May, 2024
使用双向 SSM 扩展 Diffusion Mamba 以实现高效图像和视频生成
Scaling Diffusion Mamba with Bidirectional SSMs for Efficient Image and Video Generation
Shentong Mo, Yapeng Tian
TL;DRDiffusion Mamba (DiM) is a novel architecture that effectively addresses the computational complexity of traditional diffusion transformers (DiT) in image generation tasks while maintaining linear complexity with respect to sequence length, outperforming existing techniques and establishing a new benchmark for generative models.