Doubt regarding Quantization

[kailashg26]

Hello,

I appreciate the paper content and indeed I was able to understand it perfectly with the source code. I have a doubt regarding the application of quantization you used, here are my doubts:

1. Is the quantization approach mentioned in the paper a black box approach that can be applied on any single agent and multi agent RL algorithms?
2. If the first question is true, I see that you evaluated the approach on single agent RL algorithms, can you provide some insights or any further information on how to instrument it for multi-agent reinforcement learning (MARL) problems?
   I highly appreciate your team for providing open-source code and detailed explanation on this novel quantization approach and I hope to receive the response from you.

Thanks,
Kailash