Metraipolis › archive › dc_085464fb91ce
From Local Mismatch to Global Impact: Optimizing Cache Reuse Policy for Efficient Diffusion — arXiv:2608.13043
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.13043
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-19T04:17:28.348Z
- SHA-256 of the archived text
a6c39b84cff89538f2c89c0a8e677d8f849c37506779d2325bda81e73af086f6- Stable URL of this page
https://metraipolis.com/documento/dc_085464fb91ce
How to cite
From Local Mismatch to Global Impact: Optimizing Cache Reuse Policy for Efficient Diffusion — arXiv:2608.13043 — arXiv (en), https://arxiv.org/abs/2608.13043 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_085464fb91ce (Metraipolis archive dc_085464fb91ce, sha256 a6c39b84cff89538f2c89c0a8e677d8f849c37506779d2325bda81e73af086f6)
BibTeX
@misc{dc085464fb91ce,
title = {From Local Mismatch to Global Impact: Optimizing Cache Reuse Policy for Efficient Diffusion — arXiv:2608.13043},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.13043},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_085464fb91ce; sha256 a6c39b84cff89538f2c89c0a8e677d8f849c37506779d2325bda81e73af086f6},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.13043. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Diffusion models have achieved dominant performance in visual generation but suffer from substantial inference overhead. While cache-based acceleration has emerged as a promising solution, existing policies rely on local similarity heuristics, which we identify as being significantly misaligned with final generation quality. This discrepancy stems from the non-uniform propagation and accumulation of errors along the denoising trajectory. To address this, we propose Global-Impact Cache (GCache). We first establish a rigorous theoretical characterization of the error propagation upper bound. Recognizing that this bound can be overly conservative for complex, highly non-convex diffusion models, we further reparameterize the propagation exponent with a Bernstein form and reformulate cache policy search as a bilevel optimization problem. In detail, GCache identifies an optimal reuse policy in the inner objective while aligning the error-weighting function with generation quality loss in the outer objective. This framework effectively reconciles theoretical rigor with empirical performance, learning to prioritize computation where it most impacts visual fidelity. Extensive experiments demonstrate that GCache consistently outperforms prior caching strategies on both video and image generation. Notably, on the state-of-the-art Wan2.1 video diffusion model, GCache maintains a 2.17x speedup while significantly enhancing generation quality, reducing LPIPS from 0.1095 to 0.0316. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-14 in cs.AI). Categorie arXiv: Artificial Intelligence (cs.AI) ; Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2608.13043 — 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 2026-10-19T04:13:06.566Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-14?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).