ConfigurationStabilityPredictionML

Trained model parameters [TrainedParams.zip] - RandomSeed_[random_seed]/[model_name]/Model_[num_dat]_ [id,ood_ads,ood_cat,ood_both].pth.tar - [random_seed]: 1111,2222,3333,4444,5555 (five independent trials for random extraction of training set) - [model_name]: CGCNN,SchNet,MPNN,MPNN_A,Matforer - [num_dat]: 10000(2%),46032(10%),115082(25%),230164(50%),345246(75%) *Note: Trained model parameters with whole training dataset (#: 460328) can be found in RandomSeed_1111/[model_name]/Model_460328_[id,ood_ads,ood_cat,ood_both].pth.tar WholeTrial_[2/3/4/5]/[model_name]/Model_460328_[id,ood_ads,ood_cat,ood_both].pth.tar DSTAR-feature dataset for the OC20 dataset [DSTAR_FeatureData.zip] - TrainAds.pkl: training set data - ValAds_[id,ood_ads,ood_cat,ood_both].pkl: validation set data - TestAds_[id,ood_ads,ood_cat,ood_both].pkl: test set data