Energy and Force Prediction on rMD17 Azobenzene (test)
0.7Energy (E)TensorNet
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| TensorNetLayers=2L, Parameters=770k2023.06 | 0.7 | 3.1 | — | — | — | |
| NequIP2023.06 | 0.7 | 2.9 | — | — | — | |
| BOTNet2023.06 | 0.7 | 3.3 | — | — | — | |
| TensorNetLayers=1L, Parameters=535k2023.06 | 0.9 | 3.8 | — | — | — | |
| Allegro2023.06 | 1.2 | 2.6 | — | — | — | |
| MACE2023.06 | 1.2 | 3 | — | — | — | |
| Degree-QuantBits (W/A)=8 / 82026.03 | — | — | 63.2 | 58.9 | — | |
| FP32 BaselineBits (W/A)=32 / 322026.03 | — | — | 23.2 | 21.2 | — | |
| Naive INT8Bits (W/A)=8 / 82026.03 | — | — | 118.2 | 102.39 | — | |
| Ours (GAQ)Bits (W/A)=4 / 82026.03 | — | — | 9.31 | 22.6 | — | |
| SVQ-KMeansBits (W/A)=8 / 82026.03 | — | — | 226.2 | — | — |