QM9 (Molecular Property Prediction) qm9-molecular-property-prediction Leaderboard
QM9 molecular-property regression as scored on the Papers With Code "Molecular Property Prediction" board — learned molecular representations evaluated by Mean Absolute Error (lower is better). Tracks self-supervised and 3D representation-learning models (Uni-Mol, GROVER, ChemRL-GEM, D-MPNN). · Metric: MAE (lower is better)
| # | Model | MAE | Paper |
|---|---|---|---|
| 1 | Uni-Mol: A Universal 3D Molecular Representation Learning Framework | 0.00 | — |
| 2 | ChemRL-GEM: Geometry Enhanced Molecular Representation Learning | 0.01 | — |
| 3 | Analyzing Learned Molecular Representations for Property Prediction (D-MPNN) | 0.01 | — |
| 4 | Strategies for Pre-training Graph Neural Networks (PretrainGNN) | 0.01 | — |
| 5 | N-Gram Graph: Simple Unsupervised Representation for Graphs (N-GramXGB) | 0.01 | — |
| 6 | GROVER: Self-Supervised Graph Transformer on Large-Scale Molecular Data (base) | 0.01 | — |
| 7 | GROVER: Self-Supervised Graph Transformer on Large-Scale Molecular Data (large) | 0.01 | — |
| 8 | N-Gram Graph: Simple Unsupervised Representation for Graphs (N-GramRF) | 0.01 | — |