catgrad

a categorical deep learning compiler

View on GitHub

catgrad is a deep learning framework using category theory to statically compile models into their forward and backwards passes. This means your training loop can run without needing a deep learning framework (not even catgrad!)

Installation

You can install catgrad using pip:

pip install catgrad

Research

Some papers and research which catgrad builds on:

• catgrad: A Categorical Compiler for Deep Learning
• Data-Parallel Algorithms for String Diagrams
• Categories of Differentiable Polynomial Circuits for Machine Learning
• Categorical Foundations of Gradient-Based Learning
• Reverse Derivative Ascent: A Categorical Approach to Learning Boolean Circuits