rylinnm/dqn-with-cartpole

A Deep Q-learning Network that works in the Cartpole environment

DQN-with-Cartpole

A Deep Q-learning Network that works in the Cartpole environment.

Installation

To run the code make sure the gym environment is installed by using: pip install gym and pip install gym[all]

Execution

After everything is set up correctly the code can be executed with the following command: python Cartpole.py --[implementation] [hyperparameter1] .... [hyperparameterN]

Here 'implementation' can be either two things:

-'ablation' to run our ablation study. No extra hyperparameters can be given. An example would be: python Cartpole.py --ablation

-'tune' To for tuning of hyperparameters. Each hyperparameter you want to tune can be given separately.

The hyperparameters that can be tuned consist of 'layer', 'unit', 'optimizer', 'batch_size', 'epochs', 'policy', and 'gamma'. An example would be: python Cartpole.py --tune layer optimizer batch_size or python Cartpole.py --tune gamma

Results

After the code is run, the figures will be created in the current folder.