I am stuck here,
I have derived DefaultAPPORLModule, that has it’s own implementation, there is no clear documentation on how to change to new API compatibility. The reason is that I have TCN feature extractor, and also I want to Enable LSTM for this APPO model.
Have anyone implemented, a multiagentenv with APPO for multiple agents in multienv and multiple of Multienv?
It is frustrating using rllib, they should probably diversify their talent pool. It looks very homogenic. Note that I also contributed to codes to different AI platforms.