Working with notebooks

Integrated JupyterLab

omega|ml comes pre-configured to run with JupyterLab. In this mode, every notebook that you create is automatically stored in the omega|ml jobs storage. The advantage is that you can share notebooks instantly with your colleagues - their JupyterLab workspace will show the same notebooks as you have, no matter from where they work.

Local JupyterLab

If you prefer to work within your own JupyterLab environment, you can still use omega|ml to deploy your notebooks and run the on the cloud. This is convenient, for example when you only have a certain budget to run on a CPU. Then you can prepare your notebooks locally and only run them on the GPU when it is ready.

In this case you have to store your notebook to jobs as follows. See the next section for running it on the cloud.

# store a notebook in omegaml
$ om jobs put /path/to/notebook.ipynb mynotebook

Running notebooks on the cloud

To run the notebook, waiting for its result, and store the result in jobs/result, execute the following command:

$ om runtime job mynotebook

You may also run the notebook asynchronously, as a background task, as follows. The command will print the task id for later retrieval of the job’s result:

$ om runtime job mynotebook --async

$ om runtime result 77e50b76-b4bb-4411-a407-fa1cf2162f81
<Metadata: Metadata(name=results/Untitled_2020-10-06 18:41:00.060559.ipynb,bucket=omegaml,prefix=jobs/,kind=script.ipynb,created=2020-10-06 18:41:01.942656)>

Retrieving results

$ om jobs list results
['mynotebook.ipynb', 'mynotebook2.ipynb', 'results/mynotebook_2020-09-18 13:37:56.885840.ipynb']

We can get it back or view it directly in JupyterLab:

$ om jobs get 'results/mynotebook_2020-09-18 13:37:56.885840.ipynb' mynotebook_results.ipynb