Running hyperaparameter optimization

You can run a Valohai Task from a notebook by defining Valohai parameters and running a versioned execution from the notebook.

Start by defining parameters and inputs for a Valohai step inside your Notebook

import valohai

valohai.prepare(
    step='train-model',
    image='tensorflow/tensorflow:2.6.0',
    default_inputs={
        'input': 'https://valohaidemo.blob.core.windows.net/mnist/mnist.npz'
    },
    default_parameters={
        'learning_rate': 0.001,
        'epoch': 10
    }
)

After you've defined your own inputs and parameters you can click on the "Run Remote" button in the Jupyter UI to execute the full notebook from beginning to end as a separate job and version it.

After the job has completed successfully you can launch it as a Task by opening the execution in the Valohai UI and clicking on the Task icon.