Energy and force prediction on 3BPA 1200 K (test)
29.8Energy RMSEMACE
Evaluation results
| Method | Links | ||
|---|---|---|---|
| MACEtrained_on=300 K2025.10 | 29.8 | 62 | |
| RACEtrained_on=300 K2025.10 | 37.5 | 115.3 | |
| RACE-Ensembletrained_on=300 K2025.10 | 37.7 | 119.2 | |
| NequIPtrained_on=300 K2025.10 | 38.5 | 76.2 | |
| BOTNettrained_on=300 K2025.10 | 39.1 | 81.1 | |
| Allegrotrained_on=300 K2025.10 | 42.6 | 83 | |
| RACE-DE-JEFtrained_on=300 K, loss=NLL_JEF2025.10 | 51.1 | 120.8 | |
| RACE-LAtrained_on=300 K, algorithm=Laplace Approximation2025.10 | 60.8 | 171.8 | |
| MGNNtrained_on=300 K2025.10 | 74.3 | 142.6 | |
| ICTPsym+ltNtrain=502024.05 | 78.51 | 151.37 | |
| MACENtrain=502024.05 | 83.99 | 154.46 | |
| ICTPsymNtrain=502024.05 | 86 | 153.16 | |
| ICTPfullNtrain=502024.05 | 92.16 | 157.72 | |
| RACE-DE-Etrained_on=300 K, loss=Deep Ensemble Energy-only2025.10 | 171.9 | 232.8 |