Using TensorFlow Enterprise with AI Platform Deep Learning VM Image

This page describes how to get started using TensorFlow Enterprise with a Deep Learning VM instance.

In this example, you create a TensorFlow Enterprise Deep Learning VM instance, connect to the instance using SSH, open a JupyterLab notebook, and run a classification tutorial on using neural networks with Keras.

Before you begin

  1. Sign in to your Google Account.

    If you don't already have one, sign up for a new account.

  2. In the Google Cloud Console, on the project selector page, select or create a Google Cloud project.

    Go to the project selector page

  3. Make sure that billing is enabled for your Cloud project. Learn how to confirm that billing is enabled for your project.

Create a Deep Learning VM instance

To create a TensorFlow Enterprise Deep Learning VM instance, complete these steps:

  1. Go to the Deep Learning VM Cloud Marketplace page in the Google Cloud Console.

    Go to the Deep Learning VM Cloud Marketplace page

  2. Click Launch on Compute Engine. If you see a project selection window, choose the project in which to create the instance. If this is the first time you've launched Compute Engine, you must wait for the initial API configuration process to complete.

  3. On the New Deep Learning VM deployment page, enter a Deployment name. This will be the root of your virtual machine name. Compute Engine appends -vm to this name when creating your instance.

  4. Under Number of GPUs, select None. You won't need them to complete the instructions in this guide.

  5. Under Framework, select TensorFlow Enterprise 2.1 (CUDA 10.1).

  6. For this example, you can leave the remaining settings as they are.

  7. Click Deploy.

You've just created your first instance of a Deep Learning VM. After the instance is created, the Deployment Manager opens. This is where you can manage your Deep Learning VM instance and other deployments.

Connect with SSH, open a notebook, and run a classification tutorial

Complete these steps to set up an SSH connection to your Deep Learning VM instance, open a JupyterLab notebook, and run a tutorial on using neural networks with Keras:

  1. To complete these steps, you can use either Cloud Shell or any environment where the Cloud SDK can be installed. Cloud Shell and Cloud SDK are command line tools that you can use to interface with your instance.

    • If you want to use Cloud Shell, in Google Cloud, in the upper-right corner, click the Activate Cloud Shell button.

      Google Cloud Platform console

    • If you want to use Cloud SDK, download and install Cloud SDK on your local machine.

  2. In Cloud Shell or in a local terminal window, use the following command to create an SSH connection to your instance. Replace my-project-id, my-zone, and my-instance-name with the relevant information.

    gcloud compute ssh --project my-project-id --zone my-zone \
      my-instance-name -- -L 8080:localhost:8080
    
  3. In your local browser, visit http://localhost:8080 to access a JupyterLab notebook that is included in your instance by default.

  4. In the notebook, on the left, double-click tutorials to open the folder, and navigate to and open tutorials/tf2_course/01_neural_nets_with_keras.ipynb.

  5. Click the run button to run cells of the tutorial.

What's next