Molecular Property Prediction on Tox21
90.7ROC AUCMolSight
Evaluation Results
| Method | Links | |
|---|---|---|
| MolSightInput Modality=2D image, Training Curriculum=S62026.05 | 90.7 | |
| MvMRL2026.03 | 84.5 | |
| MPNN2026.03 | 84.4 | |
| N-Gram2026.03 | 84.2 | |
| GROVER2026.03 | 84.2 | |
| DMMRL2026.03 | 84.2 | |
| SGGRL2026.03 | 83.7 | |
| GCN2026.03 | 82.9 | |
| Weave2026.03 | 82 | |
| FP-GNN2026.03 | 81.5 | |
| Transformer2026.03 | 81.3 | |
| GraSeq2026.03 | 81 | |
| RNN2026.03 | 80.6 | |
| ChemATP2025.12 | 80.23 | |
| GROVERModel Type=Graph-based2026.06 | 80 | |
| UniMol2025.12 | 79.6 | |
| MolCLRCMPNNModel Type=Graph-based2026.06 | 78.4 | |
| MOLEBLENDInput Modality=Molecular graph2026.05 | 77.8 | |
| MOLINTERACTInput Modality=Molecular graph2026.05 | 77.3 | |
| Uni-MolInput Modality=Molecular graph, uses ground-truth 3D conformers=true2026.05 | 77 | |
| MoleculeSDEInput Modality=Molecular graph2026.05 | 76.8 | |
| AttrMaskingPre-training dataset=ZINC15, Protocol=Fine-tuning2022.05 | 76.7 | |
| AttrMaskingModel=AttrMasking2026.04 | 76.7 | |
| MolCLRGINModel Type=Graph-based2026.06 | 76.7 | |
| SchNetModel Type=Graph-based2026.06 | 76.6 | |
| AD-GCLModel=AD-GCL2026.04 | 76.54 | |
| CHCLModel=CHCL2026.04 | 76.06 | |
| EdgePredModel=EdgePred2026.04 | 76 | |
| GraphMVP2025.12 | 75.9 | |
| GraphormerType=Graph-based / Full Supervision2026.07 | 75.9 | |
| ContextPredPre-training dataset=ZINC15, Protocol=Fine-tuning2022.05 | 75.7 | |
| GraphLoGPre-training dataset=ZINC15, Protocol=Fine-tuning2022.05 | 75.7 | |
| ContextPredModel=ContextPred2026.04 | 75.7 | |
| AutoGCLModel=AutoGCL2026.04 | 75.69 | |
| MoMu2025.12 | 75.6 | |
| GraphMAEPre-training dataset=ZINC15, Protocol=Fine-tuning2022.05 | 75.5 | |
| GATType=Graph-based / Full Supervision2026.07 | 75.4 | |
| InfomaxPre-training dataset=ZINC15, Protocol=Fine-tuning2022.05 | 75.3 | |
| InfomaxModel=Infomax2026.04 | 75.3 | |
| JOAOPre-training dataset=ZINC15, Protocol=Fine-tuning2022.05 | 75 | |
| JOAOModel=JOAO2026.04 | 74.98 | |
| GMCLModel=GMCL2026.04 | 74.94 | |
| GCNType=Graph-based / Full Supervision2026.07 | 74.9 | |
| LLM4SD2025.12 | 74.7 | |
| InstructMol2025.12 | 74.7 | |
| GraphMVPInput Modality=Molecular graph2026.05 | 74.5 | |
| 3D InfoMaxInput Modality=Molecular graph2026.05 | 74.5 | |
| No-pretrainPre-training dataset=None, Protocol=Supervised2022.05 | 74.3 | |
| No PretrainModel=No Pretrain2026.04 | 74 | |
| GINType=Graph-based / Full Supervision2026.07 | 74 | |
| GraphCLPre-training dataset=ZINC15, Protocol=Fine-tuning2022.05 | 73.9 | |
| GraphCLModel=GraphCL2026.04 | 73.87 | |
| MolCLRGCNModel Type=Graph-based2026.06 | 73.5 | |
| MGCNModel Type=Graph-based2026.06 | 70.7 | |
| ChemDFM14BStrategy=+MOLE-RAG2026.06 | 68.8 | |
| GPT-4oStrategy=+MOLE-RAG2026.06 | 67 | |
| MolRAG2025.12 | 66.42 | |
| Qwen34B-InstStrategy=+MOLE-RAG2026.06 | 66.2 | |
| GIMLETType=Pre-training2026.07 | 61.2 | |
| CoTPrompting Strategy=Chain-of-Thought2025.12 | 59.58 | |
| GPT-4o-miniStrategy=+MOLE-RAG2026.06 | 59.3 | |
| MolBasicShot=2-shot2026.07 | 57.9 | |
| MoMuType=Pre-training2026.07 | 57.6 | |
| MolBasicShot=4-shot2026.07 | 57.3 | |
| DAPrompting Strategy=Direct Answer2025.12 | 56.85 | |
| MolRAGShot=4-shot2026.07 | 56.6 | |
| MolBasicShot=0-shot2026.07 | 56.3 | |
| MolBasicShot=1-shot2026.07 | 56.1 | |
| ChemDFM14BStrategy=Baseline2026.06 | 55 | |
| MolRAGShot=2-shot2026.07 | 55 | |
| Mistral7B-InstStrategy=+MOLE-RAG2026.06 | 54.8 | |
| MolRAGShot=1-shot2026.07 | 54.5 | |
| MolRAGShot=0-shot2026.07 | 53.4 | |
| Mistral7B-InstStrategy=Baseline2026.06 | 53.1 | |
| GPT-4oStrategy=Baseline2026.06 | 52.7 | |
| Llama3.2-3BStrategy=+MOLE-RAG2026.06 | 52.5 | |
| GPT-4o-miniStrategy=Baseline2026.06 | 52.3 | |
| Llama3.2-3BStrategy=Baseline2026.06 | 50 | |
| Qwen34B-InstStrategy=Baseline2026.06 | 49.6 | |
| Galactica-1.3BType=Pre-training2026.07 | 49.5 | |
| KVPLMType=Pre-training2026.07 | 49.2 |