Research map / Probabilistic and scientific modeling / Variational and generative sampling
SE(3) Equivariant Augmented Coupling Flows
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| use SE(3)-equivariant EGNN with projection instead of non-rotation-equivariant transformer with augmentation (ablation) | E-ACF · QM9 positional | NLL: -76.76 -> -165.71 | Helped |
| same replacement (ablation) | E-ACF · DW4 | NLL: 10.07 -> 8.61 | Helped |
| same replacement (ablation) | E-ACF · LJ13 | NLL: 33.32 -> 30.33 | Helped |
| same replacement, alanine dipeptide (ablation) | E-ACF · Alanine dipeptide | NLL: -184.57 -> -188.57 | Helped |
| same replacement, energy-based FAB training (ablation) | E-ACF · DW4 FAB | Reverse ESS (%): 35.94 -> 80.44 | Helped |
| same replacement, energy-based FAB training (ablation) | E-ACF · LJ13 FAB | Reverse ESS (%): 5.38 -> 62.09 | Helped |
| Spherical-proj instead of Vector-proj (ablation) | E-ACF · QM9 positional | NLL: -152.23 -> -165.71 | Helped |
| Spherical-proj instead of Cartesian-proj (ablation) | E-ACF · QM9 positional | NLL: -138.62 -> -165.71 | Helped |
| Vector-proj instead of Spherical-proj (ablation) | E-ACF · LJ13 | NLL: 30.19 -> 30.33 | Hurt |
| Cartesian-proj instead of Spherical-proj (ablation) | E-ACF · Alanine dipeptide | Reverse ESS (%): 48.4 -> 52.5 | Mixed |
| E-ACF versus E-CNF FM | E-ACF · QM9 positional | NLL: -138.23 -> -165.71 | Helped |
| E-ACF versus E-CNF-Diff | E-ACF · DW4 | NLL: 8.01 -> 8.61 | Hurt |
Closest papers by meaning
- Scalable Equilibrium Sampling with Sequential Boltzmann Generators (ICML 2025 · similarity 0.86)
- BoltzNCE: Learning likelihoods for Boltzmann Generation with Stochastic Interpolants and Noise Contrastive Estimation (NeurIPS 2025 · similarity 0.86)
- High-order Equivariant Flow Matching for Density Functional Theory Hamiltonian Prediction (NeurIPS 2025 · similarity 0.86)
- Scalable Non-Equivariant 3D Molecule Generation via Rotational Alignment (ICML 2025 · similarity 0.86)
- Path Gradients after Flow Matching (NeurIPS 2025 · similarity 0.85)
- Higher-Rank Irreducible Cartesian Tensors for Equivariant Message Passing (NeurIPS 2024 · similarity 0.85)
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.