AI RESEARCH
Pyramid Forcing: Head-Aware Pyramid KV Cache Policy for High-Quality Long Video Generation
arXiv CS.CV
•
ArXi:2605.13111v1 Announce Type: new 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.