https://github.com/lightvector/KataGo/releases

https://github.com/lightvector/KataGo/releases/download/v1.15.3/katago-v1.15.3-cuda12.1-cudnn8.9.7-linux-x64.zip

https://katagotraining.org/networks/

目前最新权重
https://media.katagotraining.org/uploaded/networks/models/kata1/kata1-b28c512nbt-s9584861952-d4960414494.bin.gz

mkdir -p /kp-data/katago

cd /kp-data/katago

wget https://github.com/lightvector/KataGo/releases/download/v1.15.3/katago-v1.15.3-cuda12.1-cudnn8.9.7-linux-x64.zip

unzip katago-v1.15.3-cuda12.1-cudnn8.9.7-linux-x64.zip

下载神经网路数据到当前目录：

wget https://media.katagotraining.org/uploaded/networks/models/kata1/kata1-b28c512nbt-s9584861952-d4960414494.bin.gz

安装依赖
wget http://archive.ubuntu.com/ubuntu/pool/universe/libz/libzip/libzip5_1.5.1-0ubuntu1_amd64.deb

wget http://archive.ubuntu.com/ubuntu/pool/main/o/openssl/libssl1.1_1.1.1f-1ubuntu2_amd64.deb

dpkg -i libssl1.1_1.1.1f-1ubuntu2_amd64.deb 

dpkg -i libzip5_1.5.1-0ubuntu1_amd64.deb

安装cudnn
进入CuDNN官网，选择合适的版本下载压缩包。
https://developer.nvidia.com/cudnn-archive#

选择:
Download cuDNN v8.9.7 (December 5th, 2023), for CUDA 12.x
https://developer.nvidia.com/downloads/compute/cudnn/secure/8.9.7/local_installers/12.x/cudnn-linux-x86_64-8.9.7.29_cuda12-archive.tar.xz

解压之后，分别进入 include 以及 lib 目录中，将如下文件拷贝到之前 CUDA 的安装目录下。
sudo cp (你的目录)/include/cudnn.h /usr/local/cuda-12.4/include/
sudo cp (你的目录)/lib/libcudnn* /usr/local/cuda-12.4/lib64/
sudo chmod a+r /usr/local/cuda-12.4/include/cudnn.h 
sudo chmod a+r /usr/local/cuda-12.4/lib64/libcudnn*

验证CuDNN是否安装成功，可以结合Pytorch查看，输出如下信息则证明安装成功。
python
import torch
print(torch.backends.cudnn.version())

安装TensorRT
https://developer.nvidia.com/tensorrt/download/10x
wget https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.12.0/tars/TensorRT-10.12.0.36.Linux.x86_64-gnu.cuda-12.9.tar.gz

tar -xvf TensorRT-10.12.0.36.Linux.x86_64-gnu.cuda-12.9.tar.gz
cp -rp  TensorRT-10.12.0.36/lib/* /usr/local/cuda-12.4/lib64/

修改环境变量
vim /root/.bashrc   #文件最后增加
export PATH=/usr/local/cuda-12.4/bin${PATH:+:${PATH}}
export LD_LIBRARY_PATH="/usr/local/cuda-12.4/lib64:$LD_LIBRARY_PATH"
source /root/.bashrc

安装：
/kp-data/katago

/kp-data/katago/katago genconfig -model kata1-b28c512nbt-s9584861952-d4960414494.bin.gz -output gtp_custom.cfg

选择：（其他都默认）

Chinese

选完了程序就开始各种运行尝试，过几分钟分钟后出现done，说明运行结束。在当前前文件夹下生成文件gtp_custom.cfg,你想改成其它名字当然也没问题。

启动：
./katago gtp -model 'kata1-b28c512nbt-s9584861952-d4960414494.bin.gz' -config 'gtp_custom.cfg'

windows10为例：
系统下载plink到D:/盘
https://www.chiark.greenend.org.uk/~sgtatham/putty/latest.html  #找到plink.exe, 64-bit x86类型解压安装。

系统下载sabaki：
https://github.com/SabakiHQ/Sabaki/releases/download/v0.52.2/sabaki-v0.52.2-win-x64-setup.exe

sabaki配置语言:
file->preferences->language (选择简体中文)  #关闭重启

sabaki配置引擎：
引擎->显示引擎侧边栏->管理引擎

sabaki配置远程连接：
ikaopu  #第一行取名名字
D:\plink\plink\plink.exe    #第二行填写plink 路径
-ssh -batch -P 28511 root@ikp.ikaopu.cn -pw 9aBP9cgKFptgbSf8wXUX9cW5Nuq5Xfrp "/root/katago/katago gtp -model '/root/katago/kata1-b28c512nbt-s9584861952-d4960414494.bin.gz' -config '/root/katago/gtp_custom.cfg'"  #第三行参数
time_settings 0 5 1 #第四行也是参数，两边对阵，每5秒输入一子