Energy and force prediction on MD17 Aspirin (test)
0.02Force MAE (kcal/mol/Å)GDML
Evaluation results
| Method | Links | ||
|---|---|---|---|
| GDMLTraining Supervision=forces only, Training Set Size=50k2019.02 | 0.02 | — | |
| PhysNet-ens5Training Supervision=energies and forces, Training Set Size=50k, Evaluation Mode=ensemble of five models2019.02 | 0.04 | — | |
| PhysNetTraining Supervision=energies and forces, Training Set Size=50k, Evaluation Mode=averaged over five runs2019.02 | 0.06 | — | |
| SchNetTraining Examples (N)=50,000, Training Signal=Energy and forces2017.06 | 0.33 | 0.12 | |
| SchNetTraining Supervision=energies and forces, Training Set Size=50k2019.02 | 0.33 | — | |
| GDMLTraining Examples (N)=1,000, Training Signal=Forces only2017.06 | 0.99 | 0.27 | |
| SchNetTraining Examples (N)=1,000, Training Signal=Energy and forces2017.06 | 1.35 | 0.37 | |
| SchNetTraining Examples (N)=1,000, Training Signal=Energy only2017.06 | 23.54 | 4.2 | |
| SchNetTraining Examples (N)=50,000, Training Signal=Energy only2017.06 | — | 0.25 |