Deploy TensorFlow GPU

How to deploy a Jupyter Notebook pre-installed with TensorFlow?

NOTE: Spheron Compute offers the flexibility to create custom configurations for your instance.

Spheron allows you to deploy a password protected Jupyter Notebook instance pre-installed with TensorFlow, all set up and ready to use. To deploy a TensorFlow GPU:

  1. Upon logging in, you'll be directed to the Create Organization page, where you can give your organization name and choose Avatar. Ensure the "Compute" option is selected from the drop-down menu of the "Start With" option. Click 'Continue'.
  2. Next, you'll be taken to a new page. Click the "Create New Projects" button. Add 'Project Title' And 'Project Description' and Click 'Create'.
  3. Choose "Accelerate" to leverage GPU-powered computing for enhanced performance.
  4. Choose your desired Compute Type option under Compute Type.
  5. Click "Start from Marketplace App" and Select "TensorFlow GPU" from the marketplace.
  6. When selecting a region, we recommend starting by trying to deploy in a region closer to you. If you encounter any issues, you can consider switching to other regions. Choosing a region closer to you can improve performance and reduce latency. Click here to know more.
  7. Spheron will automatically select the recommended plan for the specific template. If you intend to move forward with the recommended plan, just Click 'Deploy' to initiate deployment.
  8. If you want you can Select the instance plan that suits your needs. You can use the "Create Custom Plan" toggle to create custom plans for your CPU based instance.
  9. You have to choose storage from the available options or the custom storage option that fits your needs. This storage will be volatile and is erased when the instance is restarted, redeployed, or shut down. Additionally, you get the option to choose Persistent Storage.
  10. Next, Spheron has made it easy and auto-filled the configuration options. You can add advanced configuration if required. Click here to know more.
  11. Click "Deploy" to initiate deployment.

Deploy Your Own

Deploy your own Jupyter Notebook pre-installed with TensorFlow with Spheron:

Deploy with Spheron (opens in a new tab)

Verify Installation

The dashboard can be accessed only after the Compute Instance is provisioned. Thus, you need to wait for the installation to complete before you can start using it.

How to connect to Jupyter Notebook?

  • Use the Connection URL provided by Spheron under Port Policy Info to access Jupyter Notebook.
  • Use the token from the instance logs to unlock the dashboard.

For more information, refer to the TensorFlow docs (opens in a new tab).

Deploy Minecraft ServerSpheron SDK