A simple Q-learning implementation in OpenAI Gym's "Taxi-v3" environment.
OpenAI Gym is a toolkit for developing and comparing reinforcement algorithms. It provides a wide range of environments with different reinforcement learning tasks.
It can be found on GitHub here and documentation is here.
Python 3 is required and can be downloaded here.
pip3 install -r requirements.txt
py agent.py
- Command line arguments to modify the amount of training episodes.
- Saving and loading the q-table.
- Tuning alpha, gamma and epsilon by decaying over training episodes.