Kai Wang

33

Papers

200

Total Citations

Papers (33)

Real-Time Video Generation with Pyramid Attention Broadcast

Efficient Online Reinforcement Learning for Diffusion Policy

A Closer Look at Time Steps is Worthy of Triple Speed-Up for Diffusion Model Training

Summarizing Stream Data for Memory-Constrained Online Continual Learning

ColorPeel: Color Prompt Learning with Diffusion Models via Color and Shape Disentanglement

FilterTS: Comprehensive Frequency Filtering for Multivariate Time Series Forecasting

REPA Works Until It Doesn’t: Early-Stopped, Holistic Alignment Supercharges Diffusion Training

CALLIC: Content Adaptive Learning for Lossless Image Compression

Drag-and-Drop LLMs: Zero-Shot Prompt-to-Weights

Free-Lunch Color-Texture Disentanglement for Stylized Image Generation

StarTrail: Concentric Ring Sequence Parallelism for Efficient Near-Infinite-Context Transformer Model Training

Info-Coevolution: An Efficient Framework for Data Model Coevolution

IMG: Calibrating Diffusion Models via Implicit Multimodal Guidance

InpDiffusion: Image Inpainting Localization via Conditional Diffusion Models

Single-View Graph Contrastive Learning with Soft Neighborhood Awareness

EnMatch: Matchmaking for Better Player Engagement via Neural Combinatorial Optimization

Unsupervised Learning for Class Distribution Mismatch

Navigating Complexity: Toward Lossless Graph Condensation via Expanding Window Matching

DiffAug: Enhance Unsupervised Contrastive Learning with Domain-Knowledge-Free Diffusion-based Data Augmentation

A Stitch in Time Saves Nine: Small VLM is a Precise Guidance for Accelerating Large VLMs

Two Heads Are Better Than One: Boosting Graph Sparse Training via Semantic and Topological Awareness

One-Way Ticket: Time-Independent Unified Encoder for Distilling Text-to-Image Diffusion Models

Optimizing for the Shortest Path in Denoising Diffusion Model

The Art of Deception: Color Visual Illusions and Diffusion Models

Emphasizing Discriminative Features for Dataset Distillation in Complex Scenarios

Distilling Long-tailed Datasets

EA-Vit: Efficient Adaptation for Elastic Vision Transformer

TAR3D: Creating High-Quality 3D Assets via Next-Part Prediction

Anchor Token Matching: Implicit Structure Locking for Training-free AR Image Editing

AR-1-to-3: Single Image to Consistent 3D Object via Next-View Prediction

Covariances for Free: Exploiting Mean Distributions for Training-free Federated Learning

LeMiCa: Lexicographic Minimax Path Caching for Efficient Diffusion-Based Video Generation

Drawing Informative Gradients from Sources: A One-stage Transfer Learning Framework for Cross-city Spatiotemporal Forecasting