def load():
"""
read input file and self storage
"""
N = 5
piece_type, previous_board, board = FileHandler.readInput(N)
q_table = FileHandler.readQTable()
player = GreedyPlayer(q_table)
action = player.get_input(piece_type, previous_board, board)
return action
def test_evaluation(self):
verbose = True
q_table = FileHandler.read_q_table(path="../QLearner/QTable.pkl")
my_player = QLearningPlayer(q_table=q_table, debug=False)
previous_board = [[0, 0, 0, 0, 0], [0, 0, 0, 2, 0], [0, 0, 1, 0, 0],
[0, 0, 0, 0, 0], [0, 0, 0, 0, 0]]
previous_board_int = GameBoard.encode_board(previous_board)
board = [[0, 0, 0, 0, 0], [0, 0, 1, 2, 0], [0, 0, 1, 0, 0],
[0, 0, 0, 0, 0], [0, 0, 0, 0, 0]]
board_int = GameBoard.encode_board(board)
valid_moves = GameBoard.get_valid_moves(previous_board_int, board_int,
2)
print(my_player._check_q_table(board_int, valid_moves))
print(my_player.get_input(2, previous_board, board))
def test_against_self(self):
num = 100
verbose = False
q_table = FileHandler.read_q_table(path="../QLearner/QTable.pkl")
my_player = QLearningPlayer(q_table=q_table, debug=False)
my_player2 = QLearningPlayer(q_table=q_table, debug=False)
print('Training MyPlayer against Greedy player for {} times......'.
format(num))
go.battle(my_player,
my_player2,
num,
learn=True,
show_result=True,
verbose=verbose)
go.battle(my_player2,
my_player,
num,
learn=True,
show_result=True,
verbose=verbose)
return False
def _store_encoded_state(self, encoded_board, piece_type, value, action):
# pass
if encoded_board in self.history:
self.history[encoded_board][piece_type] = (value, action)
else:
self.history[encoded_board] = {piece_type: (value, action)}
def _get_valid_moves(self, previous_board_int, board_int, piece_type):
moves = GameBoard.get_valid_moves(previous_board_int, board_int, piece_type)
return moves
def load():
"""
read input file and self storage
"""
N = 5
piece_type, previous_board, board = FileHandler.readInput(N)
q_table = FileHandler.readQTable()
player = GreedyPlayer(q_table)
action = player.get_input(piece_type, previous_board, board)
return action
if __name__ == "__main__":
action = load()
FileHandler.writeOutput(action)
move = (i, j)
return curr_max, move
def _check_encoded_state(self, encoded_board, piece_type):
# return False
if encoded_board in self.history:
if piece_type in self.history[encoded_board]:
return self.history[encoded_board][piece_type]
return False
def _store_encoded_state(self, encoded_board, piece_type, value, action):
# pass
if encoded_board in self.history:
self.history[encoded_board][piece_type] = (value, action)
else:
self.history[encoded_board] = {piece_type: (value, action)}
def _get_valid_moves(self, previous_board_int, board_int, piece_type):
moves = GameBoard.get_valid_moves(previous_board_int, board_int,
piece_type)
return moves
if __name__ == "__main__":
piece_type, previous_board, board = FileHandler.read_input(5)
q_table = FileHandler.read_q_table()
player = QLearningPlayer(q_table=q_table)
action = player.get_input(piece_type, previous_board, board)
# FileHandler.writeQTable(player.history)
FileHandler.write_output(action)
self.learn_from_history(self.aggressive_win_table, WIN_REWARD)
# self.learn_from_history(self.aggressive_lose_table, LOSS_REWARD)
def learn_from_history(self, history, reward, alpha=0.7, gamma=1.0):
history.reverse()
max_q_value = -1.0
for hist in history:
state, move = hist
if state == 0:
max_q_value = -1.0
self.Q(state)
q = self.q_values[state]
if max_q_value < 0:
q[move[0]][move[1]] = reward
else:
q[move[0]][move[1]] = q[move[0]][move[1]] * (
1 - alpha) + alpha * (gamma * max_q_value)
max_q_value = np.max(q)
print(max_q_value)
if __name__ == "__main__":
learner = Learner()
learner.learn_from_champion()
# learner.learn_from_aggressive()
# print(learner.q_values)
FileHandler.write_q_table(learner.q_values)
print(FileHandler.read_q_table())
print(len(learner.q_values.keys()))
def test_write_qtable(self):
temp = {6661: 456456}
FileHandler.write_q_table(temp)
def test_read_qtable2(self):
temp = FileHandler.read_q_table(path="QTable_from_run.pkl")
print(temp)
def test_read_qtable(self):
temp = FileHandler.read_q_table()
print(temp)