9 papers
FlashGaze: Training-Free Multi-Scale Patch Pruning For Efficient Video Understanding
Ziye Zhu, Yanghao Zhou, Lixing Tan +3
Multimodal Large Language Models (MLLMs) have demonstrated strong performance in video understanding, yet efficiently processing long, high-resolution videos remains challenging. S…
Causal Forcing++: Scalable Few-Step Autoregressive Diffusion Distillation for Real-Time Interactive Video Generation
Min Zhao, Hongzhou Zhu, Kaiwen Zheng +6
Real-time interactive video generation requires low-latency, streaming, and controllable rollout. Existing autoregressive (AR) diffusion distillation methods have achieved strong r…
Anatomy-Slot: Unsupervised Anatomical Factorization for Homologous Bilateral Reasoning in Retinal Diagnosis
Yingzhe Ma, Xiao Yang, Yuguo Yin +1
Retinal diagnosis is inherently bilateral: clinicians compare homologous structures across eyes (e.g., optic disc asymmetry), yet most deep models operate on monocular representati…
U-HNO: A U-shaped Hybrid Neural Operator with Sparse-Point Adaptive Routing for Non-stationary PDE Dynamics
Yingzhe Ma, Xiao Yang, Yuxin Xie +2
Solutions to many partial differential equations (PDEs) display coexisting smooth global transport and localized sharp features within a single trajectory: shock fronts, thin inter…
AdaFocus: Adaptive Relevance-Diversity Sampling with Zero-Cache Look-back for Efficient Long Video Understanding
Xiao Yang, Yingzhe Ma, Haoxuan Yu +2
Long video understanding is heavily bottlenecked by a rigid one-shot paradigm: existing methods either densely encode videos at prohibitive memory and latency costs, or aggressivel…
Helios: Real Real-Time Long Video Generation Model
Shenghai Yuan, Yuanyang Yin, Zongjian Li +3
We introduce Helios, the first 14B video generation model that runs at 19.5 FPS on a single NVIDIA H100 GPU and supports minute-scale generation while matching the quality of a str…
Adaptive 1D Video Diffusion Autoencoder
Yao Teng, Minxuan Lin, Xian Liu +3
Recent video generation models largely rely on video autoencoders that compress pixel-space videos into latent representations. However, existing video autoencoders suffer from thr…
FSVideo: Fast Speed Video Diffusion Model in a Highly-Compressed Latent Space
FSVideo Team, Qingyu Chen, Zhiyuan Fang +17
We introduce FSVideo, a fast speed transformer-based image-to-video (I2V) diffusion framework. We build our framework on the following key components: 1.) a new video autoencoder w…
IP-Prompter: Training-Free Theme-Specific Image Generation via Dynamic Visual Prompting
Yuxin Zhang, Minyan Luo, Weiming Dong +6
The stories and characters that captivate us as we grow up shape unique fantasy worlds, with images serving as the primary medium for visually experiencing these realms. Personaliz…