AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

Overclocking Electrostatic Generative Models

Daniil Shlenskii, Alexander Korotin

ICML 2026 · Image generation › Fast diffusion sampling · approach: Rectified flow straightening

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
distill with IPFM (alpha=0.5) to 4 stepsPFGM++ D=128 · CIFAR-10 (unconditional)FID: 1.92 (35 NFE teacher) vs 2.08No clear effect
distill with IPFM (alpha=0.5) to 2 stepsPFGM++ D=128 · FFHQ 64x64FID: 2.43 -> 2.12Improved
add SiD-inspired regularization (alpha 0.5 -> 1.0) (ablation)IPFM D=128 · CIFAR-10, 4 stepsFID: 2.08 -> 1.75Improved
add SiD-inspired regularization (alpha 0.5 -> 1.0) (ablation)IPFM D=128 · CIFAR-10, 1 stepFID: 5.38 -> 3.31Improved
add SiD-inspired regularization (alpha 0.5 -> 1.0) (ablation)IPFM D=128 · FFHQ 64x64, 2 stepsFID: 2.12 -> 1.72Improved
add SiD-inspired regularization (alpha 0.5 -> 1.0) (ablation)IPFM D=infinity · CIFAR-10, 4 stepsFID: 2.13 -> 1.86Improved
use finite D=128 instead of diffusion limitIPFM one-step · CIFAR-10, 1 step, 70hFID: 3.611 -> 3.216Improved
use SiD with heavy-tailed kernel instead of Gaussian (ablation)SiD D=128 · CIFAR-10, 1 step, 70hFID: 3.611 -> 3.336Improved
use full IPFM D=128 instead of SiD with heavy tails (ablation)IPFM · CIFAR-10, 1 step, 70hFID: 3.336 -> 3.216Improved
use IPFM instead of Consistency Trajectory Models (no GAN loss)distillation · PFGM++ at D=128 and D=infFIDImproved
use finite D teacher (128 vs 2048)IPFM alpha=1.0 · CIFAR-10, 4 stepsFID: 1.75 vs 1.82No 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