Open yunanyan opened 3 years ago
Thanks for your question!
To be honest, I haven't really thought about it. So I don't know exactly, but I think the major change of this code for starcraft2 are
To use recurrent network for each agent network
To change the memory which can handle time series data
Actually, I think you should to refer to the following QMIX implementation (implemented by the author of QMIX). https://github.com/oxwhirl/pymarl
I'm sorry I couldn't help you.
Regards
Hi,
If I want to apply this codebase to multiagent game environment such as Starcraft II. What part of the network should be changed?
Thank you!