# Dou Di Zhu with Combinationatorial Q-Learning
# Accepted to AIIDE 2020
## Step by step training tutorial
1. Clone the repo
```
git clone https://github.com/qq456cvb/doudizhu-C.git
```
2. Change work directory to root
```
cd doudizhu-C
```
3. Create env from environment.yml
```
conda env create -f environment.yml
```
4. Activate env
```
conda activate doudizhu
```
5. Build C++ files
```
mkdir build
cd build
cmake ..
make
```
6. Have fun training!
```
cd TensorPack/MA_Hierarchical_Q
python main.py
```
## Evaluation against other baselines
1. Download pretrained model from https://jbox.sjtu.edu.cn/l/L04d4A or [GoogleDrive](https://drive.google.com/drive/folders/1YTNR5JYNgNfQpQ9DwhQ3ClcyeXKn_17b?usp=sharing), then put it into `pretrained_model`
2. Build Monte-Carlo baseline and move the lib into root
```
git clone https://github.com/qq456cvb/doudizhu-baseline.git
cd doudizhu-baseline/doudizhu
mkdir build
cd build
cmake ..
make
mv mct.cpython-36m-x86_64-linux-gnu.so [doudizhu-C ROOT]
```
3. Run evaluation scripts in `scripts`
```
cd scripts
python experiments.py
```
## Directory Structure
* `TensorPack` contain different RL algorithms to train agents
* `experiments` contain scripts to evaluate agents' performance against other baselines
* `simulator` contain scripts to evaluate agents' performance against online gaming platform called "QQ Dou Di Zhu" (we provide it for academic use only, use it at your own risk!)
## Miscellaneous
* We provide a Monte-Carlo-Tree-Search algorithm in https://github.com/qq456cvb/doudizhu-baseline
* We provide a configured Dou Di Zhu mini-server in https://github.com/qq456cvb/doudizhu-tornado for you to play interactively. NOTE you should build the server and load pretrained model by yourself! Tutorial coming soon!
* If you meet any problems, open an issue.
## DouZero
Recently, another algorithm called DouZero (https://github.com/kwai/DouZero) has been proposed, to whom may be interested in a strong DouDizhu AI. It is also an actively maintained open-source project.
## References
See our paper https://arxiv.org/pdf/1901.08925.pdf. If you find this algorithm useful or use part of its code in your projects, please consider cite
```
@inproceedings{you2020combinatorial,
title={Combinatorial Q-Learning for Dou Di Zhu},
author={You, Yang and Li, Liangwei and Guo, Baisong and Wang, Weiming and Lu, Cewu},
booktitle={Proceedings of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment},
volume={16},
number={1},
pages={301--307},
year={2020}
}
```
没有合适的资源?快使用搜索试试~ 我知道了~
C++python用pybind11打王者(强化学习AI斗地主),AcceptedtoAIIDE-2020_Python.zip
共91个文件
py:47个
npy:19个
png:14个
1.该资源内容由用户上传,如若侵权请联系客服进行举报
2.虚拟产品一经售出概不退款(资源遇到问题,请及时私信上传者)
2.虚拟产品一经售出概不退款(资源遇到问题,请及时私信上传者)
版权申诉
0 下载量 60 浏览量
2023-04-30
10:26:09
上传
评论
收藏 12.36MB ZIP 举报
温馨提示
C++python用pybind11打王者(强化学习AI斗地主),AcceptedtoAIIDE-2020_Python.zip
资源推荐
资源详情
资源评论
收起资源包目录
C++python用pybind11打王者(强化学习AI斗地主),AcceptedtoAIIDE-2020_Python.zip (91个子文件)
doudizhu-C-master
TensorPack
PolicySL
evaluator.py 10KB
Policy_SL_v1_4.py 31KB
A3C
evaluator.py 7KB
A3Cv1_5.py 24KB
simulator.py 19KB
model_loader.py 3KB
Hierarchical_Q
evaluator.py 6KB
combination.py 4KB
DQNModel.py 8KB
DQN.py 8KB
expreplay.py 21KB
MA_Hierarchical_Q
evaluator.py 5KB
main.py 5KB
predictor.py 10KB
baseline_evaluator.py 6KB
DQNModel.py 11KB
expreplay.py 20KB
tools.py 5KB
env.py 9KB
ResNetBlock.py 3KB
ValueSL
evaluator.py 10KB
Value_SL_v1_4.py 12KB
AutoEncoder
main.py 8KB
encoding.npy 13.21MB
Vanilla_Q
evaluator.py 6KB
DQNModel.py 5KB
DQN.py 7KB
expreplay.py 12KB
A3C_FC
simulator_fc.py 20KB
evaluator_fc.py 13KB
A3cv1_5_fc.py 31KB
utils.py 31KB
card.hpp 2KB
CMakeLists.txt 312B
game.hpp 2KB
card.py 21KB
card.cpp 136KB
simulator
sim.py 14KB
preprocess.py 3KB
coordinator.py 2KB
main.py 7KB
array
bujiao.npy 551B
start.npy 551B
buqiang.npy 581B
end.npy 581B
ming_chupai.npy 581B
chupai.npy 551B
alone_chupai.npy 581B
continous_end.npy 581B
bujiabei.npy 581B
buchu.npy 551B
qiangdizhu.npy 581B
reverse.npy 551B
continuous_defeat.npy 551B
jiaodizhu.npy 551B
yaobuqi.npy 581B
addict_window.npy 731B
fail_end.npy 581B
tishi.npy 551B
predictor.py 12KB
templates
J.png 403B
Joker.png 498B
10.png 562B
9.png 534B
3.png 545B
Q.png 655B
6.png 515B
5.png 517B
K.png 539B
4.png 422B
8.png 560B
A.png 492B
7.png 420B
2.png 521B
monitor.py 3KB
manager.py 4KB
expreplay.py 11KB
tools.py 18KB
config.py 10KB
dancing_link.h 857B
main.cpp 40KB
environment.yml 1KB
.gitignore 52B
game.cpp 6KB
tools.py 5KB
README.md 2KB
dancing_link.cpp 5KB
scripts
experiments.py 2KB
mct_baseline.py 2KB
agents.py 4KB
envs.py 8KB
共 91 条
- 1
资源评论
快撑死的鱼
- 粉丝: 1w+
- 资源: 9154
上传资源 快速赚钱
- 我的内容管理 展开
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
最新资源
- 简单的Python示例,演示了如何使用TCP/IP协议进行基本的客户端和服务器通信
- 考试.sql
- keil2 + proteus + 8051.exe
- 1961ee27df03bd4595d28e24b00dde4e_744c805f7e4fb4d40fa3f695bfbab035_8(1).c
- mediapipe-0.9.0.1-cp37-cp37m-win-amd64.whl.zip
- windows注册表编辑工具
- mediapipe-0.9.0.1-cp37-cp37m-win-amd64.whl.zip
- 校园通行码预约管理系统20240522075502
- 车类型数据集6250张VOC+YOLO格式.zip
- The PyTorch implementation of STGCN.STGCN-main.zip
资源上传下载、课程学习等过程中有任何疑问或建议,欢迎提出宝贵意见哦~我们会及时处理!
点击此处反馈
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功