Using validation data separated from train data in Skorch EarlyStopping

Dear reader,

In a project implying a Sklearn pipeline, I’m trying to add my Pytorch model, therefore I have been using the usefull wrapper Skorch to make the link.

It’s working well, the problem is that I want to use a callback to perform EarlyStopping:

" callbacks = [EarlyStopping(monitor=‘valid_loss’, patience=self.classifier.patience, threshold=0.0001, threshold_mode=‘rel’, lower_is_better=True, load_best=True)],"

and doing this, I found in skorch documentation, that it was using a split of the train data, where as I already have a split train/valid made before the train loop and I would like to use this specific valid data as the validation data used in the EarlyStopping Callback, as well as the one displayed in :
" valid_loss=net.history[:, ‘valid_loss’] "

Could anyone who has any idea how to proceed with my already splitted data in this specific case, help me ?

Thanks in advance !!