Molecular Property Prediction on FreeSolv
0.796RMSEReAlignFitGCN
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ReAlignFitGCNAlignment Strategy=Representational Alignment by Attention-based Inductive Bias, Backbone=GCN2025.02 | 0.796 | — | |
| ReAlignFitAlignment Strategy=Representational Alignment by Attention-based Inductive Bias2025.02 | 0.798 | — | |
| ReAlignFitGINAlignment Strategy=Representational Alignment by Attention-based Inductive Bias, Backbone=GIN2025.02 | 0.803 | — | |
| ReAlignFitGATAlignment Strategy=Representational Alignment by Attention-based Inductive Bias, Backbone=GAT2025.02 | 0.806 | — | |
| CMRLAlignment Strategy=Representational Alignment by Attention-based Inductive Bias2025.02 | 0.815 | — | |
| ISEAlignment Strategy=Representational Alignment by Attention-based Inductive Bias2025.02 | 0.825 | — | |
| CGIBAlignment Strategy=Representational Alignment by Attention-based Inductive Bias2025.02 | 0.852 | — | |
| MMGNNAlignment Strategy=Representational Alignment by Attention-based Inductive Bias2025.02 | 0.902 | — | |
| CIGINAlignment Strategy=No Representational Alignment, Backbone=CIGIN2025.02 | 0.905 | — | |
| GINAlignment Strategy=No Representational Alignment, Backbone=GIN2025.02 | 1.015 | — | |
| MPNNAlignment Strategy=No Representational Alignment, Backbone=MPNN2025.02 | 1.159 | — | |
| GCNAlignment Strategy=No Representational Alignment, Backbone=GCN2025.02 | 1.192 | — | |
| GATAlignment Strategy=No Representational Alignment, Backbone=GAT2025.02 | 1.28 | — | |
| MolCLRGINModel Type=Graph-based2026.06 | 2.017 | — | |
| MolCLRCMPNNModel Type=Graph-based2026.06 | 2.021 | — | |
| MolCLRGCNModel Type=Graph-based2026.06 | 2.241 | — | |
| GPT-4o-miniStrategy=+MOLE-RAG2026.06 | 2.555 | — | |
| Qwen34B-InstStrategy=+MOLE-RAG2026.06 | 2.7 | — | |
| GPT-4oStrategy=+MOLE-RAG2026.06 | 2.88 | — | |
| SchNetModel Type=Graph-based2026.06 | 3.215 | — | |
| GROVERModel Type=Graph-based2026.06 | 3.235 | — | |
| ECFP + OverlapBPERegression Model=XGBoost2026.04 | 3.3409 | — | |
| MGCNModel Type=Graph-based2026.06 | 3.349 | — | |
| ChemDFM14BStrategy=+MOLE-RAG2026.06 | 3.476 | — | |
| ECFPRegression Model=XGBoost2026.04 | 3.9498 | — | |
| Mistral7B-InstStrategy=+MOLE-RAG2026.06 | 4.128 | — | |
| Llama3.2-3BStrategy=+MOLE-RAG2026.06 | 4.343 | — | |
| GPT-4o-miniStrategy=Baseline2026.06 | 5.475 | — | |
| Qwen34B-InstStrategy=Baseline2026.06 | 5.524 | — | |
| GPT-4oStrategy=Baseline2026.06 | 5.851 | — | |
| Llama3.2-3BStrategy=Baseline2026.06 | 5.995 | — | |
| ChemDFM14BStrategy=Baseline2026.06 | 6.523 | — | |
| Mistral7B-InstStrategy=Baseline2026.06 | 12.585 | — | |
| AtomPairrepetitions=5, hyperparameters=shared2026.04 | — | 1.207 | |
| HDCrepetitions=5, hyperparameters=shared2026.04 | — | 0.966 | |
| Morganrepetitions=5, hyperparameters=shared2026.04 | — | 1.185 | |
| RDKitrepetitions=5, hyperparameters=shared2026.04 | — | 0.978 | |
| Torsionrepetitions=5, hyperparameters=shared2026.04 | — | 1.84 |