Research map / Image generation / Fast diffusion sampling
Mixture of Efficient Diffusion Experts Through Automatic Interval and Sub-Network Selection
Alireza Ganjdanesh, Yan Kang, Yuchen Liu, Richard Zhang, Zhe Lin, Heng Huang
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 mixture of experts (two interval experts) to pruning (ablation) | DiffPruning, LSUN-Bedroom 50% MACs · LSUN-Bedroom | FID: 10.32 -> 9.65 | Improved |
| add Expert Routing Agent (ablation) | DiffPruning, LSUN-Bedroom 50% MACs · LSUN-Bedroom | FID: 9.65 -> 8.53 | Improved |
| add elastic depth fine-tuning (ablation) | DiffPruning, LSUN-Bedroom 50% MACs · LSUN-Bedroom | FID: 8.53 -> 8.03 | Improved |
| add elastic width fine-tuning (ablation) | DiffPruning, LSUN-Bedroom 50% MACs · LSUN-Bedroom | FID: 8.03 -> 6.73 | Improved |
| prune LDM to 50% MACs with DiffPruning | LDM · FFHQ | FID: 9.53 -> 9.90 | No clear effect |
| prune LDM to 50% MACs with DiffPruning | LDM · LSUN-Bedroom | FID: 4.39 -> 6.73 | Hurt |
| prune LDM to 30% MACs with DiffPruning | LDM · ImageNet | FID: 3.60 -> 13.18 | Hurt |
| DiffPruning 50% vs OMS-DPM on LSUN-Church | DiffPruning · LSUN-Church | FID: 11.10 -> 10.89 | Improved |
| DiffPruning 50% vs SP on ImageNet | DiffPruning · ImageNet | FID: 9.16 -> 8.45 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- OMS-DPM — OMS-DPM: Optimizing the Model Schedule for Diffusion Probabilistic Models (ICML 2023) · better on 0 of 1 shared result
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
- ALTER: All-in-One Layer Pruning and Temporal Expert Routing for Efficient Diffusion Generation (NeurIPS 2025 · similarity 0.91)
- Efficient Denoising Diffusion via Probabilistic Masking (ICML 2024 · similarity 0.90)
- Towards a Mechanistic Explanation of Diffusion Model Generalization (ICML 2025 · similarity 0.89)
- DiP-GO: A Diffusion Pruner via Few-step Gradient Optimization (NeurIPS 2024 · similarity 0.88)
- Remix-DiT: Mixing Diffusion Transformers for Multi-Expert Denoising (NeurIPS 2024 · similarity 0.88)
- Decouple-Then-Merge: Finetune Diffusion Models as Multi-Task Learning (CVPR 2025 · 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.
