Research map / Image generation / Fast diffusion sampling
Are First-Order Diffusion Samplers Really Slower? A Fast Forward-Value Approach
Yuchen Jiao, Na Li, Changxiao Cai, Gen Li
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 forward-value sampler instead of DDIM | first-order sampler · CIFAR-10 (unconditional), NFE=4 | FID: 67.03 -> 25.02 | Improved |
| use forward-value sampler instead of DDIM | first-order sampler · ImageNet512 EDM2-XS, NFE=4 | FID: 91.44 -> 52.32 | Improved |
| use forward-value sampler instead of DDIM | first-order sampler · FFHQ latent diffusion, NFE=4 | FID: 80.19 -> 15.78 | Improved |
| use forward-value first-order sampler instead of DPMSolver-3 | sampler · ImageNet512 EDM2-XXL, NFE=4 | FID: 59.34 -> 50.96 | Improved |
| use forward-value first-order sampler instead of UniPC-3 at NFE 10 | sampler · CIFAR-10 (unconditional), NFE=10 | FID: 2.86 -> 3.46 | Hurt |
| use forward-value sampler instead of DPMSolver-3 at NFE 4 | sampler · LSUN Bedroom, NFE=4 | FID: 10.13 -> 12.17 | Hurt |
| use forward-value sampler instead of DPMSolver-3 at NFE 4 | sampler · FFHQ, NFE=4 | FID: 15.65 -> 15.78 | No clear effect |
| add DPMSolver-2 to the forward-value sampler (ablation) | Ours · CIFAR-10 (conditional), NFE=6 | FID: 7.57 -> 3.66 | Improved |
| add DPMSolver-2 to the forward-value sampler (ablation) | Ours · CIFAR-10 (unconditional), NFE=4 | FID: 25.02 -> 22.30 | Improved |
| add DPMSolver-2 to the forward-value sampler (ablation) | Ours · CIFAR-10 (conditional), NFE=10 | FID: 3.18 -> 3.32 | Hurt |
| add DPMSolver-2 to the forward-value sampler (ablation) | Ours · ImageNet512, NFE=10 (hybrid table, first model column) | FID: 4.57 -> 4.90 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- UniPC-3 — UniPC: A Unified Predictor-Corrector Framework for Fast Sampling of Diffusion Models (NeurIPS 2023) · better on 3 of 5 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- A Unified Sampling Framework for Solver Searching of Diffusion Probabilistic Models (ICLR 2024 · similarity 0.93)
- Accelerating Diffusion Sampling with Optimized Time Steps (CVPR 2024 · similarity 0.93)
- PFDiff: Training-Free Acceleration of Diffusion Models Combining Past and Future Scores (ICLR 2025 · similarity 0.93)
- Enhanced Diffusion Sampling via Extrapolation with Multiple ODE Solutions (ICLR 2025 · similarity 0.92)
- Accelerating Convergence of Score-Based Diffusion Models, Provably (ICML 2024 · similarity 0.92)
- Distilling ODE Solvers of Diffusion Models into Smaller Steps (CVPR 2024 · 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.
