AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

Accelerating Diffusion Model Training under Minimal Budgets: A Condensation-Based Perspective

Rui Huang, Shitong Shao, zikai zhou, Pukun Zhao, Hangyu Guo, Tian Ye, Lichen Bai, Shuo Yang, Zeke Xie

CVPR 2026 · Image generation › Fast diffusion sampling · approach: Training-free diffusion acceleration · 7 citations (OpenAlex)

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 changedWhereMeasuredEffect
add the Select stage (diffusion difficulty plus interval sampling) to the Attach-only data (ablation)D2C on DiT-L/2 · ImageNet 256 10K budgetgFID: 37.07 -> 14.96Improved
add visual representation alignment to Select (ablation)D2C on DiT-L/2 · ImageNet 256 10K budgetgFID: 14.96 -> 10.37Improved
add DC embedding to Select (ablation)D2C on DiT-L/2 · ImageNet 256 10K budgetgFID: 14.96 -> 9.01Improved
add both DC embedding and visual embedding to Select (ablation)D2C on DiT-L/2 · ImageNet 256 10K budgetgFID: 14.96 -> 7.62Improved
add Select on top of Attach (ablation)D2C on DiT-L/2 · ImageNet 256 10K budgetgFID: 8.79 -> 7.62Improved
replace one-hot class embedding alone with text plus class dual embedding (ablation)D2C · ImageNet 256 10K budgetgFIDImproved
use interval sampling instead of lowest-loss or highest-loss selection (ablation)D2C Select · ImageNet 10K budgetgFID-10KImproved
replace MAE-L visual encoder with DINOv2-L (ablation)D2C · ImageNet 256 10K budgetgFID: 9.23 -> 7.62Improved
use a pretrained scorer instead of a scorer trained from scratch (ablation)D2C Select · ImageNet-1K SiT-XL/2gFID: 4.9 -> 4.3Improved
add Select to Attach-only training (ablation)D2C on SiT-XL/2 · ImageNet-1K 40k itergFID: 5.6 -> 4.3Improved
replace random sampling with D2CD2C · ImageNet 256 0.8% budget, 100k itergFID-50K: 35.86 -> 4.20Improved
replace random sampling with D2C at long trainingD2C · ImageNet 256 0.8% budget, 300k iter DiT-L/2gFID-50K: 4.19 -> 4.13No clear effect

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy