a look at my dataset and give some tips?

FFurkan Gözükara SECourses you need only volume disk

Stefanoid9343•10/8/23, 1:14 PM

the volume disk 100gb and the other one could be 20?

LLEGACIES can you please have a look at my dataset and give some tips?

Furkan Gözükara SECourses•10/8/23, 1:14 PM

sure pm me

SStefanoid9343 the volume disk 100gb and the other one could be 20?

Furkan Gözükara SECourses•10/8/23, 1:15 PM

yep

FFurkan Gözükara SECourses sure pm me

LEGACIES•10/8/23, 1:15 PM

done

FFurkan Gözükara SECourses yep

Stefanoid9343•10/8/23, 1:16 PM

I’m having double bodies/faces following your training tutorial also naked images using real vision the same parameters as yours

Stefanoid9343•10/8/23, 1:16 PM

canvas 768x1024 I’m also using the data images of woman. any advise?

FFurkan Gözükara SECourses look at this quality : https://twitter.com/GozukaraFurkan/status/171099576416221...

Elisoli•10/8/23, 1:17 PM

Adetailer used here? I prefer resaults without adetailer as it keeps the original texutre and style

SStefanoid9343 I’m having double bodies/faces following your training tutorial also naked image...

Furkan Gözükara SECourses•10/8/23, 1:22 PM

those models are nasty

Furkan Gözükara SECourses•10/8/23, 1:22 PM

i think i would try sdxl

EElisoli Adetailer used here? I prefer resaults without adetailer as it keeps the origina...

Furkan Gözükara SECourses•10/8/23, 1:22 PM

adetailer were used for realistic images

Furkan Gözükara SECourses•10/8/23, 1:22 PM

for styled i didnt use

Elisoli•10/8/23, 1:22 PM

ohh and not used for stylized, very good idea

EElisoli ohh and not used for stylized, very good idea

Furkan Gözükara SECourses•10/8/23, 1:23 PM

yes

Elisoli•10/8/23, 1:23 PM

also should we then use Text Encoder training or not? kept the suspence going on that twit

Elisoli•10/8/23, 1:23 PM

looks pretty identical

Furkan Gözükara SECourses•10/8/23, 1:23 PM

this one is amazing

EElisoli also should we then use Text Encoder training or not? kept the suspence going o...

Furkan Gözükara SECourses•10/8/23, 1:23 PM

i think looks like not very necessary

Furkan Gözükara SECourses•10/8/23, 1:23 PM

but can be used too :d

rafraf•10/8/23, 1:24 PM

@Dr. Furkan Gözükara does NOT-raining text encoder save memory and speed? Is it worth turning off TE learning?

rafraf•10/8/23, 1:25 PM

Oh just noticed someone asked the same question

Rrafraf @Dr. Furkan Gözükara does NOT-raining text encoder save memory and speed? Is it ...

Furkan Gözükara SECourses•10/8/23, 1:25 PM

it is turned off by default

Furkan Gözükara SECourses•10/8/23, 1:25 PM

it increases vram slighty and reduces training speed slightly

rafraf•10/8/23, 1:26 PM

Also, where do you explain the block training --block_lr 1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5

i also made a slightly more realistic merge of the model for my dog.

Similar Threads

Similar Threads

Similar Threads