Molecular property prediction on BACE
92.5ROC-AUCDMMRL
Evaluation Results
| Method | Links | |
|---|---|---|
| DMMRL2026.03 | 92.5 | |
| SGGRL2026.03 | 91.7 | |
| ChemATP2025.12 | 89.7 | |
| MvMRL2026.03 | 89.1 | |
| MolSightInput Modality=2D image, Training Curriculum=S62026.05 | 89 | |
| FP-GNN2026.03 | 88.1 | |
| UniMol2025.12 | 85.7 | |
| CHCLModel=CHCL2026.04 | 85.43 | |
| MolCLRCMPNNModel Type=Graph-based2026.06 | 85 | |
| TrimNet2026.03 | 84.1 | |
| LLM4SD2025.12 | 83.83 | |
| MOLEBLENDInput Modality=Molecular graph2026.05 | 83.7 | |
| MolCLRGINModel Type=Graph-based2026.06 | 83.5 | |
| GROVERModel Type=Graph-based2026.06 | 83.3 | |
| AutoGCLModel=AutoGCL2026.04 | 83.26 | |
| MolCLRGCNModel Type=Graph-based2026.06 | 82.6 | |
| GREAEncoder=GIN [37]2022.06 | 82.37 | |
| GMCLModel=GMCL2026.04 | 82.37 | |
| InstructMol2025.12 | 82.3 | |
| GREAEncoder=GCN [12]2022.06 | 81.91 | |
| GraphMVP2025.12 | 81.2 | |
| OOD-GNNEncoder=GCN [12]2022.06 | 80.96 | |
| ChemDFM14BStrategy=+MOLE-RAG2026.06 | 80.4 | |
| U-NETSPOOLEncoder=GIN [37]2022.06 | 80.26 | |
| GREA-REPAUGEncoder=GIN [37]2022.06 | 79.97 | |
| EdgePredModel=EdgePred2026.04 | 79.9 | |
| IRMEncoder=GIN [37]2022.06 | 79.77 | |
| 3D InfoMaxInput Modality=Molecular graph2026.05 | 79.7 | |
| ContextPredModel=ContextPred2026.04 | 79.6 | |
| MoleculeSDEInput Modality=Molecular graph2026.05 | 79.5 | |
| IRMEncoder=GCN [12]2022.06 | 79.47 | |
| AttrMaskingModel=AttrMasking2026.04 | 79.3 | |
| MOLINTERACTInput Modality=Molecular graph2026.05 | 79.1 | |
| MolEvolveBase Model=GPT-4o2026.03 | 78.75 | |
| MolEvolveBase Model=GPT-4o-mini2026.03 | 78.54 | |
| AD-GCLModel=AD-GCL2026.04 | 78.51 | |
| GPT-4oStrategy=+MOLE-RAG2026.06 | 78.3 | |
| OOD-GNNEncoder=GIN [37]2022.06 | 78 | |
| Uni-MolInput Modality=Molecular graph, uses ground-truth 3D conformers=true2026.05 | 78 | |
| MolRAGBase Model=GPT-4o-mini2026.03 | 77.62 | |
| GraphormerType=Graph-based / Full Supervision2026.07 | 77.6 | |
| U-NETSPOOLEncoder=GCN [12]2022.06 | 77.57 | |
| JOAOModel=JOAO2026.04 | 77.34 | |
| MolEvolveBase Model=Qwen2.5-7B-Instruct2026.03 | 77.05 | |
| GraphformerBase Model=GNNs2026.03 | 76.95 | |
| GraphMVPInput Modality=Molecular graph2026.05 | 76.8 | |
| Mistral7B-InstStrategy=+MOLE-RAG2026.06 | 76.8 | |
| DIR+REPAUGEncoder=GCN [12]2022.06 | 76.77 | |
| MolRAG2025.12 | 76.73 | |
| MoMu2025.12 | 76.7 | |
| MolRAGBase Model=GPT-4o2026.03 | 76.63 | |
| SchNetModel Type=Graph-based2026.06 | 76.6 | |
| GREA-REPAUGEncoder=GCN [12]2022.06 | 76.55 | |
| DIR+REPAUGEncoder=GIN [37]2022.06 | 75.9 | |
| InfomaxModel=Infomax2026.04 | 75.9 | |
| GraphCLModel=GraphCL2026.04 | 75.38 | |
| AutomolcoBase Model=GPT-4o-mini2026.03 | 73.9 | |
| SELFATTNPOOLEncoder=GCN [12]2022.06 | 73.83 | |
| GCNType=Graph-based / Full Supervision2026.07 | 73.6 | |
| SELFATTNPOOLEncoder=GIN [37]2022.06 | 73.48 | |
| MGCNModel Type=Graph-based2026.06 | 73.4 | |
| GINBase Model=GNNs2026.03 | 73.26 | |
| AutomolcoBase Model=GPT-4o2026.03 | 73.1 | |
| STABLEGNNEncoder=GIN [37]2022.06 | 72.29 | |
| Qwen34B-InstStrategy=+MOLE-RAG2026.06 | 71.1 | |
| No PretrainModel=No Pretrain2026.04 | 70.1 | |
| GINType=Graph-based / Full Supervision2026.07 | 70.1 | |
| GATType=Graph-based / Full Supervision2026.07 | 69.7 | |
| MolRAGBase Model=Qwen2.5-7B-Instruct2026.03 | 69.68 | |
| GIMLETType=Pre-training2026.07 | 69.6 | |
| ChemDFM14BStrategy=Baseline2026.06 | 68.2 | |
| CoTPrompting Strategy=Chain-of-Thought2025.12 | 67.75 | |
| DIREncoder=GCN [12]2022.06 | 67.51 | |
| AutomolcoBase Model=Qwen2.5-7B-Instruct2026.03 | 67.16 | |
| MoMuType=Pre-training2026.07 | 66.6 | |
| MolBasicShot=4-shot2026.07 | 66.1 | |
| STABLEGNNEncoder=GCN [12]2022.06 | 66.07 | |
| MolBasicShot=2-shot2026.07 | 64.9 | |
| DAPrompting Strategy=Direct Answer2025.12 | 64.28 | |
| XGBoostBase Model=Tree-based model2026.03 | 64.28 | |
| Decision TreeBase Model=Tree-based model2026.03 | 63.52 | |
| Logistic Regression (LR)Base Model=Linear model2026.03 | 63.34 | |
| MolRAGShot=4-shot2026.07 | 62.6 | |
| MolBasicShot=1-shot2026.07 | 61.9 | |
| MolRAGShot=2-shot2026.07 | 61.5 | |
| DIREncoder=GIN [37]2022.06 | 61.15 | |
| ATTRILENS-MOLBase Model=Qwen2.5-7B-Instruct2026.03 | 61.02 | |
| MolRAGShot=1-shot2026.07 | 59.4 | |
| GPT-4o-miniStrategy=+MOLE-RAG2026.06 | 57.4 | |
| Galactica-1.3BType=Pre-training2026.07 | 56.5 | |
| GPT-4oStrategy=Baseline2026.06 | 54.3 | |
| Llama3.2-3BStrategy=+MOLE-RAG2026.06 | 51.8 | |
| GPT-4o-miniStrategy=Baseline2026.06 | 51.3 | |
| KVPLMType=Pre-training2026.07 | 51.3 | |
| Llama3.2-3BStrategy=Baseline2026.06 | 50.4 | |
| MolRAGShot=0-shot2026.07 | 50.4 | |
| MolBasicShot=0-shot2026.07 | 50 | |
| Qwen34B-InstStrategy=Baseline2026.06 | 48.9 | |
| Mistral7B-InstStrategy=Baseline2026.06 | 48.6 | |
| GraphLoGPre-training dataset=ZINC15, Protocol=Fine-tuning2022.05 | 0.835 |