ConceptAI & ML
Multi-modal diffusion transformer
A diffusion transformer design variant that handles separate sets of weights for text and image modalities to remove directional biases.
Know first
Named as a prerequisite by the lectures that define this. The number is how many say so. A grey one has no page yet.
Cross-attention1Diffusion transformers1
Lecturers
1Connections
2The lectures that define it or assume it, and the concepts next to it.