Energy and force prediction on 3BPA 600 K (test)
9.7Energy RMSEMACE
Evaluation results
| Method | Links | ||
|---|---|---|---|
| MACEtrained_on=300 K2025.10 | 9.7 | 21.8 | |
| RACE-Ensembletrained_on=300 K2025.10 | 11.1 | 30.4 | |
| NequIPtrained_on=300 K2025.10 | 11.2 | 26.4 | |
| BOTNettrained_on=300 K2025.10 | 11.5 | 26.7 | |
| RACEtrained_on=300 K2025.10 | 11.7 | 31.8 | |
| Allegrotrained_on=300 K2025.10 | 12.1 | 29.2 | |
| RACE-DE-JEFtrained_on=300 K, loss=NLL_JEF2025.10 | 14.6 | 37 | |
| RACE-LAtrained_on=300 K, algorithm=Laplace Approximation2025.10 | 15.3 | 51 | |
| MGNNtrained_on=300 K2025.10 | 17.8 | 39.6 | |
| MACENtrain=502024.05 | 30.71 | 69.88 | |
| ICTPsym+ltNtrain=502024.05 | 30.74 | 69.62 | |
| ICTPsymNtrain=502024.05 | 31.63 | 68.87 | |
| ICTPfullNtrain=502024.05 | 31.68 | 69.87 | |
| RACE-DE-Etrained_on=300 K, loss=Deep Ensemble Energy-only2025.10 | 43.7 | 98 |