LoopREG: Representation Supervision Enables Test-Time Recurrent Depth in Diffusion Transformers

arXiv:2610.17877 · cs.CV · Submitted 2026-10-13 · Read on arXiv

cs.CV

Submitted: 2026-10-13

Updated: 2026-10-13

Comments: 23 pages; including appendix

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• Transformer — 1× in this paper · explained in 2 episodes

Sources

• Deep Equilibrium Models
• PonderNet: Learning to Ponder
• Training-Free Looped Transformers
• Universal Transformers
• Scaling up Test-Time Compute with Latent Reasoning: A Recurrent Depth Approach
• LoopFormer: Elastic-Depth Looped Transformers for Latent Reasoning via Shortcut Modulation
• No Other Representation Component Is Needed: Diffusion Transformers Can Provide Representation Guidance by Themselves
• Similarity of Neural Network Representations Revisited
• DINOv2: Learning Robust Visual Features without Supervision
• Looped Transformers are Better at Learning Learning Algorithms
• Representation Alignment for Generation: Training Diffusion Transformers Is Easier Than You Think
• Guiding a Diffusion Transformer with the Internal Dynamics of Itself

Related papers

• Loss Knows Best: Detecting Annotation Errors in Videos via Loss Trajectories
• AnchorWeave: World-Consistent Video Generation with Retrieved Local Spatial Memories
• Benchmarking the Robustness of Foundation Models for Mammography under Domain Shift
• MambaX-Net: Dual-Input Mamba-Enhanced Cross-Attention Network for Longitudinal MRI Segmentation
• TeleOCR: Navigating Document Parsing Across Digital and Camera-Captured Documents
• A Survey on Efficient Vision-Language-Action Models