Building efficient and trustworthy AI

cirkit is a framework for building, learning and reasoning about probabilistic machine learning models, such as circuits and tensor networks, which are tractable and expressive.

• ⚡ Exact and Efficient Inference : Automatically compile tractable models to efficient computational graphs that run on the GPU.
• Compatible: Seamlessly integrate your circuit with deep learning models; run on any device compatible with PyTorch.
• Modular and Extensible: Support for user-defined layers and parameterizations that extend the symbolic language of cirkit.
• Templates for Common Cases: Templates for constructing circuits by mixing layers and structures with a few lines of code.

Getting Started

• Build a circuit...

  ---

  from region graphs

• Learn a circuit...

  ---

  for distribution estimation

  for tensor compression

  as a (generative) multi-class classifier

  ... all of the above, with PICs

• Advanced reasoning...

  ---

  with squared circuits \((\)\()^2\)

  with logic circuits ...

  
  ...to enforce constraints in neural nets