We show how to approximate the KL divergence (in fact, any f-divergence) between implicit distributions using density ratio estimation by probabilistic classification.
We illustrate how to build complicated probability distributions in a modular fashion using the Bijector API from TensorFlow Probability.
We give an in-depth practical guide to variational autoencoders from a probabilistic perspective.
We compare NumPy's `mgrid` and `meshgrid` for building coordinate grids — what each does, why both exist, and how broadcasting often makes them optional.