Software Engineering Courses (SECourses)•10mo ago

E:\Koyha\kohya_ss\venv\lib\site-packages\torch\autograd\graph.py:825: UserWarning: cuDNN SDPA backwa

E:\Koyha\kohya_ss\venv\lib\site-packages\torch\autograd\graph.py:825: UserWarning: cuDNN SDPA backward got grad_output.strides() != output.strides(), attempting to materialize a grad_output with matching strides... (Triggered internally at C:\actions-runner_work\pytorch\pytorch\builder\windows\pytorch\aten\src\ATen\native\cudnn\MHA.cpp:676.)
return Variable._execution_engine.run_backward( # Calls into the C++ engine to run the backward pass

MetaKan•3/12/25, 8:16 AM

Hello, I have my model trained with a character, but I'm having trouble getting that face to appear in different styles. I've tried using only prompts and also various Flux Redux workflows, but I can't achieve the look I need.

Imagine transforming a human face into something like this. It doesn't need to look exactly the same, but it should retain some features and have the applied materials.

What would be your approach? Thanks!

xlads•3/12/25, 12:18 PM

Hello @Dr. Furkan Gözükara how do i stop or pause a pod in runpod?

xlads•3/12/25, 12:32 PM

is it the trash button? I REMEBER THERE USED to be a stop button

DDarth Teethius E:\Koyha\kohya_ss\venv\lib\site-packages\torch\autograd\graph.py:825: UserWarnin...

Furkan Gözükara SECourses•3/12/25, 12:35 PM

i think not an error

MMetaKan Hello, I have my model trained with a character, but I'm having trouble getting ...

Furkan Gözükara SECourses•3/12/25, 12:36 PM

yes flux really prevents stylization

Xxlads is it the trash button? I REMEBER THERE USED to be a stop button

Furkan Gözükara SECourses•3/12/25, 12:36 PM

for permanent storage yes

Furkan Gözükara SECourses•3/12/25, 12:36 PM

it will jut delete pod instance but your permanent storage will remain

Furkan Gözükara SECourses•3/12/25, 12:36 PM

i am trying to fix extra RAM usage of LoRA loading

Tria•3/12/25, 1:03 PM

@Dr. Furkan Gözükara

Tria•3/12/25, 1:04 PM

Tria•3/12/25, 1:04 PM

what am i doing wrong ?

TTria @Dr. Furkan Gözükara

Furkan Gözükara SECourses•3/12/25, 1:19 PM

you can copy paste debug logs

Furkan Gözükara SECourses•3/12/25, 1:20 PM

i fixed the lora loading ram problem with wan 2.1

Tria•3/12/25, 1:20 PM

where do find the logs ?

xlads•3/12/25, 1:23 PM

okay i terminated pod but i do have a storge hopefully it didnt delted my files or trained models

Xxlads okay i terminated pod but i do have a storge hopefully it didnt delted my files ...

Renzo•3/12/25, 1:25 PM

Did you save the files in the network storage?

Renzo•3/12/25, 1:26 PM

IIRC, the network storage is mounted at a specific location.

TTria where do find the logs ?

Furkan Gözükara SECourses•3/12/25, 1:32 PM

shown in every tutorial

Furkan Gözükara SECourses•3/12/25, 1:32 PM

i recommend watch this one

Furkan Gözükara SECourses•3/12/25, 1:32 PM

https://youtu.be/-zOKhoO9a5s

YouTubeSECourses

Stable Diffusion 3.5 Large How To Use Tutorial With Best Configurat...

Stability AI published their most power newest model Stable Diffusion 3.5 Large. This model unlike FLUX is full model not distilled and has huge potential. I have done extensive research and publishing all of it in this video regarding how to use SD 3.5 Large with the best settings. Moreover, I am sharing how to use FLUX DEV with the best possib...

Xxlads okay i terminated pod but i do have a storge hopefully it didnt delted my files ...

Furkan Gözükara SECourses•3/12/25, 1:32 PM

it wont

Furkan Gözükara SECourses•3/12/25, 1:32 PM

it was a permanent storage

vp•3/12/25, 1:43 PM

doc i have a question - trying to resume a kohya ss finetune. but when i load the last training state it still goes back to step 1 as if it is retraining from scratch

Vvp doc i have a question - trying to resume a kohya ss finetune. but when i load t...

Furkan Gözükara SECourses•3/12/25, 1:44 PM

it will show as 1

Furkan Gözükara SECourses•3/12/25, 1:44 PM

but will continue from last

Furkan Gözükara SECourses•3/12/25, 1:44 PM

so reduce number of epochs

vp•3/12/25, 1:46 PM

ah ok. thank you

MMetaKan Hello, I have my model trained with a character, but I'm having trouble getting ...

vp•3/12/25, 1:55 PM

i had a windows update kill my session overnight around 3am so if you were running this on windows it's likely the operating system started doing funky stuff and threw this off.

FFurkan Gözükara SECourses it was a permanent storage

xlads•3/12/25, 1:55 PM

which video do you reccomend i watch so that i can run swarm_ui in runpod?

Xxlads which video do you reccomend i watch so that i can run swarm_ui in runpod?

vp•3/12/25, 1:59 PM

go to his page on youtube and watch the full swarmui playlist. you have to watch the windows stuff to learn all the tidbits.

Xxlads which video do you reccomend i watch so that i can run swarm_ui in runpod?

Furkan Gözükara SECourses•3/12/25, 2:13 PM

https://youtu.be/XFUZof6Skkw

YouTubeSECourses

How to Use SwarmUI & Stable Diffusion 3 on Cloud Services Kaggle (f...

In this video, I demonstrate how to install and use #SwarmUI on cloud services. If you lack a powerful GPU or wish to harness more GPU power, this video is essential. You'll learn how to install and utilize SwarmUI, one of the most powerful Generative AI interfaces, on Massed Compute, RunPod, and Kaggle (which offers free dual T4 GPU access for ...

JaviEM•3/12/25, 2:16 PM

I'm trying wan 2.1 with FP8 and it doesn't work. I have tried everything and it still doesn't work. No torch version works, neither stable nor nightly. I even tried to install an older version of the autoinstaller with a same day commit with the same problem. I think the latest nvidia drivers have broken something. FP16 works perfectly, but it takes 1,5h to generate a video from an image.