Launch TensorFlow TensorBoard

Use the TensorBoard command line utility to launch the TensorFlow TensorBoard web service

Use the TensorBoard command line utility to launch the TensorFlow TensorBoard web service

Video Transcript


In this video, we’re going to use the TensorBoard command line utility to launch the TensorFlow TensorBoard.


To launch the TensorBoard, we have to make sure we have a log directory where the ProtoBuf event file was written.

So let’s check the graph’s directory.

ls graphs/

So we do ls graphs and we see that we have the ProtoBuf that our TensorFlow’s summary FileWriter wrote out.


Next, in the command line we use the TensorBoard command line utility.

tensorboard --logdir="./graphs"

We say our logdir is the graphs which we just saw has our file.


We can see that we’re using TensorBoard 1.8.0.

We can see that it uses sebg which is the name that I’m using here.

So it’s saying sebg.home:6006.


Because I’m on a Mac, rather than using the URL specified here, I’m going to be using the local host URL.

So I open my Chrome web browser and go to localhost:6006.

# http://localhost:6006


All right, so we see that it’s in TensorBoard, we could see graphs, and we can zoom in a little bit and we see our example from the previous video where we have a ten tensor, a twenty tensor, and an Add operation.

So we can see that the website is serving our TensorFlow ProtoBuf event file that was written and we see that it’s working.


Perfect - We were able to use the TensorBoard command line utility to launch the TensorFlow TensorBoard web service.

Receive the Data Science Weekly Newsletter every Thursday

Easy to unsubscribe at any time. Your e-mail address is safe.