Research map / Image generation / Fast diffusion sampling
Rethinking Conditional Diffusion Sampling with Progressive Guidance
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 |
|---|---|---|---|
| add ProG to classifier guidance | CADM-G · ImageNet 256x256 | IS: 188.91 -> 222.09 | Helped |
| add ProG to classifier guidance | ADM-G · ImageNet 64x64 | FID: 6.40 -> 5.16 | Helped |
| add ProG to classifier guidance | IDDPM-G · ImageNet 64x64 | FID: 13.62 -> 11.12 | Helped |
| add ProG to classifier guidance | ADM-G · CIFAR10 32x32 | FID: 2.85 -> 2.81 | No clear effect |
| add ProG to classifier guidance | CADM-G · ImageNet 64x64 | IS: 66.52 -> 65.65 | Hurt |
| add ProG to EDS-enhanced guidance | CADM-G + EDS · ImageNet 256x256 | IS: 212.51 -> 232.86 | Helped |
| add ProG to EDS-enhanced guidance | CADM-G + EDS · ImageNet 64x64 | FID: 1.85 -> 1.77 | Helped |
| add ProG to DiT-G | DiT-G · ImageNet 256x256 | FID: 2.27 -> 2.25 | Helped |
| add ProG to classifier guidance | CADM-G · ImageNet 256x256 | robustness: 79.65 -> 85.04 | Helped |
| add ProG to CADM-G + EDS | CADM-G + EDS · ImageNet 256x256 | robustness@1: 83.31 -> 86.60 | Helped |
| increase guidance scale w with ProG vs without | ADM-G · ImageNet 64x64 | FID and Recall | Helped |
| remove the schedule (gamma = 0) (ablation) | ProG · ImageNet 64x64 | FID: 5.16 -> 73.10 | Hurt |
Closest papers by meaning
- Elucidating the design space of classifier-guided diffusion generation (ICLR 2024 · similarity 0.91)
- REG: Rectified Gradient Guidance for Conditional Diffusion Models (ICML 2025 · similarity 0.89)
- Adaptive Guidance: Training-free Acceleration of Conditional Diffusion Models (AAAI 2025 · similarity 0.89)
- Rectified Diffusion Guidance for Conditional Generation (CVPR 2025 · similarity 0.89)
- CADS: Unleashing the Diversity of Diffusion Models through Condition-Annealed Sampling (ICLR 2024 · similarity 0.89)
- Provably Reliable Classifier Guidance via Cross-Entropy Control (NeurIPS 2026 · 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.