<rss xmlns:atom="http://www.w3.org/2005/Atom" version="2.0"><channel><title>Python - 标签 - HAKULA†CHANNEL</title><link>https://old.hakula.xyz/tags/python/</link><description>Python - 标签 - HAKULA†CHANNEL</description><generator>Hugo -- gohugo.io</generator><language>zh-CN</language><managingEditor>i@hakula.xyz (Hakula)</managingEditor><webMaster>i@hakula.xyz (Hakula)</webMaster><lastBuildDate>Fri, 13 Mar 2026 17:36:00 +0800</lastBuildDate><atom:link href="https://old.hakula.xyz/tags/python/" rel="self" type="application/rss+xml"/><item><title>MCP Server Guide: Building a Documentation Server for LLM Agents</title><link>https://old.hakula.xyz/posts/tutorial/llm-intro/mcp-server-guide/</link><pubDate>Fri, 13 Mar 2026 17:36:00 +0800</pubDate><author>Hakula</author><guid>https://old.hakula.xyz/posts/tutorial/llm-intro/mcp-server-guide/</guid><description><![CDATA[<div class="featured-image">
                <img src="https://hakula-1257872502.file.myqcloud.com/images/3/article-covers/21256a36-9d48-4447-862f-2ea8c59b8e39_129339284.webp" referrerpolicy="no-referrer">
            </div><p>A practical, end-to-end guide to building a custom MCP server that turns your documentation site into structured tools for LLM agents. Covers the llms.txt standard, FastMCP, caching, search, packaging, and LLM agents integration. Written for teams that have internal docs (Confluence pages, Markdown files) and want their agents to read them instead of hallucinating.</p>]]></description></item><item><title>DSP - Project: 语音识别</title><link>https://old.hakula.xyz/posts/note/dsp/speech-recognition/</link><pubDate>Fri, 03 Jun 2022 15:14:00 +0800</pubDate><author>Hakula</author><guid>https://old.hakula.xyz/posts/note/dsp/speech-recognition/</guid><description><![CDATA[<div class="featured-image">
                <img src="https://hakula-1257872502.file.myqcloud.com/images/3/article-covers/ca13be58-d114-46a1-a0e7-a7877a1474af_94819769.webp" referrerpolicy="no-referrer">
            </div><p>本项目实现了一个基础的语音识别系统，可以从 20 个给定单词中识别一段语音是其中的哪个单词。识别系统基于深度学习，以音频信号的 Mel 频率倒谱系数（MFCC）作为特征，通过一个卷积神经网络（CNN）进行训练。</p>
<p>Digital Signal Processing @ Fudan University, fall 2021.</p>]]></description></item><item><title>DSP - Lab 3: MFCC: Mel 频率的倒谱系数</title><link>https://old.hakula.xyz/posts/note/dsp/mfcc/</link><pubDate>Fri, 27 May 2022 04:29:00 +0800</pubDate><author>Hakula</author><guid>https://old.hakula.xyz/posts/note/dsp/mfcc/</guid><description><![CDATA[<div class="featured-image">
                <img src="https://hakula-1257872502.file.myqcloud.com/images/3/article-covers/b8604b7b-53dd-4620-8cc4-34417baf5ae9_95680357.webp" referrerpolicy="no-referrer">
            </div><p>本实验中，我们实现了一个端点检测算法，并构造了一个 Mel 滤波器组处理信号的能量谱，最后利用离散余弦变换（DCT）得到了信号的 MFCC 系数。</p>
<p>Digital Signal Processing @ Fudan University, fall 2021.</p>]]></description></item><item><title>DSP - Lab 2: 语谱图</title><link>https://old.hakula.xyz/posts/note/dsp/spectrogram/</link><pubDate>Thu, 12 May 2022 08:50:00 +0800</pubDate><author>Hakula</author><guid>https://old.hakula.xyz/posts/note/dsp/spectrogram/</guid><description><![CDATA[<div class="featured-image">
                <img src="https://hakula-1257872502.file.myqcloud.com/images/3/article-covers/83b43957-3908-4479-9350-68bcd4edeec4_102257565.webp" referrerpolicy="no-referrer">
            </div><p>本实验中，我们利用之前实现的 FFT 算法，生成了不同语音片段在不同窗口宽度下的语谱图。</p>
<p>Digital Signal Processing @ Fudan University, fall 2021.</p>]]></description></item><item><title>DSP - Lab 1: FFT: 快速傅立叶变换</title><link>https://old.hakula.xyz/posts/note/dsp/fft/</link><pubDate>Sat, 26 Mar 2022 01:37:00 +0800</pubDate><author>Hakula</author><guid>https://old.hakula.xyz/posts/note/dsp/fft/</guid><description><![CDATA[<div class="featured-image">
                <img src="https://hakula-1257872502.file.myqcloud.com/images/3/article-covers/d6012c5f-1973-467d-9fba-4ac6a11f4a53_102184125.webp" referrerpolicy="no-referrer">
            </div><p>本实验中，我们实现了一个基础的 FFT 算法，使用 Python 编写。</p>
<p>Digital Signal Processing @ Fudan University, fall 2021.</p>]]></description></item><item><title>PRML - Lab 3: 聚类算法</title><link>https://old.hakula.xyz/posts/note/prml/clustering/</link><pubDate>Mon, 14 Jun 2021 03:00:00 +0800</pubDate><author>Hakula</author><guid>https://old.hakula.xyz/posts/note/prml/clustering/</guid><description><![CDATA[<div class="featured-image">
                <img src="https://hakula-1257872502.file.myqcloud.com/images/3/article-covers/bd34cb2a-2392-4817-ae41-e2c37b5ab971_87631904.webp" referrerpolicy="no-referrer">
            </div><p>本次作业利用 NumPy 实现了一个 K-Means 模型和一个 GMM 模型，并利用 Gap Statistic 方法实现了数据集中聚簇数量的自动推测。</p>
<p>Pattern Recognition and Machine Learning (H) @ Fudan University, spring 2021.</p>]]></description></item><item><title>PRML - Lab 2: FNN 模型</title><link>https://old.hakula.xyz/posts/note/prml/fnn/</link><pubDate>Sun, 02 May 2021 19:30:00 +0800</pubDate><author>Hakula</author><guid>https://old.hakula.xyz/posts/note/prml/fnn/</guid><description><![CDATA[<div class="featured-image">
                <img src="https://hakula-1257872502.file.myqcloud.com/images/3/article-covers/b61460f5-786b-4041-9aae-bb993a1011c1_88775351.webp" referrerpolicy="no-referrer">
            </div><p>本次作业完成了选题 1 的实验内容，利用 NumPy 实现了一个 FNN 模型，并在 MNIST 数据集上进行了训练。</p>
<p>Pattern Recognition and Machine Learning (H) @ Fudan University, spring 2021.</p>]]></description></item><item><title>PRML - Lab 1: KNN 算法</title><link>https://old.hakula.xyz/posts/note/prml/knn/</link><pubDate>Fri, 02 Apr 2021 01:50:00 +0800</pubDate><author>Hakula</author><guid>https://old.hakula.xyz/posts/note/prml/knn/</guid><description><![CDATA[<div class="featured-image">
                <img src="https://hakula-1257872502.file.myqcloud.com/images/3/article-covers/0e3a490d-a7f0-4fea-97dd-1bc9db5cbfbf_90743556.webp" referrerpolicy="no-referrer">
            </div><p>本次作业利用 NumPy 实现了一个 KNN 模型。</p>
<p>Pattern Recognition and Machine Learning (H) @ Fudan University, spring 2021.</p>]]></description></item><item><title>Trump-bot: 一个特朗普风格的文本生成 AI</title><link>https://old.hakula.xyz/posts/project/trump-bot/</link><pubDate>Tue, 19 Jan 2021 03:28:00 +0800</pubDate><author>Hakula</author><guid>https://old.hakula.xyz/posts/project/trump-bot/</guid><description><![CDATA[<div class="featured-image">
                <img src="https://hakula-1257872502.file.myqcloud.com/images/3/article-covers/24a5e969-09b5-4da6-95a4-06984cde5c6f_87013646.webp" referrerpolicy="no-referrer">
            </div><p>一个模仿特朗普 Twitter 账号 <a href="https://twitter.com/realDonaldTrump" target="_blank" rel="noopener noreferrer ">@realDonaldTrump</a> 语言风格的简易文本生成 AI，基于 GRU 模型实现。</p>
<p>Natural Language Processing (H) @ Fudan University, fall 2020.</p>]]></description></item><item><title>Naive-FTP: 一个简易的 FTP 服务端 &amp; 客户端</title><link>https://old.hakula.xyz/posts/project/naive-ftp/</link><pubDate>Mon, 11 Jan 2021 03:08:00 +0800</pubDate><author>Hakula</author><guid>https://old.hakula.xyz/posts/project/naive-ftp/</guid><description><![CDATA[<div class="featured-image">
                <img src="https://hakula-1257872502.file.myqcloud.com/images/3/article-covers/a4be99d8-733e-477f-9772-a5f920369d17_86286793.webp" referrerpolicy="no-referrer">
            </div><p>Naive-FTP is a simple FTP server &amp; client, written in Python and TypeScript.</p>
<p>Computer Networks @ Fudan University, fall 2020.</p>]]></description></item></channel></rss>