diff --git "a/2023/10/22/pymarl\347\216\257\345\242\203/index.html" "b/2023/10/22/pymarl\347\216\257\345\242\203/index.html"
index 5cef926..aa4dd4d 100644
--- "a/2023/10/22/pymarl\347\216\257\345\242\203/index.html"
+++ "b/2023/10/22/pymarl\347\216\257\345\242\203/index.html"
@@ -25,8 +25,10 @@
 <meta property="og:locale" content="en_US">
 <meta property="og:image" content="https://gitee.com/lihaibineric/picgo/raw/master/pic/image-20231022153907123.png">
 <meta property="og:image" content="https://gitee.com/lihaibineric/picgo/raw/master/pic/image-20231022153709693.png">
+<meta property="og:image" content="https://gitee.com/lihaibineric/picgo/raw/master/pic/image-20240106145437351.png">
+<meta property="og:image" content="https://gitee.com/lihaibineric/picgo/raw/master/pic/image-20240106145218372.png">
 <meta property="article:published_time" content="2023-10-22T06:35:25.000Z">
-<meta property="article:modified_time" content="2024-01-04T13:09:58.806Z">
+<meta property="article:modified_time" content="2024-01-06T06:54:55.240Z">
 <meta property="article:author" content="Haibin Li">
 <meta property="article:tag" content="人工智能">
 <meta property="article:tag" content="深度学习">
@@ -213,7 +215,7 @@
       <span class="post-meta mr-2">
         <i class="iconfont icon-chart"></i>
         
-          2k words
+          2.7k words
         
       </span>
     
@@ -224,7 +226,7 @@
         
         
         
-          17 mins
+          23 mins
         
       </span>
     
@@ -360,6 +362,36 @@ <h2 id="额外设置">额外设置</h2>
 <figure class="highlight bash"><table><tr><td class="gutter"><pre><span class="line">1</span><br></pre></td><td class="code"><pre><code class="hljs bash">vim torch/_tensor.py<br></code></pre></td></tr></table></figure>
 <p>将757行的内容进行修改</p>
 <figure class="highlight python"><table><tr><td class="gutter"><pre><span class="line">1</span><br><span class="line">2</span><br></pre></td><td class="code"><pre><code class="hljs python">-: self.numpy()<br>+: self.cpu().numpy()<br></code></pre></td></tr></table></figure>
+<h2 id="谷歌足球环境">谷歌足球环境</h2>
+<p>Google Research
+Football：https://github.com/google-research/football</p>
+<h4 id="安装linux依赖环境">安装Linux依赖环境</h4>
+<figure class="highlight bash"><table><tr><td class="gutter"><pre><span class="line">1</span><br><span class="line">2</span><br><span class="line">3</span><br><span class="line">4</span><br><span class="line">5</span><br></pre></td><td class="code"><pre><code class="hljs bash">sudo apt-get install git cmake build-essential libgl1-mesa-dev libsdl2-dev \<br>libsdl2-image-dev libsdl2-ttf-dev libsdl2-gfx-dev libboost-all-dev \<br>libdirectfb-dev libst-dev mesa-utils xvfb x11vnc python3-pip<br><br>python3 -m pip install --upgrade pip setuptools psutil wheel<br></code></pre></td></tr></table></figure>
+<p>这一步非常重要，必须合理安装相应的包。</p>
+<h4 id="安装gfootball">安装GFootball</h4>
+<figure class="highlight python"><table><tr><td class="gutter"><pre><span class="line">1</span><br></pre></td><td class="code"><pre><code class="hljs python">python3 -m pip install gfootball<br></code></pre></td></tr></table></figure>
+<p>或者用github的仓库来进行安装</p>
+<figure class="highlight bash"><table><tr><td class="gutter"><pre><span class="line">1</span><br><span class="line">2</span><br></pre></td><td class="code"><pre><code class="hljs bash">git <span class="hljs-built_in">clone</span> https://github.com/google-research/football.git<br><span class="hljs-built_in">cd</span> football<br></code></pre></td></tr></table></figure>
+<p>可以选择使用虚拟环境:</p>
+<figure class="highlight bash"><table><tr><td class="gutter"><pre><span class="line">1</span><br><span class="line">2</span><br></pre></td><td class="code"><pre><code class="hljs bash">python3 -m venv football-env<br><span class="hljs-built_in">source</span> football-env/bin/activate<br></code></pre></td></tr></table></figure>
+<p>安装依赖包</p>
+<figure class="highlight bash"><table><tr><td class="gutter"><pre><span class="line">1</span><br></pre></td><td class="code"><pre><code class="hljs bash">python3 -m pip install .<br></code></pre></td></tr></table></figure>
+<h4 id="运行结果展示">运行结果展示：</h4>
+<p><img src="https://gitee.com/lihaibineric/picgo/raw/master/pic/image-20240106145437351.png" srcset="/img/loading.gif" lazyload alt="谷歌足球结果展示" style="zoom:50%;" /></p>
+<h4 id="遇到的问题">遇到的问题</h4>
+<p>编译错误，缺少boost的包：</p>
+<figure>
+<img
+src="https://gitee.com/lihaibineric/picgo/raw/master/pic/image-20240106145036965.png" srcset="/img/loading.gif" lazyload
+alt="编译错误" />
+<figcaption aria-hidden="true">编译错误</figcaption>
+</figure>
+<p>选择安装如下包：</p>
+<figure class="highlight bash"><table><tr><td class="gutter"><pre><span class="line">1</span><br></pre></td><td class="code"><pre><code class="hljs bash">conda install anaconda::py-boost<br></code></pre></td></tr></table></figure>
+<p>第二个错误：</p>
+<p><img src="https://gitee.com/lihaibineric/picgo/raw/master/pic/image-20240106145218372.png" srcset="/img/loading.gif" lazyload alt="错误二" style="zoom:67%;" /></p>
+<p>解决方法：</p>
+<figure class="highlight bash"><table><tr><td class="gutter"><pre><span class="line">1</span><br></pre></td><td class="code"><pre><code class="hljs bash"><span class="hljs-built_in">export</span> LD_PRELOAD=/usr/lib/x86_64-linux-gnu/libffi.so.7<br></code></pre></td></tr></table></figure>
 
                 
               </div>
diff --git a/local-search.xml b/local-search.xml
index befff72..58b87bd 100644
--- a/local-search.xml
+++ b/local-search.xml
@@ -245,7 +245,7 @@
     <link href="/2023/10/22/pymarl%E7%8E%AF%E5%A2%83/"/>
     <url>/2023/10/22/pymarl%E7%8E%AF%E5%A2%83/</url>
     
-    <content type="html"><![CDATA[<meta name="referrer" content="no-referrer"/><p>SMAC是基于暴雪星际争霸IIRTS游戏的协作多智能体强化学习领域的研究环境。SMAC利用星际争霸2机器学习API和DeepMind的PySC2为自主代理提供了一个方便的接口。</p><p>本文主要介绍在多智能体强化学习中的验证平台SAMC和Pymarl的配置和安装过程</p><p>环境地址：<ahref="https://github.com/oxwhirl/pymarl">https://github.com/oxwhirl/pymarl</a></p><p>依赖环境：<ahref="https://github.com/oxwhirl/smac">https://github.com/oxwhirl/smac</a></p><p>主要针对Ubuntu下的Pymarl环境的配置（MAC和Windows类似）</p><figure><imgsrc="https://gitee.com/lihaibineric/picgo/raw/master/pic/image-20231022145734342.png"alt="SMAC" /><figcaption aria-hidden="true">SMAC</figcaption></figure><h2 id="安装starcraftii">安装StarCraftII</h2><h3 id="windowsmac">Windows/Mac</h3><p>直接在官网下载安装最新版，需要加速器或者更换亚服：<ahref="https://sc2.blizzard.cn/landing">https://sc2.blizzard.cn/landing</a></p><p>安装地图：<ahref="https://link.zhihu.com/?target=https%3A//github.com/oxwhirl/smac/releases/download/v0.1-beta1/SMAC_Maps.zip">https://link.zhihu.com/?target=https%3A//github.com/oxwhirl/smac/releases/download/v0.1-beta1/SMAC_Maps.zip</a></p><p>将地图放在Maps文件夹下</p><h3 id="ubuntu">Ubuntu</h3><p>在Ubuntu系统中，安装需要版本的SCII即可<ahref="https://github.com/Blizzard/s2client-proto#downloads">https://github.com/Blizzard/s2client-proto#downloads</a></p><p>下载压缩包之后安装并解压放到用户名文件夹下：～/StarCraftII/</p><p>如果没有把SCII安装到用户名之下，这个地方的路径可以根据放置的位置来进行更改：</p><figure class="highlight bash"><table><tr><td class="gutter"><pre><span class="line">1</span><br></pre></td><td class="code"><pre><code class="hljs bash">vim ~/.bashrc<br></code></pre></td></tr></table></figure><p>在最后一行增加环境变量，这里可以更改星际争霸的位置索引到任何想要放置的位置</p><figure class="highlight routeros"><table><tr><td class="gutter"><pre><span class="line">1</span><br></pre></td><td class="code"><pre><code class="hljs routeros"><span class="hljs-built_in">export</span> <span class="hljs-attribute">SC2PATH</span>=~/StarCraftII/<br></code></pre></td></tr></table></figure><p>添加完之后更新环境</p><figure class="highlight bash"><table><tr><td class="gutter"><pre><span class="line">1</span><br></pre></td><td class="code"><pre><code class="hljs bash"><span class="hljs-built_in">source</span> ~/.bashrc<br></code></pre></td></tr></table></figure><p>添加SAMC的地图：Linux系统在下载SCII的时候就自带了地图</p><h2 id="安装smac">安装SMAC</h2><p>创建虚拟环境并安装pytorch</p><figure class="highlight python"><table><tr><td class="gutter"><pre><span class="line">1</span><br><span class="line">2</span><br><span class="line">3</span><br><span class="line">4</span><br></pre></td><td class="code"><pre><code class="hljs python">conda create -n pymarl python=<span class="hljs-number">3.8</span> -y<br>conda activate pymarl   <br><span class="hljs-comment"># CUDA 10.2 例子具体的的CUDA版本根据自己的服务器来定</span><br>conda install pytorch==<span class="hljs-number">1.10</span><span class="hljs-number">.0</span> torchvision==<span class="hljs-number">0.11</span><span class="hljs-number">.0</span> torchaudio==<span class="hljs-number">0.10</span><span class="hljs-number">.0</span> cudatoolkit=<span class="hljs-number">10.2</span> -c pytorch<br></code></pre></td></tr></table></figure><p>直接利用虚拟环境中的pip来安装</p><figure class="highlight python"><table><tr><td class="gutter"><pre><span class="line">1</span><br></pre></td><td class="code"><pre><code class="hljs python">pip install git+https://github.com/oxwhirl/smac.git<br></code></pre></td></tr></table></figure><p>或者可以利用git进行安装，先gitclone下载下来然后进入文件夹中进行安装</p><figure class="highlight python"><table><tr><td class="gutter"><pre><span class="line">1</span><br><span class="line">2</span><br></pre></td><td class="code"><pre><code class="hljs python">git clone https://github.com/oxwhirl/smac.git<br>pip install -e smac/<br></code></pre></td></tr></table></figure><p>如果pip install安装失败可以考虑升级pip：</p><figure class="highlight python"><table><tr><td class="gutter"><pre><span class="line">1</span><br></pre></td><td class="code"><pre><code class="hljs python">pip install --upgrade pip<br></code></pre></td></tr></table></figure><p>如果安装的内容失败了可以手动安装，比如pymarl运行所依赖的包有以下：</p><figure class="highlight python"><table><tr><td class="gutter"><pre><span class="line">1</span><br><span class="line">2</span><br></pre></td><td class="code"><pre><code class="hljs python">pip install sacred numpy scipy matplotlib seaborn pyyaml pygame pytest probscale imageio snakeviz tensorboard-logger<br><br></code></pre></td></tr></table></figure><p>或者根据<ahref="https://github.com/oxwhirl/pymarl/tree/master/docker">https://github.com/oxwhirl/pymarl/tree/master/docker</a>中的bash脚本中Dockerfile依赖内容进行安装</p><h2 id="pymarl与scii">Pymarl与SCII</h2><h3 id="采用上述安装步骤">采用上述安装步骤</h3><p>对于上述的操作完成操作之后，下载pymarl的环境包并进入</p><figure class="highlight python"><table><tr><td class="gutter"><pre><span class="line">1</span><br><span class="line">2</span><br></pre></td><td class="code"><pre><code class="hljs python">git clone git@github.com:oxwhirl/pymarl.git<br>cd pymarl<br></code></pre></td></tr></table></figure><p>将上述安装的SCII文件夹复制到pymarl的src的3dparty之下</p><h3 id="如果利用docker安装">如果利用Docker安装</h3><p>或者根据pymarl中的内容对上述环境进行自动配置（需要Docker）</p><figure class="highlight bash"><table><tr><td class="gutter"><pre><span class="line">1</span><br><span class="line">2</span><br></pre></td><td class="code"><pre><code class="hljs bash"><span class="hljs-built_in">cd</span> docker<br>bash build.sh<br></code></pre></td></tr></table></figure><p>同时安装SCII和SMAC</p><figure class="highlight bash"><table><tr><td class="gutter"><pre><span class="line">1</span><br></pre></td><td class="code"><pre><code class="hljs bash">bash install_sc2.sh<br></code></pre></td></tr></table></figure><h2 id="测试环境">测试环境</h2><h3 id="测试地图信息">测试地图信息</h3><figure class="highlight python"><table><tr><td class="gutter"><pre><span class="line">1</span><br></pre></td><td class="code"><pre><code class="hljs python">python -m smac.<span class="hljs-built_in">bin</span>.map_list <br></code></pre></td></tr></table></figure><p><img src="https://gitee.com/lihaibineric/picgo/raw/master/pic/image-20231022153907123.png" alt="Maps show" style="zoom:67%;" /></p><h3 id="测试环境信息">测试环境信息</h3><figure class="highlight python"><table><tr><td class="gutter"><pre><span class="line">1</span><br></pre></td><td class="code"><pre><code class="hljs python">python -m smac.examples.random_agents<br></code></pre></td></tr></table></figure><p><img src="https://gitee.com/lihaibineric/picgo/raw/master/pic/image-20231022153709693.png" alt="test agents" style="zoom: 50%;" /></p><h3 id="测试pymarl环境">测试Pymarl环境</h3><figure class="highlight python"><table><tr><td class="gutter"><pre><span class="line">1</span><br></pre></td><td class="code"><pre><code class="hljs python">python src/main.py --config=qmix --env-config=sc2 <span class="hljs-keyword">with</span> env_args.map_name=2s3z<br></code></pre></td></tr></table></figure><figure><imgsrc="https://gitee.com/lihaibineric/picgo/raw/master/pic/image-20231022153651418.png"alt="test pymarl" /><figcaption aria-hidden="true">test pymarl</figcaption></figure><h2 id="依赖包的版本问题">依赖包的版本问题</h2><p>当运行pymarl的时候容易因为依赖包产生问题</p><figure><imgsrc="https://gitee.com/lihaibineric/picgo/raw/master/pic/image-20240104210308852.png"alt="image-20240104210308852" /><figcaption aria-hidden="true">image-20240104210308852</figcaption></figure><p>原因是<code>sacred</code>版本不对劲，只需要将版本降低到0.7.5这个问题就能解决</p><h2 id="额外设置">额外设置</h2><p>在利用pymarl进行实验的过程中可能会出现</p><figure><imgsrc="https://gitee.com/lihaibineric/picgo/raw/master/pic/image-20231022154059365.png"alt="some bug" /><figcaption aria-hidden="true">some bug</figcaption></figure><p>需要进行设置</p><figure class="highlight bash"><table><tr><td class="gutter"><pre><span class="line">1</span><br></pre></td><td class="code"><pre><code class="hljs bash">vim torch/_tensor.py<br></code></pre></td></tr></table></figure><p>将757行的内容进行修改</p><figure class="highlight python"><table><tr><td class="gutter"><pre><span class="line">1</span><br><span class="line">2</span><br></pre></td><td class="code"><pre><code class="hljs python">-: self.numpy()<br>+: self.cpu().numpy()<br></code></pre></td></tr></table></figure>]]></content>
+    <content type="html"><![CDATA[<meta name="referrer" content="no-referrer"/><p>SMAC是基于暴雪星际争霸IIRTS游戏的协作多智能体强化学习领域的研究环境。SMAC利用星际争霸2机器学习API和DeepMind的PySC2为自主代理提供了一个方便的接口。</p><p>本文主要介绍在多智能体强化学习中的验证平台SAMC和Pymarl的配置和安装过程</p><p>环境地址：<ahref="https://github.com/oxwhirl/pymarl">https://github.com/oxwhirl/pymarl</a></p><p>依赖环境：<ahref="https://github.com/oxwhirl/smac">https://github.com/oxwhirl/smac</a></p><p>主要针对Ubuntu下的Pymarl环境的配置（MAC和Windows类似）</p><figure><imgsrc="https://gitee.com/lihaibineric/picgo/raw/master/pic/image-20231022145734342.png"alt="SMAC" /><figcaption aria-hidden="true">SMAC</figcaption></figure><h2 id="安装starcraftii">安装StarCraftII</h2><h3 id="windowsmac">Windows/Mac</h3><p>直接在官网下载安装最新版，需要加速器或者更换亚服：<ahref="https://sc2.blizzard.cn/landing">https://sc2.blizzard.cn/landing</a></p><p>安装地图：<ahref="https://link.zhihu.com/?target=https%3A//github.com/oxwhirl/smac/releases/download/v0.1-beta1/SMAC_Maps.zip">https://link.zhihu.com/?target=https%3A//github.com/oxwhirl/smac/releases/download/v0.1-beta1/SMAC_Maps.zip</a></p><p>将地图放在Maps文件夹下</p><h3 id="ubuntu">Ubuntu</h3><p>在Ubuntu系统中，安装需要版本的SCII即可<ahref="https://github.com/Blizzard/s2client-proto#downloads">https://github.com/Blizzard/s2client-proto#downloads</a></p><p>下载压缩包之后安装并解压放到用户名文件夹下：～/StarCraftII/</p><p>如果没有把SCII安装到用户名之下，这个地方的路径可以根据放置的位置来进行更改：</p><figure class="highlight bash"><table><tr><td class="gutter"><pre><span class="line">1</span><br></pre></td><td class="code"><pre><code class="hljs bash">vim ~/.bashrc<br></code></pre></td></tr></table></figure><p>在最后一行增加环境变量，这里可以更改星际争霸的位置索引到任何想要放置的位置</p><figure class="highlight routeros"><table><tr><td class="gutter"><pre><span class="line">1</span><br></pre></td><td class="code"><pre><code class="hljs routeros"><span class="hljs-built_in">export</span> <span class="hljs-attribute">SC2PATH</span>=~/StarCraftII/<br></code></pre></td></tr></table></figure><p>添加完之后更新环境</p><figure class="highlight bash"><table><tr><td class="gutter"><pre><span class="line">1</span><br></pre></td><td class="code"><pre><code class="hljs bash"><span class="hljs-built_in">source</span> ~/.bashrc<br></code></pre></td></tr></table></figure><p>添加SAMC的地图：Linux系统在下载SCII的时候就自带了地图</p><h2 id="安装smac">安装SMAC</h2><p>创建虚拟环境并安装pytorch</p><figure class="highlight python"><table><tr><td class="gutter"><pre><span class="line">1</span><br><span class="line">2</span><br><span class="line">3</span><br><span class="line">4</span><br></pre></td><td class="code"><pre><code class="hljs python">conda create -n pymarl python=<span class="hljs-number">3.8</span> -y<br>conda activate pymarl   <br><span class="hljs-comment"># CUDA 10.2 例子具体的的CUDA版本根据自己的服务器来定</span><br>conda install pytorch==<span class="hljs-number">1.10</span><span class="hljs-number">.0</span> torchvision==<span class="hljs-number">0.11</span><span class="hljs-number">.0</span> torchaudio==<span class="hljs-number">0.10</span><span class="hljs-number">.0</span> cudatoolkit=<span class="hljs-number">10.2</span> -c pytorch<br></code></pre></td></tr></table></figure><p>直接利用虚拟环境中的pip来安装</p><figure class="highlight python"><table><tr><td class="gutter"><pre><span class="line">1</span><br></pre></td><td class="code"><pre><code class="hljs python">pip install git+https://github.com/oxwhirl/smac.git<br></code></pre></td></tr></table></figure><p>或者可以利用git进行安装，先gitclone下载下来然后进入文件夹中进行安装</p><figure class="highlight python"><table><tr><td class="gutter"><pre><span class="line">1</span><br><span class="line">2</span><br></pre></td><td class="code"><pre><code class="hljs python">git clone https://github.com/oxwhirl/smac.git<br>pip install -e smac/<br></code></pre></td></tr></table></figure><p>如果pip install安装失败可以考虑升级pip：</p><figure class="highlight python"><table><tr><td class="gutter"><pre><span class="line">1</span><br></pre></td><td class="code"><pre><code class="hljs python">pip install --upgrade pip<br></code></pre></td></tr></table></figure><p>如果安装的内容失败了可以手动安装，比如pymarl运行所依赖的包有以下：</p><figure class="highlight python"><table><tr><td class="gutter"><pre><span class="line">1</span><br><span class="line">2</span><br></pre></td><td class="code"><pre><code class="hljs python">pip install sacred numpy scipy matplotlib seaborn pyyaml pygame pytest probscale imageio snakeviz tensorboard-logger<br><br></code></pre></td></tr></table></figure><p>或者根据<ahref="https://github.com/oxwhirl/pymarl/tree/master/docker">https://github.com/oxwhirl/pymarl/tree/master/docker</a>中的bash脚本中Dockerfile依赖内容进行安装</p><h2 id="pymarl与scii">Pymarl与SCII</h2><h3 id="采用上述安装步骤">采用上述安装步骤</h3><p>对于上述的操作完成操作之后，下载pymarl的环境包并进入</p><figure class="highlight python"><table><tr><td class="gutter"><pre><span class="line">1</span><br><span class="line">2</span><br></pre></td><td class="code"><pre><code class="hljs python">git clone git@github.com:oxwhirl/pymarl.git<br>cd pymarl<br></code></pre></td></tr></table></figure><p>将上述安装的SCII文件夹复制到pymarl的src的3dparty之下</p><h3 id="如果利用docker安装">如果利用Docker安装</h3><p>或者根据pymarl中的内容对上述环境进行自动配置（需要Docker）</p><figure class="highlight bash"><table><tr><td class="gutter"><pre><span class="line">1</span><br><span class="line">2</span><br></pre></td><td class="code"><pre><code class="hljs bash"><span class="hljs-built_in">cd</span> docker<br>bash build.sh<br></code></pre></td></tr></table></figure><p>同时安装SCII和SMAC</p><figure class="highlight bash"><table><tr><td class="gutter"><pre><span class="line">1</span><br></pre></td><td class="code"><pre><code class="hljs bash">bash install_sc2.sh<br></code></pre></td></tr></table></figure><h2 id="测试环境">测试环境</h2><h3 id="测试地图信息">测试地图信息</h3><figure class="highlight python"><table><tr><td class="gutter"><pre><span class="line">1</span><br></pre></td><td class="code"><pre><code class="hljs python">python -m smac.<span class="hljs-built_in">bin</span>.map_list <br></code></pre></td></tr></table></figure><p><img src="https://gitee.com/lihaibineric/picgo/raw/master/pic/image-20231022153907123.png" alt="Maps show" style="zoom:67%;" /></p><h3 id="测试环境信息">测试环境信息</h3><figure class="highlight python"><table><tr><td class="gutter"><pre><span class="line">1</span><br></pre></td><td class="code"><pre><code class="hljs python">python -m smac.examples.random_agents<br></code></pre></td></tr></table></figure><p><img src="https://gitee.com/lihaibineric/picgo/raw/master/pic/image-20231022153709693.png" alt="test agents" style="zoom: 50%;" /></p><h3 id="测试pymarl环境">测试Pymarl环境</h3><figure class="highlight python"><table><tr><td class="gutter"><pre><span class="line">1</span><br></pre></td><td class="code"><pre><code class="hljs python">python src/main.py --config=qmix --env-config=sc2 <span class="hljs-keyword">with</span> env_args.map_name=2s3z<br></code></pre></td></tr></table></figure><figure><imgsrc="https://gitee.com/lihaibineric/picgo/raw/master/pic/image-20231022153651418.png"alt="test pymarl" /><figcaption aria-hidden="true">test pymarl</figcaption></figure><h2 id="依赖包的版本问题">依赖包的版本问题</h2><p>当运行pymarl的时候容易因为依赖包产生问题</p><figure><imgsrc="https://gitee.com/lihaibineric/picgo/raw/master/pic/image-20240104210308852.png"alt="image-20240104210308852" /><figcaption aria-hidden="true">image-20240104210308852</figcaption></figure><p>原因是<code>sacred</code>版本不对劲，只需要将版本降低到0.7.5这个问题就能解决</p><h2 id="额外设置">额外设置</h2><p>在利用pymarl进行实验的过程中可能会出现</p><figure><imgsrc="https://gitee.com/lihaibineric/picgo/raw/master/pic/image-20231022154059365.png"alt="some bug" /><figcaption aria-hidden="true">some bug</figcaption></figure><p>需要进行设置</p><figure class="highlight bash"><table><tr><td class="gutter"><pre><span class="line">1</span><br></pre></td><td class="code"><pre><code class="hljs bash">vim torch/_tensor.py<br></code></pre></td></tr></table></figure><p>将757行的内容进行修改</p><figure class="highlight python"><table><tr><td class="gutter"><pre><span class="line">1</span><br><span class="line">2</span><br></pre></td><td class="code"><pre><code class="hljs python">-: self.numpy()<br>+: self.cpu().numpy()<br></code></pre></td></tr></table></figure><h2 id="谷歌足球环境">谷歌足球环境</h2><p>Google ResearchFootball：https://github.com/google-research/football</p><h4 id="安装linux依赖环境">安装Linux依赖环境</h4><figure class="highlight bash"><table><tr><td class="gutter"><pre><span class="line">1</span><br><span class="line">2</span><br><span class="line">3</span><br><span class="line">4</span><br><span class="line">5</span><br></pre></td><td class="code"><pre><code class="hljs bash">sudo apt-get install git cmake build-essential libgl1-mesa-dev libsdl2-dev \<br>libsdl2-image-dev libsdl2-ttf-dev libsdl2-gfx-dev libboost-all-dev \<br>libdirectfb-dev libst-dev mesa-utils xvfb x11vnc python3-pip<br><br>python3 -m pip install --upgrade pip setuptools psutil wheel<br></code></pre></td></tr></table></figure><p>这一步非常重要，必须合理安装相应的包。</p><h4 id="安装gfootball">安装GFootball</h4><figure class="highlight python"><table><tr><td class="gutter"><pre><span class="line">1</span><br></pre></td><td class="code"><pre><code class="hljs python">python3 -m pip install gfootball<br></code></pre></td></tr></table></figure><p>或者用github的仓库来进行安装</p><figure class="highlight bash"><table><tr><td class="gutter"><pre><span class="line">1</span><br><span class="line">2</span><br></pre></td><td class="code"><pre><code class="hljs bash">git <span class="hljs-built_in">clone</span> https://github.com/google-research/football.git<br><span class="hljs-built_in">cd</span> football<br></code></pre></td></tr></table></figure><p>可以选择使用虚拟环境:</p><figure class="highlight bash"><table><tr><td class="gutter"><pre><span class="line">1</span><br><span class="line">2</span><br></pre></td><td class="code"><pre><code class="hljs bash">python3 -m venv football-env<br><span class="hljs-built_in">source</span> football-env/bin/activate<br></code></pre></td></tr></table></figure><p>安装依赖包</p><figure class="highlight bash"><table><tr><td class="gutter"><pre><span class="line">1</span><br></pre></td><td class="code"><pre><code class="hljs bash">python3 -m pip install .<br></code></pre></td></tr></table></figure><h4 id="运行结果展示">运行结果展示：</h4><p><img src="https://gitee.com/lihaibineric/picgo/raw/master/pic/image-20240106145437351.png" alt="谷歌足球结果展示" style="zoom:50%;" /></p><h4 id="遇到的问题">遇到的问题</h4><p>编译错误，缺少boost的包：</p><figure><imgsrc="https://gitee.com/lihaibineric/picgo/raw/master/pic/image-20240106145036965.png"alt="编译错误" /><figcaption aria-hidden="true">编译错误</figcaption></figure><p>选择安装如下包：</p><figure class="highlight bash"><table><tr><td class="gutter"><pre><span class="line">1</span><br></pre></td><td class="code"><pre><code class="hljs bash">conda install anaconda::py-boost<br></code></pre></td></tr></table></figure><p>第二个错误：</p><p><img src="https://gitee.com/lihaibineric/picgo/raw/master/pic/image-20240106145218372.png" alt="错误二" style="zoom:67%;" /></p><p>解决方法：</p><figure class="highlight bash"><table><tr><td class="gutter"><pre><span class="line">1</span><br></pre></td><td class="code"><pre><code class="hljs bash"><span class="hljs-built_in">export</span> LD_PRELOAD=/usr/lib/x86_64-linux-gnu/libffi.so.7<br></code></pre></td></tr></table></figure>]]></content>
     
     
     <categories>
diff --git a/search.xml b/search.xml
index 02670be..e1c27ec 100644
--- a/search.xml
+++ b/search.xml
@@ -2405,6 +2405,36 @@ alt="some bug" />
 <figure class="highlight bash"><table><tr><td class="code"><pre><code class="hljs bash">vim torch/_tensor.py<br></code></pre></td></tr></table></figure>
 <p>将757行的内容进行修改</p>
 <figure class="highlight python"><table><tr><td class="code"><pre><code class="hljs python">-: self.numpy()<br>+: self.cpu().numpy()<br></code></pre></td></tr></table></figure>
+<h2 id="谷歌足球环境">谷歌足球环境</h2>
+<p>Google Research
+Football：https://github.com/google-research/football</p>
+<h4 id="安装linux依赖环境">安装Linux依赖环境</h4>
+<figure class="highlight bash"><table><tr><td class="code"><pre><code class="hljs bash">sudo apt-get install git cmake build-essential libgl1-mesa-dev libsdl2-dev \<br>libsdl2-image-dev libsdl2-ttf-dev libsdl2-gfx-dev libboost-all-dev \<br>libdirectfb-dev libst-dev mesa-utils xvfb x11vnc python3-pip<br><br>python3 -m pip install --upgrade pip setuptools psutil wheel<br></code></pre></td></tr></table></figure>
+<p>这一步非常重要，必须合理安装相应的包。</p>
+<h4 id="安装gfootball">安装GFootball</h4>
+<figure class="highlight python"><table><tr><td class="code"><pre><code class="hljs python">python3 -m pip install gfootball<br></code></pre></td></tr></table></figure>
+<p>或者用github的仓库来进行安装</p>
+<figure class="highlight bash"><table><tr><td class="code"><pre><code class="hljs bash">git <span class="hljs-built_in">clone</span> https://github.com/google-research/football.git<br><span class="hljs-built_in">cd</span> football<br></code></pre></td></tr></table></figure>
+<p>可以选择使用虚拟环境:</p>
+<figure class="highlight bash"><table><tr><td class="code"><pre><code class="hljs bash">python3 -m venv football-env<br><span class="hljs-built_in">source</span> football-env/bin/activate<br></code></pre></td></tr></table></figure>
+<p>安装依赖包</p>
+<figure class="highlight bash"><table><tr><td class="code"><pre><code class="hljs bash">python3 -m pip install .<br></code></pre></td></tr></table></figure>
+<h4 id="运行结果展示">运行结果展示：</h4>
+<p><img src="https://gitee.com/lihaibineric/picgo/raw/master/pic/image-20240106145437351.png" alt="谷歌足球结果展示" style="zoom:50%;" /></p>
+<h4 id="遇到的问题">遇到的问题</h4>
+<p>编译错误，缺少boost的包：</p>
+<figure>
+<img
+src="https://gitee.com/lihaibineric/picgo/raw/master/pic/image-20240106145036965.png"
+alt="编译错误" />
+<figcaption aria-hidden="true">编译错误</figcaption>
+</figure>
+<p>选择安装如下包：</p>
+<figure class="highlight bash"><table><tr><td class="code"><pre><code class="hljs bash">conda install anaconda::py-boost<br></code></pre></td></tr></table></figure>
+<p>第二个错误：</p>
+<p><img src="https://gitee.com/lihaibineric/picgo/raw/master/pic/image-20240106145218372.png" alt="错误二" style="zoom:67%;" /></p>
+<p>解决方法：</p>
+<figure class="highlight bash"><table><tr><td class="code"><pre><code class="hljs bash"><span class="hljs-built_in">export</span> LD_PRELOAD=/usr/lib/x86_64-linux-gnu/libffi.so.7<br></code></pre></td></tr></table></figure>
 ]]></content>
       <categories>
         <category>强化学习</category>