Molecular Property Prediction on MoleculeNet (test)
96.2BBBPX-Mol
Evaluation results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| X-MolPre-training molecules=1.1B2021.09 | 96.2 | 98.4 | 79.8 | — | — | — | — | — | — | — | |
| C-FREEPaiNN-3CMethod Category=MULTI, Backbone Encoder=PaiNN, Number of Conformers=3, Pre-training Dataset=GEOM, Evaluation Protocol=Full Fine-tuning2025.09 | 88.9 | 87.6 | 81.4 | 64.8 | 82.3 | 86.2 | 71.6 | 79.81 | 75.7 | — | |
| C-FREESch-1CMethod Category=MULTI, Backbone Encoder=SchNet, Number of Conformers=1, Pre-training Dataset=GEOM, Evaluation Protocol=Full Fine-tuning2025.09 | 88.6 | 73.1 | 78.9 | 61.8 | 78.8 | 84.7 | 71.3 | 76.63 | 75.9 | — | |
| C-FREESch-3CMethod Category=MULTI, Backbone Encoder=SchNet, Number of Conformers=3, Pre-training Dataset=GEOM, Evaluation Protocol=Full Fine-tuning2025.09 | 78.9 | 83.7 | 77.9 | 62.5 | 78.6 | 84.2 | 71.7 | 77.5 | 82.5 | — | |
| MM-DeaconPre-training modalities=SMILES and IUPAC, Pre-training molecules=10M2021.09 | 78.5 | 99.5 | 80.1 | 69.3 | — | — | — | — | — | — | |
| DMPArchitecture=12 layers Transformer + 12-layer GNN2021.09 | 78.1 | 95 | 81 | 69.2 | — | — | — | — | — | — | |
| MolCLR2021.09 | 73.6 | 93.2 | 80.6 | 68 | — | — | — | — | — | — | |
| MOLEBLENDMethod Category=MULTI, Pre-training Dataset=PCQM4Mv2, Evaluation Protocol=Full Fine-tuning2025.09 | 73 | 87.6 | 79 | 64.9 | 83.7 | 77.8 | 66.1 | 76.16 | 77.2 | — | |
| KernelSVM2021.09 | 72.9 | 66.9 | 79.2 | 68.2 | — | — | — | — | — | — | |
| MOLFMMethod Category=MULTI, Pre-training Dataset=PCQM4Mv2, Evaluation Protocol=Full Fine-tuning2025.09 | 72.9 | 79.7 | 78.8 | 64.2 | 83.9 | 77.2 | 64.4 | 74.62 | 76 | — | |
| UNIMOLMethod Category=MULTI, Pre-training Dataset=PCQM4Mv2, Evaluation Protocol=Full Fine-tuning2025.09 | 72.9 | 91.9 | 80.8 | 65.9 | 85.7 | 79.6 | 69.6 | 78.56 | 82.1 | — | |
| GRAPHLOGMethod Category=CL, Pre-training Dataset=PCQM4Mv2, Evaluation Protocol=Full Fine-tuning2025.09 | 72.5 | 76.7 | 77.8 | 61.2 | 83.5 | 75.7 | 63.5 | 73.4 | 76 | — | |
| GraphMVP2021.09 | 72.4 | 77.5 | 77 | 63.9 | — | — | — | — | — | — | |
| ChemRL-GEMPre-training=true2021.06 | 72.4 | 90.1 | — | 67.2 | 85.6 | 78.1 | 69.2 | 77.1 | — | — | |
| GEMMethod Category=MULTI, Pre-training Dataset=PCQM4Mv2, Evaluation Protocol=Full Fine-tuning2025.09 | 72.4 | 90.1 | 80.6 | 67.2 | 85.6 | 78.1 | 69.2 | 78.11 | 81.7 | — | |
| GRAPHMAEMethod Category=NON-CL, Pre-training Dataset=PCQM4Mv2, Evaluation Protocol=Full Fine-tuning2025.09 | 72 | 82.3 | 77.2 | 60.3 | 83.1 | 75.5 | 64.1 | 73.85 | 76.3 | — | |
| MOLE-BERTMethod Category=NON-CL, Pre-training Dataset=PCQM4Mv2, Evaluation Protocol=Full Fine-tuning2025.09 | 71.9 | 78.9 | 78.2 | 62.8 | 80.8 | 76.8 | 64.3 | 74.04 | 78.6 | — | |
| MOLECULESDEMethod Category=MULTI, Pre-training Dataset=PCQM4Mv2, Evaluation Protocol=Full Fine-tuning2025.09 | 71.8 | 87 | 78.8 | 60.8 | 79.5 | 76.8 | 65 | 75.07 | 80.9 | — | |
| RF2021.09 | 71.4 | 71.3 | 78.1 | 68.4 | — | — | — | — | — | — | |
| D-MPNNPre-training=false2021.06 | 71 | 90.6 | — | 57 | 80.9 | 75.9 | 65.5 | 73.5 | — | — | |
| Hu et al.year=20192021.09 | 70.8 | 78.9 | 80.2 | 65.2 | — | — | — | — | — | — | |
| MLM-[CLS]Pre-training molecules=10M, Fine-tuning=[CLS] token2021.09 | 70.6 | 93.2 | 77.9 | 64.8 | — | — | — | — | — | — | |
| GROVERPre-training=true, Network capacity=base2021.06 | 70 | 81.2 | — | 64.8 | 82.6 | 74.3 | 65.4 | 73 | — | — | |
| GROVERMethod Category=CL, Pre-training Dataset=PCQM4Mv2, Evaluation Protocol=Full Fine-tuning2025.09 | 70 | 81.2 | 62.5 | 64.8 | 82.6 | 74.3 | 65.4 | 71.01 | 67.3 | — | |
| N-GramPre-training=true, Downstream model=Random Forest2021.06 | 69.7 | 77.5 | — | 66.8 | 77.9 | 74.3 | — | — | — | — | |
| GRAPHCLMethod Category=CL, Pre-training Dataset=PCQM4Mv2, Evaluation Protocol=Full Fine-tuning2025.09 | 69.7 | 76 | 78.5 | 60.5 | 75.4 | 73.9 | 62.4 | 70.78 | 69.8 | — | |
| MGSSLMethod Category=NON-CL, Pre-training Dataset=PCQM4Mv2, Evaluation Protocol=Full Fine-tuning2025.09 | 69.7 | 80.7 | 78.8 | 61.8 | 79.1 | 76.5 | 64.1 | 73.7 | 78.7 | — | |
| GROVERPre-training=true, Network capacity=large2021.06 | 69.5 | 76.2 | — | 65.4 | 81 | 73.5 | 65.3 | 71.8 | — | — | |
| N-GramPre-training=true, Downstream model=XGBoost2021.06 | 69.1 | 87.5 | — | 65.5 | 79.1 | 75.8 | — | — | — | — | |
| 3D INFOMAXMethod Category=MULTI, Pre-training Dataset=PCQM4Mv2, Evaluation Protocol=Full Fine-tuning2025.09 | 69.1 | 79.9 | 76.1 | 60.6 | 79.7 | 74.5 | 64.4 | 72.34 | 74.4 | — | |
| GC2021.09 | 69 | 80.7 | 76.3 | 63.8 | — | — | — | — | — | — | |
| Multitask2021.09 | 68.8 | 77.8 | 69.8 | 66.6 | — | — | — | — | — | — | |
| PretrainGNNPre-training=true2021.06 | 68.7 | 72.6 | — | 62.7 | 84.5 | 78.1 | 65.7 | 72.1 | — | — | |
| GRAPHMVPMethod Category=MULTI, Pre-training Dataset=PCQM4Mv2, Evaluation Protocol=Full Fine-tuning2025.09 | 68.5 | 79 | 74.8 | 62.3 | 76.8 | 74.5 | 62.7 | 71.69 | 75 | — | |
| INFOGRAPHMethod Category=CL, Pre-training Dataset=PCQM4Mv2, Evaluation Protocol=Full Fine-tuning2025.09 | 67.5 | 76.5 | 75.1 | 59.9 | 77.8 | 73.2 | 63.7 | 70.98 | 74.1 | — | |
| Weave2021.09 | 67.1 | 83.2 | 70.3 | 58.1 | — | — | — | — | — | — | |
| MOLCLRMethod Category=CL, Pre-training Dataset=PCQM4Mv2, Evaluation Protocol=Full Fine-tuning2025.09 | 66.6 | 86.1 | 76.2 | 57.5 | 71.5 | 73 | 62.9 | 70.79 | 72.5 | — | |
| CONTEXTPREDMethod Category=NON-CL, Pre-training Dataset=PCQM4Mv2, Evaluation Protocol=Full Fine-tuning2025.09 | 65.7 | 77.2 | 77.1 | 62.2 | 76 | 74.2 | 62.5 | 71.28 | 75.3 | — | |
| ATTRMASKMethod Category=NON-CL, Pre-training Dataset=PCQM4Mv2, Evaluation Protocol=Full Fine-tuning2025.09 | 65 | 87.7 | 76.8 | 61.2 | 79.7 | 74.8 | 62.9 | 72.68 | 73.4 | — | |
| AttentiveFPPre-training=false2021.06 | 64.3 | 84.7 | — | 60.6 | 78.4 | 76.1 | 63.7 | 71.3 | — | — | |
| DataImprovement Axis=Data, Number of endpoints (n)=102026.06 | — | — | — | — | — | — | — | — | — | 0.1 | |
| FeatureImprovement Axis=Feature, Number of endpoints (n)=102026.06 | — | — | — | — | — | — | — | — | — | 1 | |
| ModelImprovement Axis=Model, Number of endpoints (n)=102026.06 | — | — | — | — | — | — | — | — | — | 2.3 | |
| RoutedImprovement Axis=Routed, Number of endpoints (n)=102026.06 | — | — | — | — | — | — | — | — | — | 1.1 |