def train(debug, iterations, table):
game = TicTacToe()
ai1 = Agent(True, table)
ai2 = Agent(True, table)
try:
while True:
if iterations == 0:
table.save_q_table()
break
if iterations > 0:
#if iterations < 50:
# debug = True
iterations -= 1
if iterations % 100_000 == 0:
print(iterations)
print("q_table len", len(table.table))
if game.is_board_full() or game.get_winner():
if game.get_winner() is Player.ONE:
ai1.reward(1)
ai2.reward(0)
elif game.get_winner() is Player.TWO:
ai1.reward(0)
ai2.reward(1)
else:
ai1.reward(0.1)
ai2.reward(0.5)
game.reset()
ai1.reset_history()
ai2.reset_history()
continue
ai1.iterate(game, print_q=debug) if game.get_player(
) == Player.ONE else ai2.iterate(game)
if debug:
print()
print(game.get_hash())
game.print_board()
time.sleep(0.1)
continue
except KeyboardInterrupt:
table.save_q_table()
exit()
table,
))
processes.append(p)
p.start()
for process in processes:
process.join()
else:
game = TicTacToe()
ai1 = Agent(True, table)
table.load_q_table()
while True:
print()
game.print_board()
if game.is_board_full() or game.get_winner():
if game.get_winner():
print(game.get_winner(), "has won!")
else:
print("It's a tie!")
restart = input("Do you want to restart? (Y/n) ")
if restart == "n":
break
else:
game.reset()
continue
if game.get_player() is Player.TWO:
field = input(
str(game.get_player(human=True)) +
" it's your turn! Input a field between 1-9: ")