Skip to content

Latest commit

 

History

History
52 lines (31 loc) · 1.68 KB

README.md

File metadata and controls

52 lines (31 loc) · 1.68 KB
Raw

Nim Game AI

This software implements a command-line Nim game with an AI opponent that learns through reinforcement learning. The AI uses Q-learning to improve its strategy over time.

This code was inspired by https://cs50.harvard.edu/ai/2024/projects/4/nim/

Usage

To get started with the Nim Game AI, you'll need to have Go installed. Then, you can clone this repository and build the project:

git clone https://github.com/samuelralmeida/nim.git
cd nim
go build

You can run the Nim game with the AI opponent by using the following command:

./nim -alpha=0.5 -epsilon=0.1 -n=10000

Available flags:

  • alpha: The learning rate of the AI (default: 0.5)

  • epsilon: The exploration rate of the AI (default: 0.1)

  • n: The number of training iterations for the AI (default: 10000)

Features

  • Reinforcement Learning: The AI uses Q-learning to improve its strategy.

  • Command-line Interface: Play the Nim game directly from the command line.

  • Customizable Parameters: Adjust the learning rate, exploration rate, and number of training iterations to see how they affect the AI's performance.

How it Works

The game is set up with the following initial configuration:

  • Four piles of objects: [1, 3, 5, 7]
  • Two players: human and AI

The AI learns by playing multiple games and updating its strategy based on the outcomes. The Q-learning algorithm helps the AI to balance between exploring new moves and exploiting known successful moves.

Contributing

If you have any suggestions or find any issues, please feel free to create a pull request or open an issue.

Enjoy playing and improving the AI for the Nim game!