Getting the error "num_samples=0"

[yanndubs]

Getting the error "num_samples=0"

[YannDubs]

What command are you running? It seems that you are trying to plot or train something empty

[QLaHPD]

python main.py test01 -d R:\disentangling-vae-master\data\celeba
I changed the --dataset to the correct path where are 2000 JPG 128x128 test images.

[YannDubs]

Dataset is not a path flag, but a flag with which to chose chose one of the predefined datasets (cf. The cli helper). If you want to use your own dataset you have to define a class for it.
I’m surprised you were able to give a path, it should only let you give one of the predefined datasets.

[QLaHPD]

How I should do this?

[YannDubs]

You probably want to look at basic dataset tutorials for Pytorch. It depends how exactly your data is.

If it is under a simple format, with all images in one directory then you probably want to create a new object for your dataset which closely follows the celeba one.

as it is already downloaded you can use a placeholder for the download method. e.g.

def download(self):
        return