The Weight Is Over - Interactive Diffusion on Consumer GPUs

原文：The Weight Is Over - Interactive Diffusion on Consumer GPUs

作者：Frieder Ganz, Maximilian Müller

来源：arXiv cs.LG（机器学习）

正文

Computer Science > Machine Learning

arXiv:2609.21849v1 (cs)

[Submitted on 18 Sep 2026]

Title:The Weight Is Over - Interactive Diffusion on Consumer GPUs

Authors:Frieder Ganz, Maximilian Müller

View a PDF of the paper titled The Weight Is Over - Interactive Diffusion on Consumer GPUs, by Frieder Ganz and Maximilian M"uller

View PDF

HTML (experimental)

Abstract:On-device inference is booming, but the momentum is almost all in language models. Diffusion pipelines are memory hungry, latency-sensitive, and require orchestrating an embedder, a transformer, a decoder, and often further postprocessing that is not as standardized as LLM inference loops are. We navigate the trade-off between performance, quality, and model footprint to reach as many client devices in the wild as possible. We make three contributions: an embedding translator that maps a small text encoder into a large encoder space to cut weight and latency; a reproducible sweep recipe for navigating the speed/quality/memory triangle in diffusion pipelines; and an interactive on-device image generation editor achieving sub-second TTFI on recent GPUs.

Subjects:

Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV); Performance (cs.PF)

Cite as:

arXiv:2609.21849 [cs.LG]

(or

arXiv:2609.21849v1 [cs.LG] for this version)

https://doi.org/10.48550/arXiv.2609.21849

Focus to learn more

arXiv-issued DOI via DataCite (pending registration)

主题

机器学习 · 计算机视觉 · cs.PF

由「前沿雷达」于 2026-09-18 采集。正文取自原文页面，已保留出处链接。标题与正文版权归原作者所有。