gryf/coach
1
0
Fork 0
mirror of https://github.com/gryf/coach.git synced 2025-12-18 19:50:17 +01:00
Code Issues Projects Releases Wiki Activity

remove return values from GraphManager.act

Browse Source
This commit is contained in:
Zach Dwiel
2018-10-02 12:44:34 -04:00
committed by zach dwiel
parent b02f269464
commit 01f3a0594b
1 changed files with 3 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
rl_coach/graph_managers/graph_manager.py
View File

@@ -360,8 +360,6 @@ class GraphManager(object):
:param return_on_game_over: finish acting if an episode is finished :param return_on_game_over: finish acting if an episode is finished
:param continue_until_game_over: continue playing until an episode was completed :param continue_until_game_over: continue playing until an episode was completed
:param keep_networks_in_sync: sync the network parameters with the global network before each episode :param keep_networks_in_sync: sync the network parameters with the global network before each episode
:return: the actual number of steps done, a boolean value that represent if the episode was done when finishing
the function call
""" """
self.verify_graph_was_created() self.verify_graph_was_created()
@@ -408,13 +406,7 @@ class GraphManager(object):
if keep_networks_in_sync: if keep_networks_in_sync:
self.sync_graph() self.sync_graph()
if return_on_game_over: if return_on_game_over:
return self.total_steps_counters[self.phase][EnvironmentSteps] - initial_count, True return
# return the game over status
if result:
return self.total_steps_counters[self.phase][EnvironmentSteps] - initial_count, result.game_over
else:
return self.total_steps_counters[self.phase][EnvironmentSteps] - initial_count, False
def train_and_act(self, steps: StepMethod) -> None: def train_and_act(self, steps: StepMethod) -> None:
""" """
@@ -466,7 +458,7 @@ class GraphManager(object):
# act for at least `steps`, though don't interrupt an episode # act for at least `steps`, though don't interrupt an episode
count_end = self.total_steps_counters[self.phase][steps.__class__] + steps.num_steps count_end = self.total_steps_counters[self.phase][steps.__class__] + steps.num_steps
while self.total_steps_counters[self.phase][steps.__class__] < count_end: while self.total_steps_counters[self.phase][steps.__class__] < count_end:
steps_done, _ = self.act(steps, continue_until_game_over=True, return_on_game_over=True, self.act(steps, continue_until_game_over=True, return_on_game_over=True,
keep_networks_in_sync=keep_networks_in_sync) keep_networks_in_sync=keep_networks_in_sync)
self.phase = RunPhase.UNDEFINED self.phase = RunPhase.UNDEFINED