CartPole Game Bot

This is a reinforcement learning problem. I used Deep Q-learning algorithm to build this game bot

Usage

Clone or download the repository by following command:

   git clone https://github.com/riti1302/CartPole    

Install the required libraries

    pip3 install -r requirements.txt    

To run the bot on already trained model run

    python3 main.py    

To increase the average score of each episode navigate to getTrainingData.py, update the score and run:

   python3 getTrainingData.py     

Then navigate to getModel.py and update the file_path and saved model name. Then run:

   python3 getModel.py      

Update the model name in main.py and run:

   python3 main.py    

Model Architecture

Output Graphs

Accuracy

Loss

This model is trained on five epochs. The average score after 50 episodes reached 276. For few episodes score reached upto 500.