def parse_input(game_style):
splitter = game_style.split(" ")
switcher = {
"solo": AI.SOLO(),
"minmax": AI.MINMAX(),
"mcts": AI.MCTS(),
"NN": NN.BetaOne(),
"File": Network_from_file()
}
return switcher[splitter[0]], switcher[splitter[1]]
def parse_input(game_style, view):
splitter = game_style.split(" ")
switcher = {
"solo": Solo_Ai(view),
"minmax": AI.MINMAX(),
"mcts": AI.MCTS(),
"rnd": AI.VariousRnd(),
"NN": NN.BetaOne()
}
return switcher[splitter[0]], switcher[splitter[1]]
def start_test():
"""def train_on_stored_network():
# A python file made simply to test mcts during multiple iterations
outcome = []
for i in range(0, 10):
play1, play2 = AI.Network_MCTS(f"model_V{(i + 1) * 10}.h5"), AI.VariousRnd()
res = 0
for i in range(0, 10):
start = time.time()
board_size = 3
b = Board([], board_size, 0)
player_turn = 2
while not b.is_terminal_state():
player_turn = 1 if player_turn == 2 else 2
player = play1 if player_turn == 1 else play2
player.decide_move(b, player_turn)
if b.is_winner(1):
res += 1
end = time.time()
print(f"Played {i} game and it took {end - start}")
# Creates a histogram of the results. Proves how well it does
outcome.append(res / 10)
plt.hist(outcome)
print(outcome)
plt.show()"""
"""if __name__ == '__main__':
network = Network((3, 3, 1))
buffer = np.load("tic_tac_toe_4000.npy")
plot = []
for _ in range(10):
move_sum = float(sum(len(g.history) for g in buffer))
games = np.random.choice(
buffer,
size=Config.BATCH_SIZE,
p=[len(g.history) / move_sum for g in buffer]
)
game_pos = [(g, np.random.randint(len(g.history))) for g in games]
batch = [(g.history[i], g.result) for (g, i) in game_pos]
batch_image, expected_out = np.array([i[0] for i in batch]), np.array([i[1] for i in batch])
batch_image = batch_image.reshape((-1, 3, 3, 1))
loss = network.model.fit(batch_image, expected_out, epochs=Config.TRAINING_STEPS)
plot = plot + loss.history["loss"]
plt.plot(plot)
plt.show()
network.model.save("TEST_TIC_TAC_TOE_NETWORK.h5")"""
# A python file made simply to test mcts during multiple iterations
outcome = []
play1, play2 = NN.BetaOne(), AI.VariousRnd()
# play1, play2 = NN.BetaOne(), AI.VariousRnd()
for i in range(0, 50):
#play2 = NN_MCTS(f"Output/model_V{i*Config.CHECKPOINT_INTERVAL}.h5")
start = time.time()
board_size = 4
b = Board([], board_size, 0)
player_turn = 1
while not b.is_terminal_state():
player = play1 if player_turn == 1 else play2
output = player.decide_move(b, player_turn)
player_turn = 1 if player_turn == 2 else 2
"""if not player_turn-1:
image = b.get_board().reshape((-1, b.board_size, b.board_size, 1))
res = play1.network.predict(image)
val, policy_logits = res[0][0][0], res[1][0]
print(val, "\n", policy_logits)
print(b.get_board())
print(output[1].value)"""
if b.is_winner(1):
outcome.append(1)
elif b.is_winner(2):
outcome.append(-1)
else:
outcome.append(0)
print(b.get_board())
end = time.time()
print(f"Played {i} game and it took {end - start}")
# Creates a histogram of the results. Proves how well it does
plt.hist(outcome)
#plt.plot(outcome, [i*Config.CHECKPOINT_INTERVAL for i in range(12)])
print(outcome)
plt.show()
"""play1, play2 = AI.Network_MCTS("Output/model_V3000.h5"), AI.VariousRnd()