Pytorch tutorial are great ! They introduce several concepts in differents tutorial. However, sometimes you might start a new tutorial that won't be using the concepts you've learn before in some other tutorials. It's good thing in order to focus on each tutorial-specific concepts, however sometimes one might feel like :
But here I should use a Dataset ?! Why self updating the weights since I could simply use a optimizer ? etc.
So the idea here is to share some work on some extended pytorch tutorials that get a tutorial and merge it with the other tutorial's concepts you might have previously learn. I am sharing it as I think it is a good exercice while doing a tutorial to try yourself implementing the concepts you've learn before.