Research map / Image generation / Fast diffusion sampling
Quantization-Aware Diffusion Models For Maximum Likelihood Training
Shohei Taniguchi, Masahiro Suzuki, Yutaka Matsuo
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 |
|---|---|---|---|
| replace uniform dequantization / categorical output with quantization-aware parameterization (QDPM) | VDM-style diffusion · CIFAR-10 test | NLL (bpd): 2.85 -> 0.27 | Improved |
| use QDPM instead of prior models | diffusion density estimation · ImageNet-32 (new version) | NLL (bpd): 3.43 -> 0.32 | Improved |
| add data augmentation to prior diffusion likelihood models | VDM · CIFAR-10 test | NLL (bpd): 2.65 -> 2.49 | Improved |
| use uniform dequantization instead of categorical quantized output (ablation) | VDM · CIFAR-10 test | NLL (bpd): 2.65 -> 2.85 | Hurt |
| train QDPM with maximum likelihood objective | QDPM · CIFAR-10 | FID: FID 5.60 vs 1.54 for GDD | Hurt |
Closest papers by meaning
- Towards Accurate Post-training Quantization for Diffusion Models (CVPR 2024 · similarity 0.93)
- Bridging the Gap between Gaussian Diffusion Models and Universal Quantization for Image Compression (CVPR 2025 · similarity 0.91)
- Optimizing Quantized Diffusion Models via Distillation with Cross-Timestep Error Correction (AAAI 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)
- EfficientDM: Efficient Quantization-Aware Fine-Tuning of Low-Bit Diffusion Models (ICLR 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.
