Implementation of Reinforcement Learning algorithms in CarRacing-v0 environment. Implemented algorithms:

• Deep Q-Network (DQN)
• Advantage Actor Critic (A2C)
• Asynchronous Advantage Actor Critic (A3C)

Requirements: python 3.6

To install all required dependencies run:

pip install -r requirements.txt

Start training / inference / evaluation with:

python -m run --<action> -m=<model>

Possible values for parameter action are: train, inference and evaluate.

Possible values for parameter model are: dqn, a2c and a3c.

Hyperparameters can be changed in .json files in /params directory.