Skip to content
Sagan

Paper

Quantized Keys Steal Attention: Bias Correction for KV-Cache Compression in Video Diffusion

Unreadunread

AI summary

arXiv:2605. 26266v1 Announce Type: new Abstract: Chunk-wise autoregressive video diffusion models rely on a KV cache of previously generated chunks to avoid redundant computation, but this cache quickly becomes a memory bottleneck as videos grow longer.