Graph Property Prediction on Tox21
0.8489ROC-AUCLEAP
Evaluation Results
| Method | Links | |
|---|---|---|
| LEAPPre-training Strategy=ContextPred, Shot Scenario=full-shot2025.12 | 0.8489 | |
| GPF-plusPre-training Strategy=ContextPred, Shot Scenario=full-shot2025.12 | 0.8463 | |
| SUPTsoftPre-training Strategy=ContextPred, Shot Scenario=full-shot2025.12 | 0.8436 | |
| SUPThardPre-training Strategy=ContextPred, Shot Scenario=full-shot2025.12 | 0.843 | |
| GPFPre-training Strategy=ContextPred, Shot Scenario=full-shot2025.12 | 0.842 | |
| RELIEFPre-training Strategy=ContextPred, Shot Scenario=full-shot2025.12 | 0.8417 | |
| LEAPPre-training Strategy=AttrMasking, Shot Scenario=full-shot2025.12 | 0.8394 | |
| GPFPre-training Strategy=AttrMasking, Shot Scenario=full-shot2025.12 | 0.8371 | |
| SUPTsoftPre-training Strategy=AttrMasking, Shot Scenario=full-shot2025.12 | 0.8371 | |
| LEAPPre-training Strategy=Infomax, Shot Scenario=full-shot2025.12 | 0.8365 | |
| SUPThardPre-training Strategy=AttrMasking, Shot Scenario=full-shot2025.12 | 0.8352 | |
| GPF-plusPre-training Strategy=AttrMasking, Shot Scenario=full-shot2025.12 | 0.8347 | |
| <SOGk>Backbone=LLaMA3-3B2026.02 | 0.834 | |
| GPF-plusPre-training Strategy=Infomax, Shot Scenario=full-shot2025.12 | 0.834 | |
| SUPThardPre-training Strategy=Infomax, Shot Scenario=full-shot2025.12 | 0.8336 | |
| GPFPre-training Strategy=Infomax, Shot Scenario=full-shot2025.12 | 0.8333 | |
| RELIEFPre-training Strategy=Infomax, Shot Scenario=full-shot2025.12 | 0.833 | |
| SUPTsoftPre-training Strategy=Infomax, Shot Scenario=full-shot2025.12 | 0.832 | |
| FTPre-training Strategy=ContextPred, Shot Scenario=full-shot2025.12 | 0.8311 | |
| FTPre-training Strategy=Infomax, Shot Scenario=full-shot2025.12 | 0.8309 | |
| FTPre-training Strategy=AttrMasking, Shot Scenario=full-shot2025.12 | 0.8304 | |
| RELIEFPre-training Strategy=AttrMasking, Shot Scenario=full-shot2025.12 | 0.8304 | |
| S-CGIBDomain Adaptation=true2024.12 | 0.8094 | |
| LR-GMPBackbone=GIN, Pre-training Method=EdgePred2026.04 | 0.8014 | |
| VGPBackbone=GIN, Pre-training Method=EdgePred2026.04 | 0.8008 | |
| SUPTsoftPre-training Strategy=GCL, Shot Scenario=full-shot2025.12 | 0.8005 | |
| GPF-plusBackbone=GIN, Pre-training Method=EdgePred2026.04 | 0.8004 | |
| LEAPPre-training Strategy=GCL, Shot Scenario=full-shot2025.12 | 0.8002 | |
| SUPThardPre-training Strategy=GCL, Shot Scenario=full-shot2025.12 | 0.7992 | |
| GPFBackbone=GIN, Pre-training Method=EdgePred2026.04 | 0.7974 | |
| S-CGIBDomain Adaptation=false2024.12 | 0.7952 | |
| GPFPre-training Strategy=GCL, Shot Scenario=full-shot2025.12 | 0.7942 | |
| GPF-plusPre-training Strategy=GCL, Shot Scenario=full-shot2025.12 | 0.7933 | |
| FTPre-training Strategy=GCL, Shot Scenario=full-shot2025.12 | 0.7905 | |
| RELIEFPre-training Strategy=GCL, Shot Scenario=full-shot2025.12 | 0.7895 | |
| GPPT w/o olBackbone=GIN, Pre-training Method=EdgePred2026.04 | 0.7891 | |
| FTBackbone=GIN, Pre-training Method=EdgePred2026.04 | 0.7867 | |
| MoAMa2024.12 | 0.7829 | |
| GraphFP2024.12 | 0.7735 | |
| SpecMatch-CLPre-Train dataset=ZINC 2M2025.11 | 0.7697 | |
| AttrMaskingPre-Train Dataset=ZINC 2M2021.06 | 0.767 | |
| AD-GCL-FIXPre-Train Dataset=ZINC 2M2021.06 | 0.7654 | |
| SimSGT2024.12 | 0.7623 | |
| EdgePredPre-Train Dataset=ZINC 2M2021.06 | 0.76 | |
| EdgePredPre-Train dataset=ZINC 2M2025.11 | 0.76 | |
| ContextPredPre-Train Dataset=ZINC 2M2021.06 | 0.757 | |
| SimGRACEPre-Train dataset=ZINC 2M2025.11 | 0.756 | |
| InfoGraphPre-Train Dataset=ZINC 2M2021.06 | 0.753 | |
| AttrMaskingPre-Train dataset=ZINC 2M2025.11 | 0.751 | |
| GraphCLPre-Train dataset=ZINC 2M2025.11 | 0.751 | |
| MGSSL2024.12 | 0.7482 | |
| JOAOPre-Train dataset=ZINC 2M2025.11 | 0.748 | |
| No Pre-TrainPre-Train dataset=None2025.11 | 0.746 | |
| No Pre-TrainPre-Train Dataset=None2021.06 | 0.74 | |
| JOAOv22024.12 | 0.7395 | |
| GraphCLPre-Train Dataset=ZINC 2M2021.06 | 0.7387 | |
| JOAO2024.12 | 0.7367 | |
| ContextPredPre-Train dataset=ZINC 2M2025.11 | 0.736 | |
| AttrMasking2024.12 | 0.7337 | |
| ContextPred2024.12 | 0.7326 | |
| GraphCL2024.12 | 0.7326 | |
| LEAPPre-training Strategy=AttrMasking, Shot Scenario=50-shot2025.12 | 0.7279 | |
| SUPThardPre-training Strategy=AttrMasking, Shot Scenario=50-shot2025.12 | 0.7274 | |
| GPF-plusPre-training Strategy=AttrMasking, Shot Scenario=50-shot2025.12 | 0.7271 | |
| Infomax2024.12 | 0.7266 | |
| SUPTsoftPre-training Strategy=AttrMasking, Shot Scenario=50-shot2025.12 | 0.7263 | |
| GPFPre-training Strategy=AttrMasking, Shot Scenario=50-shot2025.12 | 0.7255 | |
| RELIEFPre-training Strategy=AttrMasking, Shot Scenario=50-shot2025.12 | 0.7244 | |
| <SOGk>Backbone=LLaMA2-7B2026.02 | 0.724 | |
| FTPre-training Strategy=AttrMasking, Shot Scenario=50-shot2025.12 | 0.7236 | |
| LEAPPre-training Strategy=Infomax, Shot Scenario=50-shot2025.12 | 0.7185 | |
| MICRO-Graph2024.12 | 0.7179 | |
| GraphLoG2024.12 | 0.7164 | |
| RELIEFPre-training Strategy=Infomax, Shot Scenario=50-shot2025.12 | 0.7151 | |
| GPF-plusPre-training Strategy=Infomax, Shot Scenario=50-shot2025.12 | 0.715 | |
| FTPre-training Strategy=Infomax, Shot Scenario=50-shot2025.12 | 0.7141 | |
| SUPTsoftPre-training Strategy=Infomax, Shot Scenario=50-shot2025.12 | 0.714 | |
| SUPThardPre-training Strategy=Infomax, Shot Scenario=50-shot2025.12 | 0.7123 | |
| GPFPre-training Strategy=Infomax, Shot Scenario=50-shot2025.12 | 0.7119 | |
| LEAPPre-training Strategy=ContextPred, Shot Scenario=50-shot2025.12 | 0.7068 | |
| RELIEFPre-training Strategy=ContextPred, Shot Scenario=50-shot2025.12 | 0.7056 | |
| GPF-plusPre-training Strategy=ContextPred, Shot Scenario=50-shot2025.12 | 0.7048 | |
| GPFPre-training Strategy=ContextPred, Shot Scenario=50-shot2025.12 | 0.7042 | |
| SUPTsoftPre-training Strategy=ContextPred, Shot Scenario=50-shot2025.12 | 0.704 | |
| EdgePred2024.12 | 0.7032 | |
| SUPThardPre-training Strategy=ContextPred, Shot Scenario=50-shot2025.12 | 0.7031 | |
| FTPre-training Strategy=ContextPred, Shot Scenario=50-shot2025.12 | 0.7028 | |
| Galactica-120BProtocol=Zero-shot2026.02 | 0.689 | |
| GROVE2024.12 | 0.6859 | |
| MolCA-GSBackbone=Non-LLM2026.02 | 0.685 | |
| GraphPromptBackbone=GIN, Pre-training Method=EdgePred2026.04 | 0.6809 | |
| MolCA-SBackbone=Non-LLM2026.02 | 0.666 | |
| GPPTBackbone=GIN, Pre-training Method=EdgePred2026.04 | 0.6641 | |
| GPF-AttrMaskingBackbone=Non-LLM2026.02 | 0.66 | |
| GPF-GCLBackbone=Non-LLM2026.02 | 0.636 | |
| SUPTsoftPre-training Strategy=GCL, Shot Scenario=50-shot2025.12 | 0.6165 | |
| LEAPPre-training Strategy=GCL, Shot Scenario=50-shot2025.12 | 0.6154 | |
| SUPThardPre-training Strategy=GCL, Shot Scenario=50-shot2025.12 | 0.6152 | |
| RELIEFPre-training Strategy=GCL, Shot Scenario=50-shot2025.12 | 0.6149 | |
| FTPre-training Strategy=GCL, Shot Scenario=50-shot2025.12 | 0.6131 |