def __init__(self, args):
super(PseudoCountA3CLearner, self).__init__(args)
#more cython tuning could useful here
self.density_model = CTSDensityModel(height=args.cts_rescale_dim,
width=args.cts_rescale_dim,
num_bins=args.cts_bins,
beta=0.05)
def __init__(self, args):
super(PseudoCountQLearner, self).__init__(args)
self.cts_eta = .9
self.batch_size = 32
self.replay_memory = ReplayMemory(args.replay_size)
#more cython tuning could useful here
self.density_model = CTSDensityModel(height=args.cts_rescale_dim,
width=args.cts_rescale_dim,
num_bins=args.cts_bins,
beta=0.05)
def _init_density_model(self, args):
model_args = {
'height': args.cts_rescale_dim,
'width': args.cts_rescale_dim,
'num_bins': args.cts_bins,
'beta': args.cts_beta
}
if args.density_model == 'cts':
self.density_model = CTSDensityModel(**model_args)
else:
self.density_model = PerPixelDensityModel(**model_args)
def _init_density_model(self, args):
self.density_model_update_steps = 20 * args.q_target_update_steps
self.density_model_update_flags = args.density_model_update_flags
model_args = {
'height': args.cts_rescale_dim,
'width': args.cts_rescale_dim,
'num_bins': args.cts_bins,
'beta': args.cts_beta
}
if args.density_model == 'cts':
self.density_model = CTSDensityModel(**model_args)
else:
self.density_model = PerPixelDensityModel(**model_args)
def _init_density_model(self, args):
self.density_model_update_steps = 20 * args.q_target_update_steps
self.alg_type = args.alg_type
#self.density_model_update_flags = args.density_model_update_flags
self.density_model_update_flags = []
for x in range(0, args.num_actions):
self.density_model_update_flags.append(
args.density_model_update_flags)
#print("x is: {}".format(x))
model_args = {
'height': args.cts_rescale_dim,
'width': args.cts_rescale_dim,
'num_bins': args.cts_bins, #TODO check what this is
'beta': args.cts_beta,
}
self.density_model = []
for x in range(0, args.num_actions):
if args.density_model == 'cts':
self.density_model.append(CTSDensityModel(**model_args))
else:
self.density_model.append(PerPixelDensityModel(**model_args))
class PseudoCountA3CLearner(A3CLearner):
def __init__(self, args):
super(PseudoCountA3CLearner, self).__init__(args)
#more cython tuning could useful here
self.density_model = CTSDensityModel(height=args.cts_rescale_dim,
width=args.cts_rescale_dim,
num_bins=args.cts_bins,
beta=0.05)
def _run(self):
if not self.is_train:
return self.test()
""" Main actor learner loop for advantage actor critic learning. """
logger.debug("Actor {} resuming at Step {}".format(
self.actor_id, self.global_step.value()))
s = self.emulator.get_initial_state()
total_episode_reward = 0.0
mean_entropy = 0.0
episode_start_step = 0
while (self.global_step.value() < self.max_global_steps):
# Sync local learning net with shared mem
self.sync_net_with_shared_memory(self.local_network,
self.learning_vars)
self.save_vars()
local_step_start = self.local_step
reset_game = False
episode_over = False
bonuses = deque(maxlen=100)
rewards = list()
states = list()
actions = list()
values = list()
s_batch = list()
a_batch = list()
y_batch = list()
adv_batch = list()
while not (episode_over or (self.local_step - local_step_start
== self.max_local_steps)):
# Choose next action and execute it
a, readout_v_t, readout_pi_t = self.choose_next_action(s)
new_s, reward, episode_over = self.emulator.next(a)
total_episode_reward += reward
current_frame = new_s[..., -1]
bonus = self.density_model.update(current_frame)
bonuses.append(bonus)
if self.is_master() and (self.local_step % 200 == 0):
bonus_array = np.array(bonuses)
logger.debug(
'π_a={:.4f} / V={:.4f} / Mean Bonus={:.4f} / Max Bonus={:.4f}'
.format(readout_pi_t[a.argmax()], readout_v_t,
bonus_array.mean(), bonus_array.max()))
# Rescale or clip immediate reward
reward = self.rescale_reward(
self.rescale_reward(reward) + bonus)
rewards.append(reward)
states.append(s)
actions.append(a)
values.append(readout_v_t)
s = new_s
self.local_step += 1
self.global_step.increment()
# Calculate the value offered by critic in the new state.
if episode_over:
R = 0
else:
R = self.session.run(
self.local_network.output_layer_v,
feed_dict={self.local_network.input_ph: [new_s]})[0][0]
sel_actions = []
for i in reversed(xrange(len(states))):
R = rewards[i] + self.gamma * R
y_batch.append(R)
a_batch.append(actions[i])
s_batch.append(states[i])
adv_batch.append(R - values[i])
sel_actions.append(np.argmax(actions[i]))
# Compute gradients on the local policy/V network and apply them to shared memory
feed_dict = {
self.local_network.input_ph: s_batch,
self.local_network.critic_target_ph: y_batch,
self.local_network.selected_action_ph: a_batch,
self.local_network.adv_actor_ph: adv_batch,
}
grads, entropy = self.session.run(
[self.local_network.get_gradients, self.local_network.entropy],
feed_dict=feed_dict)
self.apply_gradients_to_shared_memory_vars(grads)
delta_old = local_step_start - episode_start_step
delta_new = self.local_step - local_step_start
mean_entropy = (mean_entropy * delta_old +
entropy * delta_new) / (delta_old + delta_new)
s, mean_entropy, episode_start_step, total_episode_reward, _ = self.prepare_state(
s, mean_entropy, episode_start_step, total_episode_reward,
self.local_step, sel_actions, episode_over)
class PseudoCountQLearner(ValueBasedLearner):
def __init__(self, args):
super(PseudoCountQLearner, self).__init__(args)
self.cts_eta = .9
self.batch_size = 32
self.replay_memory = ReplayMemory(args.replay_size)
#more cython tuning could useful here
self.density_model = CTSDensityModel(height=args.cts_rescale_dim,
width=args.cts_rescale_dim,
num_bins=args.cts_bins,
beta=0.05)
def generate_final_epsilon(self):
return 0.1
def _get_summary_vars(self):
q_vars = super(PseudoCountQLearner, self)._get_summary_vars()
bonus_q25 = tf.Variable(0., name='novelty_bonus_q25')
s1 = tf.summary.scalar('Novelty_Bonus_q25_{}'.format(self.actor_id),
bonus_q25)
bonus_q50 = tf.Variable(0., name='novelty_bonus_q50')
s2 = tf.summary.scalar('Novelty_Bonus_q50_{}'.format(self.actor_id),
bonus_q50)
bonus_q75 = tf.Variable(0., name='novelty_bonus_q75')
s3 = tf.summary.scalar('Novelty_Bonus_q75_{}'.format(self.actor_id),
bonus_q75)
return q_vars + [bonus_q25, bonus_q50, bonus_q75]
def prepare_state(self, state, total_episode_reward, steps_at_last_reward,
ep_t, episode_ave_max_q, episode_over, bonuses):
# prevent the agent from getting stuck
reset_game = False
if (self.local_step - steps_at_last_reward > 5000
or (self.emulator.get_lives() == 0
and self.emulator.game not in ONE_LIFE_GAMES)):
steps_at_last_reward = self.local_step
episode_over = True
reset_game = True
# Start a new game on reaching terminal state
if episode_over:
T = self.global_step.value()
t = self.local_step
e_prog = float(t) / self.epsilon_annealing_steps
episode_ave_max_q = episode_ave_max_q / float(ep_t)
s1 = "Q_MAX {0:.4f}".format(episode_ave_max_q)
s2 = "EPS {0:.4f}".format(self.epsilon)
self.scores.insert(0, total_episode_reward)
if len(self.scores) > 100:
self.scores.pop()
logger.info('T{0} / STEP {1} / REWARD {2} / {3} / {4}'.format(
self.actor_id, T, total_episode_reward, s1, s2))
logger.info(
'ID: {0} -- RUNNING AVG: {1:.0f} ± {2:.0f} -- BEST: {3:.0f}'.
format(
self.actor_id,
np.array(self.scores).mean(),
2 * np.array(self.scores).std(),
max(self.scores),
))
if self.is_master() and self.is_train:
stats = [
total_episode_reward,
episode_ave_max_q,
self.epsilon,
np.percentile(bonuses, 25),
np.percentile(bonuses, 50),
np.percentile(bonuses, 75),
]
feed_dict = {
self.summary_ph[i]: stats[i]
for i in range(len(stats))
}
res = self.session.run(self.update_ops + [self.summary_op],
feed_dict=feed_dict)
self.summary_writer.add_summary(res[-1],
self.global_step.value())
if reset_game or self.emulator.game in ONE_LIFE_GAMES:
state = self.emulator.get_initial_state()
ep_t = 0
total_episode_reward = 0
episode_ave_max_q = 0
episode_over = False
return state, total_episode_reward, steps_at_last_reward, ep_t, episode_ave_max_q, episode_over
def batch_update(self):
if len(self.replay_memory) < self.batch_size:
return
s_i, a_i, r_i, s_f, is_terminal = self.replay_memory.sample_batch(
self.batch_size)
q_target_values = self.session.run(
self.target_network.output_layer,
feed_dict={self.target_network.input_ph: s_f})
y_target = r_i + self.cts_eta * self.gamma * q_target_values.max(
axis=1) * (1 - is_terminal.astype(np.int))
feed_dict = {
self.local_network.input_ph: s_i,
self.local_network.target_ph: y_target,
self.local_network.selected_action_ph: a_i
}
grads = self.session.run(self.local_network.get_gradients,
feed_dict=feed_dict)
self.apply_gradients_to_shared_memory_vars(grads)
def _run(self):
""" Main actor learner loop for n-step Q learning. """
if not self.is_train:
return self.test()
logger.debug("Actor {} resuming at Step {}, {}".format(
self.actor_id, self.global_step.value(), time.ctime()))
s = self.emulator.get_initial_state()
s_batch = []
a_batch = []
y_batch = []
bonuses = deque(maxlen=100)
exec_update_target = False
total_episode_reward = 0
episode_ave_max_q = 0
episode_over = False
qmax_down = 0
qmax_up = 0
prev_qmax = -10 * 6
low_qmax = 0
ep_t = 0
t0 = time.time()
while (self.global_step.value() < self.max_global_steps):
# Sync local learning net with shared mem
self.sync_net_with_shared_memory(self.local_network,
self.learning_vars)
self.save_vars()
rewards = []
states = []
actions = []
local_step_start = self.local_step
while not episode_over:
# Choose next action and execute it
a, readout_t = self.choose_next_action(s)
new_s, reward, episode_over = self.emulator.next(a)
total_episode_reward += reward
current_frame = new_s[..., -1]
bonus = self.density_model.update(current_frame)
bonuses.append(bonus)
if self.is_master() and (self.local_step % 200 == 0):
bonus_array = np.array(bonuses)
logger.debug(
'Mean Bonus={:.4f} / Max Bonus={:.4f} / STEPS/s={}'.
format(bonus_array.mean(), bonus_array.max(),
100. / (time.time() - t0)))
t0 = time.time()
# Rescale or clip immediate reward
reward = self.rescale_reward(
self.rescale_reward(reward) + bonus)
ep_t += 1
rewards.append(reward)
states.append(s)
actions.append(a)
s = new_s
self.local_step += 1
episode_ave_max_q += np.max(readout_t)
global_step, update_target = self.global_step.increment(
self.q_target_update_steps)
if update_target:
update_target = False
exec_update_target = True
if self.local_step % 4 == 0:
self.batch_update()
self.local_network.global_step = global_step
else:
mc_returns = list()
running_total = 0.0
for r in reversed(rewards):
running_total = r + self.gamma * running_total
mc_returns.insert(0, running_total)
mixed_returns = self.cts_eta * np.array(rewards) + (
1 - self.cts_eta) * np.array(mc_returns)
states.append(new_s)
episode_length = len(rewards)
for i in range(episode_length):
self.replay_memory.append(
(states[i], actions[i], mixed_returns[i],
states[i + 1], i + 1 == episode_length))
if exec_update_target:
self.update_target()
exec_update_target = False
# Sync local tensorflow target network params with shared target network params
if self.target_update_flags.updated[self.actor_id] == 1:
self.sync_net_with_shared_memory(self.target_network,
self.target_vars)
self.target_update_flags.updated[self.actor_id] = 0
s, total_episode_reward, _, ep_t, episode_ave_max_q, episode_over = \
self.prepare_state(s, total_episode_reward, self.local_step, ep_t, episode_ave_max_q, episode_over, bonuses)
class PseudoCountQLearner(ValueBasedLearner):
def __init__(self, args):
super(PseudoCountQLearner, self).__init__(args)
self.cts_eta = .9
self.batch_size = 32
self.replay_memory = ReplayMemory(args.replay_size)
#more cython tuning could useful here
self.density_model = CTSDensityModel(height=args.cts_rescale_dim,
width=args.cts_rescale_dim,
num_bins=args.cts_bins,
beta=0.05)
def generate_final_epsilon(self):
return 0.1
def batch_update(self):
if len(self.replay_memory) < self.batch_size:
return
s_i, a_i, r_i, s_f, is_terminal = self.replay_memory.sample_batch(
self.batch_size)
q_target_values = self.session.run(
self.target_network.output_layer,
feed_dict={self.target_network.input_ph: s_f})
y_target = r_i + self.cts_eta * self.gamma * q_target_values.max(
axis=1) * (1 - is_terminal.astype(np.int))
feed_dict = {
self.local_network.input_ph: s_i,
self.local_network.target_ph: y_target,
self.local_network.selected_action_ph: a_i
}
grads = self.session.run(self.local_network.get_gradients,
feed_dict=feed_dict)
self.apply_gradients_to_shared_memory_vars(grads)
def _run(self):
""" Main actor learner loop for n-step Q learning. """
if not self.is_train:
return self.test()
logger.debug("Actor {} resuming at Step {}, {}".format(
self.actor_id, self.global_step.value(), time.ctime()))
s = self.emulator.get_initial_state()
s_batch = []
a_batch = []
y_batch = []
bonuses = deque(maxlen=100)
exec_update_target = False
total_episode_reward = 0
episode_ave_max_q = 0
episode_over = False
qmax_down = 0
qmax_up = 0
prev_qmax = -10 * 6
low_qmax = 0
ep_t = 0
while (self.global_step.value() < self.max_global_steps):
# Sync local learning net with shared mem
self.sync_net_with_shared_memory(self.local_network,
self.learning_vars)
self.save_vars()
rewards = []
states = []
actions = []
local_step_start = self.local_step
while not episode_over:
# Choose next action and execute it
a, readout_t = self.choose_next_action(s)
new_s, reward, episode_over = self.emulator.next(a)
total_episode_reward += reward
current_frame = new_s[..., -1]
bonus = self.density_model.update(current_frame)
bonuses.append(bonus)
if self.is_master() and (self.local_step % 200 == 0):
bonus_array = np.array(bonuses)
logger.debug('Mean Bonus={:.4f} / Max Bonus={:.4f}'.format(
bonus_array.mean(), bonus_array.max()))
# Rescale or clip immediate reward
# reward = self.rescale_reward(self.rescale_reward(reward) + bonus)
reward = self.rescale_reward(reward)
ep_t += 1
rewards.append(reward)
states.append(s)
actions.append(a)
s = new_s
self.local_step += 1
episode_ave_max_q += np.max(readout_t)
global_step, update_target = self.global_step.increment(
self.q_target_update_steps)
if update_target:
update_target = False
exec_update_target = True
if self.local_step % 4 == 0:
self.batch_update()
self.local_network.global_step = global_step
else:
mc_returns = list()
running_total = 0.0
for r in reversed(rewards):
running_total = r + self.gamma * running_total
mc_returns.insert(0, running_total)
mixed_returns = self.cts_eta * np.array(rewards) + (
1 - self.cts_eta) * np.array(mc_returns)
states.append(new_s)
episode_length = len(rewards)
for i in range(episode_length):
self.replay_memory.append(
(states[i], actions[i], mixed_returns[i],
states[i + 1], i + 1 == episode_length))
if exec_update_target:
self.update_target()
exec_update_target = False
# Sync local tensorflow target network params with shared target network params
if self.target_update_flags.updated[self.actor_id] == 1:
self.sync_net_with_shared_memory(self.target_network,
self.target_vars)
self.target_update_flags.updated[self.actor_id] = 0
s, total_episode_reward, _, ep_t, episode_ave_max_q, episode_over = \
self.prepare_state(s, total_episode_reward, self.local_step, ep_t, episode_ave_max_q, episode_over)