Green Tsetlin Redefining Efficiency in Tsetlin Machine Frameworks

TL;DR

Green Tsetlin introduces a production-focused TM framework that unifies training and inference while delivering hardware-agnostic performance via a C++ backend and Python orchestration. It faithfully implements Coalesced and Sparse Tsetlin Machines, includes hyperparameter search and cross-validation out of the box, and supports exporting trained TMs for embedded deployment. The framework demonstrates memory-efficient scalability through SparseTsetlin representations and competitive runtime relative to existing TM implementations, with capabilities extending to CUDA-enabled acceleration in its roadmap. Overall, GT enables practical deployment of explainable TM models across devices from microcontrollers to GPUs, aiming to broaden adoption and real-world impact.

Abstract

Green Tsetlin (GT) is a Tsetlin Machine (TM) framework developed to solve real-world problems using TMs. Several frameworks already exist that provide access to TM implementations. However, these either lack features or have a research-first focus. GT is an easy-to-use framework that aims to lower the complexity and provide a production-ready TM implementation that is great for experienced practitioners and beginners. To this end, GT establishes a clear separation between training and inference. A C++ backend with a Python interface provides competitive training and inference performance, with the option of running in pure Python. It also integrates support for critical components such as exporting trained models, hyper-parameter search, and cross-validation out-of-the-box.

Figures (1)

• Figure 1: Outline of the various components of GT. The Python box of the corresponding color orchestrates the backend blocks (in C++). For instance, the responsibility of creating & adding the ClauseBlocks to the computation graph falls to the Python Tsetlin Machine class (green color). The red box is the exportation of the TM rules into its own standalone C program.