4 ms·
this is very cool. another interesting open-source project for imperfect-information game solving is DeepMind's OpenSpiel: https://github.com/deepmind/open_sp
by currymj 6y ago
this is very cool.
another interesting open-source project for imperfect-information game solving is DeepMind's OpenSpiel:
https://github.com/deepmind/open_spiel https://github.com/deepmind/open_spiel
However as far as poker goes, I believe they only provide toy versions of the game, even though theoretically you could put in your own implementation.
I've noticed that in this part of the AI world, things tend to be closed-source for whatever reason and I think that limits the ability to learn. Seems good to see that changing.
- bupticybee 6y agoYeah I am aware of this project, and I am current trying write an alpha-star like AI for another holdem simulator https://www.pettingzoo.ml/classic https://www.pettingzoo.ml/classic, I already made some progress and I somehow believe the code for a alpha-star like texas-holdem AI will be released in sometime of the future.
- sharky6000 6y agoHi, thanks:) It does have the large poker games too. You can find generalized poker in the universal_poker game, which wraps the Annual Computer Poker Competition codebase: https://github.com/deepmind/open_spiel/blob/master/open_spiel/games/universal_poker.h https://github.com/deepmind/open_spiel/blob/master/open_spie... Check out universal_poker_test.cc for example uses. It supports limit and common abstractions as well as the full game of no-limit. There is an example of FCPA which you can play at the keyboard in examples/
- currymj 6y agooh cool, good to know!