Templates in Tensor One are Docker container images paired with configuration details. They are essential for launching Clusters with predefined settings and environments.

What is a Template?

A Template defines the container image along with:
  • Required container disk size
  • Volume size and mount paths
  • Port configurations
  • Environment variables
Templates simplify Cluster deployment by pre-configuring critical infrastructure components, making it easy to launch consistent environments.

Template Types

Tensor One supports multiple types of templates:

Managed by Tensor One (Official Templates)

  • Curated and maintained by Tensor One
  • Ideal for users who want reliable, production-ready environments
  • Automatically updated and tested

Custom Templates

  • User-defined templates for tailored Cluster configurations
  • Great for advanced users with specific tools, commands, or settings
  • Enables full customization of the runtime

Community Templates

  • Created and shared by other Tensor One users
  • Useful for collaboration, learning, and starting points
  • Can be cloned and customized

Private Templates

  • Created and shared within your team account
  • Securely stored and accessible only by your team
  • Ideal for internal tooling and workflows

Creating and Customizing Templates

You can start with an official or community template and:
  • Use it as-is
  • Or customize it to suit your specific needs

Modifying the Docker Command

The default command might look like:
bash -c "/start.sh"
To install additional tools or run pre-setup commands, modify it like so:
bash -c "apt update && apt install vim -y && /start.sh"
This allows you to:
  • Install custom packages
  • Run configuration scripts
  • Chain setup logic before launching the main app

Why Customize Templates?

Customizing templates gives you fine-grained control over the startup behavior of your Clusters. This supports complex workflows in:
  • Software development
  • Research experiments
  • AI/ML model deployment
  • Custom API environments