Errors

[Haoxiang-Wang]

When I use command bin/interactive.py --scenario simple.py, it turns out an error:

 /Users/salse/Dropbox/openai/multiagent-particle-envs/multiagent/environment.py(149)_set_action()
    148             import ipdb; ipdb.set_trace()
--> 149             action = [action]
    150 

The action seems undefined.

Also, could you give a example of training? I even don't know what to feed in env.step(). A single integer or a list of integer seems doesn't work.

[hot-cheeto]

Hello, I am having trouble as well with this. By any chance were you able to figure it out ?

[ryan-lowe]

Wasn't getting that particular bug (action is received as input into _set_action() in line 135) but I fixed some other bugs and it should work now. Apologies for the confusion.

We're working on cleaning up some code for training. In the meantime, there's an open-source MADDPG implementation here:

https://github.com/xuehy/pytorch-maddpg