MoleculeNet
FreeSolv
Structure
642Graphs
9Nodes
8Edges
1Classes
9Node Feature Dim
3Edge Feature Dim
Coverage
155Results
1Variants
23Papers
- WWW 2026: Frequency-Corrupt Based Graph Self-Supervised Learning Frequency-Corrupt Based Graph Self-Supervised Learning WWW 2026
- Hierarchical Molecular Representation Learning via Fragment-Based Self-Supervised Embedding Prediction Hierarchical Molecular Representation Learning via Fragment-Based Self-Supervised Embedding Prediction
- MolGA: Molecular Graph Adaptation with Pre-trained 2D Graph Encoder MolGA: Molecular Graph Adaptation with Pre-trained 2D Graph Encoder
- KDD 2025: Graph Positional Autoencoders as Self-supervised Learners Graph Positional Autoencoders as Self-supervised Learners KDD 2025
- ICLR 2025: NExT-Mol: 3D Diffusion Meets 1D Language Modeling for 3D Molecule Generation NExT-Mol: 3D Diffusion Meets 1D Language Modeling for 3D Molecule Generation ICLR 2025
- AAAI 2024: Pre-training Graph Neural Networks on Molecules by Using Subgraph-Conditioned Graph Information Bottleneck Pre-training Graph Neural Networks on Molecules by Using Subgraph-Conditioned Graph Information Bottleneck AAAI 2024
- NeurIPS 2024: Cross-Domain Graph Data Scaling: A Showcase with Diffusion Models Cross-Domain Graph Data Scaling: A Showcase with Diffusion Models NeurIPS 2024
- ICML 2024: UniCorn: A Unified Contrastive Learning Approach for Multi-view Molecular Representation Learning UniCorn: A Unified Contrastive Learning Approach for Multi-view Molecular Representation Learning ICML 2024
- ICML 2024: Community-Invariant Graph Contrastive Learning Community-Invariant Graph Contrastive Learning ICML 2024
- ICML 2024: Structure-Aware E(3)-Invariant Molecular Conformer Aggregation Networks Structure-Aware E(3)-Invariant Molecular Conformer Aggregation Networks ICML 2024
- NeurIPS 2023: SE(3)-Invariant Multiparameter Persistent Homology for Chiral-Sensitive Molecular Property Prediction SE(3)-Invariant Multiparameter Persistent Homology for Chiral-Sensitive Molecular Property Prediction NeurIPS 2023
- ICML 2023: Graph Neural Networks with a Distribution of Parametrized Graphs Graph Neural Networks with a Distribution of Parametrized Graphs ICML 2023
- Deep Prompt Tuning for Graph Transformers Deep Prompt Tuning for Graph Transformers
- AAAI 2023: 3D Denoisers are Good 2D Teachers: Molecular Pretraining via Denoising and Cross-Modal Distillation 3D Denoisers are Good 2D Teachers: Molecular Pretraining via Denoising and Cross-Modal Distillation AAAI 2023
- Motif-aware Attribute Masking for Molecular Graph Pre-training Motif-aware Attribute Masking for Molecular Graph Pre-training
- ICML 2023: Graph Positional and Structural Encoder Graph Positional and Structural Encoder ICML 2023
- Improving Molecular Contrastive Learning via Faulty Negative Mitigation and Decomposed Fragment Contrast Improving Molecular Contrastive Learning via Faulty Negative Mitigation and Decomposed Fragment Contrast
- OOD-GNN: Out-of-Distribution Generalized Graph Neural Network OOD-GNN: Out-of-Distribution Generalized Graph Neural Network
- ICML 2021: 3D Infomax improves GNNs for Molecular Property Prediction 3D Infomax improves GNNs for Molecular Property Prediction ICML 2021
- AAAI 2021: GeomGCL: Geometric Graph Contrastive Learning for Molecular Property Prediction GeomGCL: Geometric Graph Contrastive Learning for Molecular Property Prediction AAAI 2021
- ChemRL-GEM: Geometry Enhanced Molecular Representation Learning for Property Prediction ChemRL-GEM: Geometry Enhanced Molecular Representation Learning for Property Prediction
- Learning Graph-Level Representation for Drug Discovery Learning Graph-Level Representation for Drug Discovery
- MoleculeNet MoleculeNet
Graph regression Hydration free energy regression (kcal/mol) on 642 molecules. SMILES and labels are fixed; molecular graph featurization is implementation-defined. Random and scaffold splits are not comparable. Evaluated by RMSE.
Dataset details
Task
Hydration free energy regression (kcal/mol) on 642 molecules. The primary catalog metric is RMSE.
Data
Hydration free energy regression (kcal/mol) on 642 molecules.
Features
MoleculeNet distributes SMILES and targets, not one mandatory graph featurization. A graph implementation normally derives atom and bond fields with RDKit or DeepChem; results depend on featurizer, salt handling, stereochemistry, and invalid-molecule policy.
Splits and evaluation
The catalog records these protocols or variants: Scaffold split. Evaluation uses RMSE (lower is better). Exact masks or folds must come from the cited release.
Quirks and pitfalls
Small samples, imbalance, missing assay labels, duplicate compounds, and scaffold leakage are common. Missing labels are not negatives. Always report split algorithm, featurizer, metric aggregation, and dataset version; random and scaffold results are not comparable.
Sources
SOTA Progression
new SOTA
matches SOTA at that time
other proposed result
other baseline result
Leaderboard
| # | Model | Hydration free energy RMSE (kcal mol⁻¹) ↓ | Hydration free energy MAE (kcal mol⁻¹) | Paper |
|---|---|---|---|---|
| 1 | SE(3)-I MPPH | 0.3540
±
0.0530 | — | |
| 2 | ECFP-4 + RF | 0.5600
±
0.0660 | — | |
| 3 | GPSE + GPS | 0.6820 | — | |
| 4 | GPS | 0.7430 | — | |
| 5 | GeomGCL | 0.8660 | — | |
| 6 | GeomMPNN | 0.9130 | — | |
| 7 | dummy super node | 0.9330 | — | |
| 8 | LiGhT FT | 0.9550
±
0.0510 | — | |
| 9 | GraphConv | 0.9630 | — | |
| 10 | DimeNet++ | 0.9780 | — | |
| 11 | LiGhT DeepGPT | 0.9830
±
0.0790 | — | |
| 12 | D-MPNN | 1.0100
±
0.0640 | — | |
| 13 | AttentiveFP | 1.0340 | — | |
| 14 | MAT | 1.0590 | — | |
| 15 | DMPNN | 1.0920 | — | |
| 16 | CoMPT | 1.1030 | — | |
| 17 | MPNN | 1.1500
±
0.1200 | — | |
| 18 | HMGNN | 1.2070 | — | |
| 19 | Weave | 1.2200
±
0.2800 | — | |
| 20 | GraSPNet | 1.2320
±
0.0500 | — | |
| 21 | GraphGPS FT | 1.3800
±
0.2340 | — | |
| 22 | GC | 1.4000
±
0.1600 | — | |
| 23 | GraphGPS DeepGPT | 1.4150
±
0.2540 | — | |
| 24 | conan-FGW | 1.4230
±
0.2720 | — | |
| 25 | MoCL | 1.4780 | — | |
| 26 | UniMol | 1.4800
±
0.0480 | — | |
| 27 | conan | 1.5480
±
0.2810 | — | |
| 28 | UniCorn(DFT + RDKit) | 1.5550
±
0.0750 | — | |
| 29 | SchNet-emb | 1.5870
±
0.1360 | — | |
| 30 | MoLlama | 1.5900
±
0.0400 | — | |
| 31 | SchNet-scalar | 1.6080
±
0.1580 | — | |
| 32 | DAG | 1.6300
±
0.1800 | — | |
| 33 | S-CGIB | 1.6480
±
0.0740 | — | |
| 34 | Graphormer DeepGPT | 1.6680
±
0.1140 | — | |
| 35 | Graphormer FT | 1.6800
±
0.0130 | — | |
| 36 | MolPROP | 1.7000
±
0.0900 | — | |
| 37 | Geometry | 1.7010
±
0.1470 | — | |
| 38 | XGBoost | 1.7400
±
0.1500 | — | |
| 39 | MoleBlend | 1.8510
±
0.0680 | — | |
| 40 | GeoGNN | 1.8570
±
0.0710 | — | |
| 41 | MICRO-Graph | 1.8650
±
0.0610 | — | |
| 42 | Multitask | 1.8700
±
0.0700 | — | |
| 43 | Geometry+Graph | 1.8760
±
0.0940 | — | |
| 44 | ChemRL-GEM | 1.8770
±
0.0940 | — | |
| 45 | Graph | 1.9500
±
0.0690 | — | |
| 46 | SimSGT | 1.9530
±
0.0380 | — | |
| 47 | GEM | 1.9570
±
0.0730 | — | |
| 48 | Context+Graph | 1.9820
±
0.0980 | — | |
| 49 | ChemProp3D | 2.0140
±
0.1820 | — | |
| 50 | rf | 2.0300
±
0.2200 | — |
SOTA performance
set a new SOTA at the time
matched the SOTA at the time