20230723_figshare_DIRECT

All data required to reproduce the DIRECT sampling results in the manuscript "Robust Training of Machine Learning Interatomic Potentials with Dimensionality Reduction and Stratified Sampling".It contains the below data:Training, validation, and test data as well as model parameters of M3GNet-MS and M3GNet-DIRECT universal potentials.Training and test data as well as model parameters of MTP-AL and MTP-DIRECT for Ti-H.Pre-processed M3GNet structural features of the 1.3 million structures in <i>MPF.2021.2.8.All</i> and the 274,000 M3GNet MD structures of Ti-H.