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

What is the problem with Rlcard environment ? #9

Open hns007 opened 6 months ago

hns007 commented 6 months ago

"Rlcard environment sucks, 50bb pot, wrong pot sizes, wrong action order after flop, I don't know where to start. But it's the only environment I konw out there suitable for this task."

Question 1: "Wrong pot sizes" refers to a 50bb pot, right?
--- The pot size can be modified through configuration. for example: rlcard.make( 'no-limit-holdem', config={ 'seed': seed, 'chips_for_each': 200 } )

Quest 2: what dose "Wrong action order after flop" mean?

bupticybee commented 6 months ago

Even you can set chip number it's not good enough, bet size is different from slumbot, and in flop,BB should move first, but in rlcard env, SB moves first.

hns007 commented 6 months ago

OK, thanks. What were the results of the matches between AlphaNLHoldem and Slumbot?

bupticybee commented 6 months ago

OK, thanks. What were the results of the matches between AlphaNLHoldem and Slumbot?

Can't play with slumbot, implemted rules are different.