search

starry-sky6688 / MARL-Algorithms

Implementations of IQL, QMIX, VDN, COMA, QTRAN, MAVEN, CommNet, DyMA-CL, and G2ANet on SMAC, the decentralised micromanagement scenario of StarCraft II
1.46k stars 283 forks source link

About the system requirements to train qmix map corridor #51

Closed Sud0x67 closed 3 years ago

Sud0x67 commented 3 years ago

Hi,sir， I wonder the requirements about train qmix model. I train it with my 32GB memory machine and it run out the memory, so I want to know about how many memory it needed or what else I can do. Thanks a lot for your code anyway.

starry-sky6688 commented 3 years ago

Corridor needs about 55GB memory, if your memory is not enough, you can reduce the size of replay buffer by 'buffer_size' in arguments.py. If your GPU has enough memory, you can use the memory of GPU, but I don't know how to achieve it so you need to make some changes by yourself.