When I build a custom gymnasium
(formerly gym) environment, and I see in the memory profile the first line, I get some headache.
Where does this huge memory footprint come from? For reference, when I create gymnasium environments like "Taxi-v3", I have only ~35 MiB ...
My general questions for this thread:
- How is this first line measured?
- Which influencing factors are taken to determine the outlined 82 MiB fdor the line @profile?
Console Output:
Line # Mem usage Increment Occurrences Line Contents
=============================================================
9 82.1 MiB 82.1 MiB 1 @profile
10 82.1 MiB 0 MiB 1 def complete_walkthrough():
11
12 87.5 MiB 5.4 MiB 1 gym_env = gymnasium.make("MyEnv-v0")
13 88.4 MiB 0.9 MiB 1 initial_state, _ = gym_env.reset()
source https://stackoverflow.com/questions/76058725/openai-gymnasium-huge-memory-usage-even-before-gym-make
Comments
Post a Comment