marimo 0.1.13

A library for making reactive notebooks and apps

Next-generation Python notebooks and apps.

Docs · Discord · Examples

marimo is a Python library for creating reactive notebooks that double as apps. marimo is:

• reactive: run a cell and marimo automatically runs cells that depend on it
• interactive: interact with sliders, dropdowns, tables, and more to automatically run dependent cells
• expressive: parametrize markdown with interactive elements, plots, or anything else
• simple: no callbacks, no magical syntax
• Pythonic: cells only run Python; notebooks stored as .py files (clean git diffs!)
• performant: powered by static analysis, zero runtime overhead

marimo was built from the ground up to solve many well-known problems with traditional notebooks. marimo is not built on top of Jupyter or any other notebook or app library.

Contents.

• Getting Started
  • Installation
  • Tutorials
  • Notebooks
  • Apps
  • Convert Jupyter notebooks
  • VS Code extension
• Concepts
• Examples
• FAQ
• Contributing
• Community

Getting Started

Installing marimo gets you the marimo command-line interface (CLI), the entry point to all things marimo.

Installation

In a terminal, run

pip install marimo
marimo tutorial intro

You should see a tutorial notebook in your browser:

If that doesn't work, please open a Github issue.

Tutorials

marimo tutorial intro opens the intro tutorial. List all tutorials with

marimo tutorial --help

Notebooks

Create and edit notebooks with marimo edit.

• create a new notebook:

marimo edit

• create or edit a notebook with a given name:

marimo edit your_notebook.py

Apps

Use marimo run to serve your notebook as an app, with Python code hidden and uneditable.

marimo run your_notebook.py

Convert Jupyter notebooks

Automatically translate Jupyter notebooks to marimo notebooks with marimo convert:

marimo convert your_notebook.ipynb > your_notebook.py

Because marimo is different from traditional notebooks, your converted notebook will likely have errors that you'll need to fix. marimo will guide you through fixing them when you open it with marimo edit.

VS Code extension

If you prefer VS Code over terminal, try our VS Code extension. Use the extension to edit and run notebooks directly from VS Code, and to list all marimo notebooks in your current directory.

Concepts

marimo notebooks are reactive: they automatically react to your code changes and UI interactions and keep your notebook up-to-date (like a spreadsheet).

Cells

A marimo notebook is made of small blocks of Python code called cells. When you run a cell, marimo automatically runs all cells that read any global variables defined by that cell. This is reactive execution.

Reactive execution lets your notebooks double as interactive apps. It also guarantees that your code and program state are consistent.

Execution order. The order of cells on the page has no bearing on the order cells are executed in: execution order is completely determined by the variables cells define and the cells they read. You have full freedom over how to organize your code and tell your stories: move helper functions and other "appendices" to the bottom of your notebook, or put cells with important outputs at the top.

No hidden state. marimo notebooks have no hidden state because the program state is automatically synchronized with your code changes and UI interactions. And if you delete a cell, marimo automatically deletes that cell's variables, preventing painful bugs that arise in traditional notebooks.

No magical syntax. There's no magical syntax or API required to opt-in to reactivity: cells are Python and only Python. Behind-the-scenes, marimo statically analyzes each cell's code just once, creating a directed acyclic graph based on the global names each cell defines and reads. This is how data flows in a marimo notebook.

For more on reactive execution, open the dataflow tutorial:

marimo tutorial dataflow

The marimo library

marimo is both a notebook and a library. The marimo library lets you use markdown, interactive UI elements, layout elements, and more in your marimo notebooks.

We recommend starting each marimo notebook with a cell containing a single line of code,

import marimo as mo

Outputs

marimo visualizes the last expression of each cell as its output. Outputs can be any Python value, including markdown and interactive elements created with the marimo library, e.g., mo.md(...), mo.ui.slider(...). You can even interpolate Python values into markdown and other marimo elements to build rich composite outputs.

Thanks to reactive execution, running a cell refreshes all the relevant outputs in your notebook.

For more on outputs, try these tutorials:

marimo tutorial markdown
marimo tutorial plots
marimo tutorial layout

Interactive elements

The marimo library comes with many interactive stateful elements in marimo.ui, including simple ones like sliders, dropdowns, text fields, and file upload areas, as well as composite ones like forms, arrays, and dictionaries that can wrap other UI elements.

Using UI elements. To use a UI element, create it with marimo.ui and assign it to a global variable. When you interact with a UI element in your browser (e.g., sliding a slider), marimo sends the new value back to Python and reactively runs all cells that use the element, which you can access via its value attribute.

This combination of interactivity and reactivity is very powerful: use it to make your data tangible during exploration and to build all kinds of tools and apps.

marimo can only synchronize UI elements that are assigned to global variables. You can use composite elements like mo.ui.array and mo.ui.dictionary if the set of UI elements is not known until runtime.

For more on interactive elements, run the UI tutorial:

marimo tutorial ui

Composite elements

marimo's composite UI elements let you wrap other UI elements to create powerful UIs. For example, marimo.ui.form lets you gate elements on submission, while marimo.ui.dictionary and marimo.ui.array let you batch arbitrary collections of elements.

Layout

The marimo library also comes with layout elements, including mo.hstack, mo.vstack, and mo.tabs. See the API reference for more info.

Examples

Examples are available in the examples/ directory. Community examples can be found and shared in the marimo cookbook.

We've deployed many of these examples at our public gallery; try them out!

FAQ

See the FAQ at our docs.

Contributing

We appreciate all contributions. You don't need to be an expert to help out. Please see CONTRIBUTING.md for more details on how to get started.

Questions? Reach out to us on Discord.

Community

We're building a community. Come hang out with us!

• 🌟 Star us on GitHub
• 📧 Subscribe to our Newsletter
• 💬 Join us on Discord
• 🐦 Follow us on Twitter
• 🕴️ Follow us on LinkedIn
• ✏️ Start a GitHub Discussion