The Code
Our codebase, documentation, and experimental research code are all hosted under the Thousand Brains Project GitHub organization, making it easy to access, collaborate, and contribute to our work.
Within this organization, you will find two primary repositories:
- tbp.monty, which is our first implementation of a sensorimotor AI framework based on the Thousand Brains Theory
- monty_lab, dedicated to experimental code and exploratory research initiatives of the TBP team.
These repositories collectively represent the foundation of our open-source efforts, embodying our commitment to transparency, innovation, and collaboration.
The tbp.monty repository is an open-source implementation of a sensorimotor learning system inspired by the Thousand Brains Theory of Intelligence. Named after Vernon Mountcastle, Monty models cortical column interactions to enable AI systems to learn from and interact with their environment. Key components include sensor modules, learning modules, motor systems all interconnected via the Cortical Messaging Protocol (CMP). The project is modular, scalable, and encourages community collaboration under the MIT license. Comprehensive documentation is available, making it accessible to researchers and developers.
The monty_lab repository contains experimental files and data analysis scripts used by the team for day-to-day research activities. This repository includes various projects such as grid cell investigations, high-dimensional coincidence detection, and object behavior simulations. It also features tools like the monty_meets_world iOS app for data collection. While the code is not actively maintained, it serves as a resource for replicating experiments and understanding the project’s development.
