You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
History of State: Since Go is not completely observable from the current state of the board, the neural network also takes as input the boards from the last 7 time steps. This is a feature of the game itself, and other games such as Chess and Othello would only require the current board as input.
This is not strictly speaking true, as chess also requires knowledge of the previous states of the board, to determine for example if castling is permitted, or if you may capture en passant amongst others.
The text was updated successfully, but these errors were encountered:
From your excellent writeup (thank you!):
This is not strictly speaking true, as chess also requires knowledge of the previous states of the board, to determine for example if castling is permitted, or if you may capture en passant amongst others.
The text was updated successfully, but these errors were encountered: