Tensorlake Documentation

Connect your repositories

You need a Tensorlake project you can administer and permission to install a GitHub App for your personal account or organization.

Open your project

Open the Tensorlake dashboard, select your project, and open GitHub Actions.

Connect GitHub

Click Connect GitHub, choose your personal account or organization, and select the repositories that can use Tensorlake runners. Install the Tensorlake GitHub Actions App and approve any requested permissions.GitHub returns you to your project and connects the installation automatically. If your organization requires approval, ask an owner to approve the installation, then choose Use existing installation in your project’s connections view.

Choose a runner

Set runs-on in an existing job to a Tensorlake runner label:

runs-on: tensorlake-medium

You don’t need the self-hosted label or a Tensorlake API key in your workflow.

Each GitHub installation connects to one Tensorlake project. You can connect multiple installations to the same project.

Run your first job

Save this workflow as .github/workflows/tensorlake.yml on your repository’s default branch. In GitHub’s Actions tab, select Tensorlake runner and click Run workflow.

name: Tensorlake runner
on:
  workflow_dispatch:
permissions:
  contents: read
jobs:
  build:
    runs-on: tensorlake-small
    timeout-minutes: 10
    steps:
      - uses: actions/checkout@v6
      - run: |
          uname -m
          docker version

Open Jobs in your project’s GitHub Actions view to inspect status, runner resources, timings, and cache activity. Open a job in GitHub for its workflow logs and artifacts.

​Connect your repositories

​Run your first job

Connect your repositories

Run your first job