Hands-On Reinforcement Learning with Python

Visualization in TensorBoard

Let's visualize our network in TensorBoard. To launch TensorBoard, open your Terminal and type the following:

tensorboard --logdir=logs --port=6007 --host=127.0.0.1

This is our A3C network. We have one global network and four workers:

Let's expand our global network; you can see we have one actor and one critic:

Okay, what is really going on in workers? Let's expand our worker network. You can see how the worker nodes are performing:

What about the sync node? What is that doing? The sync node pushes the local ...