Skip to content

iliagrigorevdev/ai-snake

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

AI Snake

Screenshot

Run

Open on GitHub Pages

Description

AI plays a snake game trained using deep reinforcement learning.

For training, the toolkit OpenAI Gym and the implementation of Proximal Policy Optimization algorithm OpenAI Baselines were used.

For the board size of 6x6 cells, the neural network body consists of a rectified convolutional layer followed by a residual block, which consists of two rectified convolutional layers with a skip connection. Each convolution applies 32 filters of kernel size 3x3 with stride 1.

To show the pre-trained model in action in a browser the libraries TensorFlow.js and three.js were used.