Using TensorBoard

At this point in this book, we need to move beyond building toy examples and look to building modules or frameworks you can use to train your own agents in the future. In fact, we will use the code in this chapter for training agents to solve other challenge environments we present in later chapters. That means we need a more general way to capture our progress, preferably to log files that we can view later. Since building such frameworks is such a common task to machine learning as a whole, Google developed a very useful logging framework called TensorBoard. TensorBoard was originally developed as a subset of the other DL framework we mentioned earlier, TensorFlow. Fortunately, for us, PyTorch includes an extension that supports logging to TensorBoard. So, in this section, we are going to set up and install TensorBoard for use as a logging and graphing platform...