AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

Speculative Coupled Decoding for Training-Free Lossless Acceleration of Autoregressive Visual Generation

Junhyuk So, Hyunho Kook, Chaeyeon Jang, Eunhyeok Park

ICML 2026 · Image generation › Autoregressive and latent generation · approach: Accelerating autoregressive image generation

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
replace independent draft sampling with maximal coupling (pi_MC) in SJD (ablation)SJD on Lumina-mGPT · MS-COCO, L=64NFE: 1035.9 -> 567.7Improved
replace independent draft sampling with Gumbel coupling (pi_GS) in SJD (ablation)SJD on Lumina-mGPT · MS-COCO, L=32NFE: 1031.2 -> 652.3Improved
increase window size L with plain SJD (ablation)SJD on Lumina-mGPT · MS-COCONFE: 1058.6 (L=16) -> 1035.9 (L=64)No clear effect
increase window size L with coupling (ablation)SCD (pi_MC) on Lumina-mGPT · MS-COCONFE: 814.5 -> 567.7Improved
add coupling on top of SJD (ablation)SJD on Lumina-mGPT · MS-COCOFID: 30.81 -> 30.83No clear effect
use lossless coupling instead of lossy grouped speculative decoding (GSD)Lumina-mGPT · MS-COCOFID: 33.21 -> 30.75Improved
add coupling on top of SJD for Janus-ProSJD on Janus-Pro 7B · MS-COCO, L=32NFE: 318.01 -> 154.42Improved
add coupling on top of SJD for video ARSJD on Cosmos-1-AR-4B · Real-Estate-10k, L=128latency (s): 47.73 -> 13.60Improved
increase coupling strength alpha from 0 to 1 (ablation)SJD on Lumina-mGPT · NFE vs mean token differenceNFEImproved
use pi_GS instead of pi_MC for easy-to-draft tasks (video, low-resolution images) (ablation)SCD · video AR and Janus-ProNFE: 577.8 -> 564.4 (video L=128)Improved
increase CFG scale (ablation)SCD · Lumina-mGPTspeedupHurt

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