Software Engineering Courses (SECourses)•17mo ago

Hi, Dr Furkan, I am interested in the Flux controlnet, have you also conducted experiments on it too

Hi, Dr Furkan, I am interested in the Flux controlnet, have you also conducted experiments on it too? Is there any chance for us to train or finetune the controlnet within 24GB VRAM like the LoRAs do?

PPickUrHero Hi, Dr Furkan, I am interested in the Flux controlnet, have you also conducted e...

Furkan Gözükara SECourses•8/30/24, 3:02 PM

nope i didnt. actually i never trained a controlnet. it is not an easy task

uisato•8/30/24, 3:19 PM

Hey guys, trying to install the whole thing on my 2nd PC. Getting an error in step 1 (installing Kohya and requirements):

"Traceback (most recent call last):
File "C:\Users\Usuario\AppData\Local\Programs\Python\Python310\lib\runpy.py", line 196, in _run_module_as_main
return _run_code(code, main_globals, None,
File "C:\Users\Usuario\AppData\Local\Programs\Python\Python310\lib\runpy.py", line 86, in _run_code
exec(code, run_globals)
File "C:\Program Files\0 - KOHYA\KohyaTraining\kohya_ss\venv\Scripts\accelerate.exe__main.py", line 4, in <module>
File "C:\Program Files\0 - KOHYA\KohyaTraining\kohya_ss\venv\lib\site-packages\accelerate__init.py", line 16, in <module>
from .accelerator import Accelerator
File "C:\Program Files\0 - KOHYA\KohyaTraining\kohya_ss\venv\lib\site-packages\accelerate\accelerator.py", line 32, in <module>
import torch
File "C:\Program Files\0 - KOHYA\KohyaTraining\kohya_ss\venv\lib\site-packages\torch__init__.py", line 148, in <module>
raise err
OSError: [WinError 126] No se puede encontrar el módulo especificado. Error loading "C:\Program Files\0 - KOHYA\KohyaTraining\kohya_ss\venv\lib\site-packages\torch\lib\fbgemm.dll" or one of its dependencies.

Kohya_ss setup menu:..."

What could be going wrong?

Tacite•8/30/24, 3:26 PM

if you go to C:\Program Files\0 - KOHYA\KohyaTraining\kohya_ss\venv\lib\site-packages\torch\lib\ do you see fbgemm.dll ?

uisato•8/30/24, 3:29 PM

Yes, its there.

Uuisato Hey guys, trying to install the whole thing on my 2nd PC. Getting an error in st...

Furkan Gözükara SECourses•8/30/24, 3:32 PM

i replied you solution

Furkan Gözükara SECourses•8/30/24, 3:32 PM

this is torch 2.4 error on win 11

FFurkan Gözükara SECourses i replied you solution

uisato•8/30/24, 3:33 PM

What solution? Haven't seen it.

Furkan Gözükara SECourses•8/30/24, 3:33 PM

open this post

Furkan Gözükara SECourses•8/30/24, 3:33 PM

https://www.patreon.com/posts/109913645

Patreon

BiRefNet State Of The Art Newest Very Best Background Batch Remover...

Get more from SECourses: Tutorials, Guides, Resources, Training, FLUX, MidJourney, Voice Clone, TTS, ChatGPT, GPT, LLM, Scripts by Furkan Gözü on Patreon

Dazzastrous [4090]•8/30/24, 3:33 PM

figures win 11 is crap

Furkan Gözükara SECourses•8/30/24, 3:33 PM

read Windows Requirements

Furkan Gözükara SECourses•8/30/24, 3:33 PM

i think they gonna fix it with newer torch

uisato•8/30/24, 3:33 PM

I'm on windows 10.

Uuisato I'm on windows 10.

Furkan Gözükara SECourses•8/30/24, 3:34 PM

wow weird

Furkan Gözükara SECourses•8/30/24, 3:34 PM

still solution should work

Furkan Gözükara SECourses•8/30/24, 3:34 PM

i bet it is related to your folder path : "0 - KOHYA"

Tacite•8/30/24, 3:34 PM

i was gonna say that, spaces in folder name

uisato•8/30/24, 3:34 PM

Okay, lets try changing names.

Furkan Gözükara SECourses•8/30/24, 3:35 PM

massed compute trains 3000 steps in 1 hour

1.25$

Furkan Gözükara SECourses•8/30/24, 3:35 PM

recording video

Uuisato Okay, lets try changing names.

Furkan Gözükara SECourses•8/30/24, 3:35 PM

you need to reinstall

Furkan Gözükara SECourses•8/30/24, 3:35 PM

install into

Furkan Gözükara SECourses•8/30/24, 3:35 PM

c:/kohya_ai

uisato•8/30/24, 3:41 PM

Great! Changed names and installed said the DLL and it works. Thanks guys.

Dazzastrous [4090]•8/30/24, 3:42 PM

i have win 11 and didnt have any probs with installing it

Uuisato Great! Changed names and installed said the DLL and it works. Thanks guys.

Furkan Gözükara SECourses•8/30/24, 3:42 PM

great. i always say that dont use space char in paths

Dazzastrous [4090]•8/30/24, 3:43 PM

yeh no voids allowed

Dazzastrous [4090]•8/30/24, 3:44 PM

just facefusion

Dazzastrous [4090]•8/30/24, 3:44 PM

lol

Furkan Gözükara SECourses•8/30/24, 3:44 PM

dsienra•8/30/24, 3:55 PM

@Dr. Furkan Gözükara 20 epochs 0 resemblance to the training characters (multiple people) using regularization class person (people of all ages and genders) TE and unet lr 1e-4, Epoch 45 I'm getting some resemblance, so I increased now lr of the TE to 9e-4, lets see

Ddsienra @Dr. Furkan Gözükara 20 epochs 0 resemblance to the training characters (multip...

Furkan Gözükara SECourses•8/30/24, 3:58 PM

you need to train like 200 to see more resemblance . did you look my grids on patreon for reg images?

Furkan Gözükara SECourses•8/30/24, 3:58 PM

recording runpod and massed compute tutorial atm

FFurkan Gözükara SECourses recording runpod and massed compute tutorial atm 😄

dsienra•8/30/24, 4:02 PM

Yes I did, mixed faces, I'm playing with the TE lr to see its effects, I want to know if something fix or minimize the bleeding, also I'm not using class on my characters

dsienra•8/30/24, 4:02 PM

I'm just experimenting

dsienra•8/30/24, 4:04 PM

TE lr 9e-4 is quite aggressive but may be is what it needs idk

Furkan Gözükara SECourses•8/30/24, 4:35 PM

nice let me know

FFurkan Gözükara SECourses nice let me know

uisato•8/30/24, 4:57 PM

Do you happen to know why there's little to no improvement when using Torch 2.5? Said it a couple of messages ago.

uisato•8/30/24, 4:58 PM

Double-checked, I'm using Torch 2.5. [Turned off the validating requirements part in gui.bat]

daroth•8/30/24, 5:00 PM

wonder if this is a LORA trained on a single pic:
https://civitai.com/models/702897/crazy-face-peter-parker-meme-face

nicearoni•8/30/24, 5:51 PM

long ass post was so long, converted it to image haha

1241-photo_of_a_marching_band_practicing_outs-Fluxflux1-dev-fp8-1547391564.png

1247-photo_of_a_marching_band_practicing_outs-Fluxflux1-dev-fp8-1547391564.png

1252-a_group_of_students_wearing_red_blue_y-Fluxflux1-dev-fp8-1565880735.png

1251-a_group_of_students_wearing_red_blue_y-Fluxflux1-dev-fp8-1565880735.png

1304-a_marching_band_rehearsal_is_taking_plac-Fluxflux1-dev-fp8-1565880735.png

1254-a_marching_band_rehearsal_is_taking_plac-Fluxflux1-dev-fp8-1565880735.png

Nnicearoni long ass post was so long, converted it to image haha

Dreamer@TDFilmStudio•8/30/24, 6:00 PM

since you trained on 512X512 dataset, will your outputs be affected if you go for higher resolution image generation like full HD?

Uuisato Double-checked, I'm using Torch 2.5. [Turned off the validating requirements par...

uisato•8/30/24, 6:01 PM

Also, does anybody have a Comfy workflow for FLUX + LORAs optimized for RTX 3090? I'm getting 15.79s/it, which strikes me as there's something wrong with my workflow/configuration.

uisato•8/30/24, 6:01 PM

Using this one.

MarkuryFLUX_LoRA_v2_-_FLUX__LORA.json17.59KB

DDreamer@TDFilmStudio since you trained on 512X512 dataset, will your outputs be affected if you go fo...

nicearoni•8/30/24, 6:02 PM

all of these outputs were 1024x1024 so I don't think so? I'll try some higher resolutions now

Nnicearoni all of these outputs were 1024x1024 so I don't think so? I'll try some higher re...

Dreamer@TDFilmStudio•8/30/24, 6:05 PM

Thanks for sharing your findings. I finished with my first character Flux Lora training today. I want to try out product lora next. Do you think it would be different from your style lora experiment? Why did you remove the word style from the target trained folder name? Does that impact the final quality?

DDreamer@TDFilmStudio Thanks for sharing your findings. I finished with my first character Flux Lora t...

nicearoni•8/30/24, 6:10 PM

I removed it from the Class prompt textbox in the kohya GUI as well after I clicked 'prepare training data' and 'copy info to respective fields' and renamed the folder (not sure that textbox is used outside of preparing\copying, just wanted to be sure). I removed it because of the convos around here about training on just 'ohwx' instead of 'ohwx man'. In comparing the 25 epoch from both trainings I did, I preferred the second one that only had the instance prompt and no captions.

Nnicearoni I removed it from the Class prompt textbox in the kohya GUI as well after I clic...

Dreamer@TDFilmStudio•8/30/24, 6:11 PM

Thats really interesting. Also reduces the prep steps. Captioning is tedious job

Uuisato Also, does anybody have a Comfy workflow for FLUX + LORAs optimized for RTX 3090...

nicearoni•8/30/24, 6:13 PM

yeah that def seems high... here's what I'm getting in SwarmUI (which of course uses Comfy as its backend) for 30 steps per 1024x1024 image in Swarm with a 2GB LoRA I trained myself...

DDreamer@TDFilmStudio Thats really interesting. Also reduces the prep steps. Captioning is tedious job

nicearoni•8/30/24, 6:15 PM

definitely agree on captioning being tedious! I decided to try without after hearing some people say it didn't help or even hurt character likeness, and that flux apparently has its own internal captioning somehow when training?

Hi, Dr Furkan, I am interested in the Flux controlnet, have you also conducted experiments on it too

Similar Threads