jrodrigomg / black-ai

Playing Black Jack with Reinforcement Learning

black-ai

Playing Black Jack with Reinforcement Learning

Requisites:

• Python3
• Virtualenv

Instalation

Cloning:

$ git clone https://github.com/jrodrigomg/black-ai
$ cd black-ai/ai

Creating a virtual environment:

$ virtualenv -p python3.5 env

Load environment, this is an example for fish shell:

$ source env/bin/activate.fish

Install requirments:

$ pip install -r requirements.txt

Run the application server

Run the main file:

$ python main.py

Running the game

In the main directory go to the game carpet:

$ cd game

Run a simple http server with python(Not necesary the enviroment here)

$ python -m SimpleHTTPServer 8080

Try

• Once running both application we need to configure wherever the server ask
• go http://localhost:8080 in the browser
• If just appear "Epoch 1" then refresh the browser

Credits

• Thanks to adventuresinML for the incredible post of RL (https://adventuresinmachinelearning.com/reinforcement-learning-tutorial-python-keras/)
• Thanks to Kusnierewicz for the game of blackjack in javascript (https://github.com/Kusnierewicz/Blackjack-game-in-JS)
• For the structure of the enviroment i used some part of Vincent Dutordoir's repo (https://github.com/vdutor/TF-rex)