janosh/awesome-normalizing-flows

Curates 60+ peer-reviewed publications, 14+ framework-specific packages (PyTorch, TensorFlow, JAX, Julia), and 18 implementation repositories covering normalizing flows—a technique for constructing expressive probability distributions through chains of learnable bijective transformations. Organizes resources across applications in molecular simulation and time-series forecasting, along with foundational papers on coupling layers, autoregressive flows, and equivariant architectures. Includes video tutorials and blog posts to bridge theory and implementation across the deep learning ecosystem.

1,611 stars. No commits in the last 6 months.