🧠 Artificial Intelligence Programming

📘 Project Overview

This repository contains coursework for the Artificial Intelligence Programming course.
It consists of two main projects:

✅ Project 1: JAX Controller

Implementation of a classic PID controller and a neural-network-based controller using JAX for gradient-based optimization.

---

🧪 Project 2: MuZero Knock-Off

An educational reimplementation of MuZero, a state-of-the-art reinforcement learning algorithm by Google Deepmind that learns a model of the environment and uses MCTS for planning.

---

🛠 Usage

Commands to be run from the root directory of the repo.

🛠️ Setup Python Virtual Environment

Create a virtual environment and install dependencies:

# Create virtual environment
python3 -m venv .venv

# Activate it (Linux/macOS)
source .venv/bin/activate

# Or activate it (Windows)
.venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

🔁 Train MuZero

python3 -m project2.training.rl_manager

🎮 Play with Trained Agent & Generate GIF

python3 -m project2.playground.muzero_play_game

Make sure you have the file path for the trained model in the core/config.py file.

📊 Launch TensorBoard

python3 -m project2.training.run_tensorboard

🌲 Visualize Search Tree (Perfect Model)

python3 -m project2.playground.playground_mcts_perfect_model

✅ On Windows

You can also use "python -m project2..." as an alternative to "python3 -m project2...".