AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

One Model, Many Budgets: Elastic Latent Interfaces for Diffusion Transformers

Moayed Haji-Ali, Willi Menapace, Ivan Skorokhodov, Dogyun Park, Anil Kag, Michael Vasilkovsky, Sergey Tulyakov, Vicente Ordonez, Aliaksandr Siarohin

CVPR 2026 · Image generation › Autoregressive and latent generation · approach: High-resolution diffusion synthesis · 3 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 ELIT latent interface (Read/Write layers)DiT-XL · ImageNet-1K 512pxFID50K: 18.8 -> 11.1Improved
add multi-budget tail-drop training to ELITELIT-DiT · ImageNet-1K 512pxFID50K: 11.1 -> 10.1Improved
add multi-budget training to ELITELIT-UViT · ImageNet-1K 256px (with 0.25 CFG)FID50K: 3.8 -> 3.7No clear effect
add ELIT to DiT-XL for videoDiT-XL · Kinetics-700 256pxFVD10K: 135.9 -> 116.5Improved
use tail dropping instead of random token dropping (ablation)ELIT-DiT · ImageNet-1K 256px (25% tokens)FID10K: 38.6 -> 36.3Improved
use ELIT tail drop instead of multi-patch-size training for variable budget (ablation)DiT-B/2 · ImageNet-1K 256px (25% tokens)FID10K: 85.25 -> 36.3Improved
group size 4x4 vs 1x1 for Read/Write (ablation)ELIT-DiT · ImageNet-1K 256pxFID10K: 29.94 -> 26.53Improved
use a single group spanning the whole image (16x16) instead of 4x4 (ablation)ELIT-DiT · ImageNet-1K 256pxFID10K: 26.53 -> 30.03Hurt
add spatial head and tail blocks around the latent core (ablation)ELIT-DiT · DiT-B/2 ImageNet-1K 256pxFID10K: 33.84 -> 26.53Improved
increase head/tail to 5 each with only 2 core blocks (ablation)ELIT-DiT · DiT-B/2 ImageNet-1K 256pxFID10K: 25.34 -> 26.95Hurt
variable batch size vs constant batch size (ablation)ELIT-DiT · ImageNet-1K 256pxFID10K: 26.15 vs 26.65No clear effect
reduce inference tokens in ELIT-Qwen-Image from 100% to 12.5% (ablation)ELIT-Qwen-Image · DPG-Bench averagescore: 90.45 -> 88.02Mixed

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