← 研踪 PaperTrailarXiv:2606.13898 · v4
HiLo-Token: Input-Adaptive High-Low Frequency Token Compression for Efficient Image Editing Model Serving
首次提交 2026-06-12 04:45
版本更新 2026-11-08 09:37
雷达观测 2026-11-10 12:16
摘要
Creative image editing tools, such as Adobe Photoshop's Remove and Generative Fill buttons, are central to everyday customer use and account for a major share of traffic. However, current generative AI models face significant latency challenges, which become even more pronounced as the field transitions from convolution-based U-Nets to Diffusion Transformers (DiTs). In our evaluation on hundreds of representative image editing samples spanning a wide range of mask ratios, the DiT module alone accounts for an average of 73% of the total model latency, even after being distilled from 50 timesteps down to 8 timesteps. To tackle this challenge, we propose $\textbf{HiLo-Token}$, an input-adaptive token compression framework that allocates a larger token budget to high-frequency, rich-context regions while assigning fewer tokens to low-frequency areas. Specifically, for the editing region specified by the user mask, we retain all tokens within a dilated mask to preserve strong locality and contextual relevance. Outside the editing region, we introduce a simple yet effective high-frequency token selection strategy based on spatial frequency to capture important local details, while using tokens from a heavily downsampled image to represent low-frequency components and preserve the coarse global structure. Extensive experiments on production-level evaluation data validate the effectiveness of the proposed method, which achieves 3.13$\times$, 2.59$\times$, and 1.67$\times$ DiT speedups on A100-80GB for image editing tasks across small, medium, and large mask ratio categories with average ratios of 6.38%, 15.92%, and 35.36%, respectively, without any regression in generation quality. In production, where the editing pipeline is served as disaggregated, independently autoscaled GPU workers, these per-request savings translate into a 33% reduction in the GPUs required to serve the models.
作者备注：14 pages, 11 figures, Patent filled
观测到的版本
相邻记录可能跳过中间版本；首次发现 v2 及以上也会标为版本更新。
v4 · 2026-11-08 09:37 · 最新观测版本
HiLo-Token: Input-Adaptive High-Low Frequency Token Compression for Efficient Image Editing Model Serving
Creative image editing tools, such as Adobe Photoshop's Remove and Generative Fill buttons, are central to everyday customer use and account for a major share of traffic. However, current generative AI models face significant latency challenges, which become even more pronounced as the field transitions from convolution-based U-Nets to Diffusion Transformers (DiTs). In our evaluation on hundreds of representative image editing samples spanning a wide range of mask ratios, the DiT module alone accounts for an average of 73% of the total model latency, even after being distilled from 50 timesteps down to 8 timesteps. To tackle this challenge, we propose $\textbf{HiLo-Token}$, an input-adaptive token compression framework that allocates a larger token budget to high-frequency, rich-context regions while assigning fewer tokens to low-frequency areas. Specifically, for the editing region specified by the user mask, we retain all tokens within a dilated mask to preserve strong locality and contextual relevance. Outside the editing region, we introduce a simple yet effective high-frequency token selection strategy based on spatial frequency to capture important local details, while using tokens from a heavily downsampled image to represent low-frequency components and preserve the coarse global structure. Extensive experiments on production-level evaluation data validate the effectiveness of the proposed method, which achieves 3.13$\times$, 2.59$\times$, and 1.67$\times$ DiT speedups on A100-80GB for image editing tasks across small, medium, and large mask ratio categories with average ratios of 6.38%, 15.92%, and 35.36%, respectively, without any regression in generation quality. In production, where the editing pipeline is served as disaggregated, independently autoscaled GPU workers, these per-request savings translate into a 33% reduction in the GPUs required to serve the models.
查看此版本解读 →查看此版本原文 ↗