SD3.5-Flash：基於分佈引導的生成流蒸餾

摘要

我們推出SD3.5-Flash，這是一個高效的少步蒸餾框架，旨在將高品質的圖像生成技術帶入普及的消費級設備。我們的方法通過專門為少步生成重新設計的分佈匹配目標，來蒸餾計算上難以處理的修正流模型。我們引入了兩項關鍵創新：「時間步共享」以減少梯度噪聲，以及「分割時間步微調」來提升提示對齊效果。結合全面的管道優化，如文本編碼器重構和專用量化技術，我們的系統能夠在不同硬件配置上實現快速生成和記憶體高效部署。這使得從手機到桌面電腦的全範圍設備都能平等地使用這項技術。通過包括大規模用戶研究在內的廣泛評估，我們證明SD3.5-Flash在少步方法中始終表現優異，使先進的生成式AI真正實現了實際部署的普及化。

English

We present SD3.5-Flash, an efficient few-step distillation framework that brings high-quality image generation to accessible consumer devices. Our approach distills computationally prohibitive rectified flow models through a reformulated distribution matching objective tailored specifically for few-step generation. We introduce two key innovations: "timestep sharing" to reduce gradient noise and "split-timestep fine-tuning" to improve prompt alignment. Combined with comprehensive pipeline optimizations like text encoder restructuring and specialized quantization, our system enables both rapid generation and memory-efficient deployment across different hardware configurations. This democratizes access across the full spectrum of devices, from mobile phones to desktop computers. Through extensive evaluation including large-scale user studies, we demonstrate that SD3.5-Flash consistently outperforms existing few-step methods, making advanced generative AI truly accessible for practical deployment.

SD3.5-Flash：基於分佈引導的生成流蒸餾

SD3.5-Flash: Distribution-Guided Distillation of Generative Flows

摘要

Support