search

bupticybee / AlphaNLHoldem

An unoffical implementation of AlphaHoldem. 1v1 nl-holdem AI.
GNU Affero General Public License v3.0
66 stars 16 forks source link

Issues in RLCard environment #2

Closed daochenzha closed 1 year ago

daochenzha commented 1 year ago

Thanks for the great work!

Would you mind sharing more details about the issues in the RLCard environment? We want to fix them.

bupticybee commented 1 year ago

Thanks for the great work!

Would you mind sharing more details about the issues in the RLCard environment? We want to fix them.

Major problems:

  1. you use 50bb pot, which is not standard 100bb setting
  2. bet sizes are wrong, if you want to fix them, check it with https://www.slumbot.com/ make sure you use the same bet sizes in every branch of the game tree
  3. after a bet-call/check-call in pre-flop, IP player act first in flop, in which case you got it all wrong.

I'm thinking if we could discuss further in some way , I'm interested in discussing some of your previous works (rlcard, douzero,etc) and possible future directions, and of course also, about nl-holdem.

I'm a researcher in bytedance AI lab, located in Shenzhen. Looking forward for further discussion.

daochenzha commented 1 year ago

@bupticybee Thank you for providing the information. @ruzhwei will follow up. Yes, I am happy to chat. My WeChat is zdcwhu

bupticybee commented 1 year ago

Closing