Reinforcement Learning: Flappy Bird

This is the original version of our application. The repository is organized in three branches:

master: there is the vanilla version of our code.
image_size: there is the code used for testing different image sizes and network's parameters.
prioritized: it contains the implementation of rank-based prioritized memory.

The code is written in Python using PyTorch library.

Execution

python dqn.py <model>

For instance:

python dqn.py train