Research map / Image generation / Fast diffusion sampling
Score-Based Generative Modeling with Critically-Damped Langevin Diffusion
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 |
|---|---|---|---|
| Replace direct data-space diffusion with critically-damped Langevin diffusion | Score-based generative models · CIFAR-10 unconditional | FID: 2.25 vs 3.08 | Helped |
| Use SSCS-QS instead of EM-QS | CLD sampler · CIFAR-10 unconditional, 50 NFEs | FID: 52.7 -> 20.5 | Helped |
| Use SSCS-QS instead of EM-QS | CLD sampler · CIFAR-10 unconditional, 150 NFEs | FID: 7.00 -> 3.07 | Helped |
| Use SSCS-QS instead of EM-QS | CLD sampler · CIFAR-10 unconditional, 275 NFEs | FID: 3.24 -> 2.38 | Helped |
| Use SSCS-QS instead of EM-QS | CLD sampler · CIFAR-10 unconditional, 500 NFEs | FID: 2.41 -> 2.25 | Helped |
| Use SSCS-QS instead of EM-QS | CLD sampler · CIFAR-10 unconditional, 1000 NFEs | FID: 2.27 -> 2.30 | Hurt |
| Use SSCS-QS instead of EM-QS | CLD sampler · CIFAR-10 unconditional, 2000 NFEs | FID: 2.23 -> 2.29 | Hurt |
| Use probability-flow ODE instead of GGF | CLD sampler · CIFAR-10 unconditional, similar adaptive-solver compute | FID: 2.25 vs 2.56 | Helped |
| Vary reciprocal mass across tested values (ablation) | CLD-SGM · CIFAR-10 unconditional, smaller model | FID: 3.23; 3.14; 3.16 | No clear effect |
| Increase initial velocity variance scaling from 0.04 to 0.4 or 1 (ablation) | CLD-SGM · CIFAR-10 unconditional, smaller model | FID: 3.14 -> 3.21; 3.14 -> 3.27 | Hurt |
| Add mixed score parameterization (ablation) | CLD-SGM · CIFAR-10 unconditional, smaller model | FID: 3.56 -> 3.14 | Helped |
| Replace HSM with regular DSM (ablation) | CLD-SGM · CLD training | training stability | Hurt |
Closest papers by meaning
- Wasserstein Convergence of Critically Damped Langevin Diffusions (NeurIPS 2025 · similarity 0.93)
- Noise Conditional Variational Score Distillation (ICML 2025 · similarity 0.91)
- Denoising Diffusion Bridge Models (ICLR 2024 · similarity 0.91)
- Evaluating the design space of diffusion-based generative models (NeurIPS 2024 · similarity 0.90)
- Assessing the quality of denoising diffusion models in Wasserstein distance: noisy score and optimal bounds (NeurIPS 2025 · similarity 0.90)
- O(d/T) Convergence Theory for Diffusion Probabilistic Models under Minimal Assumptions (ICLR 2025 · similarity 0.90)
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.