def main(algo):
game = TicTacToe()
player_wins = {1: 0, 2: 0}
exp_step = EXPLORATION / (EPISODES * 0.75)
for i in range(EPISODES):
algo.exploration -= exp_step
if algo.exploration < 0:
algo.exploration = 0
player_moves = {1: [], 2: []}
winner = None
if (i % 1000 == 0):
print(f'Running episode {i}')
print(f'Exploration: {round(algo.exploration,2)}')
#print(f'Player1 wins: {player_wins[1]}\nPlayer2 wins: {player_wins[2]}')
print(f'Total wins: {player_wins[1] + player_wins[2]}')
player_wins = {1: 0, 2: 0}
nr_of_moves = 0
while game.get_gamestate() == GameState.ONGOING:
for player in range(1, 3):
if player == 2:
board = game.get_inv_board()
else:
board = game.board
legal_move = False
while not legal_move:
move = algo.make_move(board)
old_board = deepcopy(board)
legal_move = game.play(player, move)
if player == 2:
board = game.get_inv_board()
else:
board = game.board
if not legal_move:
algo.train_move((old_board, deepcopy(board), move),
illegal=True)
else:
player_moves[player].append(
(old_board, deepcopy(board), move))
if game.get_gamestate() == GameState.WON:
winner = player
player_wins[player] += 1
break
nr_of_moves += 1
if nr_of_moves >= 9:
break
if game.get_gamestate() == GameState.WON:
algo.train(player_moves[1],
winner=(winner == 1),
loser=(winner != 1))
algo.train(player_moves[2],
winner=(winner == 2),
loser=(winner != 2))
else:
algo.train(player_moves[1])
algo.train(player_moves[2])
game.clear_board()
nr_of_moves = 0
while game.get_gamestate() == GameState.ONGOING:
#Human player
print(game)
move = int(input('Move: ')) - 1
game.play(1, move)
if game.get_gamestate() == GameState.WON:
winner = 1
continue
nr_of_moves += 1
if nr_of_moves > 9:
break
#AI player
move = algo.make_move(game.get_inv_board(), exploration=False)
game.play(2, move)
if game.get_gamestate() == GameState.WON:
winner = 2
nr_of_moves += 1
if nr_of_moves > 9:
break
print(game)
