围棋AI-KataGo安装-Windows免编译

字数统计: 1.2k字 | 阅读时长: 5分

2021-05-04

围棋22

摘要: 本文记录在 Windows 下免编译安装 KataGo 的过程。时间 2021.05

【对数据分析、人工智能、金融科技、风控服务感兴趣的同学，欢迎关注我哈，阅读更多原创文章】
我的网站：潮汐朝夕的生活实验室
我的公众号：潮汐朝夕
我的知乎：潮汐朝夕
我的github：FennelDumplings
我的leetcode：FennelDumplings

KataGo 项目地址: https://github.com/lightvector/KataGo/releases

计算平台选择

最新版本为 v1.8.2

有三种版本: OpenCL vs CUDA vs Eigen

项目说明中的描述如下

KataGo has three backends, OpenCL (GPU), CUDA (GPU), and Eigen (CPU).

The quick summary is:

Use OpenCL if you have any good or decent GPU.

Use Eigen with AVX2 if you don’t have a GPU or if your GPU is too old/weak to work with OpenCL, and you just want a plain CPU KataGo.

Use Eigen without AVX2 if your CPU is old or on a low-end device that doesn’t support AVX2.

You can try CUDA you have a top-end NVIDIA FP16 + tensor-core GPU and you are willing to go through the hassle to install CUDA+CUDNN. It might or might not be faster than OpenCL, you can try it out to see.

因此对于有 GPU 的电脑来说，选择 OpenCL 比较好，如下图

下载解压后，KataGo 目录内的内容如下

模型选择

最近的有新模型发布的版本为 v1.4.5，截图如下

项目说明中的描述

Which Network Should I Use?

For weaker or mid-range GPUs, try the final 20-block network.

For top-tier GPUs and/or for the highest-quality analysis if you’re going to use many thousands and thousands of playouts and long thinking times, try the final 40-block network, which is more costly to run but should be the strongest and best.

If you care a lot about theoretical purity - no outside data, bot learns strictly on its own - use the 20 or 40 block nets from this release, which are pure in this way and still much stronger than Leela Zero, but also not quite as strong as these final nets here.

If you want some nets that are much faster to run, and each with their own interesting style of play due to their unique stages of learning, try any of the “b10c128” or “b15c192” Extended Training Nets here which are 10 block and 15 block networks from earlier in the run that are much weaker but still pro-level-and-beyond.

And if you want to see how a super ultra large/slow network performs that nobody has tested until now, try the fat 40-block 384 channel network mentioned a little up above.

总结下来就是, b20,30,40 依次变强, 有显卡的就选择b40的模型就好，如下图