Python renderer for OpenStreetMap with custom icons intended to display as many map features as possible

Free-form flows are a generative model training a pair of neural networks via maximum likelihood

A default trainable module for pytorch lightning.

Normalizing flows in PyTorch

Efficiently computes derivatives of NumPy code.

LaTeX-Kurs

Small project to make the life of tutors easier.

Quill is a modern WYSIWYG editor built for compatibility and extensibility

Interactively retrieve data from sacred experiments.

Code for artificial toy data sets used to evaluate (conditional) invertible neural networks and related methods

Efficient, lightweight variational inference and approximation bounds

Loss Patterns of Neural Networks

Framework for Easily Invertible Architectures

PyTorch AutoNEB implementation to identify minimum energy paths, e.g. in neural network loss landscapes