No data found. Please verify arguments

[LovElizabeth]

I named folder like '6_xxx' but The CMD continue gave this message to me.Anyone knows how to fix this?

D:\kohya_ss\venv\lib\site-packages\numpy\core\fromnumeric.py:3464: RuntimeWarning: Mean of empty slice. return _methods._mean(a, axis=axis, dtype=dtype, D:\kohya_ss\venv\lib\site-packages\numpy\core_methods.py:192: RuntimeWarning: invalid value encountered in scalar divide ret = ret.dtype.type(ret / rcount) mean ar error (without repeats): nan No data found. Please verify arguments (train_data_dir must be the parent of folders with images) / 画像がありません。引数指定を確認してくださ い（train_data_dirには画像があるフォルダではなく、画像があるフォルダの親フォルダを指定する必要があります）

[ThisIsCyreX]

The path to the images should point to the parent D:/training/images not into the folder itself D:/training/images/6_xxx

[LovElizabeth]

The path to the images should point to the parent D:/training/images not into the folder itself D:/training/images/6_xxx

Thanks for your reply.I notice that mistake and change the route laterly.And now I got another problem.It seems my GTX1080 not work well on these stuff.

[AbyszOne]

The path to the images should point to the parent D:/training/images not into the folder itself D:/training/images/6_xxx

Thas absurdly illogical. I have several sets into the parent folder. But more insane is that if i use the parent folder i get another error: ValueError: invalid literal for int() with base 10: 'CROP' Which is precisely the name of the folder from which it is supposed to pull the images. In fact the base dir already had a folder and I had to delete it to get around that error, only to find no data found. I took the time to install this gui to try alternatives, but the first thing I feel is that it is fucking kidding me. If I put the base dir it throws an error, if I put the parent it throws an error. Oh great, these are the things that put me in a good mood.

[andirsun]

@AbyszOne Did you solve it I have the same error:

File "/home/ubuntu/kohya_ss/venv/lib/python3.10/site-packages/anyio/_backends/_asyncio.py", line 867, in run
    result = context.run(func, *args)
  File "/home/ubuntu/kohya_ss/lora_gui.py", line 385, in train_model
    repeats = int(folder.split('_')[0])
ValueError: invalid literal for int() with base 10: ''

[AbyszOne]

@AbyszOne Did you solve it I have the same error:

File "/home/ubuntu/kohya_ss/venv/lib/python3.10/site-packages/anyio/_backends/_asyncio.py", line 867, in run
    result = context.run(func, *args)
  File "/home/ubuntu/kohya_ss/lora_gui.py", line 385, in train_model
    repeats = int(folder.split('_')[0])
ValueError: invalid literal for int() with base 10: ''

No, i just give up. I want a gui to use, not fight with.

[superboychao]

The kohya training folder structure should be like : root - img, log, model and in the img folder, there will be a folder like "xx_XXXXX", which "xx" is a number represent steps, and XXXX is the name, inside of this folder will be images and dataset txt files.

And you can also use the TOOLS - PREPARATION tab in Dreambooth Lora to help create this specific structure.

[alumcreek]

The path to the images should point to the parent D:/training/images not into the folder itself D:/training/images/6_xxx

Thank you! I had forgotten that this is pointed out in the videos that are linked under the "About" tab when running Kohya_ss. I was also trying to point directly to the folder when it really wants the enclosing folder above it.

For those who are confused by this it is because Kohya allows you to have multiple training folders. The number prefix of a folder tells Kohya how many times to itterate the images. So while you when you are starting out you might have only one instance of images to train and think it makes sense to point directly to that folder, the way Kohya sees it is that you Might have only one, or many, folders at the same level.