Research map / Probabilistic and scientific modeling / Molecules and proteins
Elign: Equivariant Diffusion Model Alignment from Foundational Machine Learned Force Fields
Yunyang Li, Lin Huang, Luojia Xia, Wenhe Zhang, Mark Gerstein
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 dense PBRS energy reward shaping (ablation) | Elign · QM9 | validity x uniqueness (%): 93.85 -> 95.31 | Improved |
| add dense PBRS energy reward shaping (ablation) | Elign · QM9 | molecule stability (%): 93.75 -> 93.70 | No clear effect |
| use larger MLFF UMA-1p1-M instead of UMA-1p1-S (ablation) | Elign · QM9 | validity x uniqueness (%): 90.81 -> 95.31 | Improved |
| use larger MLFF UMA-1p1-M instead of UMA-1p1-S (ablation) | Elign · QM9 | molecule stability (%): 94.83 -> 93.70 | Hurt |
| add force reward to energy-only reward (ablation) | Elign (no PBRS) · QM9 | molecule stability (%): 86.00 -> 93.75 | Improved |
| add energy reward to force-only reward (ablation) | Elign (no PBRS) · QM9 | validity x uniqueness (%): 90.21 -> 93.85 | Improved |
| add energy reward to force-only reward (ablation) | Elign (no PBRS) · QM9 | molecule stability (%): 94.92 -> 93.75 | Hurt |
| post-train EDM with Elign | EDM · QM9 | molecule stability (%): 82.00 -> 93.70 | Improved |
| post-train EDM with Elign | EDM · QM9 | validity x uniqueness (%): 90.70 -> 95.31 | Improved |
| post-train EDM with Elign instead of UMA run-time guidance | EDM + UMA-1p1-M Guidance · QM9 | molecule stability (%): 87.25 -> 93.70 | Improved |
| post-train EDM with Elign | EDM · GEOM-Drugs | atom stability (%): 81.3 -> 87.94 | Improved |
| post-train EDM with Elign | EDM · GEOM-Drugs | RDKit validity (%): 91.9 -> 99.40 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- UniGEM — UniGEM: A Unified Approach to Generation and Property Prediction for Molecules (ICLR 2025) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Navigating the Design Space of Equivariant Diffusion-Based Generative Models for De Novo 3D Molecule Generation (ICLR 2024 · similarity 0.89)
- Equivariant Blurring Diffusion for Hierarchical Molecular Conformer Generation (NeurIPS 2024 · similarity 0.88)
- Scalable Non-Equivariant 3D Molecule Generation via Rotational Alignment (ICML 2025 · similarity 0.88)
- ELoRA: Low-Rank Adaptation for Equivariant GNNs (ICML 2025 · similarity 0.87)
- Swallowing the Bitter Pill: Simplified Scalable Conformer Generation (ICML 2024 · similarity 0.87)
- DiSCO: Diffusion Schrödinger Bridge for Molecular Conformer Optimization (AAAI 2024 · similarity 0.87)
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.
