DL07 · Generative Models II: Diffusion and Flow Matching
A free, illustrated L3 to L4 course of 10 modules in the Deep Learning track. Modules include: Denoising diffusion models; Score matching; The SDE and ODE views.
DL07 L3 to L4 · Advanced to Expert 10 modules · about 40 lessons Planned · Ring 4
Part of the Deep Learning track. Ships in Ring 4 (Expert) of the roadmap.
Modules
- Denoising diffusion models
- Score matching
- The SDE and ODE views
- Samplers and fast sampling
- Guidance and conditioning
- Latent diffusion
- Flow matching and rectified flow
- Architectures from U-Net to diffusion transformers
- Consistency models and distillation
- Video, 3D and discrete diffusion
Signature illustrations
The pictures a learner should still see in their head a year later:
Strips of forward noising, arrows of a score field, fans of sampling trajectories, grids across guidance scale, straight versus curved flow paths.
Status
This course is planned, not written yet. It ships with Ring 4 (Expert).
Every lesson will follow the same twelve-part blueprint with at least three purpose-made figures. Watch the repository to hear when modules ship, or help write and review it.