search

decisionforce / CoPO

[NeurIPS 2021] Official implementation of paper "Learning to Simulate Self-driven Particles System with Coordinated Policy Optimization".
Apache License 2.0
117 stars 21 forks source link

Issues of installation caused by gym's vision. #47

Open SecretaryofSFA opened 6 months ago

SecretaryofSFA commented 6 months ago

I have installed the gym==0.19.0 and there is a error about the conflict between the copo==0.0.0 and ray[rllib]==2.2.0. The error just is like this: " The conflict is caused by: copo 0.0.0 depends on gym==0.19.0 ray[rllib] 2.2.0 depends on gym<0.24.0 and >=0.21.0; extra == "rllib" " issue

I have tried to uninstall the ray==2.2.0 but there were still this error... How can I solve this problem?

SecretaryofSFA commented 6 months ago

or could I change the code in setup.py? I'm afraid that it would cause more confusing bugs..

SecretaryofSFA commented 6 months ago

I changed the 'gym==0.19.0' to 'gym>=0.19.0' in setup.py then copo was installed successfully.

But I met another problem which is ''pickle5.pickle.PicklingError: Could not pickle object as excessively deep recursion required.'' when I ran the ''python./copo_code/copo/torch_copo/train_copo.py'' Could not pickle object

By the way, I just could install the metadrive-simulator==0.2.5.1 because I didn't find the 0.2.5 vision even using ''pip install git+https://github.com/metadriverse/metadrive.git@releases/0.2.5''