AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

Multi-Scale Local Speculative Decoding for Image Generation

Elia Peruzzo, Guillaume Sautière, Amirhossein Habibian

CVPR 2026 · Image generation › Autoregressive and latent generation · approach: Next-scale and token autoregressive images · 3 citations (OpenAlex)

Read the paper

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 LANTERN with MuLo-SD on Tar-7BTar-7B 1024p · MS-COCO 5k val, 1024pspeedup: 1.45x -> 5.33xImproved
replace LANTERN with MuLo-SD on Tar-1.5BTar-1.5B 1024p · MS-COCO 5k val, 1024pGenEval: 75.4 -> 76.8Improved
use MuLo-SD instead of EAGLE-2 on Tar-1.5BTar-1.5B 512p · MS-COCO 5k val, 512pspeedup: 0.72x -> 1.94xImproved
apply LANTERN to the stronger Tar models instead of LlamaGenLANTERN · MS-COCO 5k val, 512pspeedup: 1.59x -> 1.08xHurt
add MuLo-SD to LlamaGen-XLLlamaGen-XL 512p · MS-COCO 5k val, 512pspeedup: 1.40x vs 1.59x LANTERNMixed
add local rejection and resampling instead of raster-scan rejection (ablation)MuLo-SD · GenEval vs speedup, 512pGenEvalImproved
add neighborhood expansion to local verification (ablation)MuLo-SD · GenEval vs speedup, 512pGenEvalImproved
add codebook-neighbor probability pooling (ablation)MuLo-SD · GenEval vs speedup, 512pGenEvalImproved
add parallel decoding (ZipAR drafting, island resampling) (ablation)MuLo-SD · MS-COCO 5k vallatencyImproved
train up/down-samplers with pixel-space MSE and LPIPS instead of token classification loss (ablation)MuLo-SD upsampler · visual fidelityperceptual qualityImproved
add PatchGAN adversarial loss to upsampler training (ablation)MuLo-SD upsampler · visual fidelityperceptual qualityImproved
use off-the-shelf pixel-space upsampler instead of learned latent upsampler (ablation)MuLo-SD upsampler · GenEval vs speedup, 512pGenEvalHurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

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

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