Research map / Image generation / Fast diffusion sampling
How to Guide Your Flow: Few-Step Alignment via Flow Map Reward Guidance
Jerry Yuyang Huang, Justin Lin, Sheel Shah, Kartik Nair, Nicholas Matthew Boffi
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 |
|---|---|---|---|
| use flow-map Jacobian guidance (FMRG-J) instead of Euclidean gradient (FMRG-E) (ablation) | FMRG · GenEval with human preference rewards | GenEval score: 3.4 points at matched NFE | Improved |
| use FMRG-J instead of FMTT to cut function evaluations | FMTT · GenEval | NFE: FMTT 0.77 at NFE 20 matched with 70x fewer NFEs overall | Improved |
| add early stopping with t_stop = 0.25 (ablation) | FMRG · GenEval with human preference rewards | GenEval accuracy | Improved |
| add early stopping on l2 inverse problems (ablation) | FMRG · latent-space inverse problems | distortion and perception metrics | Mixed |
| use Euclidean gradient FMRG-E instead of Jacobian on l2 reconstruction (ablation) | FMRG · latent-space inverse problems | NFE-performance trade-off | Improved |
| use flow map instead of single-step denoiser for reward computation | DPS and FlowChef · style guidance | qualitative style transfer | Improved |
Closest papers by meaning
- Value Gradient Guidance for Flow Matching Alignment (NeurIPS 2025 · similarity 0.90)
- On the Guidance of Flow Matching (ICML 2025 · similarity 0.90)
- Align Your Flow: Scaling Continuous-Time Flow Map Distillation (NeurIPS 2025 · similarity 0.89)
- Aligning Flow Map Policies with Optimal $Q$-Guidance (NeurIPS 2026 · similarity 0.88)
- Flow-GRPO: Training Flow Matching Models via Online RL (NeurIPS 2025 · similarity 0.88)
- Greed is Good: A Unifying Perspective on Guided Generation (NeurIPS 2025 · 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.
