Fall 2018 Deep Learning Project Repo
libraries: torch lua luarocks luasocket
to run a match:
- connect the server.
./DeepStack-Leduc/ACPCServer/dealer <name_of_game> ./DeepStack-Leduc/ACPCServer/leduc.game <#hands> -p 20000,20001 > <name_of_output_file>
put the output file in the folder "outputs" naming convention should be number of rounds.player1.player2.txt
- in separate tab, connect player 1
- in separate tab, connect player 2
examples of how to connect a player:
pretrained deepstack: cd ./DeepStack-Leduc/Source th Player/deepstack.lua
allin_net.py (from previous year's project) -- always raises max amount: python ./code/ESMCCFR-LEDUC/allin_net.py ./DeepStack-Leduc/ACPCServer/leduc.game localhost
example_player.c (from DeepStack code) -- selects random action: ./DeepStack-Leduc/ACPCServer/example_player ./DeepStack-Leduc/ACPCServer/leduc.game localhost
always_call.py (written by us) -- always calls -- just wants to see what everyone has python ./code/ESMCCFR-LEDUC/always_call.py ./DeepStack-Leduc/ACPCServer/leduc.game localhost
bots to invent: DeepStack but bluffing DeepStack but risk-averse DeepStack trained on risk-averse self?