AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

reAR: Rethinking Visual Autoregressive Models via Token-wise Consistency Regularization

Qiyuan He, Yicong Li, Haotian Ye, Jinghao Wang, Xinyao Liao, Pheng-Ann Heng, Stefano Ermon, James Zou, Angela Yao

ICLR 2026 · Image generation › Autoregressive and latent generation · approach: Next-scale and token autoregressive images

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
add reAR (noisy context + codebook embedding regularization)raster AR-L with MaskGIT VQGAN · ImageNet-1K 256x256gFID: 3.02 -> 1.86Improved
add reAR to non-standard tokenizersTiTok and AliTok AR · ImageNet-1K 256x256gFID: TiTok 4.45 -> 4.01; AliTok 1.50 -> 1.42Improved
add decoding embedding regularization at layer 15 (predict next-token embedding) (ablation)vanilla AR (reAR-S, 80 ep) · ImageNet-256 no CFGFID: 21.32 -> 20.03Improved
place decoding regularization at early layer 10 (ablation)vanilla AR (reAR-S, 80 ep) · ImageNet-256 no CFGFID: 21.32 -> 21.29No clear effect
tie codebook embeddings to the AR model directly (ablation)vanilla AR (reAR-S, 80 ep) · ImageNet-256 no CFGFID: 21.32 -> 21.08No clear effect
add encoding regularization at layer 0 on top of DE@15 (ablation)reAR-S (80 ep) · ImageNet-256 no CFGFID: 20.03 -> 19.72Improved
apply encoding regularization at layer 5 instead of layer 0 (ablation)reAR-S (80 ep) · ImageNet-256 no CFGFID: 19.72 -> 21.36Hurt
vary regularization weight lambda (0.5, 1, 1.5) (ablation)reAR-S (80 ep) · ImageNet-256 no CFGFID: 19.79 / 19.72 / 19.74No clear effect
add fixed noisy context (eps=0.25) to embedding regularization (ablation)reAR-S (400 ep) · ImageNet-256FID: 2.12 -> 2.08Improved
raise fixed noise level to 0.5 (ablation)reAR-S (400 ep) · ImageNet-256FID: 2.12 -> 3.15Hurt
anneal the maximum noise level over training (truncated linear) (ablation)reAR-S (400 ep) · ImageNet-256FID: 2.05 -> 2.00Improved
add embedding regularization to noisy context regularization (ablation)reAR-S (400 ep) · ImageNet-256FID: 2.18 -> 2.00Improved

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