Fix validation split for subsets

[rockerBOO]

Subsets could not get a separate validation split but it is supported. Might also want to add validation_seed but I have not tested it.

[kohya-ss]

sd-scripts/library/train_util.py

Lines 2007 to 2023 in 884e07d

	# We split the dataset for the subset based on if we are doing a validation split
	# The self.is_training_dataset defines the type of dataset, training or validation
	# if self.is_training_dataset is True -> training dataset
	# if self.is_training_dataset is False -> validation dataset
	if self.validation_split > 0.0:
	# For regularization images we do not want to split this dataset.
	if subset.is_reg is True:
	# Skip any validation dataset for regularization images
	if self.is_training_dataset is False:
	img_paths = []
	sizes = []
	# Otherwise the img_paths remain as original img_paths and no split
	# required for training images dataset of regularization images
	else:
	img_paths, sizes = split_train_val(
	img_paths, sizes, self.is_training_dataset, self.validation_split, self.validation_seed
	)

This part references dataset.validation_split and is_training_dataset, so it looks like it needs some additional changes to work...

[rockerBOO]

Sorry, I had assumed I had this working. But it is not, as you point out. Making it work will be a bit more work. I updated the documentation in #2196 for how to do this with the current system.