ishikota / kyoka / 20

Committed 24 Oct 2016 - 5:58 coverage: 92.529% (+2.5%) from 90.06%

Build # 20

Build Type

Pull #16

travis-ci

Committed by

web-flow

Commit Message

Update sample script to use new save logic

Pull Request Pull Request #16: Change save logic

Run Details

116 of 122 new or added lines in 8 files covered. (95.08%)

1 existing line in 1 file now uncovered.

644 of 696 relevant lines covered (92.53%)

0.93 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

96.0

/kyoka/algorithm/base_rl_algorithm.py

from kyoka.finish_rule.watch_iteration_count import WatchIterationCount

class BaseRLAlgorithm(object):

  def setUp(self, domain, policy, value_function):
    self.domain = domain
    self.value_function = value_function
    self.value_function.setUp()
    self.policy = policy

  def save(self, save_dir_path):
    self.value_function.save(save_dir_path)
    self.save_algorithm_state(save_dir_path)

  def load(self, load_dir_path):
    self.value_function.load(load_dir_path)
    self.load_algorithm_state(load_dir_path)

  def save_algorithm_state(self, save_dir_path):
    pass

  def load_algorithm_state(self, load_dir_path):
    pass

  def update_value_function(self, domain, policy, value_function):
    err_msg = self.__build_err_msg("update_value_function")
    raise NotImplementedError(err_msg)

  def run_gpi(self, nb_iteration, finish_rules=[], callbacks=[]):
    callbacks = self.__wrap_item_if_single(callbacks)
    finish_rules = self.__wrap_item_if_single(finish_rules)
    finish_rules.append(WatchIterationCount(nb_iteration))
    iteration_counter = 0
    [callback.before_gpi_start(self.domain, self.value_function) for callback in callbacks]
    while True:
      [callback.before_update(iteration_counter, self.domain, self.value_function) for callback in callbacks]
      self.update_value_function(self.domain, self.policy, self.value_function)
      [callback.after_update(iteration_counter, self.domain, self.value_function) for callback in callbacks]
      iteration_counter += 1
      for finish_rule in finish_rules:
        if finish_rule.satisfy_condition(iteration_counter):
          finish_msg = finish_rule.generate_finish_message(iteration_counter)
          [callback.after_gpi_finish(self.domain, self.value_function) for callback in callbacks]
          return finish_msg

  def generate_episode(self, domain, value_function, policy):
    state = domain.generate_initial_state()
    episode = []
    while not domain.is_terminal_state(state):
      action = policy.choose_action(domain, value_function, state)
      next_state = domain.transit_state(state, action)
      reward = domain.calculate_reward(next_state)
      episode.append((state, action, next_state, reward))
      state = next_state
    return episode


  def __wrap_item_if_single(self, item):
    return [item] if not isinstance(item, list) else item

  def __build_err_msg(self, msg):
    return "Accessed [ {0} ] method of BaseRLAlgorithm which should be overridden".format(msg)


1	from kyoka.finish_rule.watch_iteration_count import WatchIterationCount	1×
2
3	class BaseRLAlgorithm(object):	1×
4
5	def setUp(self, domain, policy, value_function):	1×
6	self.domain = domain	1×
7	self.value_function = value_function	1×
8	self.value_function.setUp()	1×
9	self.policy = policy	1×
10
11	def save(self, save_dir_path):	1×
12	self.value_function.save(save_dir_path)	1×
13	self.save_algorithm_state(save_dir_path)	1×
14
15	def load(self, load_dir_path):	1×
16	self.value_function.load(load_dir_path)	1×
17	self.load_algorithm_state(load_dir_path)	1×
18
19	def save_algorithm_state(self, save_dir_path):	1×
NEW 20	pass	!
21
22	def load_algorithm_state(self, load_dir_path):	1×
NEW 23	pass	!
24
25	def update_value_function(self, domain, policy, value_function):	1×
26	err_msg = self.__build_err_msg("update_value_function")	1×
27	raise NotImplementedError(err_msg)	1×
28
29	def run_gpi(self, nb_iteration, finish_rules=[], callbacks=[]):	1×
30	callbacks = self.__wrap_item_if_single(callbacks)	1×
31	finish_rules = self.__wrap_item_if_single(finish_rules)	1×
32	finish_rules.append(WatchIterationCount(nb_iteration))	1×
33	iteration_counter = 0	1×
34	[callback.before_gpi_start(self.domain, self.value_function) for callback in callbacks]	1×
35	while True:	1×
36	[callback.before_update(iteration_counter, self.domain, self.value_function) for callback in callbacks]	1×
37	self.update_value_function(self.domain, self.policy, self.value_function)	1×
38	[callback.after_update(iteration_counter, self.domain, self.value_function) for callback in callbacks]	1×
39	iteration_counter += 1	1×
40	for finish_rule in finish_rules:	1×
41	if finish_rule.satisfy_condition(iteration_counter):	1×
42	finish_msg = finish_rule.generate_finish_message(iteration_counter)	1×
43	[callback.after_gpi_finish(self.domain, self.value_function) for callback in callbacks]	1×
44	return finish_msg	1×
45
46	def generate_episode(self, domain, value_function, policy):	1×
47	state = domain.generate_initial_state()	1×
48	episode = []	1×
49	while not domain.is_terminal_state(state):	1×
50	action = policy.choose_action(domain, value_function, state)	1×
51	next_state = domain.transit_state(state, action)	1×
52	reward = domain.calculate_reward(next_state)	1×
53	episode.append((state, action, next_state, reward))	1×
54	state = next_state	1×
55	return episode	1×
56
57
58	def __wrap_item_if_single(self, item):	1×
59	return [item] if not isinstance(item, list) else item	1×
60
61	def __build_err_msg(self, msg):	1×
62	return "Accessed [ {0} ] method of BaseRLAlgorithm which should be overridden".format(msg)	1×
63

ishikota / kyoka / 20

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous