Unlocking the Potential of Contrastive Regularization for Diffusion Transformer Training
cs.CV, cs.LG
Submitted: 2026-11-02
Updated: 2026-11-02
Terminology
Sources
- MDTv2: Masked Diffusion Transformer is a Strong Image Synthesizer
- LayerSync: Self-aligning Intermediate Layers
- Classifier-Free Diffusion Guidance
- TREAD: Token Routing for Efficient Architecture-agnostic Diffusion Training
- WebVision Database: Visual Learning and Understanding from Web Data
- SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis
- Movie Gen: A Cast of Media Foundation Models
- What matters for Representation Alignment: Global Information or Spatial Structure?
- Score-Based Generative Modeling through Stochastic Differential Equations
- SRA 2: Variational Autoencoder Self-Representation Alignment for Efficient Diffusion Training
- Diffuse and Disperse: Image Generation with Representation Regularization
- Representation Entanglement for Generation: Training Diffusion Transformers Is Much Easier Than You Think
- DiverseDiT: Towards Diverse Representation Learning in Diffusion Transformers
- Representation Alignment for Generation: Training Diffusion Transformers Is Easier Than You Think
- No Alignment Needed for Generation: Learning Linearly Separable Representations in Diffusion Models
- Fast Training of Diffusion Models with Masked Transformers
- SD-DiT: Unleashing the Power of Self-supervised Discrimination in 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