Accelerate.

A Garden is a full ML ecosystem for a particular topic that makes building on existing ML research significantly easier and faster while fostering community building around domain problems of interest.

The Problems with the ML Landscape Today

Reproducibility

Reproducing someone else's work is extremely time consuming in the best case scenario and impossible in the worst case (and most common) scenario. Researchers struggle to share all of their code and data in a way that other's can actually use it and reproduce their work.

Accessibility

Code and data are very rarely fully available to the public, and many require compute that most people don't have access to.

Quality

You can't tell if code runs until you spend hours getting the environment up and running to test it. That wasted time adds up after a few repos that just don't work.

Community

Finding and connecting with people working on similar or relevant problems to you can be difficult.

How Garden addresses these issues

Reproducibility

You have all the pieces to reproduce and build off of any Garden. We've found that just having a model file won't get you to where you want to go- there's a lot more code that goes into reproducing an ML pipeline to get the desired output. A Garden includes:

Links to data

Groups of models

Functions that are necessary to the ML pipeline

Automated quality testing

Citable objects for the Garden and each model

Anything else a model needs to run and be evaluated

Accessibility

All code, data, testing is available and free without any barriers to access. You can run models using our compute resources at UChicago or any Globus Compute endpoint.

Quality

We test and evaluate models on our side, so you can browse research without the time investment of doing all of that yourself.

Community

Garden is a collaborative platform that connects like-minded people. Find your ML community, collaborate, and set new benchmarks in your domain.

How Will People Use Garden

We have users in mind while creating Garden, and there is always the possibility of adding more! Here are a few users to give you an idea of how people will use Garden.

Consumer

I want to find models that address a specific problem of interest in my domain.

Publisher

I have multiple ML models that center around a problem. I want them to be published and discovered together.

Educator

My students are creating models that target a certain problem, and I want to showcase their work in one collection.

Collaborator

I have a model that aims to solve problem of interest in my domain. I want to add it to others' solutions so it can be evaluated alongside them.

Acknowledgments

Garden is funded by:

National Science Foundation - Award Abstract #2209892: “Frameworks: Garden: A FAIR Framework for Publishing and Applying AI Models for Translational Research in Science, Engineering, Education, and Industry”

This project builds upon work including:

The Materials Data Facility (MDF) - NIST-supported effort to build data services to help material scientists publish and discover data

Foundry-ML - An open source machine learning platform for scientists

Globus - Research cyberinfrastructure, developed and operated as a not-for-profit service by the University of Chicago