Deep Q-Learning (DQN) implementation for Atari pong.
Deep Q-network implementation for Pong-vo. The implementation follows from the paper - Playing Atari with Deep Reinforcement Learning and Human-level control through deep reinforcement learning.
Example:
conda create -n dqn_pong
pip install -r requirements.txt
gym.wrappers.Monitor
to record the last episodesudo apt-get install ffmpeg
python train_atari.py
python train_atari.py --load-checkpoint-file results/checkpoint_dqn_nature.pth
A video is recorded every 50 episodes. See videos in /video/
folder.