https://github.com/hanwenzhu/tictactoe-player

Reinforcement learning TicTacToe playing in pure Python
https://github.com/hanwenzhu/tictactoe-player

pure-python reinforcement-learning tictactoe

Last synced: 2 months ago
JSON representation

Reinforcement learning TicTacToe playing in pure Python

• Host: GitHub
• URL: https://github.com/hanwenzhu/tictactoe-player
• Owner: hanwenzhu
• License: mit
• Created: 2019-11-27T14:59:25.000Z (over 5 years ago)
• Default Branch: master
• Last Pushed: 2019-11-29T04:02:28.000Z (over 5 years ago)
• Last Synced: 2025-02-16T14:51:42.281Z (4 months ago)
• Topics: pure-python, reinforcement-learning, tictactoe
• Language: Python
• Size: 10.7 KB
• Stars: 0
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# tictactoe-player

Reinforcement learning TicTacToe playing in pure Python.

It uses temporal difference to learn a value function for the different states in the finite state space. That is, it updates a `dict` that sees if a move is good.

Some of this code is not well-written. But I'm not planning on improving it significantly.

You can set the decay rate higher and the number of games for training higher for better performance. You can also explore more frequently.