Welcome to Restricted Boltzmann Machines (RBM) in PyTorch

rbms is a GPU-accelerated package designed to train and analyze Restricted Boltzmann Machines (RBMs). It is intended for students and researchers who need an efficient tool for working with RBMs.

Features

  • GPU Acceleration.

  • Multiple RBM Types: Supports Bernoulli-Bernoulli RBM and Potts-Bernoulli RBM.

  • Extensible Design: Provides an abstract class RBM with methods that can be implemented for new types of RBMs, minimizing the need to reimplement training algorithms, analysis methods, and sampling methods.

Installation

To install rbms, you can use pip:

pip install rbms

What’s New

Version 0.2

  • EBM abstract class for more generic EBMs.

See Releases

Restricted Boltzmann Machines

Tutorials

API