Metraipolis › archive › dc_f0fb708ed202

DensityKV: Density-Guided KV Cache Compression for Long Video Generation — arXiv:2608.27922

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.27922
License (as archived)
CC0-1.0
Archived
2026-10-17T11:33:03.548Z
SHA-256 of the archived text
083f595de68ea465d8f810701d31a603e0e9df3dc1d34840fea267cdb9c8950b
Stable URL of this page
https://metraipolis.com/documento/dc_f0fb708ed202

How to cite

DensityKV: Density-Guided KV Cache Compression for Long Video Generation — arXiv:2608.27922 — arXiv (en), https://arxiv.org/abs/2608.27922 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f0fb708ed202 (Metraipolis archive dc_f0fb708ed202, sha256 083f595de68ea465d8f810701d31a603e0e9df3dc1d34840fea267cdb9c8950b)
BibTeX
@misc{dcf0fb708ed202,
  title = {DensityKV: Density-Guided KV Cache Compression for Long Video Generation — arXiv:2608.27922},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.27922},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f0fb708ed202; sha256 083f595de68ea465d8f810701d31a603e0e9df3dc1d34840fea267cdb9c8950b},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.27922. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Autoregressive video diffusion models enable streaming generation through sliding-window attention, but each generated block is conditioned on previously generated content, causing appearance and motion errors to propagate recursively over time. Historical key-value (KV) memory preserves earlier subject and scene states and helps maintain long-horizon consistency. However, retaining every generated state creates a historical archive that grows continuously with the rollout, while recurrent states repeatedly add redundant coverage. To address this problem, we propose DensityKV, a training-free historical KV bank management strategy. DensityKV maintains a separate token-level KV bank for each attention head and measures local redundancy among the post-RoPE keys that directly parameterize attention routing using Soft-Riesz density. By constraining neighborhood-density growth after states enter the bank, DensityKV limits repeated historical accumulation while preserving coherent states from each completed generation block. Experiments across three autoregressive video generation backbones and multiple generation lengths show that, at the same upper bound on historical KV capacity, DensityKV improves long-horizon consistency and generation stability while keeping persistent historical storage bounded independently of rollout length.

Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-08-31 in cs.CV).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2608.27922 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-17T11:10:05.280Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-08-31?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).