Skip to content

Attention-like modules and deep neural network in computer Go. 🤖

Notifications You must be signed in to change notification settings

vesran/Deep-Computer-Go

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Deep-Computer-Go

A Deep Learning Project of the master IASD. The goal is to train a multi-head network for playing the game of Go. In order to be fair about training ressources the number of parameters for the networks must be lower than 1 000 000. More info.

Top-3 model on intra-master tournament. The architecture is based on attention-like modules such as CBAM and Squeeze-and-Excitation modules as well as residual blocks and activation functions for deep networks.

Data

The data used for training comes from the Katago Go program self played games. There are 1 000 000 different games in total in the training set. The input data is composed of 21 19x19 planes (color to play, ladders, current state on two planes, two previous states on four planes). The output targets are the policy (a vector of size 361 with 1.0 for the move played, 0.0 for the other moves), and the value (1.0 if White won, 0.0 if Black won).

Authors

  • Emilie Chhean
  • Yves Tran

About

Attention-like modules and deep neural network in computer Go. 🤖

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published