def __init__(
self,
session,
num_actions,
input_formatter_info=[0, 0],
player_index=-1,
action_handler=None,
is_training=False,
optimizer=tf.train.GradientDescentOptimizer(learning_rate=0.1),
summary_writer=None,
summary_every=100,
config_file=None):
super().__init__(session,
num_actions,
input_formatter_info=input_formatter_info,
player_index=player_index,
action_handler=action_handler,
is_training=is_training,
optimizer=optimizer,
summary_writer=summary_writer,
summary_every=summary_every,
config_file=config_file)
if player_index >= 0:
self.rotating_expected_reward_buffer = live_data_util.RotatingBuffer(
player_index)
def __init__(self, name, team, index, bot_parameters=None):
self.last_frame_time = None
self.config_file = bot_parameters
self.index = index
self.load_config_file()
self.reward_manager = reward_manager.RewardManager()
config = tf.ConfigProto(device_count={'GPU': 0})
self.sess = tf.Session(config=config)
# self.sess = tf.Session()
self.actions_handler = action_factory.get_handler(
control_scheme=self.control_scheme)
self.num_actions = self.actions_handler.get_logit_size()
print('num_actions', self.num_actions)
self.model = self.model_class(self.sess,
self.num_actions,
input_formatter_info=[team, index],
player_index=self.index,
action_handler=self.actions_handler,
config_file=bot_parameters,
is_training=False)
self.model.add_summary_writer('random_packet', is_replay=True)
self.model.batch_size = 1
self.model.mini_batch_size = 1
self.model.is_graphing = self.is_graphing
self.model.is_online_training = self.is_online_training
self.model.apply_feature_creation(TensorflowFeatureCreator())
try:
self.model.create_model(self.model.get_input_placeholder())
except TypeError as e:
raise Exception('failed to create model') from e
if self.model.is_training and self.model.is_online_training:
self.model.create_reinforcement_training_model()
self.model.create_savers()
self.model.initialize_model()
if self.is_graphing:
self.rotating_real_reward_buffer = live_data_util.RotatingBuffer(
self.index + 10)
return p6, blue_curve, orng_curve
#QtGui.QApplication.setGraphicsSystem('raster')
app = QtGui.QApplication([])
#mw = QtGui.QMainWindow()
#mw.resize(800,800)
win = pg.GraphicsWindow(title="RLBot Live Plotting")
win.resize(500, 300)
win.setWindowTitle('RLBot Live Plotting')
# Enable antialiasing for prettier plots
pg.setConfigOptions(antialias=True)
# Open shared memory rotating buffer
expected_reward_blue_0 = live_data_util.RotatingBuffer(0)
expected_reward_orange_0 = live_data_util.RotatingBuffer(1)
real_reward_blue_0 = live_data_util.RotatingBuffer(10)
real_reward_orange_0 = live_data_util.RotatingBuffer(11)
realPlot, real_blue, real_orange = create_plot("RLBot Real Rewards")
expectedPlot, expected_blue, expected_orange = create_plot("RLBot Expected Rewards")
def update():
global realPlot, real_blue, real_orange, expectedPlot, expected_blue, expected_orange
expected_blue.setData(expected_reward_blue_0.get_current_buffer())
expected_orange.setData(expected_reward_orange_0.get_current_buffer())
real_blue.setData(real_reward_blue_0.get_current_buffer())
real_orange.setData(real_reward_orange_0.get_current_buffer())
timer = QtCore.QTimer()
timer.timeout.connect(update)