Convert a trained model to be used for feature loss?

I have just finished part 1 of v3 and been working on the last notebooks. I am interested in colouring sketches into colour after being inspired by lessons 7, and DeOldiy. Is it possible to train a VGG model on a large cartoon dataset and use that for the feature loss in the lesson seven? And loading the model like how it was for the feature loss in lesson seven?