Diffusion Transformer
An advanced generative architecture that replaces the traditional convolutional U-Net backbone in a diffusion model with a highly scalable transformer network.
Think of It Like This
Like upgrading a factory assembly line from specialized custom robotic arms to a unified, programmable grid of robots that can scale infinitely.
Diffusion Transformers (DiTs) apply the sequence-to-sequence modeling capabilities of transformers directly to patches of noise. This design choice benefits heavily from established scaling laws, allowing performance to predictably improve as compute and parameter count increase. DiTs are the foundation of modern high-fidelity video generators.