Setting up OpenAI Gym Environments for Reinforcement Learning

We provide Atari environments for experimenting with reinforcement learning that can be selected via the Environment component which uses OpenAI Gym as its foundation. However, dependencies must be installed prior to running PerceptiLabs. If these dependencies haven't been installed you will see the following messages in the console when running PerceptiLabs:

PerceptiLabs: Your environment does not have atari_py installed, so some functionality may not be available
PerceptiLabs: To install it, please follow the directions at https://github.com/Kojoley/atari-py and then install gym through 'pip install gym[atari]

Note

Gym environments must be installed on Windows. On Linux and Mac OS they are included in the installation package.

Follow the steps below to install the necessary dependencies within your Python environment:

1. Run pip install -f https://github.com/Kojoley/atari-py/releases atari_py

2. Run pip install gym[atari]

When you run PerceptiLabs, the above message should no longer appear in the console, and Atari environments can now be selected when using the Environment component in your model.