<?xml version="1.0" encoding="utf-8"?>
<rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/">
    <channel>
        <title>NotionNext BLOG</title>
        <link>https://hemonn.com/</link>
        <description>这是一个由NotionNext生成的站点</description>
        <lastBuildDate>Thu, 21 Mar 2024 11:00:21 GMT</lastBuildDate>
        <docs>https://validator.w3.org/feed/docs/rss2.html</docs>
        <generator>https://github.com/jpmonette/feed</generator>
        <language>zh-CN</language>
        <copyright>All rights reserved 2024, hemon</copyright>
        <item>
            <title><![CDATA[注意力机制]]></title>
            <link>https://hemonn.com/article/fbfcda56-3077-437c-8b02-dffed9820b3f</link>
            <guid>https://hemonn.com/article/fbfcda56-3077-437c-8b02-dffed9820b3f</guid>
            <pubDate>Thu, 21 Mar 2024 00:00:00 GMT</pubDate>
            <description><![CDATA[记录注意力机制知识点]]></description>
            <content:encoded><![CDATA[<div id="notion-article" class="mx-auto overflow-hidden "><main class="notion light-mode notion-page notion-block-fbfcda563077437c8b02dffed9820b3f"><div class="notion-viewport"></div><div class="notion-collection-page-properties"></div><blockquote class="notion-quote notion-block-6d5cad8af45c467eaca8dbaaa2d9be4c"><div>智慧的艺术是知道该忽视什么。</div></blockquote><h2 class="notion-h notion-h1 notion-h-indent-0 notion-block-58a08bd836a14755aa5ca5a4f07dbd22" data-id="58a08bd836a14755aa5ca5a4f07dbd22"><span><div id="58a08bd836a14755aa5ca5a4f07dbd22" class="notion-header-anchor"></div><a class="notion-hash-link" href="#58a08bd836a14755aa5ca5a4f07dbd22" title="📝 注意力基本理念"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">📝 注意力基本理念</span></span></h2><div class="notion-text notion-block-edd656a5833640c0b0ebfef699b18e46">注意力分为两种：</div><div class="notion-text notion-block-813ff45f3c394abbb1f9ac484ba9cf07">1、自上而下的有意识的注意力，称为<span class="notion-red">聚焦式注意力，</span><span class="notion-default">包含</span><span class="notion-red">自主性提示。</span></div><div class="notion-text notion-block-baf1ef4d927342449077e9e9756d4534">2、自下而上的无意识的注意力，称为<span class="notion-red">显著性注意力，</span><span class="notion-default">包含</span><span class="notion-red">非自主性提示。</span></div><div class="notion-text notion-block-52114f1d17074af591793ead9acf2643">一个和注意力有关的例子是鸡尾酒会效应，当一个人在吵闹的鸡尾酒会上和朋友聊天时，尽管周围噪音干扰很多，他还是可以听到朋友的谈话内容，而忽略其他人的声音（<span class="notion-red">聚焦式注意力</span>）．同时，如果背景声中有重要的词（比如他的名字），他会马上注意到（<span class="notion-red">显著性注意力</span>）。</div><div class="notion-text notion-block-2a9ac52ef5424334b6a89b5d7cab735c">显著性注意力偏向于感官输入，可以用简单的全连接层，或者是最大汇聚、门控机制进行模拟和实现。</div><div class="notion-text notion-block-626fb855a2e246ab8d966be0d0d820d7">注意力机制，通常指的是聚焦式注意力。自主性提示被称为查询Q（query），通过注意力汇集将选择引导至感官输入，感官输入被称为值V（value）。其中，每一个值V都和一个键K（key）配对，作为非自主提示，或是环境中的非意志信息。</div><figure class="notion-asset-wrapper notion-asset-wrapper-image notion-block-b7047462fd0742a0a6ae30d06fe2754b"><div style="position:relative;display:flex;justify-content:center;align-self:center;width:538px;max-width:100%;flex-direction:column"><img style="object-fit:cover" src="https://www.notion.so/image/https%3A%2F%2Fprod-files-secure.s3.us-west-2.amazonaws.com%2Fbc6f8139-c6c3-442c-a899-0f1954b31d51%2F0efa6011-7e93-49db-a58b-f630a923cf2c%2FUntitled.png?table=block&amp;id=b7047462-fd07-42a0-a6ae-30d06fe2754b&amp;t=b7047462-fd07-42a0-a6ae-30d06fe2754b&amp;width=538&amp;cache=v2" alt="notion image" loading="lazy" decoding="async"/></div></figure><div class="notion-blank notion-block-79c9f87c4a2c4187afc32287a3cc511b"> </div><h2 class="notion-h notion-h1 notion-h-indent-0 notion-block-99ed6c09f4a642aa870b913d21e35db9" data-id="99ed6c09f4a642aa870b913d21e35db9"><span><div id="99ed6c09f4a642aa870b913d21e35db9" class="notion-header-anchor"></div><a class="notion-hash-link" href="#99ed6c09f4a642aa870b913d21e35db9" title="😀 注意力评分函数"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">😀 <b>注意力评分函数</b></span></span></h2><div class="notion-text notion-block-34fa5215fcde4f7ab03a422d6822bb6c">查询（自主提示）和键（非自主提示）之间的交互形成了注意力汇聚。</div><div class="notion-text notion-block-13cca9660fd84e23a5cdb9a1ec94a7f7">注意力汇聚有选择地聚合了值（感官输入）以生成最终的输出。</div><div class="notion-text notion-block-eed59845b1f44386ae99141b3bfbddae">那么注意力评分函数就是用来计算查询向量Q和键向量K之间的相关性。</div><h3 class="notion-h notion-h2 notion-h-indent-1 notion-block-4be1cc9ae82744e8a247ef45f5b655a9" data-id="4be1cc9ae82744e8a247ef45f5b655a9"><span><div id="4be1cc9ae82744e8a247ef45f5b655a9" class="notion-header-anchor"></div><a class="notion-hash-link" href="#4be1cc9ae82744e8a247ef45f5b655a9" title="加性模型"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">加性模型</span></span></h3><div class="notion-callout notion-gray_background_co notion-block-94a083a371c14fa2b31ff62b5762ef40"><div class="notion-page-icon-inline notion-page-icon-span"><span class="notion-page-icon" role="img" aria-label="❗">❗</span></div><div class="notion-callout-text">当查询Q和键K的长度不同时，可以使用加性模型。</div></div><h3 class="notion-h notion-h2 notion-h-indent-1 notion-block-17151dd0a5234fd88f06e904ee00dc8b" data-id="17151dd0a5234fd88f06e904ee00dc8b"><span><div id="17151dd0a5234fd88f06e904ee00dc8b" class="notion-header-anchor"></div><a class="notion-hash-link" href="#17151dd0a5234fd88f06e904ee00dc8b" title="点积模型"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">点积模型</span></span></h3><div class="notion-text notion-block-a90ad4535a584a62b119c6f878c27689">加性模型和点积模型的复杂度差不多，但是点积模型在实现上可以更好地利用矩阵乘积，从而计算效率更高。使用点积模型的前提是qk具有相同的长度d。</div><div class="notion-text notion-block-3e6d28874ef14329b33cdb86f9f9e4e6">假设查询和键的所有元素都是独立的随机变量，并且都满足零均值和单位方差，那么两个向量的点积的均值为0，方差为1。为确保无论向量长度d如何，点积的方差在不考虑向量长度的情况下仍然是1，那么对点积除以<!-- -->，得到<b>缩放点积评分函数</b>：</div><h3 class="notion-h notion-h2 notion-h-indent-1 notion-block-447e329322eb4c25922fec49a5c668e1" data-id="447e329322eb4c25922fec49a5c668e1"><span><div id="447e329322eb4c25922fec49a5c668e1" class="notion-header-anchor"></div><a class="notion-hash-link" href="#447e329322eb4c25922fec49a5c668e1" title="小节"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">小节</span></span></h3><div class="notion-text notion-block-13dc4aa6e1bb45b19c2b569cbed50a55">缩放点积直接通过点积得到注意力分数，没有任何的学习参数。而加性注意力是通过可学习参数先进行放缩，然后进行q与k的特征相加。</div><div class="notion-text notion-block-8de39645d2314a088f5557511e3a7f1d">区别在于：</div><ol start="1" class="notion-list notion-list-numbered notion-block-bda612cb9fca4d6c9232f8acd02246f2"><li>是否可学习。</li></ol><ol start="2" class="notion-list notion-list-numbered notion-block-9c869e7b74b248d68c53ba09ee143ecd"><li>特征进行点积还是特征进行相加。</li></ol><h2 class="notion-h notion-h1 notion-h-indent-0 notion-block-489595f18b5c4cf791d2daac96ed0aa6" data-id="489595f18b5c4cf791d2daac96ed0aa6"><span><div id="489595f18b5c4cf791d2daac96ed0aa6" class="notion-header-anchor"></div><a class="notion-hash-link" href="#489595f18b5c4cf791d2daac96ed0aa6" title="😁 自注意力机制"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">😁 自注意力机制</span></span></h2><h3 class="notion-h notion-h2 notion-h-indent-1 notion-block-b2bb61fe0cb141bc93d36319cf839778" data-id="b2bb61fe0cb141bc93d36319cf839778"><span><div id="b2bb61fe0cb141bc93d36319cf839778" class="notion-header-anchor"></div><a class="notion-hash-link" href="#b2bb61fe0cb141bc93d36319cf839778" title="自注意力"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">自注意力</span></span></h3><div class="notion-text notion-block-d29e572947b7497ab555320271e15d8b">查询、键和值来自同一组输入，因此被称为自注意力（self-attention）。</div><blockquote class="notion-quote notion-block-136ae4c330ef41cbabd0c34e30cc1b2d"><div>在序列信息中，能够捕捉序列内部相关性的信息。顺序信息可通过位置编码加入。</div></blockquote><div class="notion-text notion-block-6c26fd5a3c9e473c8583c118ebb094ef">基本结构如下：</div><figure class="notion-asset-wrapper notion-asset-wrapper-image notion-block-c0992f569d75462ea0fc8c11958f0efe"><div style="position:relative;display:flex;justify-content:center;align-self:center;width:636px;max-width:100%;flex-direction:column"><img style="object-fit:cover" src="https://www.notion.so/image/https%3A%2F%2Fprod-files-secure.s3.us-west-2.amazonaws.com%2Fbc6f8139-c6c3-442c-a899-0f1954b31d51%2Fb36477f5-d10c-4d54-848a-1cffed79492e%2FUntitled.png?table=block&amp;id=c0992f56-9d75-462e-a0fc-8c11958f0efe&amp;t=c0992f56-9d75-462e-a0fc-8c11958f0efe&amp;width=636&amp;cache=v2" alt="notion image" loading="lazy" decoding="async"/></div></figure><div class="notion-text notion-block-91e246afe5a7483a9bbaa3b1df753875"><b>自注意力模型可以作为神经网络中的一层来使用，既可以用来替换卷积层和循环层，也可以和它们一起交替使用</b>（比如𝑿可以是卷积层或循环层的输出）。自注意力模型计算的权重<!-- -->只依赖于<!-- -->和<!-- --> 的相关性，而忽略了输入信息的位置信息。因此在单独使用时，自注意力模型一般需要加入位置编码信息来进行修正。</div><h3 class="notion-h notion-h2 notion-h-indent-1 notion-block-c6ef22036df8417f8ea489e4f693315a" data-id="c6ef22036df8417f8ea489e4f693315a"><span><div id="c6ef22036df8417f8ea489e4f693315a" class="notion-header-anchor"></div><a class="notion-hash-link" href="#c6ef22036df8417f8ea489e4f693315a" title="编码"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">编码</span></span></h3><div class="notion-text notion-block-2a8620522826483ba9127f37f3f7f11b">编码分为<b>绝对位置编码</b>和<b>相对位置编码</b>两种。</div><h4 class="notion-h notion-h3 notion-h-indent-2 notion-block-480edac55de14fb1bf350d0a1c976df4" data-id="480edac55de14fb1bf350d0a1c976df4"><span><div id="480edac55de14fb1bf350d0a1c976df4" class="notion-header-anchor"></div><a class="notion-hash-link" href="#480edac55de14fb1bf350d0a1c976df4" title="绝对位置编码"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">绝对位置编码</span></span></h4><div class="notion-text notion-block-93f6d5572780422fbdc5d8074ba2f0f8"><code class="notion-inline-code">后续补充</code></div><h4 class="notion-h notion-h3 notion-h-indent-2 notion-block-ef9155e236e048ae8ccfc21091098100" data-id="ef9155e236e048ae8ccfc21091098100"><span><div id="ef9155e236e048ae8ccfc21091098100" class="notion-header-anchor"></div><a class="notion-hash-link" href="#ef9155e236e048ae8ccfc21091098100" title="相对位置编码"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">相对位置编码</span></span></h4><div class="notion-text notion-block-c806de43a42c4e53b6eac6a4f6f7122d"><code class="notion-inline-code">后续补充</code></div><h4 class="notion-h notion-h3 notion-h-indent-2 notion-block-ff722484e7d64a808ca686e85d188104" data-id="ff722484e7d64a808ca686e85d188104"><span><div id="ff722484e7d64a808ca686e85d188104" class="notion-header-anchor"></div><a class="notion-hash-link" href="#ff722484e7d64a808ca686e85d188104" title="自注意力机制使用场景"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">自注意力机制使用场景</span></span></h4><div class="notion-text notion-block-f7bc30ada0764ae0a6bfc8c88f34e1aa">非序列数据：特征信息提取</div><div class="notion-text notion-block-b51c1cf17d264637a7b94b6a9b84b416">序列数据：序列内部相关信息捕捉</div><h2 class="notion-h notion-h1 notion-h-indent-0 notion-block-0645d1d410df4b4c945986cca8e42bc6" data-id="0645d1d410df4b4c945986cca8e42bc6"><span><div id="0645d1d410df4b4c945986cca8e42bc6" class="notion-header-anchor"></div><a class="notion-hash-link" href="#0645d1d410df4b4c945986cca8e42bc6" title="🤗 条件注意力机制"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">🤗 条件注意力机制</span></span></h2><h2 class="notion-h notion-h1 notion-h-indent-0 notion-block-237d842e7f194de0947775488b7abf56" data-id="237d842e7f194de0947775488b7abf56"><span><div id="237d842e7f194de0947775488b7abf56" class="notion-header-anchor"></div><a class="notion-hash-link" href="#237d842e7f194de0947775488b7abf56" title="🤗 多头注意力机制"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">🤗 多头注意力机制</span></span></h2><h2 class="notion-h notion-h1 notion-h-indent-0 notion-block-d67ee472595445d5b3bed0e92f0f544c" data-id="d67ee472595445d5b3bed0e92f0f544c"><span><div id="d67ee472595445d5b3bed0e92f0f544c" class="notion-header-anchor"></div><a class="notion-hash-link" href="#d67ee472595445d5b3bed0e92f0f544c" title="🤗 通道注意力机制"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">🤗 通道注意力机制</span></span></h2><h2 class="notion-h notion-h1 notion-h-indent-0 notion-block-773df0c655a04c82bf399dbb924150b7" data-id="773df0c655a04c82bf399dbb924150b7"><span><div id="773df0c655a04c82bf399dbb924150b7" class="notion-header-anchor"></div><a class="notion-hash-link" href="#773df0c655a04c82bf399dbb924150b7" title="🤗 空间注意力机制"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">🤗 空间注意力机制</span></span></h2><h2 class="notion-h notion-h1 notion-h-indent-0 notion-block-5f463a18cca3467c9060afe754e4c5a7" data-id="5f463a18cca3467c9060afe754e4c5a7"><span><div id="5f463a18cca3467c9060afe754e4c5a7" class="notion-header-anchor"></div><a class="notion-hash-link" href="#5f463a18cca3467c9060afe754e4c5a7" title="🤗 CBAM"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">🤗 CBAM</span></span></h2><h2 class="notion-h notion-h1 notion-h-indent-0 notion-block-587f0ea601ed4f719bb0765fba49aaad" data-id="587f0ea601ed4f719bb0765fba49aaad"><span><div id="587f0ea601ed4f719bb0765fba49aaad" class="notion-header-anchor"></div><a class="notion-hash-link" href="#587f0ea601ed4f719bb0765fba49aaad" title="📎 参考文章"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">📎 参考文章</span></span></h2><ul class="notion-list notion-list-disc notion-block-0dcc65aab9fc4abdba5544bd505e961c"><li>一些引用</li></ul><ul class="notion-list notion-list-disc notion-block-bc05ea909f0a4c1a87a8d0488b329728"><li>引用文章</li></ul><div class="notion-blank notion-block-6806aa819c064fbbb8c22a4f770e09a5"> </div><div class="notion-callout notion-gray_background_co notion-block-40152439135545bba8589bba4057fb3e"><div class="notion-page-icon-inline notion-page-icon-span"><span class="notion-page-icon" role="img" aria-label="💡">💡</span></div><div class="notion-callout-text">有关Notion安装或者使用上的问题，欢迎您在底部评论区留言，一起交流~</div></div></main></div>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[编码器和解码器]]></title>
            <link>https://hemonn.com/article/7391b105-2b8a-425e-8665-85f55e91a703</link>
            <guid>https://hemonn.com/article/7391b105-2b8a-425e-8665-85f55e91a703</guid>
            <pubDate>Thu, 21 Mar 2024 00:00:00 GMT</pubDate>
            <content:encoded><![CDATA[<div id="notion-article" class="mx-auto overflow-hidden "><main class="notion light-mode notion-page notion-block-7391b1052b8a425e866585f55e91a703"><div class="notion-viewport"></div><div class="notion-collection-page-properties"></div><div class="notion-callout notion-gray_background_co notion-block-38ae50c565ac495c93d7e787db0b3b44"><div class="notion-page-icon-inline notion-page-icon-span"><span class="notion-page-icon" role="img" aria-label="😀">😀</span></div><div class="notion-callout-text">这里写文章的前言：
一个简单的开头,简述这篇文章讨论的问题、目标、人物、背景是什么？并简述你给出的答案。<div class="notion-text notion-block-404945a466da42df925dacb371f75d69">可以说说你的故事：阻碍、努力、结果成果，意外与转折。</div></div></div><div class="notion-blank notion-block-5b2b9da35139407780f74315c96ae3f8"> </div><h2 class="notion-h notion-h1 notion-h-indent-0 notion-block-ca2c846c61944c90a76c31ac7d111c41" data-id="ca2c846c61944c90a76c31ac7d111c41"><span><div id="ca2c846c61944c90a76c31ac7d111c41" class="notion-header-anchor"></div><a class="notion-hash-link" href="#ca2c846c61944c90a76c31ac7d111c41" title="📝 主旨内容"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">📝 主旨内容</span></span></h2><h3 class="notion-h notion-h2 notion-h-indent-1 notion-block-45287f48c4b9499bb7b518c3093973a6" data-id="45287f48c4b9499bb7b518c3093973a6"><span><div id="45287f48c4b9499bb7b518c3093973a6" class="notion-header-anchor"></div><a class="notion-hash-link" href="#45287f48c4b9499bb7b518c3093973a6" title="观点1"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">观点1</span></span></h3><blockquote class="notion-quote notion-block-93a5983e20d94b5c8e70a6423c9cd740"><div>引用的话语</div></blockquote><h3 class="notion-h notion-h2 notion-h-indent-1 notion-block-63d4952095514ca9acf1662a852d8e40" data-id="63d4952095514ca9acf1662a852d8e40"><span><div id="63d4952095514ca9acf1662a852d8e40" class="notion-header-anchor"></div><a class="notion-hash-link" href="#63d4952095514ca9acf1662a852d8e40" title="观点2"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">观点2</span></span></h3><blockquote class="notion-quote notion-block-4f441c07d2c44111a1ff8dad6795dcce"><div>引用的话语</div></blockquote><h2 class="notion-h notion-h1 notion-h-indent-0 notion-block-eb4b32217e3d4a33b9270349a3f90617" data-id="eb4b32217e3d4a33b9270349a3f90617"><span><div id="eb4b32217e3d4a33b9270349a3f90617" class="notion-header-anchor"></div><a class="notion-hash-link" href="#eb4b32217e3d4a33b9270349a3f90617" title="🤗 总结归纳"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">🤗 总结归纳</span></span></h2><div class="notion-text notion-block-672ba425f4bd4704814cd8e04439dac5">总结文章的内容</div><h2 class="notion-h notion-h1 notion-h-indent-0 notion-block-a4c487a56c1a4e3ca9b4bb87f3be65a5" data-id="a4c487a56c1a4e3ca9b4bb87f3be65a5"><span><div id="a4c487a56c1a4e3ca9b4bb87f3be65a5" class="notion-header-anchor"></div><a class="notion-hash-link" href="#a4c487a56c1a4e3ca9b4bb87f3be65a5" title="📎 参考文章"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">📎 参考文章</span></span></h2><ul class="notion-list notion-list-disc notion-block-cc8b2950fce34f4ab523f35b70ce4395"><li>一些引用</li></ul><ul class="notion-list notion-list-disc notion-block-0c10ef9ae62b470a9049cebcb8f067e2"><li>引用文章</li></ul><div class="notion-blank notion-block-bc7fc4aff690485f8843b632719d5d43"> </div><div class="notion-callout notion-gray_background_co notion-block-c67f61856f824f1dbdb9a451f67d1ef7"><div class="notion-page-icon-inline notion-page-icon-span"><span class="notion-page-icon" role="img" aria-label="💡">💡</span></div><div class="notion-callout-text">有关Notion安装或者使用上的问题，欢迎您在底部评论区留言，一起交流~</div></div></main></div>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[不同层级特征的组合形式及其网络结构实现]]></title>
            <link>https://hemonn.com/article/e43aa60b-705c-4039-a5fa-cb7d325a9dbe</link>
            <guid>https://hemonn.com/article/e43aa60b-705c-4039-a5fa-cb7d325a9dbe</guid>
            <pubDate>Thu, 21 Mar 2024 00:00:00 GMT</pubDate>
            <description><![CDATA[基于业务重组特征和多级输入]]></description>
            <content:encoded><![CDATA[<div id="notion-article" class="mx-auto overflow-hidden "><main class="notion light-mode notion-page notion-block-e43aa60b705c4039a5facb7d325a9dbe"><div class="notion-viewport"></div><div class="notion-collection-page-properties"></div><div class="notion-callout notion-gray_background_co notion-block-c3030a6fa4d547ce8eeff46a66d798ae"><div class="notion-page-icon-inline notion-page-icon-span"><span class="notion-page-icon" role="img" aria-label="😀">😀</span></div><div class="notion-callout-text">这里写文章的前言：
一个简单的开头,简述这篇文章讨论的问题、目标、人物、背景是什么？并简述你给出的答案。<div class="notion-text notion-block-40c0dbb432354f938a9cb6890e37e872">可以说说你的故事：阻碍、努力、结果成果，意外与转折。</div></div></div><div class="notion-blank notion-block-4067aa1649fb4ea7a239905612ed9daf"> </div><h2 class="notion-h notion-h1 notion-h-indent-0 notion-block-1f2d2629a7784d7780a00d37e40e9cae" data-id="1f2d2629a7784d7780a00d37e40e9cae"><span><div id="1f2d2629a7784d7780a00d37e40e9cae" class="notion-header-anchor"></div><a class="notion-hash-link" href="#1f2d2629a7784d7780a00d37e40e9cae" title="📝 主旨内容"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">📝 主旨内容</span></span></h2><h3 class="notion-h notion-h2 notion-h-indent-1 notion-block-5a1b635f22b14570bc3bb822efaa069f" data-id="5a1b635f22b14570bc3bb822efaa069f"><span><div id="5a1b635f22b14570bc3bb822efaa069f" class="notion-header-anchor"></div><a class="notion-hash-link" href="#5a1b635f22b14570bc3bb822efaa069f" title="观点1"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">观点1</span></span></h3><blockquote class="notion-quote notion-block-55ef91a309994141a16d9a1d8adf126f"><div>引用的话语</div></blockquote><h3 class="notion-h notion-h2 notion-h-indent-1 notion-block-18b3018d2f234781a5e3e0dd91d87a77" data-id="18b3018d2f234781a5e3e0dd91d87a77"><span><div id="18b3018d2f234781a5e3e0dd91d87a77" class="notion-header-anchor"></div><a class="notion-hash-link" href="#18b3018d2f234781a5e3e0dd91d87a77" title="观点2"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">观点2</span></span></h3><blockquote class="notion-quote notion-block-d10749a6913b4be992372a27759888d1"><div>引用的话语</div></blockquote><h2 class="notion-h notion-h1 notion-h-indent-0 notion-block-2baeb4d85907417cb32b2590c409105d" data-id="2baeb4d85907417cb32b2590c409105d"><span><div id="2baeb4d85907417cb32b2590c409105d" class="notion-header-anchor"></div><a class="notion-hash-link" href="#2baeb4d85907417cb32b2590c409105d" title="🤗 总结归纳"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">🤗 总结归纳</span></span></h2><div class="notion-text notion-block-0b7b8ea5770d4254873ab0a3c061321c">总结文章的内容</div><h2 class="notion-h notion-h1 notion-h-indent-0 notion-block-2b1953f11fb14606adb9172e00f19482" data-id="2b1953f11fb14606adb9172e00f19482"><span><div id="2b1953f11fb14606adb9172e00f19482" class="notion-header-anchor"></div><a class="notion-hash-link" href="#2b1953f11fb14606adb9172e00f19482" title="📎 参考文章"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">📎 参考文章</span></span></h2><ul class="notion-list notion-list-disc notion-block-905dc30bc3e64e919f01058878f45f51"><li>一些引用</li></ul><ul class="notion-list notion-list-disc notion-block-111571b14a2d4784a77260e22eb14998"><li>引用文章</li></ul><div class="notion-blank notion-block-3839dc6afa9b4c029c282e1ec0df0caa"> </div><div class="notion-callout notion-gray_background_co notion-block-855fe33d85114955a653a724748df631"><div class="notion-page-icon-inline notion-page-icon-span"><span class="notion-page-icon" role="img" aria-label="💡">💡</span></div><div class="notion-callout-text">有关Notion安装或者使用上的问题，欢迎您在底部评论区留言，一起交流~</div></div></main></div>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[TODOLIST]]></title>
            <link>https://hemonn.com/article/0cd9ad8d-a261-4fe5-a41f-99cd2bfce684</link>
            <guid>https://hemonn.com/article/0cd9ad8d-a261-4fe5-a41f-99cd2bfce684</guid>
            <pubDate>Sat, 02 Mar 2024 00:00:00 GMT</pubDate>
        </item>
        <item>
            <title><![CDATA[gradio部署无法从服务器端口访问]]></title>
            <link>https://hemonn.com/article/dae3d5fc-6046-4db8-99c8-0055c00fcc28</link>
            <guid>https://hemonn.com/article/dae3d5fc-6046-4db8-99c8-0055c00fcc28</guid>
            <pubDate>Wed, 20 Mar 2024 00:00:00 GMT</pubDate>
            <content:encoded><![CDATA[<div id="notion-article" class="mx-auto overflow-hidden "><main class="notion light-mode notion-page notion-block-dae3d5fc60464db899c80055c00fcc28"><div class="notion-viewport"></div><div class="notion-collection-page-properties"></div><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-277026fa7b5a4016870085c8b68e1a14" data-id="277026fa7b5a4016870085c8b68e1a14"><span><div id="277026fa7b5a4016870085c8b68e1a14" class="notion-header-anchor"></div><a class="notion-hash-link" href="#277026fa7b5a4016870085c8b68e1a14" title="现象"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">现象</span></span></h3><div class="notion-text notion-block-b7dbeb0141794b008642e415ad988d93">在服务器通过<code class="notion-inline-code">gradio</code>的<code class="notion-inline-code">launch</code>方法</div><div class="notion-text notion-block-7e9791c60cbd4f9ea3acd5d03fa1b03a">无法从浏览器直接通过<code class="notion-inline-code">服务器地址:8083</code>访问。</div><div class="notion-text notion-block-65fd57acd48c439a8c67d3e45352add6">设置<code class="notion-inline-code">server_port=8082</code>，可以正常访问。</div><h3 class="notion-h notion-h2 notion-h-indent-0 notion-block-8a4bc992c1734fad96ce303ec7a70da8" data-id="8a4bc992c1734fad96ce303ec7a70da8"><span><div id="8a4bc992c1734fad96ce303ec7a70da8" class="notion-header-anchor"></div><a class="notion-hash-link" href="#8a4bc992c1734fad96ce303ec7a70da8" title="解决步骤"><svg viewBox="0 0 16 16" width="16" height="16"><path fill-rule="evenodd" d="M7.775 3.275a.75.75 0 001.06 1.06l1.25-1.25a2 2 0 112.83 2.83l-2.5 2.5a2 2 0 01-2.83 0 .75.75 0 00-1.06 1.06 3.5 3.5 0 004.95 0l2.5-2.5a3.5 3.5 0 00-4.95-4.95l-1.25 1.25zm-4.69 9.64a2 2 0 010-2.83l2.5-2.5a2 2 0 012.83 0 .75.75 0 001.06-1.06 3.5 3.5 0 00-4.95 0l-2.5 2.5a3.5 3.5 0 004.95 4.95l1.25-1.25a.75.75 0 00-1.06-1.06l-1.25 1.25a2 2 0 01-2.83 0z"></path></svg></a><span class="notion-h-title">解决步骤</span></span></h3><ol start="1" class="notion-list notion-list-numbered notion-block-2f0c950549e24d50b0ce8716d57ba188"><li>检查launch启动后，进程和端口是否启动</li><ol class="notion-list notion-list-numbered notion-block-2f0c950549e24d50b0ce8716d57ba188"><div class="notion-text notion-block-ffd43ed70eae41bb9e4cf67d44794d89">若未启动，检查启动代码、检查端口是否被其他进程占用</div><div class="notion-text notion-block-48f97ede146c43c7b2b9056a61c67a80">若显示启动，且状态<code class="notion-inline-code">8083</code>正在被监听，表示进程启动正常</div></ol></ol><ol start="2" class="notion-list notion-list-numbered notion-block-b44f9cc52fda4ec2af9302107f9d996c"><li>进程启动正常，但无法从本地访问。存在几种情况：</li><ol class="notion-list notion-list-numbered notion-block-b44f9cc52fda4ec2af9302107f9d996c"><li><b>防火墙或网络规则</b>：确保防火墙或网络设备没有针对8083端口设置了特殊规则，阻止外部访问该端口。有时网络设备或防火墙可能会对特定端口进行限制，从而导致无法从外部访问该端口。</li><li><b>应用程序设置</b>：有些应用程序可能有内置的设置或配置，只允许从特定的IP地址或者主机名访问。确保应用程序没有设置这样的限制，或者按照需要进行调整。</li></ol></ol><ol start="3" class="notion-list notion-list-numbered notion-block-19c5f424e172497099f20ac186a3f709"><li>防火墙检查</li><ol class="notion-list notion-list-numbered notion-block-19c5f424e172497099f20ac186a3f709"><li>如果防火墙软件是<code class="notion-inline-code">iptables</code></li><ol class="notion-list notion-list-numbered notion-block-c3f588f76f8a4d1ea8b9bbd32de1edee"></ol><li>如果防火墙软件是<code class="notion-inline-code">firewalld</code></li><ol class="notion-list notion-list-numbered notion-block-d2184650d568492eb73fab91ed292b57"></ol></ol></ol><ol start="4" class="notion-list notion-list-numbered notion-block-59a7aae7fe0e4a5b967cdf8ed96031d5"><li>网络规则检查</li><ol class="notion-list notion-list-numbered notion-block-59a7aae7fe0e4a5b967cdf8ed96031d5"><div class="notion-text notion-block-c484bfbee43842ba9a34d11a45a3481b">使用网络工具（如nmap）来扫描服务器上的端口，并查看哪些端口是开放的，确保<code class="notion-inline-code">8083</code>端口被标记为“开放”</div></ol></ol><ol start="5" class="notion-list notion-list-numbered notion-block-6dc8fe0952c5465dbe26fc458a42ae93"><li>在防火墙检查中，发现</li><ol class="notion-list notion-list-numbered notion-block-6dc8fe0952c5465dbe26fc458a42ae93"><div class="notion-text notion-block-b68c596d023243a0a3f6037f1fa86671"><code class="notion-inline-code">8083</code>端口并未被开放，外部无法访问该端口</div></ol></ol><ol start="6" class="notion-list notion-list-numbered notion-block-14b35fa9c33346c6a28fba1cb58fa6b1"><li>打开该端口</li><ol class="notion-list notion-list-numbered notion-block-14b35fa9c33346c6a28fba1cb58fa6b1"><div class="notion-text notion-block-ff94c0266dc64ebda800e48fa1907da5">这个命令将在<code class="notion-inline-code"><b>public</b></code>区域中添加8083端口的TCP规则，并将规则永久性地应用到防火墙配置中。如果希望临时性地添加规则而不是永久性地添加，可以去掉<code class="notion-inline-code"><b>--permanent</b></code>选项</div><div class="notion-text notion-block-3f8dd49b1a754d10a9a930b18018b75f">完成后重新加载防火墙配置，使其生效。</div><div class="notion-blank notion-block-6b04b61eab994368bb6ec017543acf09"> </div><div class="notion-text notion-block-33da89e0cb944885a9d6a0c1520097bb">重新启动gradio文件，8083端口成功正常访问。</div></ol></ol></main></div>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[vscode终端激活conda编码报错]]></title>
            <link>https://hemonn.com/article/1b1ff049-91f5-4fba-9a78-f8aaf308295a</link>
            <guid>https://hemonn.com/article/1b1ff049-91f5-4fba-9a78-f8aaf308295a</guid>
            <pubDate>Tue, 19 Mar 2024 00:00:00 GMT</pubDate>
            <content:encoded><![CDATA[<div id="notion-article" class="mx-auto overflow-hidden "><main class="notion light-mode notion-page notion-block-1b1ff04991f54fba9a78f8aaf308295a"><div class="notion-viewport"></div><div class="notion-collection-page-properties"></div><div class="notion-text notion-block-6d398c65ca324c809ef651d9f5ab6fa7">主要问题是&#x27;gbk&#x27; codec can&#x27;t encode character &#x27;\ue219&#x27;编码错误</div><div class="notion-text notion-block-f3198d7329a94a978d44ba41c959191c">解决方法：</div><div class="notion-text notion-block-9b7de6f5046049db9f0fa630b1de5250">在系统环境变量中添加</div><figure class="notion-asset-wrapper notion-asset-wrapper-image notion-block-d90700d2c16e42c0b00ba30e7cbcc150"><div style="position:relative;display:flex;justify-content:center;align-self:center;width:100%;max-width:100%;flex-direction:column;height:100%"><img style="object-fit:cover" src="https://www.notion.so/image/https%3A%2F%2Fprod-files-secure.s3.us-west-2.amazonaws.com%2Fbc6f8139-c6c3-442c-a899-0f1954b31d51%2F195dae1b-04b2-4b95-a38d-182bb701d7e1%2FUntitled.png?table=block&amp;id=d90700d2-c16e-42c0-b00b-a30e7cbcc150&amp;t=d90700d2-c16e-42c0-b00b-a30e7cbcc150&amp;width=1045&amp;cache=v2" alt="notion image" loading="lazy" decoding="async"/></div></figure><div class="notion-blank notion-block-984c535afb02459db287c452897688d4"> </div></main></div>]]></content:encoded>
        </item>
    </channel>
</rss>