Unconditional 3D Molecule Generation on QM9 2014 (test)
0.063FCDTrain
Evaluation results
| Method | Links | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| TrainReference Type=Training Data2026.06 | 0.063 | 0.999 | 0.988 | 0.989 | 0.989 | 0 | 0.49 | 0.992 | 0.946 | — | — | — | |
| BioMatrix-1.7BModel Size (Parameters)=1.7B, Samples=10,0002026.06 | 0.064 | 1 | 0.988 | 1 | 0.948 | 0.749 | 0.495 | 0.991 | 0.944 | — | — | — | |
| BioMatrix-4BModel Size (Parameters)=4B, Samples=10,0002026.06 | 0.066 | 1 | 0.988 | 1 | 0.949 | 0.742 | 0.494 | 0.991 | 0.942 | — | — | — | |
| NExT-MolSamples=10,0002026.06 | 0.07 | 1 | 0.989 | 1 | 0.967 | 0.802 | 0.53 | 0.992 | 0.945 | — | — | — | |
| MolGenSamples=10,0002026.06 | 0.085 | 1 | 0.988 | 1 | 0.955 | 0.479 | 0.5 | 0.988 | 0.934 | — | — | — | |
| JODOSamples=10,0002026.06 | 0.138 | 0.999 | 0.988 | 0.99 | 0.96 | 0.78 | 0.522 | 0.986 | 0.934 | — | — | — | |
| MiDiSamples=10,0002026.06 | 0.187 | 0.998 | 0.976 | 0.98 | 0.954 | 0.769 | 0.501 | 0.979 | 0.882 | — | — | — | |
| BioMatrix-4B†Model Size (Parameters)=4B, Post-hoc refinement=MMFF force-field optimization, Samples=10,0002026.06 | 0.225 | 0.985 | — | — | — | — | — | — | — | 1.16 | 0.0418 | 0.0002 | |
| BioMatrix-1.7B†Model Size (Parameters)=1.7B, Post-hoc refinement=MMFF force-field optimization, Samples=10,0002026.06 | 0.226 | 0.985 | — | — | — | — | — | — | — | 1.15 | 0.0438 | 0.0002 | |
| MolGPTSamples=10,0002026.06 | 0.461 | 0.982 | 0.976 | 0.977 | 0.937 | 0.763 | 0.523 | 0.958 | 0.923 | — | — | — | |
| CDGSSamples=10,0002026.06 | 0.798 | 0.997 | 0.951 | 0.951 | 0.936 | 0.86 | 0.493 | 0.973 | 0.784 | — | — | — | |
| TrainReference Type=Training Data2026.06 | 0.877 | 0.994 | — | — | — | — | — | — | — | 0.0005 | 0.0005 | 0.0002 | |
| NExT-MolSamples=10,0002026.06 | 0.879 | 0.993 | — | — | — | — | — | — | — | 0.115 | 0.0073 | 0.0002 | |
| JODOSamples=10,0002026.06 | 0.885 | 0.992 | — | — | — | — | — | — | — | 0.148 | 0.0121 | 0.0006 | |
| BioMatrix-4BModel Size (Parameters)=4B, Post-hoc refinement=false, Samples=10,0002026.06 | 1.012 | 0.897 | — | — | — | — | — | — | — | 1.05 | 0.0309 | 0.0006 | |
| BioMatrix-1.7BModel Size (Parameters)=1.7B, Post-hoc refinement=false, Samples=10,0002026.06 | 1.04 | 0.897 | — | — | — | — | — | — | — | 1.05 | 0.0316 | 0.0005 | |
| MiDiSamples=10,0002026.06 | 1.1 | 0.983 | — | — | — | — | — | — | — | 0.896 | 0.0208 | 0.0008 | |
| EDMSamples=10,0002026.06 | 1.285 | 0.986 | — | — | — | — | — | — | — | 0.13 | 0.0182 | 0.0007 | |
| EQGAT-diffSamples=10,0002026.06 | 1.519 | 0.988 | — | — | — | — | — | — | — | 0.409 | 0.0191 | 0.0011 | |
| EQGAT-diffSamples=10,0002026.06 | 2.157 | 1 | 0.972 | 1 | 0.996 | 0.695 | 0.479 | 0.949 | 0.707 | — | — | — | |
| G-SchNetSamples=10,0002026.06 | 2.386 | 0.957 | — | — | — | — | — | — | — | 0.362 | 0.0727 | 0.0042 | |
| MDMSamples=10,0002026.06 | 4.861 | 0.992 | — | — | — | — | — | — | — | 0.274 | 0.066 | 0.0239 | |
| G-SphereNetSamples=10,0002026.06 | 6.659 | 0.672 | — | — | — | — | — | — | — | 0.151 | 0.354 | 0.0129 |