Parallel Sequence Modeling via Generalized Spatial Propagation Network

TL;DR

The paper introduces GSPN, an attention mechanism optimized for vision tasks that directly models 2D spatial structures, improving efficiency and spatial fidelity over existing methods, and achieving state-of-the-art results.

Contribution

GSPN is a novel attention mechanism that operates on 2D spatial data with a stability condition, reducing computational complexity and enhancing performance in vision applications.

Findings

GSPN outperforms existing attention models in vision tasks.

GSPN accelerates SD-XL by over 84 times for large image generation.

GSPN achieves state-of-the-art results on ImageNet classification.

Abstract

We present the Generalized Spatial Propagation Network (GSPN), a new attention mechanism optimized for vision tasks that inherently captures 2D spatial structures. Existing attention models, including transformers, linear attention, and state-space models like Mamba, process multi-dimensional data as 1D sequences, compromising spatial coherence and efficiency. GSPN overcomes these limitations by directly operating on spatially coherent image data and forming dense pairwise connections through a line-scan approach. Central to GSPN is the Stability-Context Condition, which ensures stable, context-aware propagation across 2D sequences and reduces the effective sequence length to $\sqrt{N}$ for a square map with N elements, significantly enhancing computational efficiency. With learnable, input-dependent weights and no reliance on positional embeddings, GSPN achieves superior spatial fidelity and state-of-the-art performance in vision tasks, including ImageNet classification, class-guided image generation, and text-to-image generation. Notably, GSPN accelerates SD-XL with softmax-attention by over $84\times$ when generating 16K images.