Software Engineering Courses (SECourses)•16mo ago

Shit! I something have happened with my computer. Generating the same image with the same settings o

Shit! I something have happened with my computer. Generating the same image with the same settings on my 4090 at work takes 26 sec. On my 3090 at home it never finishes because the VRAM-usage spikes

KallamamranOP•9/13/24, 1:54 PM

At work it uses 13GB VRAM and about 23GB RAM. At home it uses all 24GB VRAM and then starts with the shared memory

KallamamranOP•9/13/24, 1:54 PM

I have no clue why

KallamamranOP•9/13/24, 1:54 PM

I have even installed a new Forge with the latest Torch to try, but it's the same thing

�𝗡𝗶𝘁𝗶𝗻 𝗣𝗮𝗿𝗲𝘃𝗮 @Dr. Furkan Gözükara I mean your khoya installer and swarm ui installer for Runp...

Furkan Gözükara SECourses•9/13/24, 1:55 PM

they install latest version

�

𝗡𝗶𝘁𝗶𝗻 𝗣𝗮𝗿𝗲𝘃𝗮•9/13/24, 1:55 PM

Great

KKallamamran Shit! I something have happened with my computer. Generating the same image with...

Furkan Gözükara SECourses•9/13/24, 1:55 PM

no idea

KallamamranOP•9/13/24, 1:55 PM

Yeah. Of course

KallamamranOP•9/13/24, 1:55 PM

Me neither

KallamamranOP•9/13/24, 1:56 PM

It's so wierd

�

𝗡𝗶𝘁𝗶𝗻 𝗣𝗮𝗿𝗲𝘃𝗮•9/13/24, 1:56 PM

did you try restarting your pc?

KallamamranOP•9/13/24, 1:56 PM

Yes

�

𝗡𝗶𝘁𝗶𝗻 𝗣𝗮𝗿𝗲𝘃𝗮•9/13/24, 1:57 PM

maybe go to the task manager and see what's eating up the memory

FFurkan Gözükara SECourses 10 second / it not bad

sergio_gonzo•9/13/24, 2:03 PM

you are right DrFurkan for the quality that brings

�

𝗡𝗶𝘁𝗶𝗻 𝗣𝗮𝗿𝗲𝘃𝗮•9/13/24, 2:12 PM

@Dr. Furkan Gözükara training t5 significantly increases vram usage as comapred to just the unet? Using a L40S on both the pods and getting 4.3 sec/it for the unet training and 4.75 sec/it on the t5

Zono50•9/13/24, 2:32 PM

i wish i could supir images without having to use comfyui, i have a linux program called upscalayr, and it has some esrgan models, but to use supir, i would have to conver it t onnx, then NCNN format, and it probably won't even work anyways

Zono50•9/13/24, 2:33 PM

@Dr. Furkan Gözükara I'd love just a simple linux or windows program you open, put your images in, and it improves using SUPIR, and also have a built in image comparison you can push back and forth

�𝗡𝗶𝘁𝗶𝗻 𝗣𝗮𝗿𝗲𝘃𝗮 @Dr. Furkan Gözükara training t5 significantly increases vram usage as comapred ...

Furkan Gözükara SECourses•9/13/24, 3:22 PM

yes increases

Furkan Gözükara SECourses•9/13/24, 3:22 PM

but speed no much impacted

ZZono50 @Dr. Furkan Gözükara I'd love just a simple linux or windows program you open, ...

Furkan Gözükara SECourses•9/13/24, 3:22 PM

we have it lol

Furkan Gözükara SECourses•9/13/24, 3:22 PM

i use batch mode

Furkan Gözükara SECourses•9/13/24, 3:23 PM

https://www.patreon.com/posts/99176057

dsienra•9/13/24, 3:35 PM

I'm still traying to train multiple subjects in flux with no success it still beeds even with t5 and also training t5 have unespected results

dsienra•9/13/24, 3:36 PM

for example the keyword associated to one of my subjects triggers this image

dsienra•9/13/24, 3:37 PM

or this

Ddsienra I'm still traying to train multiple subjects in flux with no success it still be...

Furkan Gözükara SECourses•9/13/24, 3:37 PM

ye bleeding is hard

dsienra•9/13/24, 3:37 PM

that's not even a human

Ddsienra that's not even a human

Furkan Gözükara SECourses•9/13/24, 3:37 PM

this is actually unexpected

dsienra•9/13/24, 3:38 PM

completely random

dsienra•9/13/24, 3:40 PM

hey wait is not random, is the background of one of my images.... wow

dsienra•9/13/24, 3:41 PM

in that photo it was me

dsienra•9/13/24, 3:41 PM

that picture is from my city

dsienra•9/13/24, 3:41 PM

or a very close represntation of it

dsienra•9/13/24, 3:43 PM

this only happend with T5

dsienra•9/13/24, 3:47 PM

maybe is because I increased the learning rate to very high level and I'm not using class so learned the background and not me

dsienra•9/13/24, 3:47 PM

may be using T5 i need to add the class again

Ddsienra maybe is because I increased the learning rate to very high level and I'm not us...

Furkan Gözükara SECourses•9/13/24, 4:00 PM

yep overfit

FFurkan Gözükara SECourses good question. it is a same person so may not apply

wardensc2•9/13/24, 4:05 PM

Actually I send an image here for reference but somehow discord remove it. It's the same person about same age but different make up, hair, outfit, the pose also different, some images will be lying pose 768x512, some images are portrait 1024x1024, some images are standing pose 512x768. Do you think it is insconsistent data

wardensc2•9/13/24, 4:06 PM

Maybe I will run mass compute by myself to test with caption and give you 10% fee to make it clearer right

Wwardensc2 Maybe I will run mass compute by myself to test with caption and give you 10% fe...

Furkan Gözükara SECourses•9/13/24, 4:39 PM

yep good idea

xlads•9/13/24, 4:44 PM

I’m going to take a 1-minute clip from a movie and use Joy Caption to generate captions for each frame. Then, I’ll edit those captions into prompts that fit my AI model to hopefully optimize my workflow.

Xxlads I’m going to take a 1-minute clip from a movie and use Joy Caption to generate c...

Furkan Gözükara SECourses•9/13/24, 4:48 PM

sounds like a plan

FFurkan Gözükara SECourses yep overfit

dsienra•9/13/24, 5:48 PM

yes but maybe the key is the class on this case, it learned and associated the token to the background and not me, may be training T5 I must specify the class man, so it learns me rather than the background, is like it needs guidance to what to learn, just a theory.. hehe

slipkn67•9/13/24, 5:54 PM

in the <segment:face,0.7> part what happens if I don't reduce it to 0.7? Won't it bring better result?

xlads•9/13/24, 5:57 PM

so my caption joy just complete batch processing progress and says that the output folder is same as the input folder but when i check i can see anything

xlads•9/13/24, 6:01 PM

statred processin progress again this time with oput folder pasted in ui

Vinmi•9/13/24, 6:09 PM

Anyone has any idea why this workflow is not working? It gives me an error, but it was fully working with the original Flux-Dev(8fp)

Vinmi•9/13/24, 6:09 PM

Original message was deleted

Narva•9/13/24, 6:26 PM

theres 17 gpu aliviable

Maxivi•9/13/24, 7:17 PM

@Dr. Furkan Gözükara is the lasest kohya broken for you?

DDreamer@TDFilmStudio I have trained 3 custom Flux loras till now and everytime, I am getting scan lin...

Luciani•9/13/24, 8:15 PM

I have exactly the same issue with vertical lines appearing on the results. One point is to restrict generating to 1024 and do not go above. However, often the lines are still visible on that smaller images. Did you find a solution yet? You may also send me a direct message!

Shit! I something have happened with my computer. Generating the same image with the same settings o

Similar Threads