No "Zero-Shot" Without Exponential Data: Pretraining Concept Frequency Determines Multimodal Model Performance
Paper
• 2404.04125 • Published
• 29
Vision and language
-
-
CoMat: Aligning Text-to-Image Diffusion Model with Image-to-Text Concept Matching
Paper • 2404.03653 • Published • 35 -
Cross-Attention Makes Inference Cumbersome in Text-to-Image Diffusion Models
Paper • 2404.02747 • Published • 13 -
3D Congealing: 3D-Aware Image Alignment in the Wild
Paper • 2404.02125 • Published • 10 -
BeyondScene: Higher-Resolution Human-Centric Scene Generation With Pretrained Diffusion
Paper • 2404.04544 • Published • 23 -
ControlNet++: Improving Conditional Controls with Efficient Consistency Feedback
Paper • 2404.07987 • Published • 48 -
BRAVE: Broadening the visual encoding of vision-language models
Paper • 2404.07204 • Published • 19 -
RealmDreamer: Text-Driven 3D Scene Generation with Inpainting and Depth Diffusion
Paper • 2404.07199 • Published • 27 -
Learning to Route Among Specialized Experts for Zero-Shot Generalization
Paper • 2402.05859 • Published • 5 -
Improving Explicit Spatial Relationships in Text-to-Image Generation through an Automatically Derived Dataset
Paper • 2403.00587 • Published -
ReGround: Improving Textual and Spatial Grounding at No Cost
Paper • 2403.13589 • Published • 1 -
FlexCap: Generating Rich, Localized, and Flexible Captions in Images
Paper • 2403.12026 • Published • 2 -
Scaling Rectified Flow Transformers for High-Resolution Image Synthesis
Paper • 2403.03206 • Published • 71 -
Editable Image Elements for Controllable Synthesis
Paper • 2404.16029 • Published • 12 -
Move Anything with Layered Scene Diffusion
Paper • 2404.07178 • Published -
Kaleido Diffusion: Improving Conditional Diffusion Models with Autoregressive Latent Modeling
Paper • 2405.21048 • Published • 16