深度学习anaconda、pytorch环境设置

pip源

pip install -i https://pypi.tuna.tsinghua.edu.cn/simple/   [pkgname]
pip install -i https://pypi.doubanio.com/simple  [pkgname]

# 更新包
pip install --upgrade pkg

pytorch anaconda 安装

更新conda
conda update -n base -c defaults conda 
安装pytorch
conda install pytorch torchvision cuda90 -c pytorch

环境输出到txt

pip freeze >requirements.txt

opencv安装

pip install opencv_python==3.4.2.16 
pip install opencv-contrib-python==3.4.2.16

查看cuda版本

cat /usr/local/cuda/version.txt

查看cudnn版本

cat /usr/local/cuda/include/cudnn.h | grep CUDNN_MAJOR -A 2

• pytorch查看cudnn版本

import torch
print(torch.backends.cudnn.version())

安装pytorch很慢，离线安装pytorch

https://blog.csdn.net/chenxp106/article/details/82793149
https://www.jianshu.com/p/5aec0a02b80b
安装anaconda，下载离线版本pytorch
https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/pytorch/linux-64/

conda install numpy mkl cffi
conda install --offline pytorch-1.1.0-py3.7_cuda9.0.176_cudnn7.5.1_0.tar.bz2

conda 换源

conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge 
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/msys2/
conda config --set show_channel_urls yes
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/pytorch/

conda 重置源

conda config --remove-key channels

conda 环境

# 新建环境
conda create -n envname python=3.6
# 切换、激活环境
conda info -e 
source activate envname # linux
activate envname # windows
# 删除环境
conda remove -n envname --all

启动ssh

service ssh restart

pycharm

1. ImportError: libcudart.so.10.0: cannot open shared object file: No such file or directory
   sudo ldconfig /usr/local/cuda-10.0/lib64

windows下conda命令窗口目录无法切换？

使用完整的路径运行目标安装程序 python D:/program_path/program.py

添加环境变量

export PATH=$PATH:/your_new_path
# $PATH 是指原来的环境变量，不加这个则会删除原来的，只有新的

# 运行报错 
export PATH=$PATH:/usr/local/cuda-10.0/bin
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda-10.0/lib64
export LIBRARY_PATH=$LIBRARY_PATH:/usr/local/cuda-10.0/lib64
source /etc/profile

# 永久添加环境变量
vi ~/.bashrc
export ...
source ~/bashrc

jupyter notebook

conda install nb_conda
pip install jupyter notebook
## 启动
jupyter notebook --ip 192.11..... --port 88..
根据返回的ip启动 jupyter notebook

anaconda虚拟环境下重装 cudatoolkit，cudnn

https://blog.csdn.net/qq_38163755/article/details/88583016

1. 新建虚拟环境
2. 如果已经有虚拟环境则需要删除 相关包

conda uninstall cudnn
conda uninstall cudatoolkit

3. 安装cudatoolkit和cudnn

conda install cudatoolkit=9.0
conda install cudnn