Atari Games for Deep Reinforcement Learning
The repository helps to quickly prepare an environment for deep reinforcement learning, specifically Atari games Gym environments.
The project makes use of Docker (check the
tutorial if unfamiliar). The
Dockerfile
itself is based on the Pytorch Docker image.
How to use
git clone https://github.com/masterdezign/docker-atari
cd docker-atari
First, you need to download Roms.rar to current directory.
Next, go one level up, build and run the Docker container
cd ..
docker-atari/build.sh
docker-atari/run.sh
Within the Docker container, your current directory ./ is available from /workspace.
Test e.g. running cleanrl scripts.
python cleanrl/ppo_atari.py \
--seed 1 \
--env-id BreakoutNoFrameskip-v4 \
--total-timesteps 10000000