RouteREPA: Routed Representation Alignment for Just Image Transformers
cs.CV, cs.LG
Submitted: 2026-11-02
Updated: 2026-11-02
Comments: 23 pages
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- PixelFlow: Pixel-Space Generative Models with Flow
- Classifier-Free Diffusion Guidance
- Back to Basics: Let Denoising Generative Models Denoise
- DINOv2: Learning Robust Visual Features without Supervision
- Representation Alignment for Just Image Transformers is not Easier than You Think
- Improved Baselines with Representation Autoencoders
- DDT: Decoupled Diffusion Transformer
Related papers
- TEA: Text Encoder Alignment for Robust Concept Erasure in Text-to-Image Models
- Do Egocentric Video-Language Models Capture Both Hand- and Object-Centric Cues?
- Aura: Consistent Multi-Subject Video Generation via VLM-Grounded Semantic Alignment
- PVRF: All-in-one Adverse Weather Removal via Prior-modulated and Velocity-constrained Rectified Flow
- DLWM: Diverse Latent World Models for Efficient Multimodal Reasoning
- Out of Sight, Still in Mind: Token Compression for Omni-LLMs