Pyramid Forcing: Head-Aware Pyramid KV Cache Policy for High-Quality Long Video Generation

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Chen, Jiayu, Tang, Junbei, Zhao, Wenbiao, Li, Maoliang, Luo, Jiayi, Zheng, Zihao, Yang, Jiawei, Luo, Guojie, Chen, Xiang
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866910215581990912
author Chen, Jiayu
Tang, Junbei
Zhao, Wenbiao
Li, Maoliang
Luo, Jiayi
Zheng, Zihao
Yang, Jiawei
Luo, Guojie
Chen, Xiang
author_facet Chen, Jiayu
Tang, Junbei
Zhao, Wenbiao
Li, Maoliang
Luo, Jiayi
Zheng, Zihao
Yang, Jiawei
Luo, Guojie
Chen, Xiang
contents 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, implicitly assuming homogeneous historical dependencies across attention heads. We revisit historical-frame attention and reveal three distinct head types: Anchor Heads require broad long-range context, Wave Heads exhibit periodic temporal dependencies, and Veil Heads focus on initial and adjacent frames. Based on this finding, we propose Pyramid Forcing, a head-aware pyramidal KVCache framework that identifies head types offline, assigns behavior-specific cache policies, and supports heterogeneous cache lengths via efficient ragged-cache attention. Experiments on Self Forcing and Causal Forcing show that Pyramid Forcing consistently improves long-horizon generation quality on VBench-Long, increasing the 60-second Self Forcing score from 77.87 to 81.21 while enhancing motion dynamics, visual fidelity, and semantic consistency. Project: https://if-lab-pku.github.io/Pyramid-Forcing/.
format Preprint
id arxiv_https___arxiv_org_abs_2605_13111
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Pyramid Forcing: Head-Aware Pyramid KV Cache Policy for High-Quality Long Video Generation
Chen, Jiayu
Tang, Junbei
Zhao, Wenbiao
Li, Maoliang
Luo, Jiayi
Zheng, Zihao
Yang, Jiawei
Luo, Guojie
Chen, Xiang
Computer Vision and Pattern Recognition
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, implicitly assuming homogeneous historical dependencies across attention heads. We revisit historical-frame attention and reveal three distinct head types: Anchor Heads require broad long-range context, Wave Heads exhibit periodic temporal dependencies, and Veil Heads focus on initial and adjacent frames. Based on this finding, we propose Pyramid Forcing, a head-aware pyramidal KVCache framework that identifies head types offline, assigns behavior-specific cache policies, and supports heterogeneous cache lengths via efficient ragged-cache attention. Experiments on Self Forcing and Causal Forcing show that Pyramid Forcing consistently improves long-horizon generation quality on VBench-Long, increasing the 60-second Self Forcing score from 77.87 to 81.21 while enhancing motion dynamics, visual fidelity, and semantic consistency. Project: https://if-lab-pku.github.io/Pyramid-Forcing/.
title Pyramid Forcing: Head-Aware Pyramid KV Cache Policy for High-Quality Long Video Generation
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2605.13111