AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

Generation then Reconstruction: Accelerating Masked Autoregressive Models via Two-Stage Sampling

Feihong Yan, Yao Zhu, Peiru Wang, Pang Kaiyu, Qingyan Wei, Huiqi Li, Linfeng Zhang

ICLR 2026 · Image generation › Autoregressive and latent generation · approach: Next-scale and token autoregressive images · 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
use GtR sampling instead of original 64-step MARMAR-H · ImageNet 256x256 class-conditionalspeedup / FID: 1.00 -> 3.72 speedup; FID 1.59 -> 1.59Improved
reduce sampling steps naively to 16MAR-H · ImageNet 256x256 class-conditionalFID: 1.59 -> 4.49Hurt
use GtR instead of LazyMAR at 8-step budgetMAR-H · ImageNet 256x256 class-conditionalFID: 4.06 -> 2.16Improved
use GtR sampling order instead of random order (ablation)MAR-H · ImageNet 256x256 class-conditionalFID: 1.82 -> 1.59Improved
use GtR sampling order instead of raster order (ablation)MAR-H · ImageNet 256x256 class-conditionalFID: 24.61 -> 1.59Improved
use GtR sampling order instead of 4-quadrant subsample (ablation)MAR-H · ImageNet 256x256 class-conditionalFID: 5.19 -> 1.59Improved
add GtR to the encoder-decoder (ablation)MAR-H · ImageNet 256x256 class-conditionalspeedup: 1.43 -> 2.90Improved
add GtR to the diffusion head (ablation)MAR-H · ImageNet 256x256 class-conditionalspeedup: 1.43 -> 1.90Improved
add GtR to both encoder-decoder and diffusion (ablation)MAR-H · ImageNet 256x256 class-conditionalspeedup: 1.43 -> 3.73Improved
add Frequency-Weighted Token Selection (ablation)GtR on MAR-H · ImageNet 256x256 class-conditionalFID: 1.65 -> 1.59Improved
select high-frequency tokens instead of random tokens for extra diffusion steps (ablation)GtR on MAR-H · ImageNet 256x256 class-conditionalFID: 1.64 -> 1.59Improved
select high-frequency tokens instead of low-frequency tokens (ablation)GtR on MAR-H · ImageNet 256x256 class-conditionalIS: 301.5 -> 304.4Improved

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