Molecular Property Prediction on ClinTox
97.5ROC AUCMvMRL
Evaluation Results
| Method | Links | |
|---|---|---|
| MvMRL2026.03 | 97.5 | |
| SGGRL2026.03 | 95.6 | |
| Transformer2026.03 | 95.4 | |
| TrimNet2026.03 | 94.8 | |
| DMMRL2026.03 | 93.5 | |
| GROVER2026.03 | 92.9 | |
| UniMol2025.12 | 91.9 | |
| GraSeq2026.03 | 91.8 | |
| InstructMol2025.12 | 91.5 | |
| RNN2026.03 | 91.5 | |
| LLM4SD2025.12 | 91.3 | |
| MolCLRGINModel Type=Graph-based2026.06 | 90.4 | |
| GREAEncoder=GCN [12]2022.06 | 89.61 | |
| OOD-GNNEncoder=GCN [12]2022.06 | 88.74 | |
| ChemATP2025.12 | 88.56 | |
| IRMEncoder=GCN [12]2022.06 | 88.19 | |
| MPNN2026.03 | 88.1 | |
| MolCLRCMPNNModel Type=Graph-based2026.06 | 88 | |
| GREAEncoder=GIN [37]2022.06 | 87.89 | |
| GREA-REPAUGEncoder=GCN [12]2022.06 | 87.08 | |
| DIR+REPAUGEncoder=GCN [12]2022.06 | 86.06 | |
| MolCLRGCNModel Type=Graph-based2026.06 | 85.9 | |
| GREA-REPAUGEncoder=GIN [37]2022.06 | 85.74 | |
| DIR+REPAUGEncoder=GIN [37]2022.06 | 85.61 | |
| STABLEGNNEncoder=GIN [37]2022.06 | 85.59 | |
| N-Gram2026.03 | 85.5 | |
| IRMEncoder=GIN [37]2022.06 | 84.85 | |
| U-NETSPOOLEncoder=GCN [12]2022.06 | 84.5 | |
| OOD-GNNEncoder=GIN [37]2022.06 | 84.16 | |
| FP-GNN2026.03 | 84 | |
| CHCLModel=CHCL2026.04 | 83.58 | |
| GMCLModel=GMCL2026.04 | 83.37 | |
| Weave2026.03 | 83.2 | |
| SELFATTNPOOLEncoder=GCN [12]2022.06 | 82.91 | |
| GraphMAEPre-training dataset=ZINC15, Protocol=Fine-tuning2022.05 | 82.3 | |
| U-NETSPOOLEncoder=GIN [37]2022.06 | 81.46 | |
| JOAOModel=JOAO2026.04 | 81.32 | |
| JOAOPre-training dataset=ZINC15, Protocol=Fine-tuning2022.05 | 81.3 | |
| AutoGCLModel=AutoGCL2026.04 | 80.99 | |
| GCN2026.03 | 80.7 | |
| MoMu2025.12 | 79.9 | |
| AD-GCLModel=AD-GCL2026.04 | 79.78 | |
| SELFATTNPOOLEncoder=GIN [37]2022.06 | 79.12 | |
| GraphMVP2025.12 | 79.1 | |
| STABLEGNNEncoder=GCN [12]2022.06 | 76.81 | |
| GraphLoGPre-training dataset=ZINC15, Protocol=Fine-tuning2022.05 | 76.7 | |
| GraphCLPre-training dataset=ZINC15, Protocol=Fine-tuning2022.05 | 76 | |
| GraphCLModel=GraphCL2026.04 | 75.99 | |
| AttrMaskingPre-training dataset=ZINC15, Protocol=Fine-tuning2022.05 | 71.8 | |
| AttrMaskingModel=AttrMasking2026.04 | 71.8 | |
| SchNetModel Type=Graph-based2026.06 | 71.7 | |
| GROVERModel Type=Graph-based2026.06 | 70.7 | |
| InfomaxPre-training dataset=ZINC15, Protocol=Fine-tuning2022.05 | 69.9 | |
| InfomaxModel=Infomax2026.04 | 69.9 | |
| DIREncoder=GIN [37]2022.06 | 69.11 | |
| ContextPredPre-training dataset=ZINC15, Protocol=Fine-tuning2022.05 | 65.9 | |
| ContextPredModel=ContextPred2026.04 | 65.9 | |
| EdgePredModel=EdgePred2026.04 | 64.1 | |
| MGCNModel Type=Graph-based2026.06 | 63.4 | |
| GPT-4o-miniStrategy=+MOLE-RAG2026.06 | 63.3 | |
| GPT-4oStrategy=+MOLE-RAG2026.06 | 62.7 | |
| DIREncoder=GCN [12]2022.06 | 62.51 | |
| Mistral7B-InstStrategy=+MOLE-RAG2026.06 | 62.1 | |
| ChemDFM14BStrategy=+MOLE-RAG2026.06 | 59.4 | |
| No-pretrainPre-training dataset=None, Protocol=Supervised2022.05 | 58.2 | |
| No PretrainModel=No Pretrain2026.04 | 58 | |
| Qwen34B-InstStrategy=+MOLE-RAG2026.06 | 56.6 | |
| GPT-4oStrategy=Baseline2026.06 | 52.8 | |
| GPT-4o-miniStrategy=Baseline2026.06 | 52 | |
| Qwen34B-InstStrategy=Baseline2026.06 | 50.3 | |
| DAPrompting Strategy=Direct Answer2025.12 | 50.15 | |
| Mistral7B-InstStrategy=Baseline2026.06 | 50 | |
| ChemDFM14BStrategy=Baseline2026.06 | 49.7 | |
| Llama3.2-3BStrategy=+MOLE-RAG2026.06 | 48.8 | |
| CoTPrompting Strategy=Chain-of-Thought2025.12 | 48.6 | |
| Llama3.2-3BStrategy=Baseline2026.06 | 48.3 |