Research map / Probabilistic and scientific modeling / Variational and generative sampling
Equivariant flow matching
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 equivariant OT cost (Hungarian + Kabsch alignment) instead of plain OT flow matching | Equivariant CNF · LJ13 | Path length: 2.84 -> 2.15 | Helped |
| use equivariant OT cost instead of plain OT flow matching | Equivariant CNF · LJ55 | Path length: 7.53 -> 3.52 | Helped |
| use equivariant OT cost instead of plain OT flow matching | Equivariant CNF · LJ55 | ESS: 3.74 -> 4.42 | Helped |
| use equivariant OT cost instead of plain OT flow matching | Equivariant CNF · LJ55 | NLL: -88.45 -> -89.27 | Helped |
| use equivariant OT cost instead of plain OT flow matching | Equivariant CNF · Alanine dipeptide | ESS: 1.41 -> 0.69 | Hurt |
| use equivariant OT cost instead of plain OT flow matching | Equivariant CNF · Alanine dipeptide | Path length: 10.19 -> 9.46 | Helped |
| use equivariant OT cost instead of plain OT flow matching | Equivariant CNF · DW4 | ESS: 92.37 -> 88.71 | Hurt |
| use equivariant OT cost instead of plain OT flow matching | Equivariant CNF · LJ13 | ESS: 54.36 -> 57.98 | No clear effect |
| train with OT flow matching instead of likelihood | Equivariant CNF · LJ13 | ESS: 39.78 -> 54.36 | Helped |
| train with OT flow matching instead of likelihood | Equivariant CNF · LJ13 | Path length: 5.08 -> 2.84 | Helped |
| add OT reordering to naive flow matching (ablation) | Equivariant CNF · LJ55 | Path length: 8.66 -> 7.53 | Helped |
| approximate equivariant cost by applying rotation before permutation (ablation) | Equivariant OT flow matching · Alanine dipeptide | Transport cost | Hurt |
Closest papers by meaning
- Path Gradients after Flow Matching (NeurIPS 2025 · similarity 0.90)
- BoltzNCE: Learning likelihoods for Boltzmann Generation with Stochastic Interpolants and Noise Contrastive Estimation (NeurIPS 2025 · similarity 0.89)
- Controlled Generation with Equivariant Variational Flow Matching (ICML 2025 · similarity 0.89)
- Scalable Equilibrium Sampling with Sequential Boltzmann Generators (ICML 2025 · similarity 0.88)
- Energy Matching: Unifying Flow Matching and Energy-Based Models for Generative Modeling (NeurIPS 2025 · similarity 0.88)
- ET-Flow: Equivariant Flow-Matching for Molecular Conformer Generation (NeurIPS 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.