This project utilizes two DQN models: an online model and a target model. The online model is responsible for the decision-making of the agent. In contrast, the target mode is primarily used for defining the targets and will be updated with the weights of the online model periodically. The implementation will be entirely in python using the torch framework.
Libraries used: PyTorch, Matplotlib, NumPy, Gym, tqdm
Make sure to do pip install gym[atari] to get access to all the atari environment.
Image credits: https://files.softicons.com/download/game-icons/classic-games-icons-by-thvg/ico/Pacman.ico