stanfordnmbl / osim-rl

Reinforcement learning environments with musculoskeletal models
http://osim-rl.stanford.edu/
MIT License
877 stars 248 forks source link

different behaviors for the same network on different machine #213

Open peppepel95 opened 4 years ago

peppepel95 commented 4 years ago

Hello everyone, I use opensim through osim-rl, in particular I'm training a network to make the agent walk. The problem I found is that for 2 or more machines with the same anaconda environment (python 3.6.1) with the same packages and executing the same identical code (setting the seed of both the python and numpy random) I get different results. The only difference is about the GLIBC version, in particular 2.17 and 2.29. So the problem is about the GLIBC version or there is something that I miss?