linux服务器上离线搭建深度学习神经网络环境

步骤

1.Anaconda软件下载并安装

2.搭建虚拟环境

(1)创建离线虚拟环境

conda create -n Py3_8_10 --offline

(2)进入虚拟环境

conda activate Py3_8_10

3.python离线安装

网址: [python安装]: https://repo.anaconda.com/pkgs/main/linux-64/

 conda install python-3.8.10-hdb3f193_7.tar.bz2 --offline

4.pip及setuptools工具离线安装

(1)pip安装

网址：[pip安装]: https://repo.anaconda.com/pkgs/main/linux-64/

 conda install pip-22.1.2-py38h06a4308_0.tar.bz2 --offline

:注意:安装较新版本的pip,否则使用pip工具时会报错并提示安装较新版本

(2)setuptools安装

网址：[pip安装]: https://repo.anaconda.com/pkgs/main/linux-64/

 conda install setuptools-65.5.0-py38h06a4308_0.tar.bz2 --offline

5.torch,torchversion离线安装

(1)torch安装

网址：[torch安装]: http://download.pytorch.org/whl/torch_stable.html

  pip install --no-deps torch-1.10.0+cu111-cp38-cp38-linux_x86_64.whl
```##### (2)torchversion安装
网址：[torch安装]: http://download.pytorch.org/whl/torch_stable.html
```shell
  pip install --no-deps torchvision-0.11.1+cu111-cp38-cp38-linux_x86_64.whl

6.python 第三方包安装

(1)需要什么安装包pip安装

网址：[第三方包安装]: http://mirrors.aliyun.com/pypi/simple/

  pip install --no-deps Pillow-8.1.2-cp38-cp38-manylinux1_x86_64.whl
  pip install --no-deps numpy-1.19.0-cp38-cp38-manylinux1_x86_64.whl
  pip install --no-deps matplotlib-3.4.2-cp38-cp38-manylinux1_x86_64.whl
  pip install --no-deps pyparsing-3.0.1-py3-none-any.whl
  pip install --no-deps cycler-0.11.0-py3-none-any.whl
  pip install --no-deps python_dateutil-2.8.2-py2.py3-none-any.whl
  pip install --no-deps kiwisolver-1.3.1-cp38-cp38-manylinux1_x86_64.whl
  pip install --no-deps six-1.8.0-py2.py3-none-any.wh

(2)遇到错误

a.No module named ‘typing_extensions’

解决办法:
[solve]:https://blog.youkuaiyun.com/qq_40558345/article/details/122150288
下载‘typing_extensions‘文件并放到虚拟环境中‘typing.py’同文件夹中

b.libffi.so.7: cannot open shared object file: No such file or directory

解决办法:
[solve]:https://blog.youkuaiyun.com/duguwanglong/article/details/107662421/
虚拟环境中设置软链接

ln -s libffi.so.6 libffi.so.7