Software Engineering Courses (SECourses)•17mo ago

i need help i get this error when trying to run the kohya train for flux : Traceback (most recent c

i need help i get this error when trying to run the kohya train for flux :
Traceback (most recent call last):
File "F:\kohyaflux\Kohya_GUI_Flux_Installer_v1\kohya_ss\sd-scripts\flux_train_network.py", line 411, in <module>
trainer.train(args)
File "F:\kohyaflux\Kohya_GUI_Flux_Installer_v1\kohya_ss\sd-scripts\train_network.py", line 390, in train
self.cache_text_encoder_outputs_if_needed(args, accelerator, unet, vae, text_encoders, train_dataset_group, weight_dtype)
File "F:\kohyaflux\Kohya_GUI_Flux_Installer_v1\kohya_ss\sd-scripts\flux_train_network.py", line 160, in cache_text_encoder_outputs_if_needed
unet.to("cpu")
File "F:\kohyaflux\Kohya_GUI_Flux_Installer_v1\kohya_ss\venv\lib\site-packages\torch\nn\modules\module.py", line 1174, in to
return self._apply(convert)
File "F:\kohyaflux\Kohya_GUI_Flux_Installer_v1\kohya_ss\venv\lib\site-packages\torch\nn\modules\module.py", line 780, in _apply
module._apply(fn)
File "F:\kohyaflux\Kohya_GUI_Flux_Installer_v1\kohya_ss\venv\lib\site-packages\torch\nn\modules\module.py", line 805, in _apply
param_applied = fn(param)
File "F:\kohyaflux\Kohya_GUI_Flux_Installer_v1\kohya_ss\venv\lib\site-packages\torch\nn\modules\module.py", line 1167, in convert
raise NotImplementedError(
NotImplementedError: Cannot copy out of meta tensor; no data! Please use torch.nn.Module.to_empty() instead of torch.nn.Module.to() when moving module from meta to a different device.
Traceback (most recent call last):
File "C:\Users\Surikata\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\Surikata\AppData\Local\Programs\Python\Python310\lib\runpy.py", line 86, in _run_code
exec(code, run_globals)
File "F:\kohyaflux\Kohya_GUI_Flux_Installer_v1\kohya_ss\venv\Scripts\accelerate.EXE__main__.py", line 7, in <module>
File "F:\kohyaflux\Kohya_GUI_Flux_Installer_v1\kohya_ss\venv\lib\site-packages\accelerate\commands\accelerate_cli.py", line 48, in main
args.func(args)
File "F:\kohyaflux\Kohya_GUI_Flux_Installer_v1\kohya_ss\venv\lib\site-packages\accelerate\commands\launch.py", line 1106, in launch_command
simple_launcher(args)
File "F:\kohyaflux\Kohya_GUI_Flux_Installer_v1\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 '['F:\kohyaflux\Kohya_GUI_Flux_Installer_v1\kohya_ss\venv\Scripts\python.exe', 'F:/kohyaflux/Kohya_GUI_Flux_Installer_v1/kohya_ss/sd-scripts/flux_train_network.py', '--config_file', 'F:/AISTUFF/train/model/config_lora-20240822-185636.toml']' returned non-zero exit status 1.
18:56:49-324344 INFO Training has ended.

HorridOP•8/22/24, 4:00 PM

i tried using se's config too and it also didnt work im unsure what the issue is

HorridOP•8/22/24, 4:02 PM

im downloading the fp16 dev version rn to see if it solv es iy

HHorrid i need help i get this error when trying to run the kohya train for flux : Trac...

Furkan Gözükara SECourses•8/22/24, 4:15 PM

please use dev flux fp16 as base model

Furkan Gözükara SECourses•8/22/24, 4:15 PM

the link in patreon post

Furkan Gözükara SECourses•8/22/24, 4:16 PM

also did you run Fix_For_FLUX_Step_2.bat

HorridOP•8/22/24, 4:23 PM

yes i did

HHorrid yes i did

Furkan Gözükara SECourses•8/22/24, 4:28 PM

ok fixed?

HorridOP•8/22/24, 4:42 PM

I'll have to test in a bit

mikemenders•8/22/24, 5:15 PM

I see everyone is fluxing. However, I managed to find the golden point for training ponyRealism 2.1 based on the classic SD 1.5 Lora training. So same way with reg images, Prodigy optimizer. What is interesting is that the previous Prodigy parameters hardly needed to be changed, only the d_coef value had to be lowered to 0.4 for 100 images. For about 150 images, it drops to 0.25.
What's important is that I don't use a bucket, I just work with the best 100 images, 1024x1024 resolution, and the classic 'xxxxx woman' folder name. (Where xxxxx is the name of the lora. In Furkan's case ohwx.) Another important thing is that I need 24 GB VRAM here, almost fully exploiting the training. For 100 images it's about 45-50 minutes, for 150 images it's about 1 hour 20 minutes average training.
I just trained 5 Lora, and for all of them the max_norm/keys_scaled value on Tensorboard rarely went above 2, so they became perfect person Lora. I tried it before with 16/16 dim/alpha, but 32/32 turned out to be the right value.
What's interesting is that the image generation got better with euler_a, and bemypony_Photo3 model. However, the Lora performed quite well on the other Pony models, so I'm satisfied. Here's my Kohya json file, feel free to use it!

PonyXL1-realistic-person-sfw.json4.58KB

HHorrid i need help i get this error when trying to run the kohya train for flux : Trac...