In this tutorial, we will use Jupyhai, to run a Valohai execution from your local Jupyter notebook.
- Jupyhai is a Jupyter notebook extension developed and maintained by Valohai.
- Use the Run remote button, instead of the Run cell button to run your notebook on Valohai.
- The Jupyhai addon will generate a valohai.yaml config file for each execution based on the contents of the Notebook. You don’t need to create the YAML file yourself.
- Notebook executions use a custom Docker image called
valohai/pypermill. Make sure to use it as a base for your custom Docker images.
pip install --upgrade pip pip install notebook pip install valohai-cli jupyhai jupyhai install
mkdir valohai-notebook cd valohai-notebook vh login # ... login with your username and password vh project create # give the project a name jupyter notebook
Create a new folder
valohai-notebook on your desktop and launch Jupyter in that directory.
- Create a new Python 3 notebook
- Open the new Notebook
- Click on Settings on the toolbar
- Login with your Valohai credentials
- Update the Docker image to
- Close the Settings tab
print('Hello Valohai Notebooks!')to the first cell
- Click on
Run remoteto run the notebook remotely
- View the logs from the execution by clicking on the gizmo on the right side of the page (e.g.
valohai/pypermill. You can build your own images for Notebook executions, as long as they’re based on
Check out our prebuilt set of Docker images for the most popular Notebook executions on Valohai