Hyperon-OpenPSI

This repository contains MeTTa experiments for the OpenPsi Architecture, a project to create a cognitive architecture based on the principles of self-organizing systems. OpenPsi aims to provide a framework for building intelligent agents that can learn and adapt in complex environments.

Project Structure

The project is organized into the following directories:

• main: Contains the core logic of the OpenPsi architecture, including:
  • mind-agents: Implements the core cognitive functions of the agent, such as:
    • action-planner: A hill-climbing planner that finds a sequence of actions to achieve a goal.
    • demand-updater: Manages the agent's internal "demands" or motivations.
    • feeling-updater: Updates the agent's emotional state.
    • modulator-updater: Modulates the agent's cognitive processes.
    • monitor-changes: Monitors changes in the agent's external environment.
    • perception-updater: Processes sensory information.
  • demand: Represents the agent's needs.
  • emotion: Defines the agent's emotional states.
  • modulator: Modulates the agent's cognitive processes.
  • rules: Contains the functions to work with rules that govern the agent's behavior.
• psi-utilities: Provides utility functions for the project.
• test: Contains test script to run test-cases with in every component of the project.
• use-cases: Demonstrates how to use the OpenPsi architecture to build intelligent agents, including:
  • curious-agent: An agent that is motivated by curiosity and uses a large language model (Google Gemini) to correlate its internal state with a set of rules.
  • ping-pong: A simple example of a cyclical interaction between two states.
• utilities-module: A github submodule, contains commonly used MeTTa functions for the OpenCog components.

Getting Started

1. To get started with the project, you will need to have Hyperon installed. Because this project requires specific versions of Hyperon and NumPy, it is recommended to install them directly (perhaps in your own virtual environment):

python3 -m pip install hyperon==0.2.3  numpy==2.2.1

2. Then, you need to initialize the git submodule:

git submodule update --init --recursive        
cd utilities-module/
git checkout main
git pull

3. (Optional) Run tests to verify installation

cd .. && python3 test/run-tests.py

4. Then, you can run the examples in the use-cases directory. For example, to run the ping-pong example, you can execute the following command:

metta use-cases/ping-pong/ping-pong.metta

Contributing

Contributions to the project are welcome. Please fork the repository and submit a pull request with your changes.

License

This project is licensed under the MIT License.