Research map / Image generation / Fast diffusion sampling
Sampling-Aware Quantization for Diffusion Models
Qian Zeng, Jie Song, Yuanyu Wan, Huiqiong Wang, Mingli Song
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 Mixed-Order Trajectory Alignment Calibration (ablation) | BRECQ W8A8 · ImageNet 256x256 LDM-4 (20 steps, DPM-Solver-2) | FID: 14.26 -> 10.16 | Improved |
| add Mixed-Order Trajectory Alignment Calibration (ablation) | BRECQ W8A8 · ImageNet 256x256 LDM-4 (20 steps, DPM-Solver-2) | sFID: 13.72 -> 9.89 | Improved |
| add L_COS direction constraint (ablation) | QLoRA W4A8 · ImageNet 256x256 LDM-4 (20 steps, DPM-Solver-2) | FID: 9.91 -> 9.50 | Improved |
| add L_COS direction constraint (ablation) | QLoRA W4A8 · ImageNet 256x256 LDM-4 (20 steps, DPM-Solver-2) | sFID: 8.76 -> 8.82 | No clear effect |
| add L_MOTA on top of L_COS (ablation) | QLoRA W4A8 · ImageNet 256x256 LDM-4 (20 steps, DPM-Solver-2) | FID: 9.50 -> 8.55 | Improved |
| SA-QLoRA vs EfficientDM at W4A4 | LDM-4 ImageNet · ImageNet 256x256 LDM-4 (20 steps, DPM-Solver-2) | FID: 17.28 -> 13.73 | Improved |
| SA-QLoRA vs EfficientDM at W4A4 | LDM-4 LSUN-Bedroom · LSUN-Bedroom 256x256 LDM-4 (50 steps) | FID: 19.30 -> 14.72 | Improved |
| SA-PTQ vs PTQD at W8A8 | LDM-4 LSUN-Bedroom · LSUN-Bedroom 256x256 LDM-4 (50 steps) | FID: 10.31 -> 9.83 | Improved |
| SA-QLoRA vs EfficientDM at W4A8 | SD-v1.4 MS-COCO · MS-COCO 512x512 SD-v1.4 | FID: 13.31 -> 13.27 | No clear effect |
| SA-QLoRA vs EfficientDM at W4A8 | SD-v1.4 MS-COCO · MS-COCO 512x512 SD-v1.4 | sFID: 19.92 -> 20.26 | Hurt |
| SA-PTQ vs PTQD at W8A8 | SD-v1.4 MS-COCO · MS-COCO 512x512 SD-v1.4 | FID: 13.65 -> 13.10 | Improved |
| quantized SA-QLoRA W4A8 vs full precision | LDM-4 ImageNet · ImageNet 256x256 LDM-4 (20 steps, DPM-Solver-2) | FID: 9.45 -> 8.55 | Improved |
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.
- PTQD — PTQD: Accurate Post-Training Quantization for Diffusion Models (NeurIPS 2023) · better on 5 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
- D^2-DPM: Dual Denoising for Quantized Diffusion Probabilistic Models (AAAI 2025 · similarity 0.91)
- AccuQuant: Simulating Multiple Denoising Steps for Quantizing Diffusion Models (NeurIPS 2025 · similarity 0.91)
- Optimizing Quantized Diffusion Models via Distillation with Cross-Timestep Error Correction (AAAI 2025 · similarity 0.91)
- CCS: Controllable and Constrained Sampling with Diffusion Models via Initial Noise Perturbation (NeurIPS 2025 · similarity 0.91)
- Towards Accurate Post-training Quantization for Diffusion Models (CVPR 2024 · similarity 0.90)
- A Simple Early Exiting Framework for Accelerated Sampling in Diffusion Models (ICML 2024 · 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.
