// ✦ Diffusion, Flow Matching & Generative Models
Normalizing Flows (RealNVP / Glow)
Chains invertible maps from a simple distribution to a complex one, with exact likelihood.
标准化流（RealNVP / Glow） aka RealNVP / Glow / Normalizing Flow 2016 Advanced extended
IN PLAIN WORDS
Start from a standard Gaussian and apply invertible transforms whose Jacobian determinant is cheap; run it backwards to map data to the Gaussian and get exact probabilities. RealNVP’s affine coupling layers make this fast and invertible. Diffusion overtook flows on image quality, but the idea fed into flow matching.
Used in: Exact-likelihood density estimation, physics sampling, variational inference
01Original paper
Density estimation using Real NVP
Laurent Dinh, Jascha Sohl-Dickstein, Samy Bengio · 2016
arXiv ↗ ✓ Verified · arXiv API
02Where it sits
Learn the left column first; the right column builds on this one.
Learn first
Then you can learn
- Nothing in the atlas builds on it yet
Related:Flow MatchingGenerative Adversarial Network (GAN)
Left: prerequisites · right: builds on this · dashed: related. Full map →
03Learning resources
Official docs, explainers, courses and reference code. All links checked.
04Where the course teaches it
For AI / data
- Verification: verified, checked 2026-10-22.
- JSON: /api/learn/algorithms.json (id:
normalizing-flows)