Software Engineering Courses (SECourses)•3y ago

that did it 🙂

that did it

Ppeels Is there a way to optimize learning with one or a small amount of images?

Furkan Gözükara SECourses•6/6/23, 5:59 PM

how many. and why small number?

NNom that did it 🙂

ashley•6/6/23, 6:00 PM

Did you also disable xformers?

Aashley Did you also disable xformers?

NomOP•6/6/23, 6:00 PM

do I have to? I need to do inference after

NNom no 😦 do I have to? I need to do inference after

ashley•6/6/23, 6:01 PM

Inference is in A1111, you need to disable xformers for training to get better quality

NomOP•6/6/23, 6:01 PM

ok how do I do that?

NNom ok how do I do that?

ashley•6/6/23, 6:02 PM

Under "Training parameters" expand "Advanced Configuration" and UNCHECK "Use xformers"

ashley•6/6/23, 6:04 PM

It also defaults to "cosine" as "LR scheduler" but I seem to get better results using "constant_with_warmup"

frandagostino•6/6/23, 6:11 PM

Hi @Dr. Furkan Gözükara, how are you? What do you recommend for doing some Textual Inversion training experiments?
Colab (I'm paying 10 USD plan) or Runpod?
Locally I've a a Apple M2 8-cores 16GB, which I think is super slow for training.

Ffrandagostino Hi @Dr. Furkan Gözükara, how are you? What do you recommend for doing some Textu...

Furkan Gözükara SECourses•6/6/23, 6:13 PM

do you have any particular reason for textual inversion

frandagostino•6/6/23, 6:15 PM

No, just that I've seen great models in Civitai, low size, and wanted to do some exploration

NomOP•6/6/23, 6:21 PM

@Dr. Furkan Gözükara do you have any advise on which training method to use if I'm looking to get a sketch style like this:

hand-drawn-cat-outline-illustration_23-2149299744.jpg.jpg

Eggo•6/6/23, 6:34 PM

hey!! for anyone doing lora training with 1111 dreambooth extension, how are you getting txt2img to recognize it? for me now, it wont pickup on any Lora unless i use the "sd-webui-additional-networks" extension. although i believe the previous method with "a1111-sd-webui-locon" extension was more convenient (red icon at the top under generate)? (side note, when i train with dreambooth extension, it saves the lora as .pt files not safetensors. it only saves as safetensors if i check the option "generate lora weights for extra networks" , but then i cant use those in the prompt (i receive error)

Le•6/6/23, 6:43 PM

dreambooth is installed but it is not visible as a tab.

Original message was deleted

qwerty_qwer•6/6/23, 6:44 PM

reload ui

Le•6/6/23, 6:44 PM

I did. Closed terminals, changed port. Nothing.

Le•6/6/23, 6:48 PM

does anyone know what to do here?

Le•6/6/23, 6:50 PM

and to be honest, I have dreambooth installed on my webui in my pc, and I dont see it in the tab either

Le•6/6/23, 6:51 PM

is it a python problem?

qwerty_qwer•6/6/23, 6:51 PM

@Le open extentions tab

EEggo hey!! for anyone doing lora training with 1111 dreambooth extension, how are you...

Eggo•6/6/23, 6:51 PM

but if i train a lora in Kohya SS gui, then bring it to 1111 it works just fine without using the additional networks extension (extra network) should i jsut stick with kohya instead of dreambooth? (prob is right now my kohya has a problem with xformers so traingiing is 8 hours instead of the normal 20 minutes)

Le•6/6/23, 6:52 PM

this is from runpod

Le•6/6/23, 6:54 PM

no dreambooth on both

Le•6/6/23, 6:56 PM

@Dr. Furkan Gözükara have you seen anything like this?

Le•6/6/23, 7:12 PM

Tried reseting my pod. still nothing

Ffrandagostino No, just that I've seen great models in Civitai, low size, and wanted to do some...

Furkan Gözükara SECourses•6/6/23, 7:36 PM

dreambooth and lora are better than textual inversion

NNom @Dr. Furkan Gözükara do you have any advise on which training method to use if I...

Furkan Gözükara SECourses•6/6/23, 7:36 PM

well if you can get good training dataset dreambooth best

EEggo hey!! for anyone doing lora training with 1111 dreambooth extension, how are you...

Furkan Gözükara SECourses•6/6/23, 7:36 PM

in the options there is some compability

Furkan Gözükara SECourses•6/6/23, 7:36 PM

i havent tested yet

Furkan Gözükara SECourses•6/6/23, 7:37 PM

Original message was deleted

Furkan Gözükara SECourses•6/6/23, 7:37 PM

that means install failed

Furkan Gözükara SECourses•6/6/23, 7:37 PM

if you open any desk i can quickly install

FFurkan Gözükara SECourses if you open any desk i can quickly install

Le•6/6/23, 7:39 PM

I didn't understand it

LLe I didn't understand it

Furkan Gözükara SECourses•6/6/23, 7:39 PM

i will dm you

Zono50•6/6/23, 8:13 PM

should we be using xformers for lora training or no?

EEggo but if i train a lora in Kohya SS gui, then bring it to 1111 it works just fine ...

Eggo•6/6/23, 8:31 PM

Does really not one person know? Everyone’s Lora implementation must have broken then

ZZono50 should we be using xformers for lora training or no?

Eggo•6/6/23, 8:36 PM

It makes the training go way faster (xformers and gradient checkpointing)

Zono50•6/6/23, 8:41 PM

right but i keep seeing people say not to use it, and i've heard in the past that it was botching peoples trainings

ZZono50 right but i keep seeing people say not to use it, and i've heard in the past tha...

Eggo•6/6/23, 8:49 PM

It can also depend on what version of xformers you have. Like, using 0.0.19 will give different results to 0.0.20 , or so I’ve read; so you want to find the right xformers . I think it’s mostly to reduce vram usage too. Side note, Like my problem right now is dreambooth extension is giving me issues as it always does but training takes like 20 min with xformers but kohya is more reliable but xformers isn’t working for me and same training takes 8 hours

Eggo•6/6/23, 8:49 PM

So I’d say use xformers, unless it’s bitching your photos

Eggo•6/6/23, 8:50 PM

Lastly, different versions of 1111 and kohya, extensions etc keep getting updated/ broken and it’s tough to really track down what’s up -_-

Zono50•6/6/23, 8:53 PM

Agree with that 100%, yeah not using xformers takes like 3 times as long

Astro9•6/6/23, 9:16 PM

can someone help me https://www.youtube.com/watch?v=TpuDOsuKIBo&t=1625s, i'm following tutorial and when i start setup.bat file it's different from tutorial and it's stuck here

YouTubeSECourses

Generate Studio Quality Realistic Photos By Kohya LoRA Stable Diffu...

#Kohya SS web GUI DreamBooth #LoRA training full tutorial. You don't need technical knowledge to follow this tutorial. In this tutorial I have explained how to generate professional photo studio quality portrait / self images for free with Stable Diffusion training.

Our Discord server