Skip to content

ethanscho/simple-card-game-a3c-tensorflow

Repository files navigation

Simple Card Game A3C Tensorflow

This project is an application of A3C to win a simple card game.

Purpose

A3C for solving a simple card game. In a game, agent will have 3 cards numbered 1, 2, 3 respectively. On the table, there will be 2 random cards among 1, 2, 3. The agent should hand in the same numbered card to get the cards on the table. The agent will have 3 trials. If the agent collects all cards in two turns, the agent will have +1 as a reward, if not the agent will get -1.

Result graph shows the agent with A3C will have +95% win rate after around 7k episodes. alt tag

Usage

  1. git clone https://github.com/ethanscho/simple-card-game-a3c-tensorflow.git

  2. cd simple-card-game-a3c-tensorflow

  3. python a3c.py

History

v1.0 - Apr. 03. 2017

Credits

Ethan Cho.

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages