UDT: Reconciling U-Nets and Diffusion Transformers with Data-Adaptive Token Reduction
cs.CV, cs.AI, cs.LG
Submitted: 2026-08-02
Updated: 2026-11-09
Code: https://github.com/JN-Yun/UDT
Terminology
Sources
- TREAD: Token Routing for Efficient Architecture-agnostic Diffusion Training
- Back to Basics: Let Denoising Generative Models Denoise
- Token Pooling in Vision Transformers
- GLU Variants Improve Transformer
- Diffuse and Disperse: Image Generation with Representation Regularization
- SANA: Efficient High-Resolution Image Synthesis with Linear Diffusion Transformers
- No Alignment Needed for Generation: Learning Linearly Separable Representations in Diffusion Models
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