Molecular Property Prediction (Regression) on MoleculeNet (test)
0.565ESOL ErrorMM-Deacon
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| MM-DeaconPre-training modalities=SMILES and IUPAC, Pre-training molecules=10M2021.09 | 0.565 | 0.926 | 0.65 | — | |
| X-MolPre-training molecules=1.1B2021.09 | 0.578 | 1.108 | 0.596 | — | |
| MPNN2021.09 | 0.58 | 1.15 | 0.719 | — | |
| Weave2021.09 | 0.61 | 1.22 | 0.715 | — | |
| MLM-[CLS]Pre-training molecules=10M, Fine-tuning=[CLS] token2021.09 | 0.64 | 1.21 | 0.804 | — | |
| GNNHyperparameter Selection=Best validation score, Baselines Abstracted=GCN, GIN, GAT, GraphSAGE2026.01 | 0.71 | 1.58 | 0.52 | — | |
| GNNHyperparameter selection=best on test dataset, Aggregation=10-run mean, Model selection=best-performing among GCN, GIN, GAT, GraphSAGE2026.01 | 0.71 | 1.62 | 0.52 | — | |
| HIMPHyperparameter selection=best on test dataset, Aggregation=10-run mean2026.01 | 0.79 | 1.82 | 0.54 | — | |
| HIMPHyperparameter Selection=Best validation score2026.01 | 0.8 | 1.77 | 0.52 | — | |
| XIMPHyperparameter Selection=Best validation score2026.01 | 0.82 | 1.83 | 0.52 | — | |
| XIMPHyperparameter selection=best on test dataset, Aggregation=10-run mean2026.01 | 0.83 | 1.77 | 0.52 | — | |
| GLACIER (MiniMol)Architecture Category=Multimodal Student-Teacher, Teacher Configuration=MiniMol, Evaluation Protocol=downstream fingerprinting with frozen embeddings2026.06 | 0.831 | — | 0.78 | 0.806 | |
| GLACIER (Mi-Mo)Architecture Category=Multimodal Student-Teacher, Teacher Configuration=MiniMol and MolFormer, Evaluation Protocol=downstream fingerprinting with frozen embeddings2026.06 | 0.873 | — | 0.795 | 0.834 | |
| GLACIER (MolFormer)Architecture Category=Multimodal Student-Teacher, Teacher Configuration=MolFormer, Evaluation Protocol=downstream fingerprinting with frozen embeddings2026.06 | 0.939 | — | 0.913 | 0.926 | |
| GC2021.09 | 0.97 | 1.4 | 0.655 | — | |
| GraphMVP2021.09 | 1.029 | — | 0.681 | — | |
| COATIArchitecture Category=Multimodal Models, Evaluation Protocol=downstream fingerprinting with frozen embeddings2026.06 | 1.038 | — | 0.924 | 0.981 | |
| RF2021.09 | 1.07 | 2.03 | 0.876 | — | |
| Multitask2021.09 | 1.12 | 1.87 | 0.859 | — | |
| CL-MFAPArchitecture Category=Multimodal Models, Evaluation Protocol=downstream fingerprinting with frozen embeddings2026.06 | 1.134 | — | 1.059 | 1.096 | |
| ECFPHyperparameter selection=best on test dataset, Aggregation=10-run mean2026.01 | 1.16 | 2.83 | 0.73 | — | |
| ECFPHyperparameter Selection=Best validation score2026.01 | 1.21 | 2.94 | 0.73 | — | |
| RMAT-4MArchitecture Category=Multimodal Models, Evaluation Protocol=downstream fingerprinting with frozen embeddings2026.06 | 1.305 | — | 1.029 | 1.167 | |
| MiniMolArchitecture Category=Graph Neural Networks, Evaluation Protocol=downstream fingerprinting with frozen embeddings2026.06 | 1.328 | — | 0.751 | 1.04 | |
| ChemGPT-19MArchitecture Category=Text Transformers, Evaluation Protocol=downstream fingerprinting with frozen embeddings2026.06 | 1.354 | — | 1.02 | 1.187 | |
| ChemFM-1BArchitecture Category=Text Transformers, Evaluation Protocol=downstream fingerprinting with frozen embeddings2026.06 | 1.762 | — | 1.406 | 1.584 | |
| ChemBERTa-100MArchitecture Category=Text Transformers, Evaluation Protocol=downstream fingerprinting with frozen embeddings2026.06 | 1.808 | — | 1.069 | 1.439 | |
| MolFormerArchitecture Category=Text Transformers, Evaluation Protocol=downstream fingerprinting with frozen embeddings2026.06 | 1.866 | — | 0.901 | 1.383 | |
| GIT-MolArchitecture Category=Multimodal Models, Evaluation Protocol=downstream fingerprinting with frozen embeddings2026.06 | 2.336 | — | 1.109 | 1.722 | |
| ChemeleonArchitecture Category=Graph Neural Networks, Evaluation Protocol=downstream fingerprinting with frozen embeddings2026.06 | 2.417 | — | 1.416 | 1.916 |