gryf/coach
1
0
Fork 0
mirror of https://github.com/gryf/coach.git synced 2025-12-18 03:30:19 +01:00
Code Issues Projects Releases Wiki Activity

Itaicaspi/episode reset refactoring (#105)

Browse Source 
* reordering of the episode reset operation and allowing to store episodes only when they are terminated

* reordering of the episode reset operation and allowing to store episodes only when they are terminated

* revert tensorflow-gpu to 1.9.0 + bug fix in should_train()

* tests readme file and refactoring of policy optimization agent train function

* Update README.md

* Update README.md

* additional policy optimization train function simplifications

* Updated the traces after the reordering of the environment reset

* docker and jenkins files

* updated the traces to the ones from within the docker container

* updated traces and added control suite to the docker

* updated jenkins file with the intel proxy + updated doom basic a3c test params

* updated line breaks in jenkins file

* added a missing line break in jenkins file

* refining trace tests ignored presets + adding a configurable beta entropy value

* switch the order of trace and golden tests in jenkins + fix golden tests processes not killed issue

* updated benchmarks for dueling ddqn breakout and pong

* allowing dynamic updates to the loss weights + bug fix in episode.update_returns

* remove docker and jenkins file
This commit is contained in:
Itai Caspi
2018-09-04 15:07:54 +03:00
committed by GitHub
parent 7086492127
commit 72a1d9d426
92 changed files with 9803 additions and 9740 deletions
Download Patch File Download Diff File Expand all files Collapse all files

17
rl_coach/agents/agent.py
View File

@@ -431,8 +431,13 @@ class Agent(AgentInterface):
if self.phase != RunPhase.TEST or self.ap.task_parameters.evaluate_only:
self.current_episode += 1
if self.phase != RunPhase.TEST and isinstance(self.memory, EpisodicExperienceReplay):
self.call_memory('store_episode', self.current_episode_buffer)
if self.phase != RunPhase.TEST:
if isinstance(self.memory, EpisodicExperienceReplay):
self.call_memory('store_episode', self.current_episode_buffer)
elif self.ap.algorithm.store_transitions_only_when_episodes_are_terminated:
self.current_episode_buffer.update_returns()
for transition in self.current_episode_buffer.transitions:
self.call_memory('store', transition)
if self.phase == RunPhase.TEST:
self.accumulated_rewards_across_evaluation_episodes += self.total_reward_in_current_episode
@@ -517,7 +522,7 @@ class Agent(AgentInterface):
elif step_method.__class__ == EnvironmentSteps:
should_update = (self.total_steps_counter - self.last_training_phase_step) >= step_method.num_steps
if wait_for_full_episode:
should_update = should_update and self.current_episode_steps_counter == 0
should_update = should_update and self.current_episode_buffer.is_complete
if should_update:
self.last_training_phase_step = self.total_steps_counter
else:
@@ -728,9 +733,9 @@ class Agent(AgentInterface):
if self.phase in [RunPhase.TRAIN, RunPhase.HEATUP]:
# for episodic memories we keep the transitions in a local buffer until the episode is ended.
# for regular memories we insert the transitions directly to the memory
if isinstance(self.memory, EpisodicExperienceReplay):
self.current_episode_buffer.insert(transition)
else:
self.current_episode_buffer.insert(transition)
if not isinstance(self.memory, EpisodicExperienceReplay) \
and not self.ap.algorithm.store_transitions_only_when_episodes_are_terminated:
self.call_memory('store', transition)
if self.ap.visualization.dump_in_episode_signals: