How to Use TensorBoard in Notebooks

Notebooks are a web-based Jupyter IDE with shared persistent storage for long-term development and inter-notebook collaboration, backed by accelerated compute.

TensorBoard is a visualization toolkit from TensorFlow which is useful for a variety of applications for instance visualizing metrics and histograms, and displaying images.

To get started with TensorBoard, you need to run a command in the terminal and then access a special URL.

Configuring a TensorBoard in Notebooks

To run TensorBoard in a notebook, you need to check that the notebook has the TensorBoard installed.

Next, open a notebook terminal and run the following command:

tensorboard --logdir . --bind_all

If needed, you can change the --logdir path to a different value.

The TensorBoard is available with the following URL:

https://tensorboard-NOTEBOOKID.CLUSTERID.paperspacegradient.com

You can find the NOTEBOOKID on the notebook list view in the Gradient console.

The default CLUSTERID value is clg07azjl. Private clusters have a CLUSTERID on the clusters page.