In this chapter, we covered the major areas of TensorBoard--EVENTS, HISTOGRAMS, and viewing GRAPH. We modified popular models to see the exact changes required before TensorBoard could be up and running. This should have demonstrated the fairly minimal effort required to get started with TensorBoard.

Finally, we focused on various popular models by viewing their network design. We did this by instrumenting the code with TensorBoard hooks and using the TensorBoard Graph Explorer to deep dive into the network setups.

The reader should now be able to use TensorBoard more effectively, gauge training performance, and plan runs and modify training scripts.

Next, we're going to jump into convolutional networks. We'll use parts of our prior work so we can hit the ground running. But, we'll focus on more advanced neural network setups to achieve better accuracy. The focus on training accuracy reflects the focus of most practitioner's efforts, so it is the time we face the challenge.