search

openai / random-network-distillation

Code for the paper "Exploration by Random Network Distillation"
https://openai.com/blog/reinforcement-learning-with-prediction-based-rewards/
881 stars 159 forks source link

RAM usage problem #14

Closed weak-orange-wind closed 5 years ago

weak-orange-wind commented 5 years ago

I ran python run_atari.py --ext_coeff 0.0 --int_coeff 1.0 --gamma_ext 0.999 --num_env 32 on a GTX 1080TI machine. In the beginning, it works well. But the memory(not GPU memory) usage keep improving as the program running. Finally it ran out of my 128G RAM and 32GB swap space and crashed. Any suggestions about this?

weak-orange-wind commented 5 years ago

This problem has been solved by using numpy==1.13

azouaoui-cv commented 5 years ago

Hey, did you get any good results/replays using the internal reward only?

weak-orange-wind commented 5 years ago

Hey, did you get any good results/replays using the internal reward only?

Not yet

jzstudent commented 3 years ago

This problem has been solved by using numpy==1.13

hey，I also meet this problem，my swap crashed after the program begining about 6 hours，I found using numpy==1.13 did not work. Do u have other suggestion?