更新

2025-12-16 09:23:53 +08:00
parent 19138d3cc1
commit 9e7efd0626
409 changed files with 272713 additions and 241 deletions
@@ -0,0 +1,159 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "id": "6262ada7",
+   "metadata": {
+    "origin_pos": 0
+   },
+   "source": [
+    "# 安装\n",
+    ":label:`chap_installation`\n",
+    "\n",
+    "我们需要配置一个环境来运行 Python、Jupyter Notebook、相关库以及运行本书所需的代码，以快速入门并获得动手学习经验。\n",
+    "\n",
+    "## 安装 Miniconda\n",
+    "\n",
+    "最简单的方法就是安装依赖Python 3.x的[Miniconda](https://conda.io/en/latest/miniconda.html)。\n",
+    "如果已安装conda，则可以跳过以下步骤。访问Miniconda网站，根据Python3.x版本确定适合的版本。\n",
+    "\n",
+    "如果我们使用macOS，假设Python版本是3.9（我们的测试版本），将下载名称包含字符串“MacOSX”的bash脚本，并执行以下操作：\n",
+    "\n",
+    "```bash\n",
+    "# 以Intel处理器为例，文件名可能会更改\n",
+    "sh Miniconda3-py39_4.12.0-MacOSX-x86_64.sh -b\n",
+    "```\n",
+    "\n",
+    "如果我们使用Linux，假设Python版本是3.9（我们的测试版本），将下载名称包含字符串“Linux”的bash脚本，并执行以下操作：\n",
+    "\n",
+    "```bash\n",
+    "# 文件名可能会更改\n",
+    "sh Miniconda3-py39_4.12.0-Linux-x86_64.sh -b\n",
+    "```\n",
+    "\n",
+    "接下来，初始化终端Shell，以便我们可以直接运行`conda`。\n",
+    "\n",
+    "```bash\n",
+    "~/miniconda3/bin/conda init\n",
+    "```\n",
+    "\n",
+    "现在关闭并重新打开当前的shell。并使用下面的命令创建一个新的环境：\n",
+    "\n",
+    "```bash\n",
+    "conda create --name d2l python=3.9 -y\n",
+    "```\n",
+    "\n",
+    "现在激活 `d2l` 环境：\n",
+    "\n",
+    "```bash\n",
+    "conda activate d2l\n",
+    "```\n",
+    "\n",
+    "## 安装深度学习框架和`d2l`软件包\n",
+    "\n",
+    "在安装深度学习框架之前，请先检查计算机上是否有可用的GPU。\n",
+    "例如可以查看计算机是否装有NVIDIA GPU并已安装[CUDA](https://developer.nvidia.com/cuda-downloads)。\n",
+    "如果机器没有任何GPU，没有必要担心，因为CPU在前几章完全够用。\n",
+    "但是，如果想流畅地学习全部章节，请提早获取GPU并且安装深度学习框架的GPU版本。\n"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "ed0912e5",
+   "metadata": {
+    "origin_pos": 2,
+    "tab": [
+     "pytorch"
+    ]
+   },
+   "source": [
+    "我们可以按如下方式安装PyTorch的CPU或GPU版本：\n",
+    "\n",
+    "```bash\n",
+    "pip install torch==1.12.0\n",
+    "pip install torchvision==0.13.0\n",
+    "```\n"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "4e508102",
+   "metadata": {
+    "origin_pos": 5
+   },
+   "source": [
+    "我们的下一步是安装`d2l`包，以方便调取本书中经常使用的函数和类：\n",
+    "\n",
+    "```bash\n",
+    "pip install d2l==0.17.6\n",
+    "```\n",
+    "\n",
+    "## 下载 D2L Notebook\n",
+    "\n",
+    "接下来，需要下载这本书的代码。\n",
+    "可以点击本书HTML页面顶部的“Jupyter 记事本”选项下载后解压代码，或者可以按照如下方式进行下载：\n"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "d6f62ffb",
+   "metadata": {
+    "origin_pos": 7,
+    "tab": [
+     "pytorch"
+    ]
+   },
+   "source": [
+    "```bash\n",
+    "mkdir d2l-zh && cd d2l-zh\n",
+    "curl https://zh-v2.d2l.ai/d2l-zh-2.0.0.zip -o d2l-zh.zip\n",
+    "unzip d2l-zh.zip && rm d2l-zh.zip\n",
+    "cd pytorch\n",
+    "```\n",
+    "\n",
+    "\n",
+    "注意：如果没有安装`unzip`，则可以通过运行`sudo apt install unzip`进行安装。\n"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "668ab9cb",
+   "metadata": {
+    "origin_pos": 10
+   },
+   "source": [
+    "安装完成后我们可以通过运行以下命令打开Jupyter笔记本（在Window系统的命令行窗口中运行以下命令前，需先将当前路径定位到刚下载的本书代码解压后的目录）：\n",
+    "\n",
+    "```bash\n",
+    "jupyter notebook\n",
+    "```\n",
+    "\n",
+    "现在可以在Web浏览器中打开<http://localhost:8888>（通常会自动打开）。\n",
+    "由此，我们可以运行这本书中每个部分的代码。\n",
+    "在运行书籍代码、更新深度学习框架或`d2l`软件包之前，请始终执行`conda activate d2l`以激活运行时环境。\n",
+    "要退出环境，请运行`conda deactivate`。\n"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "04be90e9",
+   "metadata": {
+    "origin_pos": 12,
+    "tab": [
+     "pytorch"
+    ]
+   },
+   "source": [
+    "[Discussions](https://discuss.d2l.ai/t/2083)\n"
+   ]
+  }
+ ],
+ "metadata": {
+  "language_info": {
+   "name": "python"
+  },
+  "required_libs": []
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}