Generation runs the same network many times, each pass removing a bit of noise under the scheduler's plan.
A diffusion model is executed in a loop: starting from pure noise, the network predicts (and the scheduler removes) a slice of noise at each step, gradually revealing the image. The scheduler defines the noise timetable and update rule - flow-matching variants learn a straighter path that needs far fewer steps. In the graph this whole loop appears as a single 'denoise step' stage that runs N times.
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 →