DFT energy prediction on KIM Si
0.3719MAE (Config Level)SchNet
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| SchNetLabel Augmentation=true, Multi-task Pretraining=true2022.10 | 0.3719 | 0.049 | 27.17 | |
| SchNetLabel Augmentation=true, Multi-task Pretraining=false2022.10 | 0.4015 | 0.0549 | 19.99 | |
| SchNetLabel Augmentation=false, Multi-task Pretraining=true2022.10 | 0.4034 | 0.0569 | 18.4 | |
| GemNetLabel Augmentation=true, Multi-task Pretraining=true2022.10 | 0.4651 | 0.0476 | 11.12 | |
| GemNetLabel Augmentation=true, Multi-task Pretraining=false2022.10 | 0.4691 | 0.0511 | 7.55 | |
| SchNetLabel Augmentation=false, Multi-task Pretraining=false2022.10 | 0.4805 | 0.0718 | — | |
| GemNetLabel Augmentation=false, Multi-task Pretraining=true2022.10 | 0.5024 | 0.0531 | 2.48 | |
| GemNetLabel Augmentation=false, Multi-task Pretraining=false2022.10 | 0.5138 | 0.0546 | — | |
| SOAPNetLabel Augmentation=true, Multi-task Pretraining=true2022.10 | 0.5307 | 0.0657 | 31.88 | |
| SOAPNetLabel Augmentation=true, Multi-task Pretraining=false2022.10 | 0.5595 | 0.0704 | 27.58 | |
| SOAPNetLabel Augmentation=false, Multi-task Pretraining=true2022.10 | 0.5717 | 0.0733 | 25.28 | |
| CGCNNLabel Augmentation=true, Multi-task Pretraining=true2022.10 | 0.7435 | 0.1005 | 24.44 | |
| CGCNNLabel Augmentation=true, Multi-task Pretraining=false2022.10 | 0.7476 | 0.105 | 22.61 | |
| SOAPNetLabel Augmentation=false, Multi-task Pretraining=false2022.10 | 0.7706 | 0.0975 | — | |
| CGCNNLabel Augmentation=false, Multi-task Pretraining=true2022.10 | 0.8457 | 0.1253 | 10.16 | |
| CGCNNLabel Augmentation=false, Multi-task Pretraining=false2022.10 | 0.9314 | 0.141 | — | |
| Best EIPLabel Augmentation=false, Multi-task Pretraining=false2022.10 | 1.6326 | 0.2524 | — |