No .eval() in tutorials

[Kaixhin]

See pytorch/vision#405 for a doc update and link to a forum thread. It seems that actually people following any of the basic tutorials will miss .eval(), which admittedly is not as easy to come across as it should be. We should update some of the basic tutorials to add this note.

[Kaixhin]

This keeps on getting missed.

[kwasnydam]

Actually I was just about opening an issue on this. In the TRANSLATION WITH A SEQUENCE TO SEQUENCE NETWORK AND ATTENTION, (https://pytorch.org/tutorials/intermediate/seq2seq_translation_tutorial.html),
under the Evaluation section the author is missing the .eval() call on the models when performing the evaluation, and he uses a dropout in his model. Is it possible to push a fix for that? Never have contributed on github yet, so do not really know the procedures.

[Kaixhin]

@kwasnydam helping fix an error you've found is very helpful, so thanks. Rough steps are as follows, :

1. Fork this repository
2. Add the missing calls in the file
3. In your forked repository, start a pull request (button should be near the clone button once you've added a commit)
4. Fill in the details and submit