def backup(self, value): if self.previous_state is not None and self.learning: self.net.fit(c4.getNeuralInput(self.previous_state).reshape( 1, 126), value, batch_size=1, nb_epoch=1)
def greedy(self, state, player=1): max_value = float("-inf") next_move = None # TODO: implemen get_possible_moves in c4 for move in range(7): if c4.isValidMove(state, move): new_state = c4.makeMove(state, player, move) val = self.net.predict(c4.getNeuralInput(new_state).reshape(1, 126), batch_size=1) if val > max_value: max_value = val next_move = move self.backup(max_value) return next_move
def greedy(self, state, player=1): max_value = float("-inf") next_move = None # TODO: implemen get_possible_moves in c4 for move in range(7): if c4.isValidMove(state, move): new_state = c4.makeMove(state, player, move) val = self.net.predict(c4.getNeuralInput(new_state).reshape( 1, 126), batch_size=1) if val > max_value: max_value = val next_move = move self.backup(max_value) return next_move
def backup(self, value): if self.previous_state is not None and self.learning: self.net.fit(c4.getNeuralInput(self.previous_state).reshape(1, 126), value, batch_size=1, nb_epoch=1)