Molecular Property Classification on MoleculeNet BBBP
89.25ROC AUCMolReFlect
Evaluation Results
| Method | Links | |
|---|---|---|
| MolReFlect2024.11 | 89.25 | |
| Mol-LLM (w/o Graph)Model Category=Generalist2025.02 | 84.3 | |
| LlaSMolModel Category=Generalist, Official Checkpoint=true2025.02 | 82.4 | |
| Mol-LLMModel Category=Generalist2025.02 | 81.1 | |
| MolXPTModel Category=Specialist2025.02 | 80 | |
| SPMMPre-training=true, Fine-tuned=true2022.11 | 74.9 | |
| MolEvolveBase Model=Qwen2.5-7B-Instruct2026.03 | 74.61 | |
| MolFormerFine-tuned=true2022.11 | 73.6 | |
| MolEvolveBase Model=GPT-4o2026.03 | 72.91 | |
| Uni-MolPre-training status=W. pre-training2026.02 | 72.9 | |
| ChemBERTa-2Fine-tuned=true2022.11 | 72.8 | |
| MolEvolveBase Model=GPT-4o-mini2026.03 | 72.68 | |
| InstructMolModel Category=Specialist2025.02 | 72.4 | |
| ChemRL-GEMFine-tuned=true2022.11 | 72.4 | |
| D-MPNNFine-tuned=true2022.11 | 71 | |
| D-MPNNPre-training status=W/o. pre-training2026.02 | 71 | |
| GraphformerBase Model=GNNs2026.03 | 70.12 | |
| MolCAModel Category=Specialist2025.02 | 70 | |
| N-GramRFFine-tuned=true2022.11 | 69.7 | |
| GINBase Model=GNNs2026.03 | 69.53 | |
| GROVER_largeFine-tuned=true2022.11 | 69.5 | |
| HyperDrop + GMTArchitecture Type=Ours, Evaluation Protocol=Standard Split2021.06 | 69.16 | |
| N-GramXGBFine-tuned=true2022.11 | 69.1 | |
| HyperDropArchitecture Type=Ours, Evaluation Protocol=Standard Split2021.06 | 69.04 | |
| ChiDeKPre-training status=W/o. pre-training2026.02 | 68.9 | |
| PretrainGNNFine-tuned=true2022.11 | 68.7 | |
| PretrainGNNPre-training status=W. pre-training2026.02 | 68.7 | |
| XGBoostBase Model=Tree-based model2026.03 | 68.64 | |
| GMTArchitecture Type=Global, Evaluation Protocol=Standard Split2021.06 | 68.31 | |
| DiffPoolArchitecture Type=Hierarchical, Evaluation Protocol=Standard Split2021.06 | 68.25 | |
| ICMA2024.11 | 67.75 | |
| GINArchitecture Type=Naive GNN, Evaluation Protocol=Standard Split2021.06 | 67.65 | |
| Logistic Regression (LR)Base Model=Linear model2026.03 | 67.27 | |
| EdgePoolArchitecture Type=Hierarchical, Evaluation Protocol=Standard Split2021.06 | 67.18 | |
| SPMM (w/o pre-train)Pre-training=false, Fine-tuned=true2022.11 | 66.6 | |
| AutomolcoBase Model=GPT-4o2026.03 | 66.37 | |
| SortPoolArchitecture Type=Global, Evaluation Protocol=Standard Split2021.06 | 65.98 | |
| MinCutPoolArchitecture Type=Hierarchical, Evaluation Protocol=Standard Split2021.06 | 65.97 | |
| GCNArchitecture Type=Naive GNN, Evaluation Protocol=Standard Split2021.06 | 65.47 | |
| TopKPoolArchitecture Type=Hierarchical, Evaluation Protocol=Standard Split2021.06 | 65.19 | |
| BioT5+Model Category=Semi-Generalist, Training Task Group=Cls. & Trans., Official Checkpoint=true2025.02 | 65.1 | |
| AutomolcoBase Model=GPT-4o-mini2026.03 | 64.35 | |
| Attentive FPPre-training status=W/o. pre-training2026.02 | 64.3 | |
| SAGPoolArchitecture Type=Hierarchical, Evaluation Protocol=Standard Split2021.06 | 63.94 | |
| MolRAGBase Model=GPT-4o-mini2026.03 | 63.92 | |
| DeepSetArchitecture Type=Set, Evaluation Protocol=Standard Split2021.06 | 63.64 | |
| ASAPArchitecture Type=Hierarchical, Evaluation Protocol=Standard Split2021.06 | 63.5 | |
| MolRAGBase Model=GPT-4o2026.03 | 62.71 | |
| GPT-4Model Category=Generalist, Evaluation Setup=zero-shot2025.02 | 61.5 | |
| AutomolcoBase Model=Qwen2.5-7B-Instruct2026.03 | 61.02 | |
| Decision TreeBase Model=Tree-based model2026.03 | 60.37 | |
| ATTRILENS-MOLBase Model=Qwen2.5-7B-Instruct2026.03 | 58.1 | |
| Mol-InstructionsModel Category=Semi-Generalist, Official Checkpoint=true2025.02 | 58 | |
| MolRAGBase Model=Qwen2.5-7B-Instruct2026.03 | 56.08 | |
| 3D-MOLMModel Category=Generalist, Official Checkpoint=true2025.02 | 53.8 | |
| GalacticaModel Category=Generalist2025.02 | 53.5 | |
| BioT5+Model Category=Semi-Generalist, Training Task Group=Reg. & React., Official Checkpoint=true2025.02 | 51.5 | |
| ChemDFMModel Category=Generalist, Official Checkpoint=true2025.02 | 50.5 | |
| Mistral-7B2024.11 | 48.29 |