An open source machine learning platform for scientists

Accessing ML-ready datasets just became the easiest part of your workflow

We make collaboration and reuse easy


dataset = f.get_dataset("foundry_stan_segmentation_v1.1")
dataset.get_as_dict()
            
                  
dendrite segmentation data snippet

Foundry-ML hosts structured, ML-ready data of any size that can be accessed programatically with our Python SDK. Once a dataset is uploaded to Foundry-ML, you can load it with just 2 lines of code. The data is loaded into a DataFrame and ready for analysis!

and save you valuable time

No need to spend time making sense of someone else's data. We collect information about each dataset to take out any guesswork on your part. Our schema requires key information for every single dataset we host, making it easy to interpret each one. Our publishing process also includes a human review to make sure it meets our data standards.

Key Type Units Description
referenceinputsource publication of the band gap value
structureinputthe structure of this compound
bandgap value (eV)targeteVtvalue of the band gap

Created for sharing large datasets

Our infrastructure was built with large datasets in mind. Share data that common infrastructures (email, DropBox, GitHub) can't. We use Globus to easily transfer data to anywhere you want to use it - from a laptop to a supercomputer.

Foundry-ML is part of the Materials Data Facility. MDF collects high-quality scientific datasets from the community and makes them easy to find and use. Foundry-ML datasets are a subset of MDF datasets that are structured and can be accessed programatically. For datasets that have less structure, or don't need to be accessed programatically, consider using MDF directly instead of Foundry-ML.

logologo

Build Foundry With Us

At Foundry, we believe accessibility and collaboration are the keys to problem solving. That's why we've made it super easy for you to share and use data. If you have a dataset you'd like to contribute, our publishing guide can set you up to do that.

Check out our documentation to see how easy it is to use Foundry, and follow our examples to do it all yourself.