Open source rebel poker AI project

[flappybird]

So, i've been diving into some poker AI research lately, and I stumbled across Facebook AI's Rebel algorithm. Got me thinking, would anyone in the ML/DL crowd be up for teaming up to implement it and maybe even build a proper game API for training? Im picturing this as a fully open-source project, part fun, part challenge and part "let's see if we can actually pull this off." If this sounds like your kind of rabbit hole, let me know!

 

[DandD]

What's your vision for the project's starting point, are you thinking of replicating Rebel's core logic first, or building the poker API framework before going into the algorithm implementation?

 

[flappybird]

What's your vision for the project's starting point, are you thinking of replicating Rebel's core logic first, or building the poker API framework before going into the algorithm implementation?

Good question, i was leaning more toward getting a solid poker API framework in place first, just so everything has a clean structure before layering in Rebel's logic. That way it'll be easier to test and swap different algorithms down the road. Do you think it makes more sense to start directly with the core decision making logic instead and scaffold the API as we go , or do you prefer having the framework locked in first?