{"id":27813,"date":"2023-03-02T05:38:56","date_gmt":"2023-09-03T14:47:52","guid":{"rendered":"https:\/\/www.silicloud.com\/zh\/blog\/%e4%bd%bf%e7%94%a8chainerrl%ef%bc%8c%e5%9c%a8minecraft%e4%b8%ad%e8%bf%9b%e8%a1%8c%e6%b7%b1%e5%ba%a6%e5%bc%ba%e5%8c%96%e5%ad%a6%e4%b9%a0%e7%9a%84%e5%ae%9e%e8%b7%b5%e3%80%82\/"},"modified":"2025-08-12T04:46:01","modified_gmt":"2025-08-11T20:46:01","slug":"%e4%bd%bf%e7%94%a8chainerrl%ef%bc%8c%e5%9c%a8minecraft%e4%b8%ad%e8%bf%9b%e8%a1%8c%e6%b7%b1%e5%ba%a6%e5%bc%ba%e5%8c%96%e5%ad%a6%e4%b9%a0%e7%9a%84%e5%ae%9e%e8%b7%b5%e3%80%82","status":"publish","type":"post","link":"https:\/\/www.silicloud.com\/zh\/blog\/%e4%bd%bf%e7%94%a8chainerrl%ef%bc%8c%e5%9c%a8minecraft%e4%b8%ad%e8%bf%9b%e8%a1%8c%e6%b7%b1%e5%ba%a6%e5%bc%ba%e5%8c%96%e5%ad%a6%e4%b9%a0%e7%9a%84%e5%ae%9e%e8%b7%b5%e3%80%82\/","title":{"rendered":"\u4f7f\u7528ChainerRL\uff0c\u5728Minecraft\u4e2d\u8fdb\u884c\u6df1\u5ea6\u5f3a\u5316\u5b66\u4e60\u7684\u5b9e\u8df5"},"content":{"rendered":"<h1>\u6ce8\u610f<\/p>\n<ul class=\"post-ul\">\u73fe\u5728\u306f\u3053\u3061\u3089\u306e\u65b9\u6cd5\u306e\u65b9\u304c\u3088\u308a\u7c21\u5358\u306b\u74b0\u5883\u69cb\u7bc9\u3067\u304d\u307e\u3059 ChainerRL + Minecraft\u3067\u6df1\u5c64\u5f37\u5316\u5b66\u7fd2 \u30cf\u30f3\u30ba\u30aa\u30f3 [MineRL version]<\/ul>\n<h1>\u6211\u7684\u4e16\u754c\u5927\u5bb6\u597d\uff0c\u4f60\u4eec\u90fd\u77e5\u9053Minecraft\u3002\u6211\u4e4b\u524d\u5e76\u4e0d\u77e5\u9053\uff0c\u4f46\u539f\u6765Minecraft\u53ef\u4ee5\u7b80\u79f0\u4e3a&#8221;\u30de\u30a4\u30af\u30e9&#8221;\u3002\u4f5c\u4e3a\u5b98\u65b9\u7684\u5ba3\u4f20\u89c6\u9891\uff0c\u4e5f\u6709\u4e0b\u9762\u8fd9\u6837\u7684\u5185\u5bb9\u3002<\/p>\n<blockquote><p>\u8bf7\u7528\u4e2d\u6587\u7ed9\u4ee5\u4e0b\u94fe\u63a5\u8fdb\u884c\u91ca\u4e49\uff1ahttps:\/\/t.co\/gITUej4Tmb\u2014 Keisuke Umezawa (Chainer Evangelist) (@kmechann) 2018\u5e7412\u670812\u65e5<\/p><\/blockquote>\n<p><script><\/script><\/p>\n<p>\u6b64\u5916\uff0c\u5728Minecraft Japan Wiki\u4e0a\u6709\u4ee5\u4e0b\u7684\u89e3\u91ca\u3002<\/p>\n<blockquote><p>Minecraft\u306f\u30012009\u5e745\u670810\u65e5\u306bNotch\u6c0f(\u672c\u540d:Markus Persson)\u304c\u958b\u767a\u3092\u59cb\u3081\u305f\u3001\u30b5\u30f3\u30c9\u30dc\u30c3\u30af\u30b9\u578b\u306e\u3082\u306e\u3065\u304f\u308a\u30b2\u30fc\u30e0\u3067\u3059\u3002<br \/>\n\u30ec\u30c8\u30ed\u30b2\u30fc\u3092\u60f3\u8d77\u3055\u305b\u308b\u30c9\u30c3\u30c8\u30c6\u30a4\u30b9\u30c8\u306e\u30d6\u30ed\u30c3\u30af\u304c\u6ea2\u308c\u308b\u4e16\u754c\u3067\u3001\u30d7\u30ec\u30a4\u30e4\u30fc\u306f\u5efa\u7269\u3084\u305d\u306e\u4ed6\u306e\u3082\u306e\u3092\u81ea\u7531\u306b\u5275\u9020\u3059\u308b\u3053\u3068\u304c\u51fa\u6765\u307e\u3059\u3002<br \/>\n\u672a\u958b\u306e\u571f\u5730\u3092\u63a2\u7d22\u3057\u305f\u308a\u3001\u6d1e\u7a9f\u3092\u63a2\u691c\u3057\u305f\u308a\u3001\u30e2\u30f3\u30b9\u30bf\u30fc\u3068\u6226\u3063\u305f\u308a\u3001\u690d\u7269\u3092\u80b2\u3066\u305f\u308a\u3001\u65b0\u3057\u3044\u30d6\u30ed\u30c3\u30af\u3092\u624b\u306b\u5165\u308c\u914d\u7f6e\u3059\u308b\u3053\u3068\u3067\u3001\u69d8\u3005\u306a\u3082\u306e\u3092\u4f5c\u308b\u3053\u3068\u304c\u3067\u304d\u307e\u3059\u3002<br \/>\n\u5275\u9020\uff08\u60f3\u50cf\uff09\u529b\u6b21\u7b2c\u3067\u5c0f\u3055\u306a\u5bb6\u304b\u3089\u3001\u30c9\u30c3\u30c8\u7d75\u3001\u5730\u4e0b\u57fa\u5730\u3001\u5de8\u5927\u306a\u57ce\u307e\u3067\u4f55\u3067\u3082\u4f5c\u308b\u3053\u3068\u304c\u3067\u304d\u307e\u3059\u3002<\/p>\n<p>\u30de\u30eb\u30c1\u30d7\u30ec\u30a4\u3067\u306f\u3001\u5354\u529b\u3057\u3066\u5de8\u5927\u306a\u5efa\u7bc9\u7269\u3092\u4f5c\u3063\u305f\u308a\u3001Player VS Player(PvP)\u3067\u4ed6\u30d7\u30ec\u30a4\u30e4\u30fc\u3068\u6226\u3046\u3053\u3068\u3082\u51fa\u6765\u307e\u3059\u3002<\/p><\/blockquote>\n<p>\u6839\u636e\u8fd9\u4e2a\u8bf4\u660e\uff0c\u6709\u4e00\u4e9b\u9002\u5408\u5f53\u524d\u5f3a\u5316\u5b66\u4e60\u7684\u5173\u952e\u8bcd\u3002<\/p>\n<ul class=\"post-ul\">\n<li style=\"list-style-type: none;\">\n<ul class=\"post-ul\">\u5efa\u7269\u3084\u305d\u306e\u4ed6\u306e\u3082\u306e\u3092\u81ea\u7531\u306b\u4f5c\u308b\u3053\u3068\u3067\u304d\u308b<\/ul>\n<\/li>\n<\/ul>\n<p>&nbsp;<\/p>\n<ul class=\"post-ul\">\n<li style=\"list-style-type: none;\">\n<ul class=\"post-ul\">\u8272\u3005\u306a\u30b2\u30fc\u30e0\u304c\u3067\u304d\u308b<\/ul>\n<\/li>\n<\/ul>\n<p>&nbsp;<\/p>\n<ul class=\"post-ul\">\u30de\u30eb\u30c1\u30d7\u30ec\u30a4\u3067\u304d\u308b<\/ul>\n<p>\u6211\u8ba4\u4e3a\u6b63\u56e0\u4e3aMinecraft\u5177\u6709\u8fd9\u4e9b\u7279\u6027\uff0c\u6240\u4ee5\u624d\u4ea7\u751f\u4e86\u5728Minecraft\u8fdb\u884c\u5f3a\u5316\u5b66\u4e60\u7684\u52a8\u673a\u3002<\/p>\n<h1>\u9a6c\u6d1b\u9879\u76ee\u5728\u8fd9\u4e2a\u5de5\u4f5c\u574a\u4e2d\uff0c\u6211\u4eec\u5c06\u4f7f\u7528Minecraft\u4f5c\u4e3a\u6a21\u62df\u73af\u5883\u8fdb\u884c\u6df1\u5ea6\u5f3a\u5316\u5b66\u4e60\u3002\u76ee\u524d\uff0c\u6709\u4e00\u4e2a\u540d\u4e3aMARLO\u7684\u6df1\u5ea6\u5f3a\u5316\u5b66\u4e60\u7ade\u8d5b\u6b63\u5728\u4f7f\u7528Minecraft\u8fdb\u884c\u3002\u8fd9\u6b21\u6211\u4eec\u5c06\u4f7f\u7528\u4e00\u4e2a\u540d\u4e3amarLo\u7684OpenAI&#8217;s Gym\u517c\u5bb9\u7684Minecraft\u73af\u5883\u6765\u53c2\u52a0\u8fd9\u4e2a\u7ade\u8d5b\u3002\u7531\u4e8e\u517c\u5bb9OpenAI&#8217;s Gym\uff0c\u6211\u4eec\u53ef\u4ee5\u8f7b\u677e\u5730\u4f7f\u7528\u5f3a\u5316\u5b66\u4e60\u6846\u67b6\u5982ChainerRL\uff08\u867d\u7136\u5e76\u4e0d\u5b8c\u5168\uff0c\u6bd4\u5982\u4e0d\u80fd\u4f7f\u7528\u4fdd\u5b58\u89c6\u9891\u7684\u5305\u88c5\u5668\uff09\u3002<\/p>\n<p>\u5728MarLo\u4e2d\uff0c\u5df2\u7ecf\u63d0\u4f9b\u4e86\u4ee5\u4e0b\u73af\u5883\uff0c\u4f8b\u5982\u53ef\u4ee5\u4f7f\u7528\u6df1\u5ea6\u5f3a\u5316\u5b66\u4e60\u6765\u521b\u5efa\u4e00\u4e2a\u80fd\u591f\u5728\u7194\u5ca9\u4e0a\u8d70\u7684AI\u3002\u672c\u6b21\u6211\u4eec\u5c06\u4f7f\u7528MarLo-FindTheGoal-v0\u8fdb\u884c\u8bf4\u660e\uff0c\u7136\u540e\u8bf7\u5927\u5bb6\u6311\u6218\u65b0\u7684\u4efb\u52a1\u3002<\/p>\n<div><img decoding=\"async\" class=\"post-images\" title=\"\" src=\"https:\/\/cdn.silicloud.com\/blog-img\/blog\/img\/657cf8cd37434c4406bb6cf3\/14-10.gif\" alt=\"\" \/><\/p>\n<h1>\u30cf\u30f3\u30ba\u30aa\u30f3\u306e\u5185\u5bb9\u203b \u3053\u306e\u30cf\u30f3\u30ba\u30aa\u30f3\u306f\u3001https:\/\/github.com\/keisuke-umezawa\/marlo-handson \u3092\u5143\u306b\u4f5c\u6210\u3057\u3066\u3044\u307e\u3059\u3002<\/p>\n<h2>\u9700\u6c42\u622a\u81f32018\u5e7412\u670814\u65e5\uff0c\u9700\u8981\u4ee5\u4e0b\u5185\u5bb9\u3002<\/p>\n<ul class=\"post-ul\">\n<li style=\"list-style-type: none;\">\n<ul class=\"post-ul\">Python 3.5+ environment with<\/ul>\n<\/li>\n<\/ul>\n<p>Chainer v5.0.0<br \/>\nCuPy v5.0.0<br \/>\nChainerRL v0.4.0<br \/>\nmarlo v0.0.1.dev23<\/p>\n<h2>\u4f7f\u7528Azure\u8fdb\u884c\u73af\u5883\u8bbe\u7f6e\u3002\u4e3a\u4e86\u6309\u7167\u4ee5\u4e0b\u6b65\u9aa4\u8fdb\u884c\u64cd\u4f5c\uff0c\u9700\u8981\u4e00\u4e2aAzure\u8ba2\u9605\u3002<br \/>\n\u8bf7\u53c2\u8003\u5176\u4ed6\u73af\u5883\u5efa\u8bbe\u65b9\u6cd5\u7684\u6587\u7ae0\u3002<\/p>\n<h3>1. \u5b89\u88c5Azure CLI\u8bf7\u6839\u636e\u73af\u5883\u9009\u62e9\u4ee5\u4e0b\u9009\u9879\u3002<\/p>\n<p>Windows\u306e\u5834\u5408<\/p>\n<p>Install Azure CLI on Windows<\/p>\n<p>Homebrew(macOS)\u306e\u5834\u5408<\/p>\n<p>$ brew update &amp;&amp; brew install azure-cli<\/p>\n<p>python\u3092\u4f7f\u7528\u3059\u308b\u5834\u5408<\/p>\n<p>$ pip install azure-cli<\/p>\n<h3>2. \u767b\u5f55 Azure<\/p>\n<pre class=\"post-pre\"><code>$ az login\r\n<\/code><\/pre>\n<h3>3. \u8ba2\u9605\u9009\u62e9\u4ee5\u4e0b\u306e\u30b3\u30de\u30f3\u30c9\u3067\u3001\u3042\u306a\u305f\u304c\u304a\u6301\u3061\u306esubscription\u3092\u30ea\u30b9\u30c8\u30a2\u30c3\u30d7\u3067\u304d\u307e\u3059\u3002<\/p>\n<pre class=\"post-pre\"><code>$ az account list --all\r\n<\/code><\/pre>\n<p>\u8ba9\u6211\u4eec\u5728\u60a8\u7684\u8d26\u6237\u4e2d\u8bbe\u7f6e\u6240\u9009\u7684\u8ba2\u9605\u3002\u5f53\u7136\uff0c[A SUBSCRIPTION ID]\u8bf7\u66ff\u6362\u4e3a\u60a8\u7684\u8ba2\u9605ID\u3002<\/p>\n<pre class=\"post-pre\"><code>$ az account set --subscription [A SUBSCRIPTION ID]\r\n<\/code><\/pre>\n<h3>4. \u542f\u52a8GPU\u865a\u62df\u673a\u63a5\u4e0b\u6765\uff0c\u4f7f\u7528\u4ee5\u4e0b\u547d\u4ee4\u521b\u5efa\u6570\u636e\u79d1\u5b66VM\u3002\u53c2\u6570&#8211;generate-ssh-keys\u4f1a\u81ea\u52a8\u521b\u5efa\u7528\u4e8e\u8fde\u63a5VM\u7684\u5bc6\u94a5\uff0c\u5e76\u5c06\u79c1\u94a5\u4fdd\u5b58\u5728~\/.ssh\/\u76ee\u5f55\u4e0b\u7684id_rsa\u6587\u4ef6\u4e2d\uff0c\u5c06\u516c\u94a5\u4fdd\u5b58\u5728id_rsa.pub\u6587\u4ef6\u4e2d\u3002<\/p>\n<pre class=\"post-pre\"><code>$ AZ_USER=[\u597d\u304d\u306auser\u540d e.g. kumezawa]\r\n$ AZ_LOCATION=[resource-group\u3092\u4f5c\u3063\u305flocation e.g. eastus]\r\n$ AZ_RESOURCE_GROUP=[resource-group\u3092\u4f5c\u3063\u305flocation e.g. marmo]\r\n$ az vm create \\\r\n--location ${AZ_LOCATION} \\\r\n--resource-group ${AZ_RESOURCE_GROUP} \\\r\n--name ${AZ_USER}-vm \\\r\n--admin-username ${AZ_USER} \\\r\n--public-ip-address-dns-name ${AZ_USER} \\\r\n--image microsoft-ads:linux-data-science-vm-ubuntu:linuxdsvmubuntu:latest \\\r\n--size Standard_NC6 \\\r\n--generate-ssh-keys\r\n<\/code><\/pre>\n<p>\u5982\u679c\u987a\u5229\u7684\u8bdd\uff0c\u5e94\u8be5\u4f1a\u51fa\u73b0\u4ee5\u4e0b\u7c7b\u4f3c\u7684\u6d88\u606f\u3002<\/p>\n<pre class=\"post-pre\"><code>{\r\n  \"fqdns\": \"[YOUR USERNAME].eastus.cloudapp.azure.com\",\r\n  \"id\": \"\/subscriptions\/[YOUR SUBSCRIPTION ID]\/resourceGroups\/marLo-handson\/providers\/Microsoft.Compute\/virtualMachines\/vm\",\r\n  \"location\": \"eastus\",\r\n  \"macAddress\": \"AA-BB-CC-DD-EE-FF\",\r\n  \"powerState\": \"VM running\",\r\n  \"privateIpAddress\": \"10.0.0.4\",\r\n  \"publicIpAddress\": \"123.456.78.910\",\r\n  \"resourceGroup\": \"marLo-handson\",\r\n  \"zones\": \"\"\r\n}\r\n<\/code><\/pre>\n<p>\u8bf7\u8bb0\u4f4f\u4ee5\u4e0b\u6b65\u9aa4\uff0c\u4ee5\u4fbf\u4f7f\u7528\u516c\u5171IP\u5730\u5740\u3002<\/p>\n<h4>\u6ce8\u610f\u5982\u679c\u5728~\/.ssh\/\u76ee\u5f55\u4e0b\u5b58\u5728\u79c1\u94a5id_rsa\u548c\u516c\u94a5id_rsa.pub\uff0c\u5219\u4e0a\u8ff0\u547d\u4ee4\u5c06\u5f15\u53d1\u9519\u8bef\u3002<br \/>\n\u5728\u8fd9\u79cd\u60c5\u51b5\u4e0b\uff0c\u60a8\u53ef\u4ee5\u81ea\u5df1\u521b\u5efa\u5bc6\u94a5\u5e76\u6307\u5b9a\u5176\u521b\u5efa\u65b9\u5f0f\u3002<\/p>\n<pre class=\"post-pre\"><code>$ az vm create \\\r\n--location ${AZ_LOCATION} \\\r\n--resource-group ${AZ_RESOURCE_GROUP} \\\r\n--name ${AZ_USER}-vm \\\r\n--admin-username ${AZ_USERER} \\\r\n--public-ip-address-dns-name ${AZ_USER} \\\r\n--image microsoft-ads:linux-data-science-vm-ubuntu:linuxdsvmubuntu:latest \\\r\n--size Standard_NC6 \\\r\n--ssh-key-value [\u516c\u958b\u9375\u306epath e.g. ~\/.ssh\/id_rsa.pub]\r\n<\/code><\/pre>\n<h4>\u6ce8\u610f\u5982\u679c\u60a8\u5e0c\u671b\u5728CPU\u5b9e\u4f8b\u800c\u4e0d\u662fGPU\u4e0a\u542f\u52a8\uff0c\u53ef\u4ee5\u4f7f\u7528\u4f8b\u5982&#8211;size Standard_D2s_v3\u7684\u9009\u9879\u3002\u5982\u679c\u60a8\u60f3\u8981\u4e86\u89e3\u5176\u4ed6\u53ef\u7528\u7684\u865a\u62df\u673a\u5927\u5c0f\uff0c\u53ef\u4ee5\u901a\u8fc7\u4ee5\u4e0b\u65b9\u5f0f\u8fdb\u884c\u67e5\u8be2\u3002<\/p>\n<pre class=\"post-pre\"><code>az vm list-sizes --location eastus --output table\r\n<\/code><\/pre>\n<h3>\u9700\u8981\u6253\u5f00\u7684\u7aef\u53e3\u4ee5\u8bbf\u95ee\u3002<\/p>\n<pre class=\"post-pre\"><code>$ az vm open-port --resource-group ${AZ_RESOURCE_GROUP} --name ${AZ_USER}-VM --port 8000 --priority 1010 \\\r\n&amp;&amp; az vm open-port --resource-group ${AZ_RESOURCE_GROUP} --name ${AZ_USER}-VM --port 8001 --priority 1020 \\\r\n&amp;&amp; az vm open-port --resource-group ${AZ_RESOURCE_GROUP} --name ${AZ_USER}-VM --port 6080 --priority 1030\r\n<\/code><\/pre>\n<h3>6. SSH\u8fde\u63a5\u5230\u865a\u62df\u673a<\/p>\n<pre class=\"post-pre\"><code>$ AZ_IP=[\u3042\u306a\u305f\u306evm\u306eIP\u30a2\u30c9\u30ec\u30b9 e.g. \"40.121.36.99\"]\r\n$ ssh ${AZ_USER}@${AZ_IP} -i ~\/.ssh\/id_rsa\r\n<\/code><\/pre>\n<h3>\u4e3aMarLo\u521b\u5efa\u4e00\u4e2aConda\u73af\u5883\u3002<br \/>\n\u8bf7\u5728VM\u73af\u5883\u4e2d\u6267\u884c\u4ee5\u4e0b\u547d\u4ee4\u3002<\/p>\n<pre class=\"post-pre\"><code>$ curl -s -L https:\/\/nvidia.github.io\/nvidia-docker\/gpgkey | sudo apt-key add - \\\r\n&amp;&amp; distribution=$(. \/etc\/os-release;echo $ID$VERSION_ID) \\\r\n&amp;&amp; curl -s -L https:\/\/nvidia.github.io\/nvidia-docker\/$distribution\/nvidia-docker.list | sudo tee \/etc\/apt\/sources.list.d\/nvidia-docker.list \\\r\n&amp;&amp; sudo apt-get update \\\r\n&amp;&amp; sudo apt-get install -y libopenal-dev\r\n<\/code><\/pre>\n<pre class=\"post-pre\"><code>$ conda config --set always_yes yes \\\r\n&amp;&amp; conda create python=3.6 --name marlo \\\r\n&amp;&amp; conda config --add channels conda-forge \\\r\n&amp;&amp; conda activate marlo \\\r\n&amp;&amp; conda install -c crowdai malmo matplotlib ipython numpy scipy opencv \\\r\n&amp;&amp; pip install git+https:\/\/github.com\/crowdAI\/marLo.git\r\n<\/code><\/pre>\n<p>\u6839\u636e\u5f53\u524d\u5b89\u88c5\u7684CUDA\u7248\u672c\uff0c\u5b89\u88c5\u76f8\u5e94\u7684CuPy\u548cChainer\u3002\u4f8b\u5982\uff0c\u5bf9\u4e8eCUDA 9.0\uff0c\u5b89\u88c5CuPy-cuda90\u3002<\/p>\n<pre class=\"post-pre\"><code>$ nvcc --version\r\n<\/code><\/pre>\n<pre class=\"post-pre\"><code>$ pip install chainer==5.1.0 cupy-cuda90==5.1.0 chainerrl==0.5.0\r\n<\/code><\/pre>\n<h3>8. \u901a\u8fc7Docker\u542f\u52a8Minecraft\u5ba2\u6237\u7aef\u3002<\/p>\n<pre class=\"post-pre\"><code>$ sudo docker pull ikeyasu\/marlo:latest\r\n$ VNC_PW=[\u9069\u5f53\u306apassword]\r\n$ sudo docker run -it --net host --rm --name robo6082 -d -p 6080:6080 -p 10000:10000 -e VNC_PW=${VNC_PW} ikeyasu\/marlo:latest\r\n<\/code><\/pre>\n<p>\u8bf7\u8bbf\u95ee\u60a8VM\u7684IP\u5730\u5740\uff1a6080\uff0c\u5e76\u5c1d\u8bd5\u8f93\u5165\u60a8\u81ea\u5df1\u8bbe\u7f6e\u7684\u5bc6\u7801${VNC_PW}\u3002<\/p>\n<p>\u901a\u8fc7VNC\u8fde\u63a5\u53ef\u4ee5\u8fde\u63a5\u5230\u8fdc\u7a0b\u73af\u5883\uff0c\u5e76\u5728\u5176\u4e2d\u542f\u52a8Minecraft\uff0c\u60a8\u5e94\u8be5\u53ef\u4ee5\u786e\u8ba4\u8fd9\u4e00\u70b9\uff01<\/p>\n<div><img decoding=\"async\" class=\"post-images\" title=\"\" src=\"https:\/\/cdn.silicloud.com\/blog-img\/blog\/img\/657cf8cd37434c4406bb6cf3\/59-0.png\" alt=\"image.png\" \/><\/p>\n<h2>\u8ba9\u6211\u4eec\u5f00\u59cb\u52a8\u624b\u5b9e\u8df5\u5427<\/p>\n<h3>0. \u786e\u8ba4\u662f\u5426\u6fc0\u6d3b\u4e86Conda\u73af\u5883\u3002<\/p>\n<pre class=\"post-pre\"><code>$ conda info -e\r\n# conda environments:\r\n#\r\nbase                     \/anaconda\r\nmarlo                 *  \/anaconda\/envs\/marlo\r\npy35                     \/anaconda\/envs\/py35\r\npy36                     \/anaconda\/envs\/py36\r\n<\/code><\/pre>\n<p>\u5982\u679c\u4e4b\u524d\u521b\u5efa\u7684marlo\u5c1a\u672a\u88ab\u6fc0\u6d3b\uff0c\u8bf7\u6267\u884c\u4e0b\u9762\u7684\u547d\u4ee4\u3002<\/p>\n<pre class=\"post-pre\"><code>$ conda activate marlo\r\n<\/code><\/pre>\n<h3>1. \u4eb2\u81ea\u64cd\u4f5c\uff0c\u514b\u9686\u4ee3\u7801\u5e93\u3002<\/p>\n<pre class=\"post-pre\"><code>$ git clone https:\/\/github.com\/keisuke-umezawa\/marlo-handson.git\r\n$ cd marlo-handson\r\n<\/code><\/pre>\n<h3>2. \u8fd0\u884cmalro\u6d4b\u8bd5\u811a\u672c<\/p>\n<pre class=\"post-pre\"><code>$ python test_malmo.py\r\n<\/code><\/pre>\n<p>\u7a0d\u7b49\u4e00\u4e0b\uff0c\u63a5\u4e0b\u6765\u5e94\u8be5\u4f1a\u663e\u793a\u4ee5\u4e0b\u753b\u9762\u3002<\/p>\n<div><img decoding=\"async\" class=\"post-images\" title=\"\" src=\"https:\/\/cdn.silicloud.com\/blog-img\/blog\/img\/657cf8cd37434c4406bb6cf3\/70-0.png\" alt=\"\u30b9\u30af\u30ea\u30fc\u30f3\u30b7\u30e7\u30c3\u30c8 2018-12-13 21.20.41.png\" \/><br \/>\n\u6211\u5c06\u6267\u884c\u4ee5\u4e0b\u7684Python\u811a\u672c\u3002<\/p>\n<p>\u6d4b\u8bd5_malmo.py<\/p>\n<pre class=\"post-pre\"><code><span class=\"kn\">import<\/span> <span class=\"nn\">marlo<\/span>\r\n\r\n\r\n<span class=\"k\">def<\/span> <span class=\"nf\">make_env<\/span><span class=\"p\">(<\/span><span class=\"n\">env_seed<\/span><span class=\"o\">=<\/span><span class=\"mi\">0<\/span><span class=\"p\">):<\/span>\r\n    <span class=\"n\">join_tokens<\/span> <span class=\"o\">=<\/span> <span class=\"n\">marlo<\/span><span class=\"p\">.<\/span><span class=\"n\">make<\/span><span class=\"p\">(<\/span>\r\n        <span class=\"s\">\"MarLo-FindTheGoal-v0\"<\/span><span class=\"p\">,<\/span>\r\n        <span class=\"n\">params<\/span><span class=\"o\">=<\/span><span class=\"nb\">dict<\/span><span class=\"p\">(<\/span>\r\n            <span class=\"n\">allowContinuousMovement<\/span><span class=\"o\">=<\/span><span class=\"p\">[<\/span><span class=\"s\">\"move\"<\/span><span class=\"p\">,<\/span> <span class=\"s\">\"turn\"<\/span><span class=\"p\">],<\/span>\r\n            <span class=\"n\">videoResolution<\/span><span class=\"o\">=<\/span><span class=\"p\">[<\/span><span class=\"mi\">336<\/span><span class=\"p\">,<\/span> <span class=\"mi\">336<\/span><span class=\"p\">],<\/span>\r\n            <span class=\"n\">kill_clients_after_num_rounds<\/span><span class=\"o\">=<\/span><span class=\"mi\">500<\/span>\r\n        <span class=\"p\">))<\/span>\r\n    <span class=\"n\">env<\/span> <span class=\"o\">=<\/span> <span class=\"n\">marlo<\/span><span class=\"p\">.<\/span><span class=\"n\">init<\/span><span class=\"p\">(<\/span><span class=\"n\">join_tokens<\/span><span class=\"p\">[<\/span><span class=\"mi\">0<\/span><span class=\"p\">])<\/span>\r\n\r\n    <span class=\"n\">obs<\/span> <span class=\"o\">=<\/span> <span class=\"n\">env<\/span><span class=\"p\">.<\/span><span class=\"n\">reset<\/span><span class=\"p\">()<\/span>\r\n    <span class=\"n\">action<\/span> <span class=\"o\">=<\/span> <span class=\"n\">env<\/span><span class=\"p\">.<\/span><span class=\"n\">action_space<\/span><span class=\"p\">.<\/span><span class=\"n\">sample<\/span><span class=\"p\">()<\/span>\r\n    <span class=\"n\">obs<\/span><span class=\"p\">,<\/span> <span class=\"n\">r<\/span><span class=\"p\">,<\/span> <span class=\"n\">done<\/span><span class=\"p\">,<\/span> <span class=\"n\">info<\/span> <span class=\"o\">=<\/span> <span class=\"n\">env<\/span><span class=\"p\">.<\/span><span class=\"n\">step<\/span><span class=\"p\">(<\/span><span class=\"n\">action<\/span><span class=\"p\">)<\/span>\r\n    <span class=\"n\">env<\/span><span class=\"p\">.<\/span><span class=\"n\">seed<\/span><span class=\"p\">(<\/span><span class=\"nb\">int<\/span><span class=\"p\">(<\/span><span class=\"n\">env_seed<\/span><span class=\"p\">))<\/span>\r\n    <span class=\"k\">return<\/span> <span class=\"n\">env<\/span>\r\n\r\n\r\n<span class=\"n\">env<\/span> <span class=\"o\">=<\/span> <span class=\"n\">make_env<\/span><span class=\"p\">()<\/span>\r\n<span class=\"n\">obs<\/span> <span class=\"o\">=<\/span> <span class=\"n\">env<\/span><span class=\"p\">.<\/span><span class=\"n\">reset<\/span><span class=\"p\">()<\/span>\r\n\r\n\r\n<span class=\"k\">for<\/span> <span class=\"n\">i<\/span> <span class=\"ow\">in<\/span> <span class=\"nb\">range<\/span><span class=\"p\">(<\/span><span class=\"mi\">10<\/span><span class=\"p\">):<\/span>\r\n    <span class=\"n\">action<\/span> <span class=\"o\">=<\/span> <span class=\"n\">env<\/span><span class=\"p\">.<\/span><span class=\"n\">action_space<\/span><span class=\"p\">.<\/span><span class=\"n\">sample<\/span><span class=\"p\">()<\/span>\r\n    <span class=\"n\">obs<\/span><span class=\"p\">,<\/span> <span class=\"n\">r<\/span><span class=\"p\">,<\/span> <span class=\"n\">done<\/span><span class=\"p\">,<\/span> <span class=\"n\">info<\/span> <span class=\"o\">=<\/span> <span class=\"n\">env<\/span><span class=\"p\">.<\/span><span class=\"n\">step<\/span><span class=\"p\">(<\/span><span class=\"n\">action<\/span><span class=\"p\">)<\/span>\r\n    <span class=\"k\">print<\/span><span class=\"p\">(<\/span><span class=\"n\">r<\/span><span class=\"p\">,<\/span> <span class=\"n\">done<\/span><span class=\"p\">,<\/span> <span class=\"n\">info<\/span><span class=\"p\">)<\/span>\r\n<\/code><\/pre>\n<p>&#8220;MarLo-FindTheGoal-v0&#8243;\u3092\u5909\u66f4\u3059\u308b\u3068\u74b0\u5883\u3092\u5909\u66f4\u3067\u304d\u307e\u3059<\/p>\n<p>obs\u306fnumpy\u5f62\u5f0f\u306e\u753b\u50cf\u30c7\u30fc\u30bf\u3067\u3059\u3002<\/p>\n<p>r\u306f\u524d\u56de\u306e\u884c\u52d5\u306b\u5bfe\u3059\u308b\u5831\u916c\u3067\u3059<\/p>\n<p>done\u306f\u30b2\u30fc\u30e0\u304c\u7d42\u4e86\u3057\u305f\u304b\u771f\u507d\u5024\u3067\u3059\u3002<\/p>\n<p>info\u306f\u305d\u306e\u4ed6\u60c5\u5831\u304c\u5165\u3063\u3066\u3044\u307e\u3059\u3002<\/p>\n<h4>\u8bfe\u98981\uff1a\u8bf7\u786e\u8ba4\u5e76\u5bf9\u4e0a\u8ff0\u5185\u5bb9\u8fdb\u884c\u4fee\u6539\uff0c\u5e76\u5c1d\u8bd5\u73a9\u800d\u3002<\/p>\n<h3>3. \u8fd0\u884c\u4f7f\u7528ChainerRL\u8fdb\u884cDQN\u8bad\u7ec3\u7684\u811a\u672c\u3002\u8bf7\u5148\u6267\u884c\u4ee5\u4e0b\u547d\u4ee4\u3002\u8fd9\u6837\uff0c\u60a8\u5c31\u53ef\u4ee5\u4ece\u96f6\u5f00\u59cb\u4f7f\u7528ChainerRL\u8bad\u7ec3DQN\u5f3a\u5316\u5b66\u4e60\u6a21\u578b\u4e86\u3002<\/p>\n<pre class=\"post-pre\"><code>$ python train_DQN.py\r\n<\/code><\/pre>\n<p>\u8bf7\u5728\u9002\u5f53\u7684\u65f6\u5019\u6309\u4e0bCTRL+C\uff0c\u505c\u6b62\u811a\u672c\u3002\u8fd9\u6837\u5c31\u5e94\u8be5\u521b\u5efa\u4e86\u4ee5\u4e0b\u76ee\u5f55\uff1axxxx_except\uff0c\u5176\u4e2d\u4fdd\u5b58\u4e86\u6211\u4eec\u8fc4\u4eca\u4e3a\u6b62\u5b66\u4e60\u7684\u6a21\u578b\u3002<\/p>\n<pre class=\"post-pre\"><code>$ ls results\/\r\n3765_except  scores.txt\r\n<\/code><\/pre>\n<h4>Note\u5982\u679c\u60a8\u53ea\u60f3\u4f7f\u7528CPU\u8fd0\u884c\uff0c\u8bf7\u5c1d\u8bd5\u6dfb\u52a0\u4ee5\u4e0b\u9009\u9879\u5e76\u8fd0\u884c\u3002<\/p>\n<pre class=\"post-pre\"><code>$ python train_DQN.py --gpu -1\r\n<\/code><\/pre>\n<h3>3. \u9a8c\u8bc1\u4fdd\u5b58\u7684\u6a21\u578b\u662f\u5426\u6b63\u5e38\u8fd0\u884c\u3002\u4ee5\u4e0b\u306e\u30b3\u30de\u30f3\u30c9\u3067\u3001\u5b66\u7fd2\u3057\u305f\u30e2\u30c7\u30eb\u306e\u30ed\u30fc\u30c9\u3068\u3001\u305d\u306e\u52d5\u4f5c\u78ba\u8a8d\u304c\u3067\u304d\u307e\u3059\u3002<\/p>\n<pre class=\"post-pre\"><code>$ python train_DQN.py --load results\/3765_except --demo\r\n<\/code><\/pre>\n<p>\u8bf7\u95ee\u600e\u4e48\u6837\u5462\uff1f\u6211\u5e76\u6ca1\u6709\u8ba9\u5b83\u8fdb\u884c\u8fc7\u5f88\u591a\u5b66\u4e60\uff0c\u6240\u4ee5\u5e94\u8be5\u4e0d\u80fd\u6b63\u5e38\u8fd0\u884c\u3002<\/p>\n<h3>\u4ece\u4fdd\u5b58\u7684\u6a21\u578b\u5f00\u59cb\u8bad\u7ec3\u3002<br \/>\n\u53ea\u9700\u8981\u4e00\u4e2a\u9009\u9879\uff0c\u5bf9\u4ee5\u4e0b\u547d\u4ee4\u8fdb\u884c\u6c49\u8bed\u7684\u540c\u4e49\u8f6c\u8ff0\uff1a<br \/>\n\u901a\u8fc7\u4f7f\u7528\u4ee5\u4e0b\u547d\u4ee4\uff0c\u60a8\u53ef\u4ee5\u4ece\u4e4b\u524d\u4fdd\u5b58\u7684\u6a21\u578b\u4e2d\u6062\u590d\u5b66\u4e60\u3002<\/p>\n<pre class=\"post-pre\"><code>$ python train_DQN.py --load results\/3765_except\r\n<\/code><\/pre>\n<p>\u307e\u305f\u3001\u3053\u3061\u3089\u3067\u3042\u308b\u7a0b\u5ea6\u8a13\u7df4\u3057\u305f\u30e2\u30c7\u30eb\u3092\u7528\u610f\u3057\u305f\u306e\u3067\u3001\u305d\u3053\u304b\u3089\u5b66\u7fd2\u3092\u958b\u59cb\u3059\u308b\u3053\u3068\u3082\u3067\u304d\u307e\u3059\u3002\u305f\u3060\u3057\u3001\u5358\u306bpython train_DQN.py\u3092\u5b9f\u884c\u3057\u3066\u4f5c\u6210\u3055\u308c\u305f\u30e2\u30c7\u30eb\u3067\u3059\u306e\u3067\u3001\u30b3\u30fc\u30c9\u306b\u5909\u66f4\u3092\u52a0\u3048\u305f\u308a\u3001\u74b0\u5883\u3092\u5909\u66f4\u3057\u305f\u308a\u3057\u305f\u5834\u5408\u52d5\u304b\u306a\u3044\u3053\u3068\u304c\u3042\u308a\u307e\u3059\u3002<\/p>\n<pre class=\"post-pre\"><code>$ wget https:\/\/github.com\/keisuke-umezawa\/marlo-handson\/releases\/download\/v0.2\/157850_except.tar.gz\r\n$ tar -xvzf 157850_except.tar.gz\r\n$ python train_DQN.py --load 157850_except\r\n<\/code><\/pre>\n<h4>\u5b9e\u73b0\u4ee5\u4e0bparaphrase\uff1a\u9898\u76ee2\uff1a\u8ba9\u6211\u4eec\u6765\u9605\u8bfbtrain_DQN.py\u7684\u4ee3\u7801\u3002<\/p>\n<p>\u8fd9\u4e2a\u811a\u672c\u672c\u8eab\u5982\u4e0b\u6240\u793a\u3002<\/p>\n<p>\u8bad\u7ec3_DQN.py<\/p>\n<pre class=\"post-pre\"><code>\r\n<span class=\"c1\"># ...\u7701\u7565...\r\n<\/span>\r\n<span class=\"k\">def<\/span> <span class=\"nf\">main<\/span><span class=\"p\">():<\/span>\r\n    <span class=\"n\">parser<\/span> <span class=\"o\">=<\/span> <span class=\"n\">argparse<\/span><span class=\"p\">.<\/span><span class=\"n\">ArgumentParser<\/span><span class=\"p\">()<\/span>\r\n\r\n    <span class=\"c1\"># ...\u7701\u7565...\r\n<\/span>\r\n    <span class=\"c1\"># Set a random seed used in ChainerRL.\r\n<\/span>    <span class=\"n\">misc<\/span><span class=\"p\">.<\/span><span class=\"n\">set_random_seed<\/span><span class=\"p\">(<\/span><span class=\"n\">args<\/span><span class=\"p\">.<\/span><span class=\"n\">seed<\/span><span class=\"p\">,<\/span> <span class=\"n\">gpus<\/span><span class=\"o\">=<\/span><span class=\"p\">(<\/span><span class=\"n\">args<\/span><span class=\"p\">.<\/span><span class=\"n\">gpu<\/span><span class=\"p\">,))<\/span>\r\n\r\n    <span class=\"k\">if<\/span> <span class=\"ow\">not<\/span> <span class=\"n\">os<\/span><span class=\"p\">.<\/span><span class=\"n\">path<\/span><span class=\"p\">.<\/span><span class=\"n\">exists<\/span><span class=\"p\">(<\/span><span class=\"n\">args<\/span><span class=\"p\">.<\/span><span class=\"n\">out_dir<\/span><span class=\"p\">):<\/span>\r\n        <span class=\"n\">os<\/span><span class=\"p\">.<\/span><span class=\"n\">makedirs<\/span><span class=\"p\">(<\/span><span class=\"n\">args<\/span><span class=\"p\">.<\/span><span class=\"n\">out_dir<\/span><span class=\"p\">)<\/span>\r\n\r\n    <span class=\"n\">experiments<\/span><span class=\"p\">.<\/span><span class=\"n\">set_log_base_dir<\/span><span class=\"p\">(<\/span><span class=\"n\">args<\/span><span class=\"p\">.<\/span><span class=\"n\">out_dir<\/span><span class=\"p\">)<\/span>\r\n    <span class=\"k\">print<\/span><span class=\"p\">(<\/span><span class=\"s\">'Output files are saved in {}'<\/span><span class=\"p\">.<\/span><span class=\"nb\">format<\/span><span class=\"p\">(<\/span><span class=\"n\">args<\/span><span class=\"p\">.<\/span><span class=\"n\">out_dir<\/span><span class=\"p\">))<\/span>\r\n\r\n    <span class=\"n\">env<\/span> <span class=\"o\">=<\/span> <span class=\"n\">make_env<\/span><span class=\"p\">(<\/span><span class=\"n\">env_seed<\/span><span class=\"o\">=<\/span><span class=\"n\">args<\/span><span class=\"p\">.<\/span><span class=\"n\">seed<\/span><span class=\"p\">)<\/span>\r\n\r\n    <span class=\"n\">n_actions<\/span> <span class=\"o\">=<\/span> <span class=\"n\">env<\/span><span class=\"p\">.<\/span><span class=\"n\">action_space<\/span><span class=\"p\">.<\/span><span class=\"n\">n<\/span>\r\n\r\n    <span class=\"n\">q_func<\/span> <span class=\"o\">=<\/span> <span class=\"n\">links<\/span><span class=\"p\">.<\/span><span class=\"n\">Sequence<\/span><span class=\"p\">(<\/span>\r\n        <span class=\"n\">links<\/span><span class=\"p\">.<\/span><span class=\"n\">NatureDQNHead<\/span><span class=\"p\">(<\/span><span class=\"n\">n_input_channels<\/span><span class=\"o\">=<\/span><span class=\"mi\">3<\/span><span class=\"p\">),<\/span>\r\n        <span class=\"n\">L<\/span><span class=\"p\">.<\/span><span class=\"n\">Linear<\/span><span class=\"p\">(<\/span><span class=\"mi\">512<\/span><span class=\"p\">,<\/span> <span class=\"n\">n_actions<\/span><span class=\"p\">),<\/span>\r\n        <span class=\"n\">DiscreteActionValue<\/span>\r\n    <span class=\"p\">)<\/span>\r\n\r\n    <span class=\"c1\"># Use the same hyper parameters as the Nature paper's\r\n<\/span>    <span class=\"n\">opt<\/span> <span class=\"o\">=<\/span> <span class=\"n\">optimizers<\/span><span class=\"p\">.<\/span><span class=\"n\">RMSpropGraves<\/span><span class=\"p\">(<\/span>\r\n        <span class=\"n\">lr<\/span><span class=\"o\">=<\/span><span class=\"n\">args<\/span><span class=\"p\">.<\/span><span class=\"n\">lr<\/span><span class=\"p\">,<\/span> <span class=\"n\">alpha<\/span><span class=\"o\">=<\/span><span class=\"mf\">0.95<\/span><span class=\"p\">,<\/span> <span class=\"n\">momentum<\/span><span class=\"o\">=<\/span><span class=\"mf\">0.0<\/span><span class=\"p\">,<\/span> <span class=\"n\">eps<\/span><span class=\"o\">=<\/span><span class=\"mf\">1e-2<\/span><span class=\"p\">)<\/span>\r\n\r\n    <span class=\"n\">opt<\/span><span class=\"p\">.<\/span><span class=\"n\">setup<\/span><span class=\"p\">(<\/span><span class=\"n\">q_func<\/span><span class=\"p\">)<\/span>\r\n\r\n    <span class=\"n\">rbuf<\/span> <span class=\"o\">=<\/span> <span class=\"n\">replay_buffer<\/span><span class=\"p\">.<\/span><span class=\"n\">ReplayBuffer<\/span><span class=\"p\">(<\/span><span class=\"mi\">10<\/span> <span class=\"o\">**<\/span> <span class=\"mi\">6<\/span><span class=\"p\">)<\/span>\r\n\r\n    <span class=\"n\">explorer<\/span> <span class=\"o\">=<\/span> <span class=\"n\">explorers<\/span><span class=\"p\">.<\/span><span class=\"n\">LinearDecayEpsilonGreedy<\/span><span class=\"p\">(<\/span>\r\n        <span class=\"mf\">1.0<\/span><span class=\"p\">,<\/span> <span class=\"n\">args<\/span><span class=\"p\">.<\/span><span class=\"n\">final_epsilon<\/span><span class=\"p\">,<\/span>\r\n        <span class=\"n\">args<\/span><span class=\"p\">.<\/span><span class=\"n\">final_exploration_frames<\/span><span class=\"p\">,<\/span>\r\n        <span class=\"k\">lambda<\/span><span class=\"p\">:<\/span> <span class=\"n\">np<\/span><span class=\"p\">.<\/span><span class=\"n\">random<\/span><span class=\"p\">.<\/span><span class=\"n\">randint<\/span><span class=\"p\">(<\/span><span class=\"n\">n_actions<\/span><span class=\"p\">))<\/span>\r\n\r\n    <span class=\"k\">def<\/span> <span class=\"nf\">phi<\/span><span class=\"p\">(<\/span><span class=\"n\">x<\/span><span class=\"p\">):<\/span>\r\n        <span class=\"c1\"># Feature extractor\r\n<\/span>        <span class=\"n\">x<\/span> <span class=\"o\">=<\/span> <span class=\"n\">x<\/span><span class=\"p\">.<\/span><span class=\"n\">transpose<\/span><span class=\"p\">(<\/span><span class=\"mi\">2<\/span><span class=\"p\">,<\/span> <span class=\"mi\">0<\/span><span class=\"p\">,<\/span> <span class=\"mi\">1<\/span><span class=\"p\">)<\/span>\r\n        <span class=\"k\">return<\/span> <span class=\"n\">np<\/span><span class=\"p\">.<\/span><span class=\"n\">asarray<\/span><span class=\"p\">(<\/span><span class=\"n\">x<\/span><span class=\"p\">,<\/span> <span class=\"n\">dtype<\/span><span class=\"o\">=<\/span><span class=\"n\">np<\/span><span class=\"p\">.<\/span><span class=\"n\">float32<\/span><span class=\"p\">)<\/span> <span class=\"o\">\/<\/span> <span class=\"mi\">255<\/span>\r\n\r\n    <span class=\"n\">agent<\/span> <span class=\"o\">=<\/span> <span class=\"n\">agents<\/span><span class=\"p\">.<\/span><span class=\"n\">DQN<\/span><span class=\"p\">(<\/span>\r\n        <span class=\"n\">q_func<\/span><span class=\"p\">,<\/span>\r\n        <span class=\"n\">opt<\/span><span class=\"p\">,<\/span>\r\n        <span class=\"n\">rbuf<\/span><span class=\"p\">,<\/span>\r\n        <span class=\"n\">gpu<\/span><span class=\"o\">=<\/span><span class=\"n\">args<\/span><span class=\"p\">.<\/span><span class=\"n\">gpu<\/span><span class=\"p\">,<\/span>\r\n        <span class=\"n\">gamma<\/span><span class=\"o\">=<\/span><span class=\"mf\">0.99<\/span><span class=\"p\">,<\/span>\r\n        <span class=\"n\">explorer<\/span><span class=\"o\">=<\/span><span class=\"n\">explorer<\/span><span class=\"p\">,<\/span>\r\n        <span class=\"n\">replay_start_size<\/span><span class=\"o\">=<\/span><span class=\"n\">args<\/span><span class=\"p\">.<\/span><span class=\"n\">replay_start_size<\/span><span class=\"p\">,<\/span>\r\n        <span class=\"n\">target_update_interval<\/span><span class=\"o\">=<\/span><span class=\"n\">args<\/span><span class=\"p\">.<\/span><span class=\"n\">target_update_interval<\/span><span class=\"p\">,<\/span>\r\n        <span class=\"n\">update_interval<\/span><span class=\"o\">=<\/span><span class=\"n\">args<\/span><span class=\"p\">.<\/span><span class=\"n\">update_interval<\/span><span class=\"p\">,<\/span>\r\n        <span class=\"n\">batch_accumulator<\/span><span class=\"o\">=<\/span><span class=\"s\">'sum'<\/span><span class=\"p\">,<\/span>\r\n        <span class=\"n\">phi<\/span><span class=\"o\">=<\/span><span class=\"n\">phi<\/span>\r\n    <span class=\"p\">)<\/span>\r\n\r\n    <span class=\"k\">if<\/span> <span class=\"n\">args<\/span><span class=\"p\">.<\/span><span class=\"n\">load<\/span><span class=\"p\">:<\/span>\r\n        <span class=\"n\">agent<\/span><span class=\"p\">.<\/span><span class=\"n\">load<\/span><span class=\"p\">(<\/span><span class=\"n\">args<\/span><span class=\"p\">.<\/span><span class=\"n\">load<\/span><span class=\"p\">)<\/span>\r\n\r\n    <span class=\"k\">if<\/span> <span class=\"n\">args<\/span><span class=\"p\">.<\/span><span class=\"n\">demo<\/span><span class=\"p\">:<\/span>\r\n        <span class=\"n\">eval_stats<\/span> <span class=\"o\">=<\/span> <span class=\"n\">experiments<\/span><span class=\"p\">.<\/span><span class=\"n\">eval_performance<\/span><span class=\"p\">(<\/span>\r\n            <span class=\"n\">env<\/span><span class=\"o\">=<\/span><span class=\"n\">env<\/span><span class=\"p\">,<\/span>\r\n            <span class=\"n\">agent<\/span><span class=\"o\">=<\/span><span class=\"n\">agent<\/span><span class=\"p\">,<\/span>\r\n            <span class=\"n\">n_runs<\/span><span class=\"o\">=<\/span><span class=\"n\">args<\/span><span class=\"p\">.<\/span><span class=\"n\">eval_n_runs<\/span><span class=\"p\">)<\/span>\r\n        <span class=\"k\">print<\/span><span class=\"p\">(<\/span><span class=\"s\">'n_runs: {} mean: {} median: {} stdev {}'<\/span><span class=\"p\">.<\/span><span class=\"nb\">format<\/span><span class=\"p\">(<\/span>\r\n            <span class=\"n\">args<\/span><span class=\"p\">.<\/span><span class=\"n\">eval_n_runs<\/span><span class=\"p\">,<\/span> <span class=\"n\">eval_stats<\/span><span class=\"p\">[<\/span><span class=\"s\">'mean'<\/span><span class=\"p\">],<\/span> <span class=\"n\">eval_stats<\/span><span class=\"p\">[<\/span><span class=\"s\">'median'<\/span><span class=\"p\">],<\/span>\r\n            <span class=\"n\">eval_stats<\/span><span class=\"p\">[<\/span><span class=\"s\">'stdev'<\/span><span class=\"p\">]))<\/span>\r\n    <span class=\"k\">else<\/span><span class=\"p\">:<\/span>\r\n        <span class=\"n\">experiments<\/span><span class=\"p\">.<\/span><span class=\"n\">train_agent_with_evaluation<\/span><span class=\"p\">(<\/span>\r\n            <span class=\"n\">agent<\/span><span class=\"o\">=<\/span><span class=\"n\">agent<\/span><span class=\"p\">,<\/span>\r\n            <span class=\"n\">env<\/span><span class=\"o\">=<\/span><span class=\"n\">env<\/span><span class=\"p\">,<\/span>\r\n            <span class=\"n\">steps<\/span><span class=\"o\">=<\/span><span class=\"n\">args<\/span><span class=\"p\">.<\/span><span class=\"n\">steps<\/span><span class=\"p\">,<\/span>\r\n            <span class=\"n\">eval_n_runs<\/span><span class=\"o\">=<\/span><span class=\"n\">args<\/span><span class=\"p\">.<\/span><span class=\"n\">eval_n_runs<\/span><span class=\"p\">,<\/span>\r\n            <span class=\"n\">eval_interval<\/span><span class=\"o\">=<\/span><span class=\"n\">args<\/span><span class=\"p\">.<\/span><span class=\"n\">eval_interval<\/span><span class=\"p\">,<\/span>\r\n            <span class=\"n\">outdir<\/span><span class=\"o\">=<\/span><span class=\"n\">args<\/span><span class=\"p\">.<\/span><span class=\"n\">out_dir<\/span><span class=\"p\">,<\/span>\r\n            <span class=\"n\">save_best_so_far_agent<\/span><span class=\"o\">=<\/span><span class=\"bp\">False<\/span><span class=\"p\">,<\/span>\r\n            <span class=\"n\">max_episode_len<\/span><span class=\"o\">=<\/span><span class=\"n\">args<\/span><span class=\"p\">.<\/span><span class=\"n\">max_episode_len<\/span><span class=\"p\">,<\/span>\r\n            <span class=\"n\">eval_env<\/span><span class=\"o\">=<\/span><span class=\"n\">env<\/span><span class=\"p\">,<\/span>\r\n        <span class=\"p\">)<\/span>\r\n\r\n\r\n<span class=\"k\">if<\/span> <span class=\"n\">__name__<\/span> <span class=\"o\">==<\/span> <span class=\"s\">'__main__'<\/span><span class=\"p\">:<\/span>\r\n    <span class=\"n\">main<\/span><span class=\"p\">()<\/span>\r\n<\/code><\/pre>\n<ul class=\"post-ul\">\n<li style=\"list-style-type: none;\">\n<ul class=\"post-ul\">\u53c2\u8003\u30ea\u30f3\u30af<\/ul>\n<\/li>\n<\/ul>\n<p>NatureDQNHead<br \/>\nDiscreteActionValue<br \/>\nReplayBuffer<br \/>\nLinearDecayEpsilonGreedy<br \/>\nDQN<\/p>\n<h4>\u95ee\u98983\uff1a\u63d0\u9ad8\u6027\u80fd<br \/>\n\u8981\u6539\u8fdb\u5f3a\u5316\u5b66\u4e60\u6027\u80fd\uff0c\u6709\u4ee5\u4e0b\u65b9\u5f0f\u53ef\u4f9b\u9009\u62e9\u3002<\/p>\n<ul class=\"post-ul\">\n<li style=\"list-style-type: none;\">\n<ul class=\"post-ul\">\u30e2\u30c7\u30eb\u306e\u5909\u66f4<\/ul>\n<\/li>\n<\/ul>\n<p>&nbsp;<\/p>\n<ul class=\"post-ul\">\n<li style=\"list-style-type: none;\">\n<ul class=\"post-ul\">ReplayBuffer\u306e\u5909\u66f4<\/ul>\n<\/li>\n<\/ul>\n<p>&nbsp;<\/p>\n<ul class=\"post-ul\">\u305d\u306e\u305f\u30d1\u30e9\u30e1\u30fc\u30bf\u306e\u5909\u66f4<\/ul>\n<p>\u53e6\u5916\uff0c\u8fd8\u6709\u4e00\u7bc7\u540d\u4e3a\u300aRainbow\uff1a\u7ed3\u5408\u6df1\u5ea6\u5f3a\u5316\u5b66\u4e60\u7684\u6539\u8fdb\u65b9\u6cd5\u6765\u8bc4\u4f30\u6027\u80fd\u63d0\u5347\u300b\u7684\u8bba\u6587\uff0c\u5c1d\u8bd5\u4e86\u5404\u79cd\u5f3a\u5316\u5b66\u4e60\u65b9\u6cd5\uff0c\u5e76\u8bc4\u4f30\u4e86\u5b83\u4eec\u5728\u6027\u80fd\u4e0a\u7684\u63d0\u5347\u3002\u6839\u636e\u8be5\u8bba\u6587\uff0c\u53ef\u4ee5\u5c1d\u8bd5\u4ee5\u4e0b\u65b9\u5f0f\u6765\u63d0\u9ad8\u6027\u80fd\u3002<\/p>\n<ul class=\"post-ul\">\n<li style=\"list-style-type: none;\">\n<ul class=\"post-ul\">PrioritizedReplayBuffer\u306e\u4f7f\u7528<\/ul>\n<\/li>\n<\/ul>\n<p>&nbsp;<\/p>\n<ul class=\"post-ul\">\n<li style=\"list-style-type: none;\">\n<ul class=\"post-ul\">DDQN\u306e\u4f7f\u7528<\/ul>\n<\/li>\n<\/ul>\n<p>&nbsp;<\/p>\n<ul class=\"post-ul\">etc<\/ul>\n","protected":false},"excerpt":{"rendered":"<p>\u6ce8\u610f \u73fe\u5728\u306f\u3053\u3061\u3089\u306e\u65b9\u6cd5\u306e\u65b9\u304c\u3088\u308a\u7c21\u5358\u306b\u74b0\u5883\u69cb\u7bc9\u3067\u304d\u307e\u3059 ChainerRL + Minecraft\u3067\u6df1\u5c64\u5f37\u5316 [&hellip;]<\/p>\n","protected":false},"author":4,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[227],"class_list":["post-27813","post","type-post","status-publish","format-standard","hentry","category-uncategorized","tag-227"],"yoast_head":"<!-- This site is optimized with the Yoast SEO Premium plugin v21.5 (Yoast SEO v21.5) - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>\u4f7f\u7528ChainerRL\uff0c\u5728Minecraft\u4e2d\u8fdb\u884c\u6df1\u5ea6\u5f3a\u5316\u5b66\u4e60\u7684\u5b9e\u8df5 - Blog - Silicon Cloud<\/title>\n<meta name=\"description\" content=\"\u5173\u4e8e\u4f7f\u7528ChainerRL\uff0c\u5728Minecraft\u4e2d\u8fdb\u884c\u6df1\u5ea6\u5f3a\u5316\u5b66\u4e60\u7684\u5b9e\u8df5\u7684\u6280\u672f\u6587\u7ae0\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/www.silicloud.com\/zh\/blog\/\u4f7f\u7528chainerrl\uff0c\u5728minecraft\u4e2d\u8fdb\u884c\u6df1\u5ea6\u5f3a\u5316\u5b66\u4e60\u7684\u5b9e\u8df5\u3002\/\" \/>\n<meta property=\"og:locale\" content=\"zh_CN\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"\u4f7f\u7528ChainerRL\uff0c\u5728Minecraft\u4e2d\u8fdb\u884c\u6df1\u5ea6\u5f3a\u5316\u5b66\u4e60\u7684\u5b9e\u8df5\" \/>\n<meta property=\"og:description\" content=\"\u5173\u4e8e\u4f7f\u7528ChainerRL\uff0c\u5728Minecraft\u4e2d\u8fdb\u884c\u6df1\u5ea6\u5f3a\u5316\u5b66\u4e60\u7684\u5b9e\u8df5\u7684\u6280\u672f\u6587\u7ae0\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.silicloud.com\/zh\/blog\/\u4f7f\u7528chainerrl\uff0c\u5728minecraft\u4e2d\u8fdb\u884c\u6df1\u5ea6\u5f3a\u5316\u5b66\u4e60\u7684\u5b9e\u8df5\u3002\/\" \/>\n<meta property=\"og:site_name\" content=\"Blog - Silicon Cloud\" \/>\n<meta property=\"article:published_time\" content=\"2023-09-03T14:47:52+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2025-08-11T20:46:01+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/cdn.silicloud.com\/blog-img\/blog\/img\/657cf8cd37434c4406bb6cf3\/14-10.gif\" \/>\n<meta name=\"author\" content=\"\u79d1, \u96c5\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"\u4f5c\u8005\" \/>\n\t<meta name=\"twitter:data1\" content=\"\u79d1, \u96c5\" \/>\n\t<meta name=\"twitter:label2\" content=\"\u9884\u8ba1\u9605\u8bfb\u65f6\u95f4\" \/>\n\t<meta name=\"twitter:data2\" content=\"2 \u5206\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/www.silicloud.com\/zh\/blog\/%e4%bd%bf%e7%94%a8chainerrl%ef%bc%8c%e5%9c%a8minecraft%e4%b8%ad%e8%bf%9b%e8%a1%8c%e6%b7%b1%e5%ba%a6%e5%bc%ba%e5%8c%96%e5%ad%a6%e4%b9%a0%e7%9a%84%e5%ae%9e%e8%b7%b5%e3%80%82\/\",\"url\":\"https:\/\/www.silicloud.com\/zh\/blog\/%e4%bd%bf%e7%94%a8chainerrl%ef%bc%8c%e5%9c%a8minecraft%e4%b8%ad%e8%bf%9b%e8%a1%8c%e6%b7%b1%e5%ba%a6%e5%bc%ba%e5%8c%96%e5%ad%a6%e4%b9%a0%e7%9a%84%e5%ae%9e%e8%b7%b5%e3%80%82\/\",\"name\":\"\u4f7f\u7528ChainerRL\uff0c\u5728Minecraft\u4e2d\u8fdb\u884c\u6df1\u5ea6\u5f3a\u5316\u5b66\u4e60\u7684\u5b9e\u8df5 - Blog - Silicon Cloud\",\"isPartOf\":{\"@id\":\"https:\/\/www.silicloud.com\/zh\/blog\/#website\"},\"datePublished\":\"2023-09-03T14:47:52+00:00\",\"dateModified\":\"2025-08-11T20:46:01+00:00\",\"author\":{\"@id\":\"https:\/\/www.silicloud.com\/zh\/blog\/#\/schema\/person\/41e222757cdd2a3365361328bd79970a\"},\"description\":\"\u5173\u4e8e\u4f7f\u7528ChainerRL\uff0c\u5728Minecraft\u4e2d\u8fdb\u884c\u6df1\u5ea6\u5f3a\u5316\u5b66\u4e60\u7684\u5b9e\u8df5\u7684\u6280\u672f\u6587\u7ae0\",\"breadcrumb\":{\"@id\":\"https:\/\/www.silicloud.com\/zh\/blog\/%e4%bd%bf%e7%94%a8chainerrl%ef%bc%8c%e5%9c%a8minecraft%e4%b8%ad%e8%bf%9b%e8%a1%8c%e6%b7%b1%e5%ba%a6%e5%bc%ba%e5%8c%96%e5%ad%a6%e4%b9%a0%e7%9a%84%e5%ae%9e%e8%b7%b5%e3%80%82\/#breadcrumb\"},\"inLanguage\":\"zh-Hans\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/www.silicloud.com\/zh\/blog\/%e4%bd%bf%e7%94%a8chainerrl%ef%bc%8c%e5%9c%a8minecraft%e4%b8%ad%e8%bf%9b%e8%a1%8c%e6%b7%b1%e5%ba%a6%e5%bc%ba%e5%8c%96%e5%ad%a6%e4%b9%a0%e7%9a%84%e5%ae%9e%e8%b7%b5%e3%80%82\/\"]}]},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/www.silicloud.com\/zh\/blog\/%e4%bd%bf%e7%94%a8chainerrl%ef%bc%8c%e5%9c%a8minecraft%e4%b8%ad%e8%bf%9b%e8%a1%8c%e6%b7%b1%e5%ba%a6%e5%bc%ba%e5%8c%96%e5%ad%a6%e4%b9%a0%e7%9a%84%e5%ae%9e%e8%b7%b5%e3%80%82\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"\u9996\u9875\",\"item\":\"https:\/\/www.silicloud.com\/zh\/blog\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"\u4f7f\u7528ChainerRL\uff0c\u5728Minecraft\u4e2d\u8fdb\u884c\u6df1\u5ea6\u5f3a\u5316\u5b66\u4e60\u7684\u5b9e\u8df5\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/www.silicloud.com\/zh\/blog\/#website\",\"url\":\"https:\/\/www.silicloud.com\/zh\/blog\/\",\"name\":\"Blog - Silicon Cloud\",\"description\":\"\",\"inLanguage\":\"zh-Hans\"},{\"@type\":\"Person\",\"@id\":\"https:\/\/www.silicloud.com\/zh\/blog\/#\/schema\/person\/41e222757cdd2a3365361328bd79970a\",\"name\":\"\u79d1, \u96c5\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"https:\/\/www.silicloud.com\/zh\/blog\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/secure.gravatar.com\/avatar\/1b2d3e00a7df03689797ebd4af8c5827ba5af936849a71050ec331f4cf902c5d?s=96&d=mm&r=g\",\"contentUrl\":\"https:\/\/secure.gravatar.com\/avatar\/1b2d3e00a7df03689797ebd4af8c5827ba5af936849a71050ec331f4cf902c5d?s=96&d=mm&r=g\",\"caption\":\"\u79d1, \u96c5\"},\"url\":\"https:\/\/www.silicloud.com\/zh\/blog\/author\/keya\/\"},{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"https:\/\/www.silicloud.com\/zh\/blog\/%e4%bd%bf%e7%94%a8chainerrl%ef%bc%8c%e5%9c%a8minecraft%e4%b8%ad%e8%bf%9b%e8%a1%8c%e6%b7%b1%e5%ba%a6%e5%bc%ba%e5%8c%96%e5%ad%a6%e4%b9%a0%e7%9a%84%e5%ae%9e%e8%b7%b5%e3%80%82\/#local-main-organization-logo\",\"url\":\"\",\"contentUrl\":\"\",\"caption\":\"Blog - Silicon Cloud\"}]}<\/script>\n<!-- \/ Yoast SEO Premium plugin. -->","yoast_head_json":{"title":"\u4f7f\u7528ChainerRL\uff0c\u5728Minecraft\u4e2d\u8fdb\u884c\u6df1\u5ea6\u5f3a\u5316\u5b66\u4e60\u7684\u5b9e\u8df5 - Blog - Silicon Cloud","description":"\u5173\u4e8e\u4f7f\u7528ChainerRL\uff0c\u5728Minecraft\u4e2d\u8fdb\u884c\u6df1\u5ea6\u5f3a\u5316\u5b66\u4e60\u7684\u5b9e\u8df5\u7684\u6280\u672f\u6587\u7ae0","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/www.silicloud.com\/zh\/blog\/\u4f7f\u7528chainerrl\uff0c\u5728minecraft\u4e2d\u8fdb\u884c\u6df1\u5ea6\u5f3a\u5316\u5b66\u4e60\u7684\u5b9e\u8df5\u3002\/","og_locale":"zh_CN","og_type":"article","og_title":"\u4f7f\u7528ChainerRL\uff0c\u5728Minecraft\u4e2d\u8fdb\u884c\u6df1\u5ea6\u5f3a\u5316\u5b66\u4e60\u7684\u5b9e\u8df5","og_description":"\u5173\u4e8e\u4f7f\u7528ChainerRL\uff0c\u5728Minecraft\u4e2d\u8fdb\u884c\u6df1\u5ea6\u5f3a\u5316\u5b66\u4e60\u7684\u5b9e\u8df5\u7684\u6280\u672f\u6587\u7ae0","og_url":"https:\/\/www.silicloud.com\/zh\/blog\/\u4f7f\u7528chainerrl\uff0c\u5728minecraft\u4e2d\u8fdb\u884c\u6df1\u5ea6\u5f3a\u5316\u5b66\u4e60\u7684\u5b9e\u8df5\u3002\/","og_site_name":"Blog - Silicon Cloud","article_published_time":"2023-09-03T14:47:52+00:00","article_modified_time":"2025-08-11T20:46:01+00:00","og_image":[{"url":"https:\/\/cdn.silicloud.com\/blog-img\/blog\/img\/657cf8cd37434c4406bb6cf3\/14-10.gif"}],"author":"\u79d1, \u96c5","twitter_card":"summary_large_image","twitter_misc":{"\u4f5c\u8005":"\u79d1, \u96c5","\u9884\u8ba1\u9605\u8bfb\u65f6\u95f4":"2 \u5206"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/www.silicloud.com\/zh\/blog\/%e4%bd%bf%e7%94%a8chainerrl%ef%bc%8c%e5%9c%a8minecraft%e4%b8%ad%e8%bf%9b%e8%a1%8c%e6%b7%b1%e5%ba%a6%e5%bc%ba%e5%8c%96%e5%ad%a6%e4%b9%a0%e7%9a%84%e5%ae%9e%e8%b7%b5%e3%80%82\/","url":"https:\/\/www.silicloud.com\/zh\/blog\/%e4%bd%bf%e7%94%a8chainerrl%ef%bc%8c%e5%9c%a8minecraft%e4%b8%ad%e8%bf%9b%e8%a1%8c%e6%b7%b1%e5%ba%a6%e5%bc%ba%e5%8c%96%e5%ad%a6%e4%b9%a0%e7%9a%84%e5%ae%9e%e8%b7%b5%e3%80%82\/","name":"\u4f7f\u7528ChainerRL\uff0c\u5728Minecraft\u4e2d\u8fdb\u884c\u6df1\u5ea6\u5f3a\u5316\u5b66\u4e60\u7684\u5b9e\u8df5 - Blog - Silicon Cloud","isPartOf":{"@id":"https:\/\/www.silicloud.com\/zh\/blog\/#website"},"datePublished":"2023-09-03T14:47:52+00:00","dateModified":"2025-08-11T20:46:01+00:00","author":{"@id":"https:\/\/www.silicloud.com\/zh\/blog\/#\/schema\/person\/41e222757cdd2a3365361328bd79970a"},"description":"\u5173\u4e8e\u4f7f\u7528ChainerRL\uff0c\u5728Minecraft\u4e2d\u8fdb\u884c\u6df1\u5ea6\u5f3a\u5316\u5b66\u4e60\u7684\u5b9e\u8df5\u7684\u6280\u672f\u6587\u7ae0","breadcrumb":{"@id":"https:\/\/www.silicloud.com\/zh\/blog\/%e4%bd%bf%e7%94%a8chainerrl%ef%bc%8c%e5%9c%a8minecraft%e4%b8%ad%e8%bf%9b%e8%a1%8c%e6%b7%b1%e5%ba%a6%e5%bc%ba%e5%8c%96%e5%ad%a6%e4%b9%a0%e7%9a%84%e5%ae%9e%e8%b7%b5%e3%80%82\/#breadcrumb"},"inLanguage":"zh-Hans","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.silicloud.com\/zh\/blog\/%e4%bd%bf%e7%94%a8chainerrl%ef%bc%8c%e5%9c%a8minecraft%e4%b8%ad%e8%bf%9b%e8%a1%8c%e6%b7%b1%e5%ba%a6%e5%bc%ba%e5%8c%96%e5%ad%a6%e4%b9%a0%e7%9a%84%e5%ae%9e%e8%b7%b5%e3%80%82\/"]}]},{"@type":"BreadcrumbList","@id":"https:\/\/www.silicloud.com\/zh\/blog\/%e4%bd%bf%e7%94%a8chainerrl%ef%bc%8c%e5%9c%a8minecraft%e4%b8%ad%e8%bf%9b%e8%a1%8c%e6%b7%b1%e5%ba%a6%e5%bc%ba%e5%8c%96%e5%ad%a6%e4%b9%a0%e7%9a%84%e5%ae%9e%e8%b7%b5%e3%80%82\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"\u9996\u9875","item":"https:\/\/www.silicloud.com\/zh\/blog\/"},{"@type":"ListItem","position":2,"name":"\u4f7f\u7528ChainerRL\uff0c\u5728Minecraft\u4e2d\u8fdb\u884c\u6df1\u5ea6\u5f3a\u5316\u5b66\u4e60\u7684\u5b9e\u8df5"}]},{"@type":"WebSite","@id":"https:\/\/www.silicloud.com\/zh\/blog\/#website","url":"https:\/\/www.silicloud.com\/zh\/blog\/","name":"Blog - Silicon Cloud","description":"","inLanguage":"zh-Hans"},{"@type":"Person","@id":"https:\/\/www.silicloud.com\/zh\/blog\/#\/schema\/person\/41e222757cdd2a3365361328bd79970a","name":"\u79d1, \u96c5","image":{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"https:\/\/www.silicloud.com\/zh\/blog\/#\/schema\/person\/image\/","url":"https:\/\/secure.gravatar.com\/avatar\/1b2d3e00a7df03689797ebd4af8c5827ba5af936849a71050ec331f4cf902c5d?s=96&d=mm&r=g","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/1b2d3e00a7df03689797ebd4af8c5827ba5af936849a71050ec331f4cf902c5d?s=96&d=mm&r=g","caption":"\u79d1, \u96c5"},"url":"https:\/\/www.silicloud.com\/zh\/blog\/author\/keya\/"},{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"https:\/\/www.silicloud.com\/zh\/blog\/%e4%bd%bf%e7%94%a8chainerrl%ef%bc%8c%e5%9c%a8minecraft%e4%b8%ad%e8%bf%9b%e8%a1%8c%e6%b7%b1%e5%ba%a6%e5%bc%ba%e5%8c%96%e5%ad%a6%e4%b9%a0%e7%9a%84%e5%ae%9e%e8%b7%b5%e3%80%82\/#local-main-organization-logo","url":"","contentUrl":"","caption":"Blog - Silicon Cloud"}]}},"_links":{"self":[{"href":"https:\/\/www.silicloud.com\/zh\/blog\/wp-json\/wp\/v2\/posts\/27813","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.silicloud.com\/zh\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.silicloud.com\/zh\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.silicloud.com\/zh\/blog\/wp-json\/wp\/v2\/users\/4"}],"replies":[{"embeddable":true,"href":"https:\/\/www.silicloud.com\/zh\/blog\/wp-json\/wp\/v2\/comments?post=27813"}],"version-history":[{"count":3,"href":"https:\/\/www.silicloud.com\/zh\/blog\/wp-json\/wp\/v2\/posts\/27813\/revisions"}],"predecessor-version":[{"id":111204,"href":"https:\/\/www.silicloud.com\/zh\/blog\/wp-json\/wp\/v2\/posts\/27813\/revisions\/111204"}],"wp:attachment":[{"href":"https:\/\/www.silicloud.com\/zh\/blog\/wp-json\/wp\/v2\/media?parent=27813"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.silicloud.com\/zh\/blog\/wp-json\/wp\/v2\/categories?post=27813"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.silicloud.com\/zh\/blog\/wp-json\/wp\/v2\/tags?post=27813"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}