Error with parallelized RL training with PPO
4 views (last 30 days)
Show older comments
Hello,
At the end of my parallelized RL training, i am getting the following warning, which is then causing one of the parallel pool's workers to automatically shut down. Is there a known cause for this? It says it is an error inside of the "onCleanup" class, but i am not using any custom functions for this (it is the one already implemented in MATLAB).
It seems to be happening only for one worker for some reason, and no crash dump is being generated.
Warning: The following error was caught while executing 'onCleanup' class
destructor:
Error using rl.train.parallel.AbstractParallelTrainer/cleanupPostRun_
Invalid input for argument 2 (rhs2):
Value must be a scalar.
Error in rl.train.parallel.AbstractParallelTrainer>@()cleanupPostRun_(this)
(line 21)
cln1 = onCleanup(@() cleanupPostRun_(this));
Error in onCleanup/delete (line 25)
obj.task();
Error in rl.train.parallel.AbstractParallelTrainer/run (line 141)
end
Error in rl.train.TrainingManager/train (line 479)
run(trainer);
Error in rl.train.TrainingManager/run (line 233)
train(this);
Error in rl.agent.AbstractAgent/train (line 136)
trainingResult = run(trainMgr,checkpoint);
1 Comment
Drew Davis
on 30 Jun 2023
Hi Matteo
Do you have reproduction steps you could share, specifically your MATLAB environment class?
Drew
Answers (0)
See Also
Categories
Find more on Sequence and Numeric Feature Data Workflows in Help Center and File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!