Research map / Probabilistic and scientific modeling / Variational and generative sampling
Flow Matching for Generative Modeling
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 the OT path instead of the diffusion path (ablation) | Flow Matching · CIFAR-10 | NLL (BPD): 3.10 -> 2.99 | Helped |
| Use the OT path instead of the diffusion path (ablation) | Flow Matching · CIFAR-10 | FID: 8.06 -> 6.35 | Helped |
| Use the OT path instead of the diffusion path (ablation) | Flow Matching · CIFAR-10 | NFE: 183 -> 142 | Helped |
| Use the OT path instead of the diffusion path (ablation) | Flow Matching · ImageNet 32×32 | NLL (BPD): 3.54 -> 3.53 | Helped |
| Use the OT path instead of the diffusion path (ablation) | Flow Matching · ImageNet 32×32 | FID: 6.37 -> 5.02 | Helped |
| Use the OT path instead of the diffusion path (ablation) | Flow Matching · ImageNet 32×32 | NFE: 193 -> 122 | Helped |
| Use the OT path instead of the diffusion path (ablation) | Flow Matching · ImageNet 64×64 | NLL (BPD): 3.33 -> 3.31 | Helped |
| Use the OT path instead of the diffusion path (ablation) | Flow Matching · ImageNet 64×64 | FID: 16.88 -> 14.45 | Helped |
| Use the OT path instead of the diffusion path (ablation) | Flow Matching · ImageNet 64×64 | NFE: 187 -> 138 | Helped |
| Train with Flow Matching using the OT path rather than DDPM | ImageNet 128×128 unconditional generation · ImageNet 128×128 | FID: 25.3 -> 20.9 | Helped |
| Use Flow Matching with the OT path | ImageNet super-resolution · ImageNet validation 64×64→256×256 | FID: 5.2 -> 3.4 | Helped |
| Use Flow Matching with the OT path | ImageNet super-resolution · ImageNet validation 64×64→256×256 | PSNR: 26.4 -> 24.7 | Hurt |
Closest papers by meaning
- An Error Analysis of Flow Matching for Deep Generative Modeling (ICML 2025 · similarity 0.93)
- Reflected Flow Matching (ICML 2024 · similarity 0.92)
- Optimal Flow Matching: Learning Straight Trajectories in Just One Step (NeurIPS 2024 · similarity 0.91)
- Discrete Flow Matching (NeurIPS 2024 · similarity 0.91)
- On the Closed-Form of Flow Matching: Generalization Does Not Arise from Target Stochasticity (NeurIPS 2025 · similarity 0.91)
- Improving Flow Matching by Aligning Flow Divergence (ICML 2025 · similarity 0.91)
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.