Classic
AQSOL
Structure
9,823Graphs
18Nodes
18Edges
NoneClasses
65Node Feature Dim
5Edge Feature Dim
Coverage
27Results
1Variants
3Papers
- ICML 2025: Graph Alignment for Benchmarking Graph Neural Networks and Learning Positional Encodings Graph Alignment for Benchmarking Graph Neural Networks and Learning Positional Encodings ICML 2025
- JMLR 2023: GNNBenchmark GNNBenchmark JMLR 2023
- TMLR 2022: Rewiring with Positional Encodings for Graph Neural Networks Rewiring with Positional Encodings for Graph Neural Networks TMLR 2022
Graph regression on molecular graphs, predicting measured aqueous solubility. Atom and bond types form the graph input. Evaluated by MAE.
Dataset details
Task
Aqueous solubility regression on 9,982 molecular graphs from the LRGB paper. The primary catalog metric is MAE.
Data
Aqueous solubility regression on 9,982 molecular graphs from the LRGB paper.
Features
The representation follows the cited PyG loader or benchmark release. Node and edge fields are dataset-specific; preprocessing, graph direction, and any feature normalization must be kept fixed when reproducing a result.
Splits and evaluation
The catalog records these protocols or variants: Standard split. Evaluation uses MAE (lower is better). Exact masks or folds must come from the cited release.
Quirks and pitfalls
The same short dataset name is used for incompatible processed artifacts and split protocols. Report the loader/version, directedness, feature preprocessing, and exact masks; transductive results should not be described as inductive.
Sources
SOTA Progression
new SOTA
matches SOTA at that time
other proposed result
other baseline result
Leaderboard
| # | Model | MAE ↓ | Paper |
|---|---|---|---|
| 1 | Ours(100K) | 0.9200
±
0.0090 | |
| 2 | GatedGCN-PE/E(500K) | 0.9960
±
0.0080 | |
| 3 | GatedGCN-E-PE | 0.9960
±
0.0080 | |
| 4 | 3WLGNN-E | 1.0420
±
0.0640 | |
| 5 | GAPE + RWPE | 1.0690
±
0.0090 | |
| 6 | GAPE | 1.0850
±
0.0090 | |
| 7 | RWPE | 1.1200
±
0.0040 | |
| 8 | 3WLGNN | 1.1540
±
0.0500 | |
| 9 | R-LAP | 1.2300
±
0.0070 | |
| 10 | GAT | 1.2400
±
0.0110 | |
| 11 | SignNet | 1.2600
±
0.0060 | |
| 12 | ABS-LAP | 1.2700
±
0.0090 | |
| 13 | GCN | 1.2800
±
0.0110 | |
| 14 | GatedGCN-E | 1.2950
±
0.0160 | |
| 15 | LAP | 1.3100
±
0.0060 | |
| 16 | LaplacianPE | 1.3100
±
0.0390 | |
| 17 | GatedGCN | 1.3520
±
0.0340 | |
| 18 | GCN | 1.3720
±
0.0200 | |
| 19 | MoNet | 1.3950
±
0.0270 | |
| 20 | GraphSAGE | 1.4310
±
0.0100 | |
| 21 | GAT | 1.4410
±
0.0230 | |
| 22 | vanilla GCN | 1.4830
±
0.0140 | |
| 23 | None | 1.7100
±
0.0500 | |
| 24 | MLP | 1.7440
±
0.0160 | |
| 25 | GIN | 1.8940
±
0.0240 | |
| 26 | RingGNN-E | 3.7690
±
1.0120 | |
| 27 | RingGNN | 20.2640
±
7.5490 |
SOTA performance
set a new SOTA at the time
matched the SOTA at the time