Skip to content

Jupyter AI Assistant

Summary

Jupyter AI Assistant is an interactive app which uses local LLM inference to assist users with programming tasks.
This app is very similar to the jupyterlab app, except that it comes with a customized Jupyter AI extension.

Usage

First, navigate to the OOD dashboard and click on "Jupyter AI Assistant" under interactive apps: Interactive Apps Next you will be presented with a form for selecting the version of Python or Anaconda to use as well as the choice to specify an optional environment. For information on creating a custom environment for a jupyterlab session, please reference the jupyterlab page: JupyterLab
Once in the interactivate app session, you can click on the message icon on the left side to interact with the Jupyter AI extension: Message Bubble in Interactive Session If it is your first time using the extension, a page will be displayed with a button that says, "Start Here" which you need to click. After this you will see a page which looks like: Start Screen From here, all you need to do is select the HPRC provider from the completions model dropdown: Jupyter AI provider and click, "Save Changes"
Afterwards, click the back arrow at the top, and you will land on this page:

The commands here are somewhat self-explanatory. Use /ask followed by text for an answer to a question, for instance:

/ask How do I write lambda functions in Python?
Note that the model is better with instruction based prompts rather than direct questions do due to the way the llama instruct models were trained.
you can also use /fix for helping with errors in a selected cell in your opened notebook file: