Why my RL training abruptly stopping before the total EpisodeCount?

I have made a custom RL environment based on the answer provided in (full description of the env described there along with the commented attached code)--

https://in.mathworks.com/matlabcentral/answers/619138-custom-rl-environment-creation?s_tid=srchtitle

But its training is stopping at 500th episode before the total episode count (5000).

I am sharing both the codes (env and main training). Don't bother about hyperparameters.

I am unable to understand what I am doing wrong.

I have implemented another more complex environment based on this logic that stops at the 500th episode even though I am fixing training for 5000.

Any suggestion?

Thanks.

Answer 1

[Emmanouil Tzorakoleftherakis]

Please take a look at this doc page. While you are selecting "episodecount" as the termination criterion, you don't set the stopping value. The default is "500" which explains why training stops after 500 episodes.

The "MaxEpisodes" field is checked after the "StopTrainingCriteria" one.

Hope that helps.

[laha_M]

Thanks.