Jupyter is a collaborative tool that data scientists use to develop and execute code, documentation, and visualization in their ML model development process. For more information, see the Jupyter website.
To create and manage notebook servers in your Kubeflow deployment:
A notebook management window opens, as shown in Figure 4.
Figure 4. Jupyter Notebook Servers window
Figure 5 shows the menu that is now available for configuring the notebook server with various options. The user should at a minimum
A pod is deployed using the TensorFlow container image that you specified, and it can be verified by opening the Notebook section of the Kubeflow dashboard, as shown Figure 6.
Figure 5. Creating a Jupyter Notebook Server
A notebook server is created using your selected options, as shown in the following figure:
Figure 6. Configured Notebook Server
The Jupyter notebook opens, as shown in Figure 7.
Figure 7. Jupyter notebook
An empty code cell opens in which you can run Python code to display the TensorFlow version that is installed in the notebook. Figure 8 shows the expected version, v1.13.1:
Figure 8. Verifying the TensorFlow installation