Understanding lstm tutorial

In the LSTM tutorial here : http://pytorch.org/tutorials/beginner/nlp/sequence_models_tutorial.html

Shouldn’t init_hidden be called before predicting at the final trained model:

That would seem logical.