def measure(name, iters=5000, **settings):
print(name)
for k, v in settings.items():
print("\t{}: {}".format(k, v))
# Vizdoom wrapper
doom_wrapper = VizdoomWrapper(**settings)
start = time()
for _ in trange(iters, leave=False):
current_img, current_misc = doom_wrapper.get_current_state()
action_index = randint(0, doom_wrapper.actions_num - 1)
doom_wrapper.make_action(action_index)
if doom_wrapper.is_terminal():
doom_wrapper.reset()
end = time()
wrapper_t = (end - start)
# Vanilla vizdoom:
doom = vzd.DoomGame()
if "scenarios_path" not in settings:
scenarios_path = vzd.__path__[0] + "/scenarios"
else:
scenarios_path = settings["scenarios_path"]
config_file = scenarios_path + "/" + settings["config_file"]
doom.load_config(config_file)
doom.set_window_visible(False)
doom.set_screen_format(vzd.ScreenFormat.GRAY8)
doom.set_screen_resolution(vzd.ScreenResolution.RES_160X120)
doom.init()
actions = [
list(a)
for a in it.product([0, 1],
repeat=len(doom.get_available_game_variables()))
]
start = time()
frame_skip = settings["frame_skip"]
for _ in trange(iters, leave=False):
if doom.is_episode_finished():
doom.new_episode()
doom.make_action(choice(actions), frame_skip)
end = time()
vanilla_t = end - start
print(green("\twrapper: {:0.2f} steps/s".format(iters / wrapper_t)))
print(
green("\twrapper: {:0.2f} s/1000 steps".format(wrapper_t / iters *
1000)))
print(blue("\tvanilla: {:0.2f} steps/s".format(iters / vanilla_t)))
print(
blue("\tvanilla: {:0.2f} s/1000 steps\n".format(vanilla_t / iters *
1000)))
class DQN(object):
def __init__(self,
scenario_tag=None,
run_id_string=None,
network_type="networks.DQNNet",
write_summaries=True,
tf_logdir="tensorboard_logs",
epochs=100,
train_steps_per_epoch=1000000,
test_episodes_per_epoch=100,
run_tests=True,
initial_epsilon=1.0,
final_epsilon=0.0000,
epsilon_decay_steps=10e07,
epsilon_decay_start_step=2e05,
frozen_steps=5000,
batchsize=32,
memory_capacity=10000,
update_pattern=(4, 4),
prioritized_memory=False,
enable_progress_bar=True,
save_interval=1,
writer_max_queue=10,
writer_flush_secs=120,
dynamic_frameskips=None,
**settings):
if prioritized_memory:
raise NotImplementedError("Prioritized memory not implemented. Maybe some day.")
# TODO maybe some day ...
pass
if dynamic_frameskips:
if isinstance(dynamic_frameskips, (list, tuple)):
self.frameskips = list(dynamic_frameskips)
elif isinstance(dynamic_frameskips, int):
self.frameskips = list(range(1, dynamic_frameskips + 1))
else:
self.frameskips = [None]
self.update_pattern = update_pattern
self.write_summaries = write_summaries
self._settings = settings
self.run_id_string = run_id_string
self.train_steps_per_epoch = train_steps_per_epoch
self._run_tests = test_episodes_per_epoch > 0 and run_tests
self.test_episodes_per_epoch = test_episodes_per_epoch
self._epochs = np.float32(epochs)
self.doom_wrapper = VizdoomWrapper(**settings)
misc_len = self.doom_wrapper.misc_len
img_shape = self.doom_wrapper.img_shape
self.use_misc = self.doom_wrapper.use_misc
self.actions_num = self.doom_wrapper.actions_num
self.replay_memory = ReplayMemory(img_shape, misc_len, batch_size=batchsize, capacity=memory_capacity)
self.network = eval(network_type)(actions_num=self.actions_num * len(self.frameskips), img_shape=img_shape,
misc_len=misc_len,
**settings)
self.batchsize = batchsize
self.frozen_steps = frozen_steps
self.save_interval = save_interval
self._model_savefile = settings["models_path"] + "/" + self.run_id_string
## TODO move summaries somewhere so they are consistent between dqn and asyncs
if self.write_summaries:
assert tf_logdir is not None
if not os.path.isdir(tf_logdir):
os.makedirs(tf_logdir)
self.scores_placeholder, summaries = setup_vector_summaries(scenario_tag + "/scores")
self._summaries = tf.summary.merge(summaries)
self._train_writer = tf.summary.FileWriter("{}/{}/{}".format(tf_logdir, self.run_id_string, "train"),
flush_secs=writer_flush_secs, max_queue=writer_max_queue)
self._test_writer = tf.summary.FileWriter("{}/{}/{}".format(tf_logdir, self.run_id_string, "test"),
flush_secs=writer_flush_secs, max_queue=writer_max_queue)
else:
self._train_writer = None
self._test_writer = None
self._summaries = None
self.steps = 0
# TODO epoch as tf variable?
self._epoch = 1
# Epsilon
self.epsilon_decay_rate = (initial_epsilon - final_epsilon) / epsilon_decay_steps
self.epsilon_decay_start_step = epsilon_decay_start_step
self.initial_epsilon = initial_epsilon
self.final_epsilon = final_epsilon
self.enable_progress_bar = enable_progress_bar
def get_current_epsilon(self):
eps = self.initial_epsilon - (self.steps - self.epsilon_decay_start_step) * self.epsilon_decay_rate
return np.clip(eps, self.final_epsilon, 1.0)
def get_action_and_frameskip(self, ai):
action = ai % self.actions_num
frameskip = self.frameskips[ai // self.actions_num]
return action, frameskip
@staticmethod
def print_epoch_log(prefix, scores, steps, epoch_time):
mean_score = np.mean(scores)
score_std = np.std(scores)
min_score = np.min(scores)
max_score = np.max(scores)
episodes = len(scores)
steps_per_sec = steps / epoch_time
mil_steps_per_hour = steps_per_sec * 3600 / 1000000.0
log(
"{}: Episodes: {}, mean: {}, min: {}, max: {}, "
" Speed: {:.0f} STEPS/s, {:.2f}M STEPS/hour, time: {}".format(
prefix,
episodes,
green("{:0.3f}±{:0.2f}".format(mean_score, score_std)),
red("{:0.3f}".format(min_score)),
blue("{:0.3f}".format(max_score)),
steps_per_sec,
mil_steps_per_hour,
sec_to_str(epoch_time)
))
def save_model(self, session, savefile=None):
if savefile is None:
savefile = self._model_savefile
savedir = os.path.dirname(savefile)
if not os.path.exists(savedir):
log("Creating directory: {}".format(savedir))
os.makedirs(savedir)
log("Saving model to: {}".format(savefile))
saver = tf.train.Saver()
saver.save(session, savefile)
def load_model(self, session, savefile):
saver = tf.train.Saver()
log("Loading model from: {}".format(savefile))
saver.restore(session, savefile)
log("Loaded model.")
def train(self, session):
# Prefill replay memory:
for _ in trange(self.replay_memory.capacity, desc="Filling replay memory",
leave=False, disable=not self.enable_progress_bar, file=sys.stdout):
if self.doom_wrapper.is_terminal():
self.doom_wrapper.reset()
s1 = self.doom_wrapper.get_current_state()
action_frameskip_index = randint(0, self.actions_num * len(self.frameskips) - 1)
action_index, frameskip = self.get_action_and_frameskip(action_frameskip_index)
reward = self.doom_wrapper.make_action(action_index, frameskip)
terminal = self.doom_wrapper.is_terminal()
s2 = self.doom_wrapper.get_current_state()
self.replay_memory.add_transition(s1, action_frameskip_index, s2, reward, terminal)
overall_start_time = time()
self.network.update_target_network(session)
log(green("Starting training.\n"))
while self._epoch <= self._epochs:
self.doom_wrapper.reset()
train_scores = []
test_scores = []
train_start_time = time()
for _ in trange(self.train_steps_per_epoch, desc="Training, epoch {}".format(self._epoch),
leave=False, disable=not self.enable_progress_bar, file=sys.stdout):
self.steps += 1
s1 = self.doom_wrapper.get_current_state()
if random() <= self.get_current_epsilon():
action_frameskip_index = randint(0, self.actions_num*len(self.frameskips) - 1)
action_index, frameskip = self.get_action_and_frameskip(action_frameskip_index)
else:
action_frameskip_index = self.network.get_action(session, s1)
action_index, frameskip = self.get_action_and_frameskip(action_frameskip_index)
reward = self.doom_wrapper.make_action(action_index, frameskip)
terminal = self.doom_wrapper.is_terminal()
s2 = self.doom_wrapper.get_current_state()
self.replay_memory.add_transition(s1, action_frameskip_index, s2, reward, terminal)
if self.steps % self.update_pattern[0] == 0:
for _ in range(self.update_pattern[1]):
self.network.train_batch(session, self.replay_memory.get_sample())
if terminal:
train_scores.append(self.doom_wrapper.get_total_reward())
self.doom_wrapper.reset()
if self.steps % self.frozen_steps == 0:
self.network.update_target_network(session)
train_time = time() - train_start_time
log("Epoch {}".format(self._epoch))
log("Training steps: {}, epsilon: {}".format(self.steps, self.get_current_epsilon()))
self.print_epoch_log("TRAIN", train_scores, self.train_steps_per_epoch, train_time)
test_start_time = time()
test_steps = 0
# TESTING
for _ in trange(self.test_episodes_per_epoch, desc="Testing, epoch {}".format(self._epoch),
leave=False, disable=not self.enable_progress_bar, file=sys.stdout):
self.doom_wrapper.reset()
while not self.doom_wrapper.is_terminal():
test_steps += 1
state = self.doom_wrapper.get_current_state()
action_frameskip_index = self.network.get_action(session, state)
action_index, frameskip = self.get_action_and_frameskip(action_frameskip_index)
self.doom_wrapper.make_action(action_index, frameskip)
test_scores.append(self.doom_wrapper.get_total_reward())
test_time = time() - test_start_time
self.print_epoch_log("TEST", test_scores, test_steps, test_time)
if self.write_summaries:
log("Writing summaries.")
train_summary = session.run(self._summaries, {self.scores_placeholder: train_scores})
self._train_writer.add_summary(train_summary, self.steps)
if self._run_tests:
test_summary = session.run(self._summaries, {self.scores_placeholder: test_scores})
self._test_writer.add_summary(test_summary, self.steps)
# Save model
if self._epoch % self.save_interval == 0:
savedir = os.path.dirname(self._model_savefile)
if not os.path.exists(savedir):
log("Creating directory: {}".format(savedir))
os.makedirs(savedir)
log("Saving model to: {}".format(self._model_savefile))
saver = tf.train.Saver()
saver.save(session, self._model_savefile)
overall_time = time() - overall_start_time
log("Total elapsed time: {}\n".format(sec_to_str(overall_time)))
self._epoch += 1
def run_test_episode(self, session):
self.doom_wrapper.reset()
while not self.doom_wrapper.is_terminal():
state = self.doom_wrapper.get_current_state()
action_frameskip_index = self.network.get_action(session, state)
action_index, frameskip = self.get_action_and_frameskip(action_frameskip_index)
self.doom_wrapper.make_action(action_index, frameskip)
reward = self.doom_wrapper.get_total_reward()
return reward
class A3CLearner(Thread):
def __init__(self,
thread_index=0,
game=None,
model_savefile=None,
network_class="ACLstmNet",
global_steps_counter=None,
scenario_tag=None,
run_id_string=None,
session=None,
tf_logdir=None,
global_network=None,
optimizer=None,
learning_rate=None,
test_only=False,
test_interval=1,
write_summaries=True,
enable_progress_bar=True,
deterministic_testing=True,
save_interval=1,
writer_max_queue=10,
writer_flush_secs=120,
gamma_compensation=False,
figar_gamma=False,
gamma=0.99,
show_heatmaps=True,
**settings):
super(A3CLearner, self).__init__()
log("Creating actor-learner #{}.".format(thread_index))
self.thread_index = thread_index
self._global_steps_counter = global_steps_counter
self.write_summaries = write_summaries
self.save_interval = save_interval
self.enable_progress_bar = enable_progress_bar
self._model_savefile = None
self._train_writer = None
self._test_writer = None
self._summaries = None
self._session = session
self.deterministic_testing = deterministic_testing
self.local_steps = 0
# TODO epoch as tf variable?
self._epoch = 1
self.train_scores = []
self.train_actions = []
self.train_frameskips = []
self.show_heatmaps = show_heatmaps
self.test_interval = test_interval
self.local_steps_per_epoch = settings["local_steps_per_epoch"]
self._run_tests = settings["test_episodes_per_epoch"] > 0 and settings["run_tests"]
self.test_episodes_per_epoch = settings["test_episodes_per_epoch"]
self._epochs = np.float32(settings["epochs"])
self.max_remembered_steps = settings["max_remembered_steps"]
assert not (gamma_compensation and figar_gamma)
gamma = np.float32(gamma)
if gamma_compensation:
self.scale_gamma = lambda fskip: ((1 - gamma ** fskip) / (1 - gamma), gamma ** fskip)
elif figar_gamma:
self.scale_gamma = lambda fskip: (1.0, gamma ** fskip)
else:
self.scale_gamma = lambda _: (1.0, gamma)
if self.write_summaries and thread_index == 0 and not test_only:
assert tf_logdir is not None
self.run_id_string = run_id_string
self.tf_models_path = settings["models_path"]
create_directory(tf_logdir)
if self.tf_models_path is not None:
create_directory(self.tf_models_path)
if game is None:
self.doom_wrapper = VizdoomWrapper(**settings)
else:
self.doom_wrapper = game
misc_len = self.doom_wrapper.misc_len
img_shape = self.doom_wrapper.img_shape
self.use_misc = self.doom_wrapper.use_misc
self.actions_num = self.doom_wrapper.actions_num
self.local_network = getattr(networks, network_class)(actions_num=self.actions_num, img_shape=img_shape,
misc_len=misc_len,
thread=thread_index, **settings)
if not test_only:
self.learning_rate = learning_rate
# TODO check gate_gradients != Optimizer.GATE_OP
grads_and_vars = optimizer.compute_gradients(self.local_network.ops.loss,
var_list=self.local_network.get_params())
grads, local_vars = zip(*grads_and_vars)
grads_and_global_vars = zip(grads, global_network.get_params())
self.train_op = optimizer.apply_gradients(grads_and_global_vars, global_step=tf.train.get_global_step())
self.global_network = global_network
self.local_network.prepare_sync_op(global_network)
if self.thread_index == 0 and not test_only:
self._model_savefile = model_savefile
if self.write_summaries:
self.actions_placeholder = tf.placeholder(tf.int32, None)
self.frameskips_placeholder = tf.placeholder(tf.int32, None)
self.scores_placeholder, summaries = setup_vector_summaries(scenario_tag + "/scores")
# TODO remove scenario_tag from histograms
a_histogram = tf.summary.histogram(scenario_tag + "/actions", self.actions_placeholder)
fs_histogram = tf.summary.histogram(scenario_tag + "/frameskips", self.frameskips_placeholder)
score_histogram = tf.summary.histogram(scenario_tag + "/scores", self.scores_placeholder)
lr_summary = tf.summary.scalar(scenario_tag + "/learning_rate", self.learning_rate)
summaries.append(lr_summary)
summaries.append(a_histogram)
summaries.append(fs_histogram)
summaries.append(score_histogram)
self._summaries = tf.summary.merge(summaries)
self._train_writer = tf.summary.FileWriter("{}/{}/{}".format(tf_logdir, self.run_id_string, "train"),
flush_secs=writer_flush_secs, max_queue=writer_max_queue)
self._test_writer = tf.summary.FileWriter("{}/{}/{}".format(tf_logdir, self.run_id_string, "test"),
flush_secs=writer_flush_secs, max_queue=writer_max_queue)
def heatmap(self, actions, frameskips):
min_frameskip = np.min(frameskips)
max_frameskip = np.max(frameskips)
fs_values = range(min_frameskip, max_frameskip + 1)
a_labels = [str(a) for a in self.doom_wrapper.actions]
mat = np.zeros((self.actions_num, (len(fs_values))))
for f, a in zip(frameskips, actions):
mat[a, f - min_frameskip] += 1
return string_heatmap(mat, fs_values, a_labels)
@staticmethod
def choose_best_index(policy, deterministic=True):
if deterministic:
return np.argmax(policy)
r = random.random()
cummulative_sum = 0.0
for i, p in enumerate(policy):
cummulative_sum += p
if r <= cummulative_sum:
return i
return len(policy) - 1
def make_training_step(self):
states_img = []
states_misc = []
actions = []
rewards_reversed = []
Rs = []
self._session.run(self.local_network.ops.sync)
initial_network_state = None
if self.local_network.has_state():
initial_network_state = self.local_network.get_current_network_state()
terminal = None
steps_performed = 0
for _ in range(self.max_remembered_steps):
steps_performed += 1
current_state = self.doom_wrapper.get_current_state()
policy = self.local_network.get_policy(self._session, current_state)
action_index = A3CLearner.choose_best_index(policy, deterministic=False)
states_img.append(current_state[0])
states_misc.append(current_state[1])
actions.append(action_index)
reward = self.doom_wrapper.make_action(action_index)
terminal = self.doom_wrapper.is_terminal()
rewards_reversed.insert(0, reward)
self.local_steps += 1
if terminal:
if self.thread_index == 0:
self.train_scores.append(self.doom_wrapper.get_total_reward())
self.doom_wrapper.reset()
if self.local_network.has_state():
self.local_network.reset_state()
break
self.train_actions += actions
self.train_frameskips += [self.doom_wrapper.frameskip] * len(actions)
if terminal:
R = 0.0
else:
R = self.local_network.get_value(self._session, self.doom_wrapper.get_current_state())
# #TODO this could be handles smarter ....
for ri in rewards_reversed:
scale, gamma = self.scale_gamma(self.doom_wrapper.frameskip)
R = scale * ri + gamma * R
Rs.insert(0, R)
train_op_feed_dict = {
self.local_network.vars.state_img: states_img,
self.local_network.vars.a: actions,
self.local_network.vars.R: Rs
}
if self.use_misc:
train_op_feed_dict[self.local_network.vars.state_misc] = states_misc
if self.local_network.has_state():
train_op_feed_dict[self.local_network.vars.initial_network_state] = initial_network_state
train_op_feed_dict[self.local_network.vars.sequence_length] = [len(actions)]
self._session.run(self.train_op, feed_dict=train_op_feed_dict)
return steps_performed
def run_episode(self, deterministic=True, return_stats=False):
self.doom_wrapper.reset()
if self.local_network.has_state():
self.local_network.reset_state()
actions = []
frameskips = []
rewards = []
while not self.doom_wrapper.is_terminal():
current_state = self.doom_wrapper.get_current_state()
action_index, frameskip = self._get_best_action(self._session, current_state, deterministic=deterministic)
reward = self.doom_wrapper.make_action(action_index, frameskip)
if return_stats:
actions.append(action_index)
if frameskip is None:
frameskip = self.doom_wrapper.frameskip
frameskips.append(frameskip)
rewards.append(reward)
total_reward = self.doom_wrapper.get_total_reward()
if return_stats:
return total_reward, actions, frameskips, rewards
else:
return total_reward
def test(self, episodes_num=None, deterministic=True):
if episodes_num is None:
episodes_num = self.test_episodes_per_epoch
test_start_time = time.time()
test_rewards = []
test_actions = []
test_frameskips = []
for _ in trange(episodes_num, desc="Testing", file=sys.stdout,
leave=False, disable=not self.enable_progress_bar):
total_reward, actions, frameskips, _ = self.run_episode(deterministic=deterministic, return_stats=True)
test_rewards.append(total_reward)
test_actions += actions
test_frameskips += frameskips
self.doom_wrapper.reset()
if self.local_network.has_state():
self.local_network.reset_state()
test_end_time = time.time()
test_duration = test_end_time - test_start_time
min_score = np.min(test_rewards)
max_score = np.max(test_rewards)
mean_score = np.mean(test_rewards)
score_std = np.std(test_rewards)
log(
"TEST: mean: {}, min: {}, max: {}, test time: {}".format(
green("{:0.3f}±{:0.2f}".format(mean_score, score_std)),
red("{:0.3f}".format(min_score)),
blue("{:0.3f}".format(max_score)),
sec_to_str(test_duration)))
return test_rewards, test_actions, test_frameskips
def _print_train_log(self, scores, overall_start_time, last_log_time, steps):
current_time = time.time()
mean_score = np.mean(scores)
score_std = np.std(scores)
min_score = np.min(scores)
max_score = np.max(scores)
elapsed_time = time.time() - overall_start_time
global_steps = self._global_steps_counter.get()
local_steps_per_sec = steps / (current_time - last_log_time)
global_steps_per_sec = global_steps / elapsed_time
global_mil_steps_per_hour = global_steps_per_sec * 3600 / 1000000.0
log(
"TRAIN: {}(GlobalSteps), {} episodes, mean: {}, min: {}, max: {}, "
"\nLocalSpd: {:.0f} STEPS/s GlobalSpd: "
"{} STEPS/s, {:.2f}M STEPS/hour, total elapsed time: {}".format(
global_steps,
len(scores),
green("{:0.3f}±{:0.2f}".format(mean_score, score_std)),
red("{:0.3f}".format(min_score)),
blue("{:0.3f}".format(max_score)),
local_steps_per_sec,
blue("{:.0f}".format(
global_steps_per_sec)),
global_mil_steps_per_hour,
sec_to_str(elapsed_time)
))
def run(self):
# TODO this method is ugly, make it nicer
try:
overall_start_time = time.time()
last_log_time = overall_start_time
local_steps_for_log = 0
while self._epoch <= self._epochs:
steps = self.make_training_step()
local_steps_for_log += steps
global_steps = self._global_steps_counter.inc(steps)
# Logs & tests
if self.local_steps_per_epoch * self._epoch <= self.local_steps:
self._epoch += 1
if self.thread_index == 0:
log("EPOCH {}".format(self._epoch - 1))
self._print_train_log(
self.train_scores, overall_start_time, last_log_time, local_steps_for_log)
run_test_this_epoch = ((self._epoch - 1) % self.test_interval) == 0
if self._run_tests and run_test_this_epoch:
test_scores, test_actions, test_frameskips = self.test(
deterministic=self.deterministic_testing)
if self.write_summaries:
train_summary = self._session.run(self._summaries,
{self.scores_placeholder: self.train_scores,
self.actions_placeholder: self.train_actions,
self.frameskips_placeholder: self.train_frameskips})
self._train_writer.add_summary(train_summary, global_steps)
if self._run_tests and run_test_this_epoch:
test_summary = self._session.run(self._summaries,
{self.scores_placeholder: test_scores,
self.actions_placeholder: test_actions,
self.frameskips_placeholder: test_frameskips})
self._test_writer.add_summary(test_summary, global_steps)
last_log_time = time.time()
local_steps_for_log = 0
log("Learning rate: {}".format(self._session.run(self.learning_rate)))
# Saves model
if self._epoch % self.save_interval == 0:
self.save_model()
now = datetime.datetime.now()
log("Time: {:2d}:{:02d}".format(now.hour, now.minute))
if self.show_heatmaps:
log("Train heatmaps:")
log(self.heatmap(self.train_actions, self.train_frameskips))
log("")
if run_test_this_epoch:
log("Test heatmaps:")
log(self.heatmap(test_actions, test_frameskips))
log("")
self.train_scores = []
self.train_actions = []
self.train_frameskips = []
threadsafe_print("Thread {} finished.".format(self.thread_index))
except (SignalException, ViZDoomUnexpectedExitException):
threadsafe_print(red("Thread #{} aborting(ViZDoom killed).".format(self.thread_index)))
def run_training(self, session):
self._session = session
self.start()
def save_model(self):
ensure_parent_directories(self._model_savefile)
log("Saving model to: {}".format(self._model_savefile))
saver = tf.train.Saver(self.local_network.get_params())
saver.save(self._session, self._model_savefile)
def load_model(self, session, savefile):
saver = tf.train.Saver(self.local_network.get_params())
log("Loading model from: {}".format(savefile))
saver.restore(session, savefile)
log("Loaded model.")
def _get_best_action(self, sess, state, deterministic=True):
policy = self.local_network.get_policy(sess, state)
action_index = self.choose_best_index(policy, deterministic=deterministic)
frameskip = None
return action_index, frameskip
class A3CLearner(Thread):
def __init__(self,
thread_index,
network_type,
global_steps_counter,
scenario_tag=None,
run_id_string=None,
session=None,
tf_logdir=None,
global_network=None,
optimizer=None,
learning_rate=None,
test_only=False,
write_summaries=True,
enable_progress_bar=True,
deterministic_testing=True,
save_interval=1,
writer_max_queue=10,
writer_flush_secs=120,
**settings):
super(A3CLearner, self).__init__()
log("Creating actor-learner #{}.".format(thread_index))
self.thread_index = thread_index
self._global_steps_counter = global_steps_counter
self.write_summaries = write_summaries
self.save_interval = save_interval
self.enable_progress_bar = enable_progress_bar
self._model_savefile = None
self._train_writer = None
self._test_writer = None
self._summaries = None
self._session = session
self.deterministic_testing = deterministic_testing
self.local_steps = 0
# TODO epoch as tf variable?
self._epoch = 1
self.train_scores = []
self.local_steps_per_epoch = settings["local_steps_per_epoch"]
self._run_tests = settings["test_episodes_per_epoch"] > 0 and settings[
"run_tests"]
self.test_episodes_per_epoch = settings["test_episodes_per_epoch"]
self._epochs = np.float32(settings["epochs"])
self.max_remembered_steps = settings["max_remembered_steps"]
self.gamma = np.float32(settings["gamma"])
if self.write_summaries and thread_index == 0 and not test_only:
assert tf_logdir is not None
self.run_id_string = run_id_string
self.tf_models_path = settings["models_path"]
if not os.path.isdir(tf_logdir):
os.makedirs(tf_logdir)
if self.tf_models_path is not None:
if not os.path.isdir(settings["models_path"]):
os.makedirs(settings["models_path"])
self.doom_wrapper = VizdoomWrapper(**settings)
misc_len = self.doom_wrapper.misc_len
img_shape = self.doom_wrapper.img_shape
self.use_misc = self.doom_wrapper.use_misc
self.actions_num = self.doom_wrapper.actions_num
# TODO add debug log
self.local_network = eval(network_type)(actions_num=self.actions_num,
img_shape=img_shape,
misc_len=misc_len,
thread=thread_index,
**settings)
if not test_only:
self.learning_rate = learning_rate
# TODO check gate_gradients != Optimizer.GATE_OP
grads_and_vars = optimizer.compute_gradients(
self.local_network.ops.loss,
var_list=self.local_network.get_params())
grads, local_vars = zip(*grads_and_vars)
grads_and_global_vars = zip(grads, global_network.get_params())
self.train_op = optimizer.apply_gradients(
grads_and_global_vars, global_step=tf.train.get_global_step())
self.global_network = global_network
self.local_network.prepare_sync_op(global_network)
if self.thread_index == 0 and not test_only:
self._model_savefile = settings[
"models_path"] + "/" + self.run_id_string
if self.write_summaries:
self.scores_placeholder, summaries = setup_vector_summaries(
scenario_tag + "/scores")
lr_summary = tf.summary.scalar(scenario_tag + "/learning_rate",
self.learning_rate)
summaries.append(lr_summary)
self._summaries = tf.summary.merge(summaries)
self._train_writer = tf.summary.FileWriter(
"{}/{}/{}".format(tf_logdir, self.run_id_string, "train"),
flush_secs=writer_flush_secs,
max_queue=writer_max_queue)
self._test_writer = tf.summary.FileWriter(
"{}/{}/{}".format(tf_logdir, self.run_id_string, "test"),
flush_secs=writer_flush_secs,
max_queue=writer_max_queue)
@staticmethod
def choose_action_index(policy, deterministic=False):
if deterministic:
return np.argmax(policy)
r = random.random()
cummulative_sum = 0.0
for i, p in enumerate(policy):
cummulative_sum += p
if r <= cummulative_sum:
return i
return len(policy) - 1
def make_training_step(self):
states_img = []
states_misc = []
actions = []
rewards_reversed = []
values_reversed = []
advantages = []
Rs = []
# TODO check how default session works
self._session.run(self.local_network.ops.sync)
initial_network_state = None
if self.local_network.has_state():
initial_network_state = self.local_network.get_current_network_state(
)
terminal = None
steps_performed = 0
for _ in range(self.max_remembered_steps):
steps_performed += 1
current_img, current_misc = self.doom_wrapper.get_current_state()
policy, state_value = self.local_network.get_policy_and_value(
self._session, [current_img, current_misc])
action_index = A3CLearner.choose_action_index(policy)
values_reversed.insert(0, state_value)
states_img.append(current_img)
states_misc.append(current_misc)
actions.append(action_index)
reward = self.doom_wrapper.make_action(action_index)
terminal = self.doom_wrapper.is_terminal()
rewards_reversed.insert(0, reward)
self.local_steps += 1
if terminal:
if self.thread_index == 0:
self.train_scores.append(
self.doom_wrapper.get_total_reward())
self.doom_wrapper.reset()
if self.local_network.has_state():
self.local_network.reset_state()
break
if terminal:
R = 0.0
else:
R = self.local_network.get_value(
self._session, self.doom_wrapper.get_current_state())
for ri, Vi in zip(rewards_reversed, values_reversed):
R = ri + self.gamma * R
advantages.insert(0, R - Vi)
Rs.insert(0, R)
train_op_feed_dict = {
self.local_network.vars.state_img: states_img,
self.local_network.vars.a: actions,
self.local_network.vars.advantage: advantages,
self.local_network.vars.R: Rs
}
if self.use_misc:
train_op_feed_dict[
self.local_network.vars.state_misc] = states_misc
if self.local_network.has_state():
train_op_feed_dict[self.local_network.vars.
initial_network_state] = initial_network_state
train_op_feed_dict[self.local_network.vars.sequence_length] = [
len(actions)
]
self._session.run(self.train_op, feed_dict=train_op_feed_dict)
return steps_performed
def run_episode(self, deterministic=True):
self.doom_wrapper.reset()
if self.local_network.has_state():
self.local_network.reset_state()
while not self.doom_wrapper.is_terminal():
current_state = self.doom_wrapper.get_current_state()
action_index = self._get_best_action(self._session,
current_state,
deterministic=deterministic)
self.doom_wrapper.make_action(action_index)
total_reward = self.doom_wrapper.get_total_reward()
return total_reward
def test(self, episodes_num=None, deterministic=True):
if episodes_num is None:
episodes_num = self.test_episodes_per_epoch
test_start_time = time.time()
test_rewards = []
for _ in trange(episodes_num,
desc="Testing",
file=sys.stdout,
leave=False,
disable=not self.enable_progress_bar):
total_reward = self.run_episode(
deterministic=self.deterministic_testing)
test_rewards.append(total_reward)
self.doom_wrapper.reset()
if self.local_network.has_state():
self.local_network.reset_state()
test_end_time = time.time()
test_duration = test_end_time - test_start_time
min_score = np.min(test_rewards)
max_score = np.max(test_rewards)
mean_score = np.mean(test_rewards)
score_std = np.std(test_rewards)
log("TEST: mean: {}, min: {}, max: {}, test time: {}".format(
green("{:0.3f}±{:0.2f}".format(mean_score, score_std)),
red("{:0.3f}".format(min_score)),
blue("{:0.3f}".format(max_score)), sec_to_str(test_duration)))
return test_rewards
def _print_train_log(self, scores, overall_start_time, last_log_time,
steps):
current_time = time.time()
mean_score = np.mean(scores)
score_std = np.std(scores)
min_score = np.min(scores)
max_score = np.max(scores)
elapsed_time = time.time() - overall_start_time
global_steps = self._global_steps_counter.get()
local_steps_per_sec = steps / (current_time - last_log_time)
global_steps_per_sec = global_steps / elapsed_time
global_mil_steps_per_hour = global_steps_per_sec * 3600 / 1000000.0
log("TRAIN: {}(GlobalSteps), mean: {}, min: {}, max: {}, "
" LocalSpd: {:.0f} STEPS/s GlobalSpd: "
"{} STEPS/s, {:.2f}M STEPS/hour, total elapsed time: {}".format(
global_steps,
green("{:0.3f}±{:0.2f}".format(mean_score, score_std)),
red("{:0.3f}".format(min_score)),
blue("{:0.3f}".format(max_score)), local_steps_per_sec,
blue("{:.0f}".format(global_steps_per_sec)),
global_mil_steps_per_hour, sec_to_str(elapsed_time)))
def run(self):
# TODO this method is ugly, make it nicer
try:
overall_start_time = time.time()
last_log_time = overall_start_time
local_steps_for_log = 0
while self._epoch <= self._epochs:
steps = self.make_training_step()
local_steps_for_log += steps
global_steps = self._global_steps_counter.inc(steps)
# Logs & tests
if self.local_steps_per_epoch * self._epoch <= self.local_steps:
self._epoch += 1
if self.thread_index == 0:
self._print_train_log(self.train_scores,
overall_start_time,
last_log_time,
local_steps_for_log)
if self._run_tests:
test_scores = self.test(
deterministic=self.deterministic_testing)
if self.write_summaries:
train_summary = self._session.run(
self._summaries,
{self.scores_placeholder: self.train_scores})
self._train_writer.add_summary(
train_summary, global_steps)
if self._run_tests:
test_summary = self._session.run(
self._summaries,
{self.scores_placeholder: test_scores})
self._test_writer.add_summary(
test_summary, global_steps)
last_log_time = time.time()
local_steps_for_log = 0
log("Learning rate: {}".format(
self._session.run(self.learning_rate)))
# Saves model
if self._epoch % self.save_interval == 0:
self.save_model()
log("")
self.train_scores = []
except (SignalException, ViZDoomUnexpectedExitException):
threadsafe_print(
red("Thread #{} aborting(ViZDoom killed).".format(
self.thread_index)))
def run_training(self, session):
self._session = session
self.start()
def save_model(self):
savedir = os.path.dirname(self._model_savefile)
if not os.path.exists(savedir):
log("Creating directory: {}".format(savedir))
os.makedirs(savedir)
log("Saving model to: {}".format(self._model_savefile))
saver = tf.train.Saver(self.local_network.get_params())
saver.save(self._session, self._model_savefile)
def load_model(self, session, savefile):
saver = tf.train.Saver(self.local_network.get_params())
log("Loading model from: {}".format(savefile))
saver.restore(session, savefile)
log("Loaded model.")
def _get_best_action(self, sess, state, deterministic=True):
policy = self.local_network.get_policy(sess, state)
action_index = self.choose_action_index(policy,
deterministic=deterministic)
return action_index