Skip to content

Model for Ge FF energy

  • Description: This is an AI benchmark to evaluate how accurately a machine learning force-field (MLFF) can predict the total energy of Ge using the relaxation trajectories (energy and forces of intermediate steps) of the mlearn dataset, calculated with the PBE density functional. The dataset contains different types of chemical formula and atomic structures. Here we use mean absolute error (MAE) to compare MLFFs with respect to DFT (PBE) accuracy. External links: https://github.com/materialsvirtuallab/mlearn


Reference(s): https://pubs.rsc.org/en/content/articlehtml/2023/dd/d2dd00096b, https://github.com/materialsvirtuallab/maml, https://www.nature.com/articles/s43588-022-00349-3, https://github.com/materialsvirtuallab/m3gnet, https://doi.org/10.1021/acs.jpca.9b08723, https://github.com/CederGroupHub/chgnet

Model benchmarks

Model nameDataset Accuracy Team name Dataset size Date submitted Notes
matgl_pretrainedmlearn_Ge13.9598Matgl25301-14-2023CSV, JSON, run.sh, Info
alignnff_mlearn_wt1mlearn_Ge10.9181JARVIS25301-14-2023CSV, JSON, run.sh, Info
chgnet_pretrainedmlearn_Ge9.3009CHGNET25308-07-2023CSV, JSON, run.sh, Info
m3gnet_pretrainedmlearn_Ge16.1668M3GNET25301-14-2023CSV, JSON, run.sh, Info
snap_mlearnmlearn_Ge2.4512JARVIS25301-14-2023CSV, JSON, run.sh, Info
alignnff_mlearn_all_wt1mlearn_Ge1.5227JARVIS25301-14-2023CSV, JSON, run.sh, Info