A transformer block operating on image-latent tokens, modulated by the diffusion timestep.
Diffusion transformers replace the classic U-Net with a stack of transformer blocks over patchified latent tokens. Each block is conditioned on the timestep (and prompt) through adaptive layer-norm modulation - the timestep embedding shifts and scales activations inside every block. DiTs scale more predictably than U-Nets and power most recent image and video generators.
Open any of these on hfviewer to find this block in the interactive architecture graph.
hfviewer renders the full architecture of 2,300+ Hugging Face models as interactive graphs - hover any block to see what it does, with this model's real numbers.
Browse all model graphs →