Can someone explain me how PyMC3 Model is leveraging the Theano static graph to boost the speed?

There’s a great piece of documentation here: PyMC3 and Theano — PyMC3 3.10.0 documentation. Are there particular questions or issues that you are especially curious about?

Thanks for the information.

Actually I was trying to understand how `aesara`

(theano) generates static graph for the model to calculate gradient with respect to free variable. Is there a way to visualize the parent-child paradigm of the graph?

This page in the theano docs might help you get a better understanding: Printing/Drawing Theano graphs — Theano 1.1.2+29.g8b2825658.dirty documentation

1 Like