中文
相关论文

相关论文: Focused Forcing: Content-Aware Per-Frame KV Select…

200 篇论文

Autoregressive video diffusion models support real-time synthesis but suffer from error accumulation and context loss over long horizons. We discover that attention heads in AR video diffusion transformers serve functionally distinct roles…

计算机视觉与模式识别 · 计算机科学 2026-05-15 Jiahao Tian , Yiwei Wang , Gang Yu , Chi Zhang

Autoregressive (AR) video diffusion models adopt a streaming generation framework, enabling long-horizon video generation with real-time responsiveness, as exemplified by the Self Forcing training paradigm. However, existing AR video…

计算机视觉与模式识别 · 计算机科学 2026-05-12 Yicheng Ji , Zhizhou Zhong , Jun Zhang , Qin Yang , XiTai Jin , Ying Qin , Wenhan Luo , Shuiyang Mao , Wei Liu , Huan Li

Autoregressive video generation enables streaming and open-ended long video synthesis, but still suffers from long-term degradation caused by accumulated errors. Existing KVCache strategies usually apply unified historical-frame retention,…

计算机视觉与模式识别 · 计算机科学 2026-05-14 Jiayu Chen , Junbei Tang , Wenbiao Zhao , Maoliang Li , Jiayi Luo , Zihao Zheng , Jiawei Yang , Guojie Luo , Xiang Chen

Autoregressive (AR) video generation has emerged as a promising paradigm for long-horizon video synthesis, where each frame is generated conditioned on previously generated tokens. To accelerate inference, the KV cache is used to avoid…

计算机视觉与模式识别 · 计算机科学 2026-05-29 Jiayi Luo , Qiyan Liu , Tengyang Wang , JunHao Liu , Jiayu Chen , Cong Wang , Hanxin Zhu , Chen Gao , Xiaobin Hu , Qingyun Sun , Zhibo Chen

The autoregressive video diffusion model has recently gained considerable research interest due to its causal modeling and iterative denoising. In this work, we identify that the multi-head self-attention in these models under-utilizes…

计算机视觉与模式识别 · 计算机科学 2026-01-29 Hang Guo , Zhaoyang Jia , Jiahao Li , Bin Li , Yuanhao Cai , Jiangshan Wang , Yawei Li , Yan Lu

Autoregressive video diffusion models enable open-ended generation through local attention and KV caching. However, existing training-free long-video optimization methods mainly focus on stable extension under a single prompt, making them…

计算机视觉与模式识别 · 计算机科学 2026-05-18 Mingqiang Wu , Weilun Feng , Zhefeng Zhang , Haotong Qin , Yuqi Li , Guoxin Fan , Xiaokun Liu , Zhulin An , Libo Huang , Yongjun Xu , Chuanguang Yang

Autoregressive (AR) visual generation has achieved remarkable performance but suffers from high memory usage and low throughput, as it requires caching previously generated visual tokens. Recent research has shown that retaining only a few…

计算机视觉与模式识别 · 计算机科学 2026-05-21 Guotao Liang , Baoquan Zhang , Zhiyuan Wen , Yunming Ye

We introduce Self Forcing, a novel training paradigm for autoregressive video diffusion models. It addresses the longstanding issue of exposure bias, where models trained on ground-truth context must generate sequences conditioned on their…

计算机视觉与模式识别 · 计算机科学 2025-11-11 Xun Huang , Zhengqi Li , Guande He , Mingyuan Zhou , Eli Shechtman

We introduce Sparse Forcing, a training-and-inference paradigm for autoregressive video diffusion models that improves long-horizon generation quality while reducing decoding latency. Sparse Forcing is motivated by an empirical observation…

计算机视觉与模式识别 · 计算机科学 2026-04-24 Boxun Xu , Yuming Du , Zichang Liu , Siyu Yang , Ziyang Jiang , Siqi Yan , Rajasi Saha , Albert Pumarola , Wenchen Wang , Peng Li

Autoregressive (AR) video diffusion has recently emerged as a promising paradigm for long video generation, enabling causal synthesis beyond the limits of bidirectional models. To address training-inference mismatch, a series of…

计算机视觉与模式识别 · 计算机科学 2026-03-24 Zengqun Zhao , Yanzuo Lu , Ziquan Liu , Jifei Song , Jiankang Deng , Ioannis Patras

Recent advances in autoregressive video diffusion have enabled real-time frame streaming, yet existing solutions still suffer from temporal repetition, drift, and motion deceleration. We find that naively applying StreamingLLM-style…

计算机视觉与模式识别 · 计算机科学 2025-12-05 Jung Yi , Wooseok Jang , Paul Hyunbin Cho , Jisu Nam , Heeji Yoon , Seungryong Kim

Autoregressive video diffusion models have demonstrated remarkable progress, yet they remain bottlenecked by intractable linear KV-cache growth, temporal repetition, and compounding errors during long-video generation. To address these…

计算机视觉与模式识别 · 计算机科学 2026-03-27 Xiaofeng Mao , Shaohao Rui , Kaining Ying , Bo Zheng , Chuanhao Li , Mingmin Chi , Kaipeng Zhang

Self-forcing video generation extends a short-horizon video model to longer rollouts by repeatedly feeding generated content back in as context. This scaling path immediately exposes a systems bottleneck: the key-value (KV) cache grows with…

机器学习 · 计算机科学 2026-03-31 Suraj Ranganath , Vaishak Menon , Anish Patnaik

Advanced autoregressive (AR) video generation models have improved visual fidelity and interactivity, but the quadratic complexity of attention remains a primary bottleneck for efficient deployment. While existing sparse attention solutions…

计算机视觉与模式识别 · 计算机科学 2026-05-26 Chengtao Lv , Yumeng Shi , Yushi Huang , Ruihao Gong , Shen Ren , Wenya Wang

A unified autoregressive model is a Transformer-based framework that addresses diverse multimodal tasks (e.g., text, image, video) as a single sequence modeling problem under a shared token space. Such models rely on the KV-cache mechanism…

计算机视觉与模式识别 · 计算机科学 2026-01-09 Kunyang Li , Mubarak Shah , Yuzhang Shang

Current frontier video diffusion models have demonstrated remarkable results at generating high-quality videos. However, they can only generate short video clips, normally around 10 seconds or 240 frames, due to computation limitations…

计算机视觉与模式识别 · 计算机科学 2025-05-20 Desai Xie , Zhan Xu , Yicong Hong , Hao Tan , Difan Liu , Feng Liu , Arie Kaufman , Yang Zhou

Visual Autoregressive (VAR) models adopt a next-scale prediction paradigm, offering high-quality content generation with substantially fewer decoding steps. However, existing VAR models suffer from significant attention complexity and…

计算机视觉与模式识别 · 计算机科学 2025-12-09 Ziran Qin , Youru Lv , Mingbao Lin , Hang Guo , Zeren Zhang , Danping Zou , Weiyao Lin

Video generation is pivotal to digital media creation, and recent advances in autoregressive video generation have markedly enhanced the efficiency of real-time video synthesis. However, existing approaches generally rely on heuristic KV…

计算机视觉与模式识别 · 计算机科学 2026-02-05 Hanmo Chen , Chenghao Xu , Xu Yang , Xuan Chen , Cheng Deng

Streaming video generation, as one fundamental component in interactive world models and neural game engines, aims to generate high-quality, low-latency, and temporally coherent long video streams. However, most existing work suffers from…

计算机视觉与模式识别 · 计算机科学 2025-09-30 Kunhao Liu , Wenbo Hu , Jiale Xu , Ying Shan , Shijian Lu

Long-context video modeling is essential for enabling generative models to function as world simulators, as they must maintain temporal coherence over extended time spans. However, most existing models are trained on short clips, limiting…

计算机视觉与模式识别 · 计算机科学 2025-05-20 Yuchao Gu , Weijia Mao , Mike Zheng Shou
‹ 上一页 1 2 3 10 下一页 ›