API and Documentation


Most Edward random variables are light wrappers around distributions in TensorFlow. Therefore the list of available random variables depends on the TensorFlow version installed. For TensorFlow 1.1.0, the following are available:


Inference is mostly comprised of classes. They are organized in a class hierarchy, where methods are shared via parent classes and Inference is the top-most base class.


Criticism is comprised of functions. They either operate on random variables in the model or NumPy arrays representing values drawn from the random variables.