MDF Spotlight

Datasets worth a closer look.

Spotlight takes one release at a time and slows down on it: how the data were produced, what is actually inside, and how to get it onto your own machine. Written for the researcher who is about to use it.

Releases: 01
Latest: 2025-10-07
Combined volume: ~500 TB
Access: Open, CC-BY-4.0

Latest release

01 / 01

Scanning electron micrograph of a superconducting nanowire single-photon detector array — NASA/JPL · public domain

Specimen — SNSPD array, SEM500 TB

New release

Quantum chemistry and machine learning

OMol25: a large-scale electronic structure dataset for molecular machine learning

Half a petabyte of open molecular data — raw DFT outputs, electronic densities, wavefunctions, and molecular orbitals from more than four million high-accuracy calculations, served over a high-performance Globus endpoint at Argonne.

Dataset volume: ~500 TB
DFT calculations: >4M
Published: 2025-10-07
License: CC-BY-4.0

Read the spotlight

More in preparation

Spotlight is deliberately slow. Releases are chosen for scale, reuse, and the odds that someone else can build something on top of them. If you are sitting on a dataset like that, publish it with MDF and tell us — the next one could be yours.

Series 01 · 01 published · next release TBD

Publish with MDF Search all datasets