Is the trainable parameter saved when saving a tensorflow model?

Question

I have a model that have some of its first layers frozen and the other ones not frozen. Then I save this model using model.save(path). When I'll load it using load_model(path), will the correct layers still be frozen or not ?

Answer 1

Not clear from the documentation but my best guess is the layers are frozen. You could test this by loading the saved model then try:

for layer.model.layers:
    print(layer.name, layer.trainable)