Research map / Probabilistic and scientific modeling / Variational and generative sampling
Preconditioned Flow Matching
Shadab Ahamed, Eshed Gal, Md Shahriar Rahim Siddiqui, Simon Ghyselincks, Moshe Eliasof, Eldad Haber
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
placeholder
Problem
placeholder
Approach
placeholder
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add NF preconditioner | latent FM · MNIST | FID: 5.97 -> 2.74 | Improved |
| add FM preconditioner | latent FM · MNIST | FID: 5.97 -> 3.98 | Improved |
| add ZCA whitening | latent FM · MNIST | FID: 5.97 -> 4.20 | Improved |
| enlarge unpreconditioned FM to 3.37M params (compute matched) (ablation) | latent FM · MNIST | FID: 5.97 -> 5.42 | Improved |
| add FM preconditioner vs compute-matched baseline (ablation) | latent FM · MNIST | FID: 5.42 -> 3.98 | Improved |
| add FM preconditioner | flow matching · LSUN Churches 256x256 | FID: 19.53 -> 14.47 | Improved |
| add FM preconditioner | flow matching · Oxford Flowers-102 256x256 | FID: 23.82 -> 22.64 | Improved |
| add FM preconditioner | flow matching · AFHQ Cats 512x512 | FID: 7.11 -> 4.86 | Improved |
| add FM preconditioner | flow matching · CIFAR-10 32x32 | FID: 4.15 -> 4.02 | No clear effect |
| add FM preconditioner | flow matching · ImageNet-1k 64x64 | FID: 8.66 -> 4.98 | Improved |
| add FM preconditioner | flow matching · CIFAR-10 32x32 | Precision: 0.8097 -> 0.7856 | Hurt |
| add FM preconditioner | flow matching · AFHQ Cats 512x512 | Recall: 0.5396 -> 0.5297 | Hurt |
Closest papers by meaning
- Coreset-Induced Conditional Velocity Flow Matching (NeurIPS 2026 · similarity 0.90)
- Stream-level Flow Matching with Gaussian Processes (ICML 2025 · similarity 0.89)
- A Theory on Flow Matching with Neural Networks (NeurIPS 2026 · similarity 0.89)
- Value Gradient Guidance for Flow Matching Alignment (NeurIPS 2025 · similarity 0.89)
- Variational Rectified Flow Matching (ICML 2025 · similarity 0.89)
- Gaussian Mixture Flow Matching Models (ICML 2025 · similarity 0.89)
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.
