AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

What matters for Representation Alignment: Global Information or Spatial Structure?

Jaskirat Singh, Xingjian Leng, Zongze Wu, Liang Zheng, Richard Zhang, Eli Shechtman, Saining Xie

ICLR 2026 · Label-efficient and robust learning › Recognition and long-tail · approach: Correspondence and causal representation learning

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 a target encoder with higher ImageNet-1K accuracy (PE-Core-G vs PE-Spatial-B)REPA · ImageNet 256 generation, SiT-XL/2, 100KFID: FID 32.3 (82.8% acc) vs 21.0 (53.1% acc)Hurt
use larger encoder within same family (WebSSL-1B higher accuracy)REPA · ImageNet 256 generation, SiT-XL/2, 100KFID: FID 26.1 vs 21.0 (acc 76.0% vs 53.1%)Hurt
inject global info into patch tokens via CLS mixing (alpha 0 to 0.5) (ablation)REPA with DINOv2 target · ImageNet 256 generation, SiT-XL/2FID: 19.2 -> 25.4 (linear probe 70.7% -> 78.5%)Hurt
replace MLP projection with 3x3 convolution and add spatial normalization (iREPA)REPA, SiT-XL/2 DINOv2-B · ImageNet 256, 100K steps, no CFGFID: 19.06 -> 16.96Improved
add iREPA to weak-spatial-structure target PE-Core-GREPA, SiT-XL/2 · ImageNet 256, 100K steps, no CFGFID: 32.35 -> 18.19Improved
add spatial normalization layer (to conv projection) (ablation)iREPA · ImageNet 256, 100K steps, SiT-XL/2, DINOv2-BFID: 18.52 -> 16.96Improved
add convolution projection (to spatial normalization) (ablation)iREPA · ImageNet 256, 100K steps, SiT-XL/2, DINOv2-BFID: 17.66 -> 16.96Improved
add iREPA across encoder sizes PE-B, PE-L, PE-GREPA, SiT-XL/2 · ImageNet 256, 100K stepsFID: PE-B 22.5 -> 17.5; PE-L 28.7 -> 17.6; PE-G 32.3 -> 19.5Improved
add iREPA across diffusion model sizes SiT-B, L, XLREPA, DINOv2 target · ImageNet 256, 100K stepsFID: SiT-B 49.50 -> 43.37; SiT-L 24.10 -> 20.28; SiT-XL 19.06 -> 16.96Improved
add iREPA at alignment layers 4, 6, 8REPA, SiT-B/2 · ImageNet 256, 100K stepsFID: Layer 4: 49.0 -> 36.0; Layer 8: 54.8 -> 38.9Improved
add iREPA to REPA-EREPA-E, SiT-XL/2 · ImageNet 256, 100K stepsFID: WebSSL-1B 26.5 -> 13.2; DINOv2-B 12.9 -> 12.1Improved
add iREPA to MeanFlow w/ REPAMeanFlow w/ REPA, SiT-B/2 · ImageNet 256, 100K steps, 1 NFE w/ CFGFID: DINOv3-B 22.55 -> 17.32Improved

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