cgnicholls/rlpoker

Reinforcement learning algorithms to play Poker

rlpoker

Reinforcement learning algorithms to play Poker.

Create a game of Leduc Hold em with 3 cards:

leduc = Leduc.create_game(3)

Run NFSP on this:

python rlpoker/nfsp.py

Solve Leduc Hold Em using cfr

strategy = cfr(leduc, num_iters=100000, use_chance_sampling=True)

You can also use external sampling cfr instead:

python -m examples.cfr --game Leduc

You can also use external sampling cfr instead:

python -m examples.cfr --cfr_algorithm external --game Leduc

Solve Leduc Hold Em using deep cfr

python -m examples.deep_cfr --game Leduc

Installation

Use the rlpoker conda environment.