AlphaGo.jl is pure Julia implementation of AlphaGo Zero using Flux.jl.
To install this package simply run
pkg> add https://github.com/tejank10/AlphaGo.jlusing AlphaGoMaking an environment of Go is simple
env = GoEnv(9)
Here 9 is the size of board i.e., a 9x9 board is created.
A B C D E F G H J
9 . . . . . . . . . 9
8 . . . . . . . . . 8
7 . . . . . . . . . 7
6 . . . . . . . . . 6
5 . . . . . . . . . 5
4 . . . . . . . . . 4
3 . . . . . . . . . 3
2 . . . . . . . . . 2
1 . . . . . . . . . 1
A B C D E F G H J
Move: 0. Captures X: 0 O: 0
To Play: X(BLACK) Training is done using train() method. train() method is used by the user to train the model based on the following parameters:
envnum_games: Number of self-play games to be played
Optional arguments:memory_size: Size of the memory bufferbatch_sizeepochs: Number of epochs to train the data onckp_freq: Frequecy of saving the model and weightstower_height: AlphaGo Zero Architecture uses residual networks stacked together. This is called a tower of residual networks. tower_height specifies how many residual blocks to be stacked.model: Object of type NeuralNetreadouts: number of readouts by MCTSPlayerstart_training_after: Number of games after which training will be startedThe network can be tested to play against humans by using the play() method. play() takes following arguments:
envnn: an object of type NeuralNettower_heightnum_readoutsmode: It specifies human will play as Black or white. If mode is 0 then human is Black, else White.