Skip to content

Model for formation_energy_peratom

  • Description: This is a benchmark to evaluate how accurately an AI model can predict the formation energy per atom using the JARVIS-DFT (dft_3d) dataset. The dataset contains different types of chemical formula and atomic structures. Here we use mean absolute error (MAE) to compare models with respect to DFT accuracy.


Reference(s): https://www.nature.com/articles/s41524-023-01012-9;https://hackingmaterials.lbl.gov/matminer, https://hackingmaterials.lbl.gov/matminer/, https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.120.145301, https://www.nature.com/articles/s41524-021-00650-1, https://github.com/divelab/AIRS/tree/main/OpenMat/Matformer, https://doi.org/10.1103/PhysRevMaterials.2.083801, https://doi.org/10.48550/arXiv.2305.11842, https://www.nature.com/articles/s41524-020-00440-1, https://github.com/aimat-lab/gcnn_keras, https://github.com/divelab/AIRS/tree/main/OpenMat/PotNet

Model benchmarks

Model nameDataset MAE Team name Dataset size Date submitted Notes
matformer_256dft_3d0.0322DIVE@TAMU5571306-01-2023CSV, JSON, run.sh, Info
kgcnn_schnetdft_3d0.0365kgcnn5571309-26-2023CSV, JSON, run.sh, Info
kgcnn_coNGNdft_3d0.0291kgcnn5571305-06-2023CSV, JSON, run.sh, Info
matminer_xgboostdft_3d0.0734UofT5571305-22-2023CSV, JSON, run.sh, Info
alignn_modeldft_3d0.0331ALIGNN5571301-14-2023CSV, JSON, run.sh, Info
potnetdft_3d0.0293DIVE@TAMU5571306-02-2023CSV, JSON, run.sh, Info
kgcnn_megnetdft_3d0.0423kgcnn5571305-06-2023CSV, JSON, run.sh, Info
cfid_chemdft_3d0.2226JARVIS5571301-14-2023CSV, JSON, run.sh, Info
matminer_lgbmdft_3d0.1023Matminer5571301-14-2023CSV, JSON, run.sh, Info
matminer_rfdft_3d0.096UofT5571305-22-2023CSV, JSON, run.sh, Info
kgcnn_coGNdft_3d0.0271kgcnn5571305-06-2023CSV, JSON, run.sh, Info
cgcnn_modeldft_3d0.0625CGCNN5571301-14-2023CSV, JSON, run.sh, Info
cfiddft_3d0.1419JARVIS5571301-14-2023CSV, JSON, run.sh, Info
kgcnn_dimenetPPdft_3d0.0528kgcnn5571305-06-2023CSV, JSON, run.sh, Info
kgcnn_cgcnndft_3d0.0551kgcnn5571309-26-2023CSV, JSON, run.sh, Info