Research map / Probabilistic and scientific modeling / Molecules and proteins
A Kinetic Energy Perspective of Flow Matching
Ziyun Li, Huancheng Hu, Soon Hoe Lim, Xuyu Li, Fei Gao, Enmao Diao, Zezhen Ding, Michalis Vazirgiannis, Henrik Boström
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 |
|---|---|---|---|
| add Kinetic Soft Landing (b0=0.02) to FM sampling (ablation) | FM · CelebA 32x32 | F_mem (%): 37.34 -> 19.36 | Improved |
| add Kinetic Soft Landing (b0=0.02) alone (ablation) | FM · CelebA 32x32 | FID@10k: 16.68 -> 86.56 | Hurt |
| add Kinetic Launch (a0=0.02) alone (ablation) | FM · CelebA 32x32 | FID@10k: 16.68 -> 11.27 | Improved |
| add Kinetic Launch (a0=0.02) alone (ablation) | FM · CelebA 32x32 | F_mem (%): 37.34 -> 36.78 | No clear effect |
| add both launch and landing (a0=b0=0.01) (ablation) | FM · CelebA 32x32 | FID@10k: 16.68 -> 14.35 | Improved |
| add both launch and landing (a0=b0=0.01) (ablation) | FM · CelebA 32x32 | F_mem (%): 37.34 -> 31.22 | Improved |
| add Kinetic Launch a0=0.05 alone (ablation) | FM · ImageNet-256 | FID@10K: 11.70 -> 11.59 | Improved |
| add Kinetic Launch a0=0.05 alone (ablation) | FM · ImageNet-256 | Recall: 0.655 -> 0.630 | Hurt |
| add Kinetic Soft Landing b0=0.05 alone (ablation) | FM · ImageNet-256 | Recall: 0.655 -> 0.657 | Improved |
| move tau_split from 0.2 to 0.6 (ablation) | KTS a0=b0=0.01 · CelebA 32x32 | FID@10k: 60.31 -> 14.35 | Improved |
| move tau_split from 0.8 to 0.6 (ablation) | KTS a0=b0=0.01 · CelebA 32x32 | F_mem (%): 34.30 -> 31.22 | Improved |
| use exponential landing instead of linear/constant landing (ablation) | KTS a0=b0=0.01 · CelebA 32x32 | F_mem (%): 35.50 -> 31.22 | Improved |
Closest papers by meaning
- Flow Matching with General Discrete Paths: A Kinetic-Optimal Perspective (ICLR 2025 · similarity 0.89)
- Energy Matching: Unifying Flow Matching and Energy-Based Models for Generative Modeling (NeurIPS 2025 · similarity 0.89)
- Physics-Constrained Flow Matching: Sampling Generative Models with Hard Constraints (NeurIPS 2025 · similarity 0.89)
- Trajectory as the Teacher: Few-Step Discrete Flow Matching via Energy-Navigated Distillation (NeurIPS 2026 · similarity 0.89)
- Shortcutting Pre-trained Flow Matching Diffusion Models is Almost Free Lunch (NeurIPS 2025 · similarity 0.88)
- Optimal Flow Matching: Learning Straight Trajectories in Just One Step (NeurIPS 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.
