Skip to content

A training environment for machine learning agents to learn Tetris.

Notifications You must be signed in to change notification settings

TomasBorsje/TetrisGym

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

TetrisGym - An ML Agent Training Environment

TetrisGym

A training environment for machine learning agents to learn Tetris built with Unity & Unity's ml-agents package.

This project also includes a rudimentary machine learning agent implementation & the agent's trained model.

Configuration

Each instance of the TetrisTrainingGym has its own TetrisAgent and TetrisGame instance. By default, games are not rendered on the UI. To render a game on the UI, enable the 'Render GUI' checkbox on the TetrisGame instance. Make sure there is only 1 game instance rendering at a time.

Training

Currently the project is set up to display the actions of the included agent. To train your own agent, make sure to remove the model in the Model field of the only active TetrisAgent.

Parallel Training

This project is set up to support parallel training. To train multiple instances of the neural net at once, simply add more instances of the TetrisTrainingGym prefab. There are already an extra 9 TetrisTrainingGym instances present in the scene, but they are not active by default.

About

A training environment for machine learning agents to learn Tetris.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages