Using Platform Notebooks

Learning Objectives

The learning objectives of this article are:

• How to create a notebook in the platform
• Differences between Jupyter Notebook and Abacus Notebook
• What it means to edit code using a notebook
• This is used for: Agents, Python Feature Groups, custom modules, Pipelines, and other Abacus objects

Abacus and Jupyter Notebooks

Abacus has two types of notebooks:

1. Abacus Notebooks: They spin up immediately, but have limited resources
2. Jupyter Notebooks: They take time to spin up, but have more resources

Visually, both notebooks will be similar, but a user will be able to tell the difference by look and feel.

By default, when you are trying to edit objects that have code embedded such as:

• Pipelines
• Python Feature Groups
• Custom Modules
• AI Agents
• etc

An Abacus notebook will be opened. If you want to change the Abacus notebook to a Jupyter Notebook, just click on the “Jupyter Notebook” icon on the top right.

Creating a new Notebook

To create a new Jupyter notebook, click on “Notebooks” and “Create Notebook”:

Choose the settings that you want, and then a Jupyter Notebook will launch in a minute.

If you want to open an existing notebook as an Abacus Notebook, you can click on “Actions” to the right of each notebook.

This will give you the ability to change the memory of the individual notebook, stop it, or even rename it. To open the notebook as an Abacus Notebook, click on “Details” and then “Open in Abacus Notebook”

Editing Code Objects Using a Notebook

Throughout the platform, there are multiple objects that are tied to code. For example, Pipelines, Custom Modules, and Python Feature Groups are all generated using code.

All of these objects have an “EDIT” button. Here is how that edit button looks for a Python feature group, for example:

When you click on "Edit Code and Details," Abacus automatically retrieves the latest deployed version of the code and creates a special notebook named "EDIT python_feature_group_name." This notebook is fully populated with the current code and details.

If you make changes to this notebook without deploying them to the feature group and then click "Edit Code and Details" again, the notebook will be recreated, and any changes you made that were not deployed will be lost. It's important to be aware of this behavior to avoid unexpected results.

This functionality is true for all objects that have code related to them, whether it’s a Python feature group, an agent, a pipeline, etc. The reason it exists is to ensure that when you are editing a code object, you are always using the latest version of it.

You can always create other sub-notebooks where you test and iterate.