Does model.train() put every thing in train mode in pytorch even sub-networks?

Question

I wrote a neural network in PyTorch, which uses ResNet as feature generator, and it is fined tuned with the whole network.

My model consists of Resnet and several lay