Software Engineering Courses (SECourses)•10mo ago

nice, btw your wan 2.1 1.3B gives me better results than kijai nodes on comfy

TThe90sPope nice, btw your wan 2.1 1.3B gives me better results than kijai nodes on comfy

Furkan Gözükara SECourses•3/6/25, 8:13 PM

nice. yes kijai does so much optimizations and trade of is quality

Furkan Gözükara SECourses•3/6/25, 8:13 PM

i am waiting official teacache because of this reason also

The90sPopeOP•3/6/25, 8:14 PM

I just but flow to 8 instead of 6 since devs say 8 to 12 is best on 1.3b

TThe90sPope I just but flow to 8 instead of 6 since devs say 8 to 12 is best on 1.3b

Furkan Gözükara SECourses•3/6/25, 8:14 PM

i can change default

Furkan Gözükara SECourses•3/6/25, 8:14 PM

which one is best you think?

Furkan Gözükara SECourses•3/6/25, 8:14 PM

The90sPopeOP•3/6/25, 8:14 PM

I use 8 all the time, but it really depends, lottery

TThe90sPope I use 8 all the time, but it really depends, lottery

Furkan Gözükara SECourses•3/6/25, 8:14 PM

i see thanks

Furkan Gözükara SECourses•3/6/25, 8:14 PM

we added configs so you can always save and load

The90sPopeOP•3/6/25, 8:15 PM

yeah, i love that, i already made my preset, also have a really nice negative prompt which helps

The90sPopeOP•3/6/25, 8:15 PM

Overexposure, static, blurred details, subtitles, paintings, pictures, still, overall gray, worst quality, low quality, JPEG compression residue, ugly, mutilated, redundant fingers, poorly painted hands, poorly painted faces, deformed, disfigured, deformed limbs, fused fingers, cluttered background, three legs, a lot of people in the background, upside down, inaccurate, distorted, motion smear, motion blur, motion stutter, video noise, video artifacts, warped, motionless

TThe90sPope Overexposure, static, blurred details, subtitles, paintings, pictures, still, ov...

Furkan Gözükara SECourses•3/6/25, 8:19 PM

nice

The90sPopeOP•3/6/25, 8:19 PM

10 works!

The90sPopeOP•3/6/25, 8:19 PM

great

xlads•3/6/25, 8:48 PM

started training and got this error "Traceback (most recent call last):
File "C:\Python310\lib\runpy.py", line 196, in _run_module_as_main
return _run_code(code, main_globals, None,
File "C:\Python310\lib\runpy.py", line 86, in _run_code
exec(code, run_globals)
File "C:\Kohya_FLUX_DreamBooth_v17\kohya_ss\venv\Scripts\accelerate.EXE__main__.py", line 7, in <module>
sys.exit(main())
File "C:\Kohya_FLUX_DreamBooth_v17\kohya_ss\venv\lib\site-packages\accelerate\commands\accelerate_cli.py", line 48, in main
args.func(args)
File "C:\Kohya_FLUX_DreamBooth_v17\kohya_ss\venv\lib\site-packages\accelerate\commands\launch.py", line 1106, in launch_command
simple_launcher(args)
File "C:\Kohya_FLUX_DreamBooth_v17\kohya_ss\venv\lib\site-packages\accelerate\commands\launch.py", line 704, in simple_launcher
raise subprocess.CalledProcessError(returncode=process.returncode, cmd=cmd)
subprocess.CalledProcessError: Command '['C:\Kohya_FLUX_DreamBooth_v17\kohya_ss\venv\Scripts\python.exe', 'C:/Kohya_FLUX_DreamBooth_v17/kohya_ss/sd-scripts/flux_train.py', '--config_file', 'C:\swarm_ui\SwarmUI\Models\diffusion_models\model/config_dreambooth-20250306-124033.toml']' returned non-zero exit status 1."

xlads•3/6/25, 8:53 PM

getting "CUDA out-of-memory error"

TThe90sPope 10 works!

Furkan Gözükara SECourses•3/6/25, 8:58 PM

awesome

Xxlads started training and got this error "Traceback (most recent call last): File "...

Furkan Gözükara SECourses•3/6/25, 8:58 PM

tell me gpu and config

xlads•3/6/25, 9:01 PM

6GB_GPU_5400MB_14.1_second_it_Tier_3_512px.json

FFurkan Gözükara SECourses tell me gpu and config

xlads•3/6/25, 9:02 PM

Xxlads 6GB_GPU_5400MB_14.1_second_it_Tier_3_512px.json

Furkan Gözükara SECourses•3/6/25, 9:22 PM

in that case you loaded it into inaccurate tab

Xxlads 6GB_GPU_5400MB_14.1_second_it_Tier_3_512px.json

Furkan Gözükara SECourses•3/6/25, 9:23 PM

it is dreambooth config so get a fresh config and load into dreambooth tab

Furkan Gözükara SECourses•3/6/25, 9:23 PM

ah your problem is ram memory

Furkan Gözükara SECourses•3/6/25, 9:23 PM

you need 64 gb . but still give it a try

FFurkan Gözükara SECourses you can use as smartphone photo

julius•3/6/25, 9:37 PM

you mean the class prompt would be "smartphone photo" (2 words)?

Jjulius you mean the class prompt would be "smartphone photo" (2 words)?

Furkan Gözükara SECourses•3/6/25, 9:44 PM

yep

FFurkan Gözükara SECourses it is dreambooth config so get a fresh config and load into dreambooth tab

xlads•3/6/25, 9:44 PM

yes it is fresh config... I gave it another try and it failed again with same error, inside of the dreambooth tab

FFurkan Gözükara SECourses yep

julius•3/6/25, 10:00 PM

Do you know if, in the dataset, it's best to include people or not?

Xxlads yes it is fresh config... I gave it another try and it failed again with same er...

Furkan Gözükara SECourses•3/6/25, 10:08 PM

ye 64 gb ram

can you upgrade?

Jjulius Do you know if, in the dataset, it's best to include people or not?

Furkan Gözükara SECourses•3/6/25, 10:09 PM

yes you can include but not repeat same person

FFurkan Gözükara SECourses ye 64 gb ram 😄 can you upgrade?

xlads•3/6/25, 10:14 PM

Not at the moment. Will have to stick with this tutorial for the moment https://youtu.be/nySGu12Y05k?si=hxVTHXjehERPPGY7

YouTubeSECourses

FLUX LoRA Training Simplified: From Zero to Hero with Kohya SS GUI ...

Ultimate Kohya GUI FLUX LoRA training tutorial. This tutorial is product of non-stop 9 days research and training. I have trained over 73 FLUX LoRA models and analyzed all to prepare this tutorial video. The research still going on and hopefully the results will be significantly improved and latest configs and findings will be shared. Please wat...