AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

(1D) Ordered Tokens Enable Efficient Test-Time Search

Zhitong Gao, Parham Rezaei, Ali Cy, Mingqiao Ye, Nataša Jovanović, Jesse Allardice, Afshin Dehghan, Amir Zamir, Roman Bachmann, Oğuzhan Fatih Kar

ICML 2026 · Image generation › Autoregressive and latent generation · approach: Next-scale and token autoregressive images · 0 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 beam search on 1D ordered tokens (FlexTok) versus 2D grid tokens (ablation)AR text-to-image models with matched data, architecture and compute · COCO Karpathy val (300-image subset)CLIPScore vs inference computeImproved
use best-of-N and lookahead search on 1D ordered vs 2D grid tokens (ablation)AR text-to-image models · COCO Karpathy val (300-image subset)CLIPScore vs inference computeNo clear effect
add beam search with DreamSim verifier over first 32 tokensFlexTok · DreamBench++DINO-I: 32.5 -> 50.9Improved
add lookahead search with DreamSim verifierJanus-1.3B · DreamBench++DINO-I: 34.8 -> 40.7Improved
add beam search with DreamSim verifierFlexTok · DreamBench++CLIP-T (prompt following): 34.1 -> 33.1No clear effect
add search on top of conditional prior (ablation)FlexTok · GenEval subset, two objectaccuracy (%): 48 -> 81Improved
replace conditional AR prior with uniform prior in search (ablation)FlexTok · GenEval subset, two objectaccuracy (%): 81 -> 32Hurt
replace uniform prior with unconditional AR prior in search (ablation)FlexTok · GenEval subset, single objectaccuracy (%): 79 -> 85Improved
scale AR model size with best-of-N under limited vs large inference compute (ablation)FlexTok AR (212M to 3.4B) · GenEvaloverall accuracyMixed
use an ensemble of eight verifiers instead of a single verifier (ablation)FlexTok beam search · GenEvalaverage rank across metrics: 2.12 average rankImproved
use any verifier in beam search versus no search (ablation)FlexTok · GenEvalevaluation metricsImproved

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