This PR demonstrates how one can write thin python wrappers over the existing sarsa libraries. Such wrappers can allow one to code up high level agents in python that can directly call the fast C++ sarsa library functions.
As an example, this PR also contains a high level sarsa agent written in python. This agent is a python paraphrase of the existing high level sarsa offense agent written in C++. A script that runs this agent is also included.
In comparison to the C++ agent, the python agent obtains < 0.5 % difference in mean goal scoring % when averaged over 20 runs. This is when the fullstate flag is ON.
However, in case of partial observability, the python agent obtains 0 - 5% less goal scoring percentage when compared to the C++ agent. Despite my Best efforts, I am unable to pin down the reason for this.
It would be great if the maintainers and the broader community could replicate this issue and find ways to overcome it.
This PR demonstrates how one can write thin python wrappers over the existing sarsa libraries. Such wrappers can allow one to code up high level agents in python that can directly call the fast C++ sarsa library functions.
As an example, this PR also contains a high level sarsa agent written in python. This agent is a python paraphrase of the existing high level sarsa offense agent written in C++. A script that runs this agent is also included.
In comparison to the C++ agent, the python agent obtains < 0.5 % difference in mean goal scoring % when averaged over 20 runs. This is when the fullstate flag is ON.
However, in case of partial observability, the python agent obtains 0 - 5% less goal scoring percentage when compared to the C++ agent. Despite my Best efforts, I am unable to pin down the reason for this.
It would be great if the maintainers and the broader community could replicate this issue and find ways to overcome it.