Open neuronphysics opened 1 year ago
Unfortunately at the moment the stable baselines example is broken due to an inherent incompatibility between RLLib, Gym and Gymnasium. We are working in a way around this, but currently our priority is RLLib (because of the upcoming Melting Pot Competition).
The numpy issue seems to be a version incompatibility problem. At some point we had to pin the numpy version beacuse it broke RLLib. I think we might be OK unpinning now, if that would be useful.
Is there any update on this? or any workaround with specific pinned versions?
I am trying to do the same thing with a torch codebase that works with gym environment. Getting stuck at TypeError: unhashable type: 'ConfigDict'
Hello, I am attempting to integrate the MeltingPot environment as a benchmark for my Multi-Agent Reinforcement Learning (MARL) algorithm. The repository appears to be updated frequently. I ran the example script
python3 -m meltingpot.examples.pettingzoo.sb3_train
and I got the following error messageMoreover, I tried to test the meltingpot environment work on Google Colab using the provided code. However, I continuously receive the following error:
This error message