1. Set Up Mantik Account

This section focuses on setting up your Machine Learning project on the Mantik platform.

1.1. Registration

First, in order to use Mantik, you need to create an account.

1.2. Create Project

Note

If this is not your first tutorial you may skip this section since you should already have a project for tutorials.

Once logged in, you need to create a Project:

  • click on the Projects section of the top-bar

  • click on the + CREATE PROJECT button

  • fill the form with:

    • name

    • short description

    • long description

    • public/private

    Example Inputs name: My first Project

    short description: Tutorials

    long description: A tutorial project created when following the step-by-step guide from the Mantik official docs.

    private (In order to keep the project private, do not click on Make this a Public Project)

  • click on CREATE

1.3. Create Experiment

Now that you created a project it should appear in the list of projects accessible to you, displayed in the Projects section of the top-bar.

A project is divided into Experiments, each Experiment bundles some Runs together, and can be used to divide Runs related to different research questions.

Note

A Run must be part of an Experiment and an Experiment must be part of a Project.

To create an Experiment:

  • click on the VIEW DETAILS on your project

  • click on Experiments on the side menu

  • click on the + ADD button (it is all the way to the right)

  • give a name to your experiment e.g. JUWELS Tutorial

  • click on CREATE

1.4. Create Code Repository

Code repositories are references to GitLab or GitHub repositories used in the project.

In this tutorial, you will use a demo stored in our public tutorials repo on GitLab: https://gitlab.com/mantik-ai/tutorials, called Wine Quality Estimator. This demo enables you to execute training of a linear model to predict wine quality from wine-related input features such as acidity.

To create a Code Repository:

  • click on the VIEW DETAILS on your project

  • click on Code on the side menu

  • click on the + ADD button (it is all the way to the right)

  • fill the form with:

  • click on CREATE

1.5. Store Connection

As mentioned in the tutorial’s overview, to follow this tutorial you need to have your own credentials to one of the compute facilities () from the Jülich Supercomputing Center (JSC) or the Swiss National Supercomputing Center (CSCS).

These credentials are going to be used each time that you submit a new Run to the supercomputer and have to be stored as a Connection:

  • click on your avatar icon in the top right corner and click on Settings

  • click on Connections on the side menu

  • click on the + ADD button (it is all the way to the right)

  • fill the form with:

    • Name: the name you want to give the credential

    • Provider: The provider of the credentials

    • Auth method: Username-Password

    • Login name: Your JuDoor username

    • Password: Your JuDoor password

    Example

    • Name: JULICH credentials

    • Provider: Jülich Supercomputing Centre (JSC)

    • Auth method: Username-Password

    • Login name: mustermann1

    • Password: MyAwesomePassword!1

    • Name: the name you want to give the credential

    • Provider: The provider of the credentials

    • Auth method: Username-Password

    • Login name: OIDC client ID

    • Password: OIDC client secret

    In order to create the OIDC client ID-secret pair, go to the CSCS OIDC dashboard, create a client and copy its name and secret.

    Example

    • Name: CSCS credentials

    • Provider: Swiss National Supercomuting Center (CSCS)

    • Auth method: Username-Password

    • Login name: my-firecrest-client

    • Password: !lj#lA19A_k*.

  • click on SUBMIT

Now your credentials are stored securely on Mantik and can be used when submitting runs or when using Mantik CLI commands.