OMat24 is a large-scale open source dataset of Open Materials 2024 released by Meta in 2024. The related paper results are "Open Materials 2024 (OMat24) Inorganic Materials Dataset and Models".

The OMat24 dataset contains more than 110 million density functional theory (DFT) calculation results focusing on structural and compositional diversity, covering different atomic configurations sampled from equilibrium and non-equilibrium structures, making it one of the largest open source datasets in this field.

The OMat24 dataset is currently the largest open source dataset for training DFT substitution models for materials. The dataset consists of DFT single-point calculations, structural relaxations, and molecular dynamic trajectories for a series of inorganic bulk materials. The researchers calculated a total of about 118 million structures annotated with total energy, forces norm, and unit cell stress, using more than 400 million core hours of computing resources.

In addition, the OMat24 dataset also provides some sub-datasets, including training sets and validation sets, each of which contains multiple sub-datasets based on different input generation strategies. The sizes of these sub-datasets range from millions to tens of millions, covering different structures and element combinations, providing researchers with rich material property data.