Paper · arXiv 2403.13802
ZigMa: Zigzag Mamba Diffusion Model
Vincent Tao Hu, Stefan Andreas Baumann, Ming Gui, Olga Grebenkova, Pingchuan Ma, Johannes Fischer, Bjorn Ommer
Zigzag Mamba, an improvement over Mamba, enhances visual data generation with better spatial continuity, speed, and memory usage, and is scalable with Stochastic Interpolant on large-resolution datasets.
Abstract
The diffusion model has long been plagued by scalability and quadratic complexity issues, especially within transformer-based structures. In this study, we aim to leverage the long sequence modeling capability of a State-Space Model called Mamba to extend its applicability to visual data generation. Firstly, we identify a critical oversight in most current Mamba-based vision methods, namely the lack of consideration for spatial continuity in the scan scheme of Mamba. Secondly, building upon this insight, we introduce a simple, plug-and-play, zero-parameter method named Zigzag Mamba, which outperforms Mamba-based baselines and demonstrates improved speed and memory utilization compared to transformer-based baselines. Lastly, we integrate Zigzag Mamba with the Stochastic Interpolant framework to investigate the scalability of the model on large-resolution visual datasets, such as FacesHQ 1024times 1024 and UCF101, MultiModal-CelebA-HQ, and MS COCO 256times 256. Code will be released at https://taohu.me/zigma/