深度学习环境部署

配置conda

虚拟环境可以使用，conda，主要是为了环境隔离，当然用pyenv效果也是一样的

conda create -n -y myenv python=3.12
conda activate myenv

配置镜像源

可以在~/.condarc,也可以在conda安装目录下，优先读取~/.condarc

channels:
  - defaults
show_channel_urls: true
channel_alias: https://mirrors.tuna.tsinghua.edu.cn/anaconda
default_channels:
  - https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
  - https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free
  - https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/r
  - https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/pro
  - https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/msys2
custom_channels:
  conda-forge: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud
  msys2: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud
  bioconda: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud
  menpo: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud
  pytorch: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud
  simpleitk: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud

Conda 命令

conda install --name 环境名  包名

配置运行环境

装包

安装深度学习环境需要的包

pip install -y jupyter d2l torch torchvision rise pandas jupyter_contrib_nbextensions

配置jupyter

生成配置文件

jupyter notebook --generate-config`, 生成的配置文件在`~/.jupyter/jupyter_notebook_config.py

编辑配置文件

c = get_config()  #noqa
c.ExtensionApp.open_browser = False
c.ServerApp.allow_origin = '*'
c.ServerApp.allow_remote_access = True
c.ServerApp.base_url = '/deeplearning'
c.ServerApp.custom_display_url = '/deeplearning'
c.ServerApp.disable_check_xsrf = True
c.ServerApp.ip = '*'
c.ServerApp.notebook_dir = '/data/homework/deeplearning'
c.ServerApp.port = 1111 
c.ServerApp.token = ''
c.JupyterAppNotebookApp.token=''

配置解释：

c = get_config()  #noqa
c.ExtensionApp.open_browser = False  #关闭浏览器
c.ServerApp.allow_origin = '*'  #解决跨域
c.ServerApp.allow_remote_access = True  #允许远程访问
c.ServerApp.base_url = '/deeplearning'  #设置路由第一段path，默认有很多path单第一段各不相同，对nginx配置不友好
c.ServerApp.custom_display_url = '/deeplearning'  #同上
c.ServerApp.disable_check_xsrf = True  # 关闭xsrf
c.ServerApp.ip = '*'  # 监听0.0.0.0
c.ServerApp.notebook_dir = '/data/homework/deeplearning'  # 设置工作目录，按个人实际情况，后续新建文件会在这个目录下
c.ServerApp.port = 1111  #自定义端口
c.ServerApp.token = ''  #关闭token验证，不然每次访问都要输入token，每次启动token都变
c.JupyterAppNotebookApp.token=''

开机启动

vim /etc/systemd/system/jupyter-notebook.service
# 内容，按需修改
[Unit]
Description=Jupyter Notebook NO.1
After=network.target

[Service]
Type=simple
User=username
EnvironmentFile=/home/username/.jupyter_env
WorkingDirectory=/home/username/
Restart=on-failure
RestartSec=10

ExecStart=/usr/local/anaconda3/bin/jupyter notebook

StandardOutput=file:/path2output/output.log
StandardError=file:/path2error/error.log

[Install]
WantedBy=multi-user.target

vim /home/username/.jupyter_env

__conda_setup="$(CONDA_REPORT_ERRORS=false '/usr/local/anaconda3/bin/conda' shell.bash hook 2> /dev/null)"
if [ $? -eq 0 ]; then
    \eval "$__conda_setup"
else
    if [ -f "/usr/local/anaconda3/etc/profile.d/conda.sh" ]; then
        . "/usr/local/anaconda3/etc/profile.d/conda.sh"
        CONDA_CHANGEPS1=false conda activate base
    else
        \export PATH="/usr/local/anaconda3/bin:$PATH"
    fi
fi
unset __conda_setup
conda activate deeplearning

配置学习环境

下载代码包

## 下载官方代码包
wget https://zh-v2.d2l.ai/d2l-zh.zip
unzip d2l-zh.zip
## 下载ppt代码包
git clone https://github.com/d2l-ai/d2l-zh-pytorch-slides.git 

启动项目jupyter notebook ，默认走配置文件启动

访问

nginx反向代理

    # /deeplearning是和jupyter配置中的base_url 一致；端口同上
    location  /deeplearning {
        proxy_pass http://127.0.0.1:1111;
        proxy_http_version 1.1; #必须设置，否则400
        proxy_redirect http:// https://;
        proxy_set_header Host $host:443/path;
        proxy_set_header Upgrade $http_upgrade;
        proxy_set_header Connection upgrade;
        proxy_set_header Accept-Encoding gzip;

    }

1. 可以将域名指向nginx，通过域名访问，如www.abc.com/deeplearning/tree
2. 可以vscode 配置remote server 访问，url:www.abc.com/deeplearning?token=