Question Halt training process issue when enable / disable ml agents.

I have a problem using ml agent.

Issue is that disabling and enabling agents on Unity causes error.

When that happens training stops with below errors. I need to switch the agent from disabled to enabled. for example when the agent dies from receiving attacks then respawns after a while, and between that death and respawn time the agent becomes disabled. Meaning the gameobject of the agent becomes SetActive(false). Mlagents-learn halts with below error message

Traceback (most recent call last):
File "C:\Program Files\Python37\lib\runpy.py", line 193, in _run_module_as_main
"__main__", mod_spec)
File "C:\Program Files\Python37\lib\runpy.py", line 85, in _run_code
exec(code, run_globals)
File "D:\UnityProjects\zGame\python-env\release6\Scripts\mlagents-learn.exe\__main__.py", line 7, in <module>
File "d:\unityprojects\zgame\python-env\release6\lib\site-packages\mlagents\trainers\learn.py", line 276, in main
run_cli(parse_command_line())
File "d:\unityprojects\zgame\python-env\release6\lib\site-packages\mlagents\trainers\learn.py", line 272, in run_cli
run_training(run_seed, options)
File "d:\unityprojects\zgame\python-env\release6\lib\site-packages\mlagents\trainers\learn.py", line 149, in run_training
tc.start_learning(env_manager)
File "d:\unityprojects\zgame\python-env\release6\lib\site-packages\mlagents_envs\timers.py", line 305, in wrapped
return func(*args, **kwargs)
File "d:\unityprojects\zgame\python-env\release6\lib\site-packages\mlagents\trainers\trainer_controller.py", line 181, in start_learning
self.reset_env_if_ready(env_manager)
File "d:\unityprojects\zgame\python-env\release6\lib\site-packages\mlagents\trainers\trainer_controller.py", line 229, in reset_env_if_ready
self.end_trainer_episodes()
File "d:\unityprojects\zgame\python-env\release6\lib\site-packages\mlagents\trainers\trainer_controller.py", line 210, in end_trainer_episodes
trainer.end_episode()
File "d:\unityprojects\zgame\python-env\release6\lib\site-packages\mlagents\trainers\ghost\trainer.py", line 298, in end_episode
self.trainer.end_episode()
File "d:\unityprojects\zgame\python-env\release6\lib\site-packages\mlagents\trainers\trainer\rl_trainer.py", line 56, in end_episode
for agent_id in rewards:
RuntimeError: dictionary changed size during iteration


Initially I called agent's endEpisode then deactivated the gameobject of the agent. But with the error I tried not calling endEpisode, but both of the cases had the same error but it wasn't consistent at all. Usually occurred between 10M steps to 20M steps. Again, is this a bug and will there be a fix to this?



Thanks!

 

Sure. I don't get error when does not calling disable/enable agent. ( don't call gameobject SetActive(flase) ).
that error occurs randomly.
Same error occured create and destroy agent.