Formation Energy On Qm9
Métriques
MAE
Résultats
Résultats de performance de divers modèles sur ce benchmark
Tableau comparatif
| Nom du modèle | MAE |
|---|---|
| physnet-a-neural-network-for-predicting | 0.19 |
| tensornet-cartesian-tensor-representations | 0.09 |
| neural-message-passing-with-edge-updates-for | 0.314 |
| hierarchical-modeling-of-molecular-energies | 0.256 |
| wigner-kernels-body-ordered-equivariant | 0.100 ± 0.003 |
| a-universal-framework-for-accurate-and-1 | 0.136 |
| molecular-mechanics-driven-graph-neural | 0.137 |
| graph-networks-as-a-universal-machine | 0.21 |
| heterogeneous-molecular-graph-neural-networks | 0.138 |
| atomistic-line-graph-neural-network-for | 0.30 |
| directional-message-passing-for-molecular-1 | 0.185 |
| neural-message-passing-with-edge-updates-for | 0.242 |
| neural-message-passing-for-quantum-chemistry | 0.49 |
| machine-learning-prediction-errors-better | 0.58 |
| transferable-multi-level-attention-neural | 0.141 |
| graph-networks-as-a-universal-machine | 0.28 |
| physnet-a-neural-network-for-predicting | 0.14 |
| scalable-gaussian-process-regression-enables | 0.167 |