Open Materials 2024 (OMat24)

OMat24 is Meta AI's large-scale open dataset for training machine-learning interatomic potentials (MLIPs) on inorganic materials. It comprises roughly 100 million DFT calculations — both single-point evaluations and structural relaxations — annotated with total energies, per-atom forces, and cell stresses, spanning broad structural and compositional diversity. Distributed as LMDB files compatible with the FAIRChem package. The companion EquiformerV2 models trained on OMat24 lead the Matbench Discovery leaderboard.

Category
Chemistry & Materials
Provider
Meta AI
License
cc-by-4.0
Hosted on
Hugging Face, bucket facebook/OMAT24, us
Access
Free and public. No account with the data provider is needed.
Source
https://arxiv.org/abs/2410.12771
Data structure and loading code

Files

Example file: references/omat-elemental-reference-compounds.json.gz

Load the data

from huggingface_hub import hf_hub_download
path = hf_hub_download(repo_id="facebook/OMAT24", filename="references/omat-elemental-reference-compounds.json.gz", repo_type="dataset")
print(path)

Written for a Scigantic notebook, where /mnt/archive is the mounted dataset.

Analyze this dataset without downloading it

On Scigantic, Open Materials 2024 (OMat24) opens in a hosted JupyterLab notebook with the data already mounted, so you can read files where they are stored instead of copying them first. Each notebook includes an AI assistant that can see the mounted files and write and run analysis code. Open the dataset page to launch a notebook, or browse the full catalog.

Related Chemistry & Materials datasets