suragnair / alpha-zero-general

A clean implementation based on AlphaZero for any game in any framework + tutorial + Othello/Gobang/TicTacToe/Connect4 and more
MIT License
3.9k stars 1.04k forks source link

mini-othello not trained right? #312

Open elephantpanda opened 1 year ago

elephantpanda commented 1 year ago

In the mini Othello, the output of the neural network model always gives -0.999999 for the first value v of the board when it is X's turn. In the normal 8x8 Othello it gives more of a range of numbers in the range [-1,1]

Also, in the 8x8 Othello game, sometimes it predicts with high probability an invalid move. It seems like it can't understand moves longer than 3-4 squares long. Probably needs more training.