Research map / Probabilistic and scientific modeling / Variational and generative sampling
Flow Sampling : Learning to Sample from Unnormalized Densities via Denoising Conditional Processes
Aaron J Havens, Brian Karrer, Neta Shaul
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 |
|---|---|---|---|
| reduce training NFE from 1024 to 64 (ablation) | ASBS · Ala2 | energy W2: 8.650 -> 6.0e7 | Hurt |
| reduce training NFE from 1024 to 64 (ablation) | Flow Sampling · Ala2 | energy W2: 0.637 -> 4.113 | Hurt |
| reduce training NFE from 256 to 64 (ablation) | Flow Sampling · SPICE conformers | coverage recall: 91.89 -> 90.13 | No clear effect |
| reduce training NFE from 256 to 64 (ablation) | AS · SPICE conformers | coverage recall: 88.60 -> 29.13 | Hurt |
| scale network from 5 to 10 layers (XL) (ablation) | Flow Sampling · LJ-55 | energy W2: 21.32 -> 16.29 | Improved |
| scale network from 5 to 10 layers (XL) (ablation) | Flow Sampling · LJ-13 | energy W2: 0.97 -> 0.81 | Improved |
| use Flow Sampling instead of ASBS | ASBS · DW-4 | energy W2: 0.20 -> 0.11 | Improved |
| use Flow Sampling instead of ASBS | ASBS · LJ-55 | energy W2: 28.10 -> 21.32 | Improved |
| use Flow Sampling instead of ASBS | ASBS · Ala2 (train NFE 1024) | energy W2: 8.650 -> 0.637 | Improved |
| use Flow Sampling instead of ASBS (no auxiliary network) | ASBS · SPICE conformers | coverage recall: 89.66 -> 91.89 | Improved |
| use Flow Sampling instead of ASBS | ASBS · GEOM-DRUGS conformers | coverage recall: 74.50 -> 72.61 | Mixed |
Closest papers by meaning
- The Geometry of Noise: Why Diffusion Models Don't Need Noise Conditioning (NeurIPS 2026 · similarity 0.90)
- Adjoint Sampling: Highly Scalable Diffusion Samplers via Adjoint Matching (ICML 2025 · similarity 0.90)
- On scalable and efficient training of diffusion samplers (NeurIPS 2025 · similarity 0.90)
- Elucidating Flow Matching ODE Dynamics via Data Geometry and Denoisers (ICML 2025 · similarity 0.89)
- Diffusion Generative Flow Samplers: Improving learning signals through partial trajectory optimization (ICLR 2024 · similarity 0.89)
- Generative Modeling on Manifolds Through Mixture of Riemannian Diffusion Processes (ICML 2024 · 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.
