search

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

several issues about training a no-limit hold'em agent #314

Open hzwudi2014 opened 6 months ago

hzwudi2014 commented 6 months ago
  1. How can I modify the origin env such as players' origin chip num or game settings like ante(every player have to put some chips into pot which is different from a standard game).
  2. As my understanding, if I want to train a good no-limit hold'em angent which can achive human-level, I should initiate the env with 8 agents and all these agents should use cfr algorithm ( in the example code, you use cfr agent playing against random agent, while as I see this is for a instant effect proving the RL algorighm runs well)
  3. After the model is trained, can I integrate it in my code if I input a certain situation in a well-suited format?