❔ About

I recently saw this tweet on my thread:

👉 Quite immediately I was excited to give it a try.

Sometimes, you may need to seamlessly call various AI inferences... so it may look like en embedded pair-programming that keeps your notebook clean, compact, smart and still highly understandable** so people (including yourself) can focus on the storytelling.

🪝 Teaser

This notebook is dedicated to a (not so) short jupyterlab/jupyter-ai unboxing so anyone can enjoy this kind of magic (and much much more):

%%ai gpt4 -f markdown
Write a haiku about Python.

.. and get the output in the next cell:

Python code so clean,
Logic flowing like a stream,
In data, we glean.

🧙 Tricks you don't want to miss

... and many other cool tricks like:

• ↔️ Deal with input and output variables (aka. PROMPT templates)
• 𝕏 Write nice looking LaTeX like formulae
• 🐍 Code generation... and execution
• 📊 Dataviz by plotting data
• 🗣️ 🤖 Chat with a dataframe
• 🌟 📜 Restaurant customer experience disruption

🍿 Demo time

🔖 Resources

• jupyterlab/jupyter-ai
• https://jupyter-ai.readthedocs.io/en/latest/
• Jupyter AI supported on Kaggle
• Interpolating in prompts
• Formatting the output
• The %ai and %%ai magic commands