Notebooks

Inside a Notebook we have access to a variety of tools that will help you interact with your data et develop your AI use-case. You will be able also to use specific features to convert your notebook into reusable code (modules) and execute it (pipelines & jobs).

Currently, Allonia platform supports these versions of Jupyter:

  • JupyterHub : 3.0

  • JupyterLab : 4.0.4

Create and open a Notebook

To create a notebook, just simply click “Create New Notebook” from the workspace:

create notebook

You will then be asked on which environment you want the Notebook to run.

To open a notebook, click on it in the workspace or click "…" > open.

notebook

Manage your notebook compute environment

We aim to provide a flexibility around resources that you could use with your notebook.

Each notebook you created can be launched each time with a specific computing environment.

notebook compute

Here are the detailed computing ressources specifications for notebooks:

Computing environment CPU core Memory (GB) GPU Core GPU memory (GB)

Python notebook small

1

4

0

0

Python notebook large

4

16

0

0

Python notebook gpu

8

40

1 (P100)

16

Develop with a notebook

Most of the development will remains the same as any work done inside a Jupyter Notebook, so you could refer at any tutorial available on the internet.

However, there will remain 2 things that will be specific to Allonia Platform with dedicated utilities that will help you in your Allonia journey:

  • Allonia Python utilities (Aleialib) : a library that will help you accessing & managing various objects:

    • Datasources (files, db connectors)

    • Specific partner algorithms & features (geo data, annotation algo, classification, …​)

  • Allonia Python model management (Aleiamodel) : a framework that will help you manage the whole model lifecycle:

    • Model architecture definition

    • Model training & performance tracking

    • Model evaluation

    • Model deployment

    • Model documentation tracking

More content are available through dedicated documentation part provided with examples:

  • Allonia Python utilities : here

  • Allonia Python model management : here

Convert to a module

You can convert a Notebook into a module, selecting one or multiple cells. It will let you export class & functions you will want to reuse in any other Python code (notebooks, modules, services).

notebook module

For more details about module management, you can refer to the dedicated part about Modules.

Convert to a pipeline & build a job

You can convert a Notebook into a pipeline. It will do two things:

  • Export the notebook code into a module, as one single monolithic code to be executed

  • Create a pipeline file with the related module referenced inside

export notebook pipeline