鬥地主RL項目編譯環境說明

Dou Di Zhu with Combinational Q-Learning

Combinational Q-Learning for Dou Di Zhu

上交的一篇關於鬥地主論文及相關代碼。其中值得注意的環境配置

 

1. conda的安裝與使用

wget -c https://repo.continuum.io/miniconda/Miniconda3-latest-Linux-x86_64.sh

2 安裝pybind11  pybind11 安裝

2.1 安裝依賴的pytest

pip install pytest

2.2 Get pybind11, from website: https://github.com/pybind/pybind11 to get the source package.

wget https://github.com/pybind/pybind11/archive/v2.4.3.tar.gz

$ tar xvf pybind11-2.4.3.tar.gz

$cd pybind11-2.4.3

$ mkdir build

$ cd build

$ cmake ..

$ make check -j 4  (make and check, not necessary)

$ sudo make install  (maybe should disable python3 in /usr/bin/ if you use python2)

3. Step by step training tutorial

 3.1 Clone the repo

git clone https://github.com/qq456cvb/doudizhu-C.git

  3.2 Change work directory to root

cd doudizhu-C

 3.3 Create env from environment.yml

conda env create -f environment.yml

 3.4 Activate env

conda activate doudizhu

 3.5 Build C++ files

mkdir build
cd build
cmake ..
make

 3.6 Have fun training!

cd TensorPack/MA_Hierarchical_Q
python main.py