AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

Autoregressive Image Generation with Randomized Parallel Decoding

Haopeng Li, Jinyue Yang, Guoqi Li, Huan Wang

ICLR 2026 · Image generation › Autoregressive and latent generation · approach: Next-scale and token autoregressive images · 16 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 block-wise causal attention at inference instead of causal (ARPG-L, 16 steps) (ablation)ARPG-L · ImageNet 256x256FID: 3.51 -> 2.67Improved
use block-wise attention at 256 steps (ablation)ARPG-L · ImageNet 256x256FID: 2.43 -> 2.44No clear effect
reduce decoding steps from 256 to 32 (ablation)ARPG-L · ImageNet 256x256throughput (img/s): 17.45 -> 130.14Improved
reduce decoding steps from 256 to 32 (block-wise) (ablation)ARPG-XL · ImageNet 256x256FID: 2.01 -> 1.93Improved
replace RoPE with cosine positional encoding (ablation)ARPG-L · ImageNet 256 ablationFID: 3.51 -> 3.61Hurt
remove shared KV (separate KV projections) (ablation)ARPG-L · ImageNet 256 ablationFID: 3.51 -> 3.46Mixed
make all layers second-pass decoders (0+24) (ablation)ARPG-L · ImageNet 256 ablationFID: 3.51 -> 4.57Hurt
use fewer pass-2 decoder layers (18+6) (ablation)ARPG-L · ImageNet 256 ablationthroughput (img/s): 67.47 -> 62.35Hurt
add the two-pass decoder (vs none, standard AR with 24+0 layers) (ablation)ARPG-L · ImageNet 256 ablationthroughput (img/s): 11.70 -> 67.47Improved
random generation order instead of raster order (ablation)ARPG-L · ImageNet 256x256FID: 2.49 -> 2.44Improved
use spiral-out order instead of raster (ablation)ARPG-L · ImageNet 256x256FID: 2.49 -> 4.11Hurt
ARPG-XXL vs LlamaGen-XXLARPG-XXL · ImageNet 256x256FID: 2.62 -> 1.83Improved

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