Research map / Architectures and efficiency / Quantization
DiffPro: Joint Timestep and Layer-Wise Precision Optimization for Efficient Diffusion Inference
Farhana Amin, Sabiha Afroz, Kanchon Gharami, Mona Moghadampanah, Dimitrios Nikolopoulos
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 dynamic activation quantization (DAQ) (ablation) | DiffPro (DiT-XL/2) · ImageNet 256 | FID: 31.17 -> 28.05 | Improved |
| use learned per-layer bit plan instead of uniform W8/G32 (ablation) | DiffPro (DiT-XL/2) · ImageNet 256 | FID: 31.84 -> 28.05 | Improved |
| add drift-guided timestep pruning (1000 -> about 500 steps) (ablation) | DiffPro (DiT-XL/2) · ImageNet 256 | FID: FID 30.38 (no prune) vs 28.05; latency 2.65 -> 2.55 s | Mixed |
| add drift-guided timestep pruning (ablation) | DiffPro (DiT-XL/2) · ImageNet 256 | latency (sec/image): 2.65 -> 2.55 | Improved |
| apply joint INT8 quantization plan plus pruning vs full precision | DiT-XL/2 · ImageNet 256 | latency and size: latency 6.88 -> 2.55 s; size 2575.42 -> 397.241 MB; FID 20 -> 28.05 | Mixed |
| compare with Q-DiT and DiffuseVAE on FID | DiffPro (DiT) · ImageNet 256 | FID: 28.05 vs 21.59 (QDiT), 23.20 (DiffuseVAE) | Hurt |
Closest papers by meaning
- TCAQ-DM: Timestep-Channel Adaptive Quantization for Diffusion Models (AAAI 2025 · similarity 0.91)
- Faster Diffusion: Rethinking the Role of the Encoder for Diffusion Model Inference (NeurIPS 2024 · similarity 0.91)
- Optimizing Quantized Diffusion Models via Distillation with Cross-Timestep Error Correction (AAAI 2025 · similarity 0.91)
- PFDiff: Training-Free Acceleration of Diffusion Models Combining Past and Future Scores (ICLR 2025 · similarity 0.91)
- Q-DiT: Accurate Post-Training Quantization for Diffusion Transformers (CVPR 2025 · similarity 0.90)
- EfficientDM: Efficient Quantization-Aware Fine-Tuning of Low-Bit Diffusion Models (ICLR 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.
