mofresh 0.2.3

A Python package for fresh data processing

Mofresh

Widgets, designed for marimo, that can automatically refresh

Installation

You can install the package using pip:

uv pip install mofresh

Usage

The goal of this project is to offer a few tools that make it easy for you to refresh charts in marimo. This can be useful during a PyTorch training loop where you might want to update a chart on every iteration, but there are many other use-cases for this too.

Widgets

The library provides three widgets:

1. ImageRefreshWidget - Displays images that can be updated dynamically. Perfect for refreshing matplotlib plots or any image content.
2. HTMLRefreshWidget - Renders HTML content that can be updated on the fly. Great for Altair charts, plotly visualizations, or custom HTML.
3. ProgressBar - A modern progress bar with dark mode support. Ideal for tracking training loops or long-running operations.

How it works

The trick to get updating charts to work is to leverage anywidget. These widgets have a loop that is independant of the marimo cells which means that you can update a chart even if the cell hasn't completed running. The goal of this library is to make it easy to use this pattern by giving you a few utilities.

Effectively that means you can expect to see stuff like this in marimo:

Live demo

If you want to dive deep and experience the API, the best way is to explore the live notebook on Github pages.

Go to live docs.