Config num_players for holdem environments

datamllab / rlcard

Reinforcement Learning / AI Bots in Card (Poker) Games - Blackjack, Leduc, Texas, DouDizhu, Mahjong, UNO.

http://www.rlcard.org

MIT License

2.87k stars 619 forks source link

Config num_players for holdem environments #230

Closed rodrigodelazcano closed 3 years ago

rodrigodelazcano commented 3 years ago

This PR fixes two holdem games for adding extra players:

Leduc Holdem: the reward judger for leduc was only considering two player games. After this fixes more than two players can be added to the game.
Nolimitholdem: when adding extra players the default configuration for chips for each player didn't change and would only consider two player games as well. This PR expands the default chips list to the desired number of players

jkterry1 commented 3 years ago

@daochenzha just following up on this.

daochenzha commented 3 years ago

@jkterry1 It is merged, thanks for the the reminder

jkterry1 commented 3 years ago

@daochenzha can you please cut a new release to PyPI with this?

daochenzha commented 3 years ago

@jkterry1 @rodrigodelazcano Found a small bug on this. The current design cannot allow the config of num_players to be unspecified. If game_num_players is not in config, it will throw errors. I have fixed this by still using one value for chips_for_each. I instead make it as a list in the configure function of the game. Please see https://github.com/datamllab/rlcard/commit/955ed2b4ce39d4b3f1ee97382bfde451eeedf659

Could you confirm the fix is OK?

jkterry1 commented 3 years ago

@daochenzha that'll work, please cut a release

hsywhu commented 3 years ago

@daochenzha that'll work, please cut a release

The new version 1.0.4 has been released