PAROAttention: Pattern-Aware ReOrdering for Efficient Sparse and Quantized Attention in Visual Generation Models

• URL: http://arxiv.org/abs/2506.16054v1
• Date: Thu, 19 Jun 2025 06:25:02 GMT
• Title: PAROAttention: Pattern-Aware ReOrdering for Efficient Sparse and Quantized Attention in Visual Generation Models
• Authors: Tianchen Zhao, Ke Hong, Xinhao Yang, Xuefeng Xiao, Huixia Li, Feng Ling, Ruiqi Xie, Siqi Chen, Hongyu Zhu, Yichong Zhang, Yu Wang,
• Abstract summary: In visual generation, the quadratic complexity of attention mechanisms results in high memory and computational costs.<n>We propose an alternative strategy: *reorganizing* the attention pattern to alleviate the challenges.<n>Inspired by the local aggregation nature of visual feature extraction, we design a novel **Pattern-Aware token ReOrdering (PARO)** technique.
• Score: 14.14413223631804
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: In visual generation, the quadratic complexity of attention mechanisms results in high memory and computational costs, especially for longer token sequences required in high-resolution image or multi-frame video generation. To address this, prior research has explored techniques such as sparsification and quantization. However, these techniques face significant challenges under low density and reduced bitwidths. Through systematic analysis, we identify that the core difficulty stems from the dispersed and irregular characteristics of visual attention patterns. Therefore, instead of introducing specialized sparsification and quantization design to accommodate such patterns, we propose an alternative strategy: *reorganizing* the attention pattern to alleviate the challenges. Inspired by the local aggregation nature of visual feature extraction, we design a novel **Pattern-Aware token ReOrdering (PARO)** technique, which unifies the diverse attention patterns into a hardware-friendly block-wise pattern. This unification substantially simplifies and enhances both sparsification and quantization. We evaluate the performance-efficiency trade-offs of various design choices and finalize a methodology tailored for the unified pattern. Our approach, **PAROAttention**, achieves video and image generation with lossless metrics, and nearly identical results from full-precision (FP) baselines, while operating at notably lower density (~20%-30%) and bitwidth (**INT8/INT4**), achieving a **1.9x** to **2.7x** end-to-end latency speedup.

Related papers

• GreedyPrune: Retenting Critical Visual Token Set for Large Vision Language Models [5.025353943896242]
  GreedyPrune is a training-free visual token pruning algorithm designed to optimize semantic saliency and visual diversity.<n>We show that GreedyPrune achieves state-of-the-art accuracy across various multimodal tasks and models while significantly reducing end-to-end inference latency.
  arXiv  Detail & Related papers  (2025-06-16T07:21:11Z)
• Re-ttention: Ultra Sparse Visual Generation via Attention Statistical Reshape [23.01286982392074]
  A huge bottleneck is the attention mechanism where complexity scales quadratically with resolution and video length.<n>Existing techniques fail to preserve visual quality at extremely high sparsity levels and might even incur non-negligible compute overheads.<n>We propose Re-ttention, which implements very high sparse attention for visual generation models.
  arXiv  Detail & Related papers  (2025-05-28T22:39:12Z)
• Task-Oriented Feature Compression for Multimodal Understanding via Device-Edge Co-Inference [49.77734021302196]
  We propose a task-oriented feature compression (TOFC) method for multimodal understanding in a device-edge co-inference framework.<n>To enhance compression efficiency, multiple entropy models are adaptively selected based on the characteristics of the visual features.<n>Results show that TOFC achieves up to 60% reduction in data transmission overhead and 50% reduction in system latency.
  arXiv  Detail & Related papers  (2025-03-17T08:37:22Z)
• "Principal Components" Enable A New Language of Images [79.45806370905775]
  We introduce a novel visual tokenization framework that embeds a provable PCA-like structure into the latent token space.<n>Our approach achieves state-of-the-art reconstruction performance and enables better interpretability to align with the human vision system.
  arXiv  Detail & Related papers  (2025-03-11T17:59:41Z)
• Training-free and Adaptive Sparse Attention for Efficient Long Video Generation [31.615453637053793]
  generating high-fidelity long videos with Diffusion Transformers (DiTs) is often hindered by significant latency.<n>We propose AdaSpa, the first Dynamic Pattern and Online Precise Search sparse attention method.<n>AdaSpa is implemented as an adaptive, plug-and-play solution and can be integrated seamlessly with existing DiTs.
  arXiv  Detail & Related papers  (2025-02-28T14:11:20Z)
• Dynamic Token Reduction during Generation for Vision Language Models [11.376359442815986]
  We introduce a dynamic pruning strategy tailored for Vision-Language Models (VLMs)<n>Our approach enables flexible adjustment of pruning rates based on the attention distribution.<n>Our experimental results demonstrate that our method not only reduces computational demands but also maintains the quality of responses.
  arXiv  Detail & Related papers  (2025-01-24T03:20:37Z)
• MCGS: Multiview Consistency Enhancement for Sparse-View 3D Gaussian Radiance Fields [73.49548565633123]
  Radiance fields represented by 3D Gaussians excel at synthesizing novel views, offering both high training efficiency and fast rendering. Existing methods often incorporate depth priors from dense estimation networks but overlook the inherent multi-view consistency in input images. We propose a view framework based on 3D Gaussian Splatting, named MCGS, enabling scene reconstruction from sparse input views.
  arXiv  Detail & Related papers  (2024-10-15T08:39:05Z)
• Temporal Feature Matters: A Framework for Diffusion Model Quantization [105.3033493564844]
  Diffusion models rely on the time-step for the multi-round denoising.<n>We introduce a novel quantization framework that includes three strategies.<n>This framework preserves most of the temporal information and ensures high-quality end-to-end generation.
  arXiv  Detail & Related papers  (2024-07-28T17:46:15Z)
• Hybrid Dynamic Pruning: A Pathway to Efficient Transformer Inference [1.0919012968294923]
  We introduce a novel algorithm-architecture co-design approach that accelerates transformers using head sparsity, block sparsity and approximation opportunities to reduce computations in attention and reduce memory access. With the observation of the huge redundancy in attention scores and attention heads, we propose a novel integer-based row-balanced block pruning to prune unimportant blocks in the attention matrix at run time. Also propose integer-based head pruning to detect and prune unimportant heads at an early stage at run time.
  arXiv  Detail & Related papers  (2024-07-17T11:15:16Z)
• Revisiting Point Cloud Simplification: A Learnable Feature Preserving Approach [57.67932970472768]
  Mesh and Point Cloud simplification methods aim to reduce the complexity of 3D models while retaining visual quality and relevant salient features. We propose a fast point cloud simplification method by learning to sample salient points. The proposed method relies on a graph neural network architecture trained to select an arbitrary, user-defined, number of points from the input space and to re-arrange their positions so as to minimize the visual perception error.
  arXiv  Detail & Related papers  (2021-09-30T10:23:55Z)
• An Image Enhancing Pattern-based Sparsity for Real-time Inference on Mobile Devices [58.62801151916888]
  We introduce a new sparsity dimension, namely pattern-based sparsity that comprises pattern and connectivity sparsity, and becoming both highly accurate and hardware friendly. Our approach on the new pattern-based sparsity naturally fits into compiler optimization for highly efficient DNN execution on mobile platforms.
  arXiv  Detail & Related papers  (2020-01-20T16:17:36Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.