# De Novo 3D Molecule Generation on GEOM-Drugs Source: [De Novo 3D Molecule Generation on GEOM-Drugs]() Collection index: [llms.txt]() Data updated: 2026-06-15T07:03:13.752Z Evaluation of generative models for de novo 3D molecule generation using the GEOM-Drugs dataset, assessed using a comprehensive suite of metrics including FCD/FCD3D, 2D/3D stability, validity checks (V&C, V&U, V&U&N), and structural fidelity via MMD analysis of bond and angle distributions. A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | de novo 3D molecule generation | | Dataset (including any stated split) | GEOM-Drugs | | Metric | Atom Stability (2D) | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 23 of 23 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [Atom Stability (2D)](): Higher is better; unit: PERCENT. - [FCD](): Lower is better; unit: SCALAR. - [Mol Stability (2D)](): Higher is better; unit: PERCENT. - [Validity (Connected)](): Higher is better; unit: PERCENT. - [Validity (Unique)](): Higher is better; unit: PERCENT. - [Validity (Novel)](): Higher is better; unit: PERCENT. - [SNN Similarity](): Higher is better; unit: PERCENT. - [Fragment Coverage](): Higher is better; unit: PERCENT. - [Scaffold Coverage](): Higher is better; unit: PERCENT. - [FCD3D](): Lower is better; unit: SCALAR. - [Atom Stability (3D)](): Higher is better; unit: PERCENT. - [Mol Stability (3D)](): Higher is better; unit: PERCENT. - [MMD Bond Length](): Lower is better; unit: SCALAR. - [MMD Bond Angle](): Lower is better; unit: SCALAR. - [MMD Dihedral Angle](): Lower is better; unit: SCALAR. - [FCD (2D)](): Lower is better; unit: SCALAR. ## Recorded results | Method | Atom Stability (2D) | Result date | Report paper | | --- | --- | --- | --- | | Train | 100 | Not recorded | 693f960e6cfd488761a819fc | | JODO | 100 | Not recorded | 693f960e6cfd488761a819fc | | EQGAT-diff | 100 | Not recorded | 693f960e6cfd488761a819fc | | NExT-Mol | 100 | Not recorded | 693f960e6cfd488761a819fc | | UDM-3D | 100 | Not recorded | 693f960e6cfd488761a819fc | | MolSculpt | 100 | Not recorded | 693f960e6cfd488761a819fc | | JODO | 100 | Not recorded | 6a2f871dd33dfb23e783c01a | | EQGAT-diff | 100 | Not recorded | 6a2f871dd33dfb23e783c01a | | UDM-3D | 100 | Not recorded | 6a2f871dd33dfb23e783c01a | | TopVAE+DiT | 100 | Not recorded | 6a2f871dd33dfb23e783c01a | | GeoLDM | 99.6 | Not recorded | 6a2f871dd33dfb23e783c01a | | CDGS | 99.1 | Not recorded | 693f960e6cfd488761a819fc | | EDM | 99.1 | Not recorded | 6a2f871dd33dfb23e783c01a | | CDGS | 99.1 | Not recorded | 6a2f871dd33dfb23e783c01a | | MiDi | 96.8 | Not recorded | 693f960e6cfd488761a819fc | | MiDi | 96.8 | Not recorded | 6a2f871dd33dfb23e783c01a | ### Train Atom Stability (2D): 100. Result ID: 699977f85698665ec4e6e2b5. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### JODO Atom Stability (2D): 100. Result ID: 699977f85698665ec4e6e2b7. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### EQGAT-diff Atom Stability (2D): 100. Result ID: 699977f85698665ec4e6e2b9. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### NExT-Mol Atom Stability (2D): 100. Result ID: 699977f85698665ec4e6e2ba. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### UDM-3D Atom Stability (2D): 100. Result ID: 699977f85698665ec4e6e2bb. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### MolSculpt Atom Stability (2D): 100. Result ID: 699977f85698665ec4e6e2bc. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### JODO Atom Stability (2D): 100. Result ID: 6a2fa39bfbc373592c2b6742. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"n":"10,000"}. This record does not assert independent reproduction. ### EQGAT-diff Atom Stability (2D): 100. Result ID: 6a2fa39bfbc373592c2b6743. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"n":"10,000"}. This record does not assert independent reproduction. ### UDM-3D Atom Stability (2D): 100. Result ID: 6a2fa39bfbc373592c2b6744. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"n":"10,000"}. This record does not assert independent reproduction. ### TopVAE+DiT Atom Stability (2D): 100. Result ID: 6a2fa39bfbc373592c2b6745. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"n":"10,000"}. This record does not assert independent reproduction. ### GeoLDM Atom Stability (2D): 99.6. Result ID: 6a2fa39bfbc373592c2b6740. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"n":"10,000"}. This record does not assert independent reproduction. ### CDGS Atom Stability (2D): 99.1. Result ID: 699977f85698665ec4e6e2b6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### EDM Atom Stability (2D): 99.1. Result ID: 6a2fa39bfbc373592c2b673e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"n":"10,000"}. This record does not assert independent reproduction. ### CDGS Atom Stability (2D): 99.1. Result ID: 6a2fa39bfbc373592c2b673f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"n":"10,000"}. This record does not assert independent reproduction. ### MiDi Atom Stability (2D): 96.8. Result ID: 699977f85698665ec4e6e2b8. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### MiDi Atom Stability (2D): 96.8. Result ID: 6a2fa39bfbc373592c2b6741. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"n":"10,000"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.