A few notebooks on deep learning with PyTorch. Instructions from installing PyTorch on your (non-Windows) platform can be found at http://pytorch.org/.
Part 1: PyTorch basics, you'll build a simple feed forward network to classify handwritten digits (MNIST dataset).
Part 2: You'll learn how to train networks with PyTorch's autograd module. You'll also build a convolutional network which improves performance on image problems.
Autoencoder exercise: A notebook that leads you through building an autoencoder with PyTorch, as well as well as exercise solutions.
Here's a PyTorch implementation of a character-level RNN that can generate text.