Research map / Image generation / Fast diffusion sampling
Absorbing Quantization Error by Deformable Noise Scheduler for Diffusion Models
Mingrui Yang, Wei Huang, Hao Nan SHENG, Donglin Yang, Jichang Yang, Xin Yu, Huining Yu, Yuzhong Jiao, Zhongrui Wang, Xiaojuan Qi
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 DNS to Q-Diffusion | LDM-4 W4A8 · ImageNet 256 | FID: 10.68 -> 9.81 | Improved |
| add DNS to Q-Diffusion | LDM-4 W4A8 · ImageNet 256 | Image Reward: -0.014 -> 0.015 | Improved |
| add PTQD to Q-Diffusion | LDM-4 W4A8 · ImageNet 256 | FID: 10.68 -> 10.32 | Improved |
| add DNS to PTQ4DiT | DiT-XL W4A8 · ImageNet 256 | FID: 9.83 -> 8.51 | Improved |
| add PTQD to PTQ4DiT | DiT-XL W4A8 · ImageNet 256 | FID: 9.83 -> 14.01 | Hurt |
| add DNS to SVDQuant with W16 LoRA | FLUX.1 W4A4 · MJHQ-30K | FID: 22.95 -> 22.78 | Improved |
| add DNS to SVDQuant with W16 LoRA | FLUX.1 W4A4 · MJHQ-30K | Image Reward: 0.877 -> 0.890 | Improved |
| add DNS to SVDQuant with 6-bit LoRA | FLUX.1 W4A4 · MJHQ-30K | FID: 27.16 -> 26.22 | Improved |
| add DNS to SVDQuant with 6-bit LoRA | FLUX.1 W4A4 · MJHQ-30K | Image Reward: 0.693 -> 0.709 | Improved |
| add DNS to Q-Diffusion | LDM-4 W4A8 · ImageNet 256 | PSNR: 22.84 -> 23.10 | No clear effect |
| add DNS to SVDQuant with 6-bit LoRA | FLUX.1 W4A4 · MJHQ-30K | PSNR: 16.17 -> 15.87 | Hurt |
| quantize SVDQuant LoRA branch to 6 bits | FLUX.1 W4A4 · MJHQ-30K | FID: 22.64 -> 27.16 | Hurt |
Closest papers by meaning
- Towards Accurate Post-training Quantization for Diffusion Models (CVPR 2024 · similarity 0.93)
- Optimizing Quantized Diffusion Models via Distillation with Cross-Timestep Error Correction (AAAI 2025 · similarity 0.92)
- AccuQuant: Simulating Multiple Denoising Steps for Quantizing Diffusion Models (NeurIPS 2025 · similarity 0.91)
- TCAQ-DM: Timestep-Channel Adaptive Quantization for Diffusion Models (AAAI 2025 · similarity 0.91)
- Q-DiT: Accurate Post-Training Quantization for Diffusion Transformers (CVPR 2025 · similarity 0.91)
- D^2-DPM: Dual Denoising for Quantized Diffusion Probabilistic Models (AAAI 2025 · 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.
