Launch Your First Notebook

Get a cloud- or onprem-hosted Jupyter environment running on Valohai.

Launch a Notebook

Navigate to your project and follow these steps:

1. Go to the Notebooks tab

2. Click Create Notebook Execution

3. Select your environment and Docker image

Valohai will spin up a machine and install Jupyter if your image doesn't include it. This takes 2–5 minutes depending on your environment.

💡 Tip: Use standard images like python:3.12 to get started quickly. Valohai handles Jupyter installation automatically.

Optional Configuration

You can also set:

• Title and tags to find your notebook later

• Inputs to download files from your data stores

• Git commit to load code from a specific version

Once the notebook shows "Ready," click Open Notebook to access your Jupyter environment.

Run Your First Cell

Your notebook is now running on cloud infrastructure. Test it:

import sys
print(f"Python version: {sys.version}")
print("Hello from Valohai!")

Run the cell. You're now executing code on a real cloud machine with access to your configured storage and compute resources.

Stop the Notebook

Critical: Your notebook keeps running (and billing) until you manually stop it.

When you're done working:

1. Go back to the Valohai UI

2. Navigate to Notebooks

3. Click Stop on your running notebook

The machine will be released and billing will stop. Your notebook's final state is automatically saved and versioned.

⚠️ Always stop notebooks when done. Leaving a notebook running costs the same as an active training job—even if you're not using it.

Next Steps

Now that you have a notebook running, you can:

• Save Your Work — Version outputs and models

• Push to Git — Commit changes directly from Jupyter

• Convert to Production Script — Move from experimentation to execution