PyTorch provides both a low- and medium-level interface to building DL networks/computational graphs. As much as we build DL systems as networks with neurons connected in layers, the actual implementation of a neural network is through a computational graph. Computational graphs reside at the heart of all DL frameworks, and TensorFlow is no exception. However, Keras abstracts away any concept of computational graphs from the user, which makes it easier to learn but does not provide flexibility like PyTorch. Before we begin building computational graphs with PyTorch though, let's first install PyTorch in the next exercise:
- Navigate your browser to pytorch.org, and scroll down to the Run this Command section, as shown in the following screenshot:
Generating a PyTorch installation command
- Select the Stable version and then your specific OS (Linux, Mac...
