MicroRTS-Py
MicroRTS-Py copied to clipboard
Support wall observation and multiple improvements
This PR does a few things.
- supports the wall observation.
- fixes https://github.com/Farama-Foundation/MicroRTS-Py/pull/34#issuecomment-1012722258
- one side effect is that the
agent_sota.pt
would no longer work. Hence we deprecate theppo_gridnet_large.py
model. People who are interested in using theagent_sota.pt
should use the https://github.com/Farama-Foundation/MicroRTS-Py/tree/v0.6.0.
- one side effect is that the
- update to the latest pytorch==1.13.0
- added some profiling tools showing the time spent on rollout, inference, and training.
- update docs
Running training here: https://wandb.ai/costa-huang/gym-microrts/runs/36ssyk73?workspace=user-costa-huang
Because this PR was relatively old and there had been numerous other changes in master
since it was created, I re-did the support for wall observations outside of this PR. That was in https://github.com/Farama-Foundation/MicroRTS-Py/pull/129, which I have just merged, so I assume this PR can be closed?