AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

Representation Alignment for Just Image Transformers is not Easier than You Think

Jaeyo Shin, Jiwook Kim, Hyunjung Shim

ECCV 2026 · Image generation › Autoregressive and latent generation · approach: High-resolution diffusion synthesis · 18 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
apply standard REPA (MLP projection) to JiTJiT-B/16 · ImageNet 256x256 (200 epochs)FID: 4.37 -> 5.14Hurt
replace REPA MLP head with Masked Transformer Adapter (PixelREPA)JiT-B/16 + REPA · ImageNet 256x256 (200 epochs)FID: 5.14 -> 4.00Improved
add shallow transformer adapter without masking in place of the MLP head (ablation)JiT-B/16 + REPA · ImageNet 256x256 (200 epochs)FID: 5.14 -> 4.68Improved
add partial token masking (ratio 0.2) to the adapter input (ablation)PixelREPA adapter · ImageNet 256x256 (200 epochs)FID: 4.68 -> 4.00Improved
PixelREPA without masking vs vanilla JiT (ablation)JiT-B/16 · ImageNet 256x256 (200 epochs)FID: 4.37 -> 4.68Hurt
raise mask ratio from 0.2 to 0.5 (ablation)PixelREPA-B/16 · ImageNet 256x256 (200 epochs)FID: 4.00 -> 4.58Hurt
lower mask ratio from 0.2 to 0.1 (ablation)PixelREPA-B/16 · ImageNet 256x256 (200 epochs)FID: 4.00 -> 4.26Hurt
apply PixelREPA to JiT-B/16JiT-B/16 · ImageNet 256x256FID: 3.66 -> 3.17Improved
apply PixelREPA to JiT-B/16JiT-B/16 · ImageNet 256x256Inception Score: 275.1 -> 284.6Improved
apply PixelREPA to JiT-L/16JiT-L/16 · ImageNet 256x256FID: 2.36 -> 2.11Improved
apply PixelREPA to JiT-H/16JiT-H/16 · ImageNet 256x256FID: 1.86 -> 1.81Improved
apply REPA to JiT at low resolutionJiT · ImageNet 32x32FIDImproved

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