Research map / Image generation / Autoregressive and latent generation
VEDA: Generation of 3D Molecules via Variance-Exploding Diffusion with Annealing
Peining Zhang, Jinbo Bi, Minghu Song
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add LMMSE preconditioning to VEDA (ablation) | VEDA-S · GEOM-DRUGS | relaxation energy (kcal/mol): 2.14 -> 1.72 | Improved |
| add LMMSE preconditioning to VEDA (ablation) | VEDA-S · GEOM-DRUGS | validity & connectivity (%): 97.9 -> 98.8 | Improved |
| add amplified noise injection (annealing) to VEDA (ablation) | VEDA-S · GEOM-DRUGS | relaxation energy (kcal/mol): 6.29 -> 1.72 | Improved |
| add amplified noise injection (annealing) to VEDA (ablation) | VEDA-S · GEOM-DRUGS | validity & connectivity (%): 95.9 -> 98.8 | Improved |
| add preconditioning when noise injection is already removed (ablation) | VEDA-S · GEOM-DRUGS | relaxation energy (kcal/mol): 10.1 -> 6.29 | Improved |
| add OT alignment to a no-noise, no-preconditioning variant (ablation) | VEDA-S · GEOM-DRUGS | relaxation energy (kcal/mol): 10.1 -> 23.5 | Hurt |
| add OT alignment to a no-noise, no-preconditioning variant (ablation) | VEDA-S · GEOM-DRUGS | RMSD: 0.409 -> 0.557 | Hurt |
| add OT alignment to a no-noise, no-preconditioning variant (ablation) | VEDA-S · GEOM-DRUGS | validity & connectivity (%): 93.8 -> 91.7 | Hurt |
| add noise injection to SemlaFlow sampling (gamma sweep) (ablation) | SemlaFlow · GEOM-DRUGS | median relaxation energy (kcal/mol): 32.3 -> 31.6 | No clear effect |
| uniform random re-masking instead of confidence-based re-masking (ablation) | VEDA-S · GEOM-DRUGS | energy, RMSD, validity | Improved |
| use variable eta_t schedule instead of fixed eta_t in discrete sampler (ablation) | VEDA-S · GEOM-DRUGS | energy, RMSD, validity | No clear effect |
| increase sampling steps from 50 to 100 (ablation) | VEDA-S · GEOM-DRUGS | median relaxation energy (kcal/mol): 2.99 -> 1.72 | Improved |
Closest papers by meaning
- Equivariant Blurring Diffusion for Hierarchical Molecular Conformer Generation (NeurIPS 2024 · similarity 0.91)
- Navigating the Design Space of Equivariant Diffusion-Based Generative Models for De Novo 3D Molecule Generation (ICLR 2024 · similarity 0.90)
- DiSCO: Diffusion Schrödinger Bridge for Molecular Conformer Optimization (AAAI 2024 · similarity 0.88)
- Scalable Non-Equivariant 3D Molecule Generation via Rotational Alignment (ICML 2025 · similarity 0.88)
- Geometric-Facilitated Denoising Diffusion Model for 3D Molecule Generation (AAAI 2024 · similarity 0.88)
- Towards Unified and Lossless Latent Space for 3D Molecular Latent Diffusion Modeling (NeurIPS 2025 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
