Ubuntu16.04+GTX060+Pytorch+Anaconda+Python3.6环境配置的各种坑

1参考的博客

包含各种问题，各种折腾
1.https://blog.youkuaiyun.com/qq_31531635/article/details/79963188
2.https://blog.youkuaiyun.com/wanzhen4330/article/details/81699769
3.https://blog.youkuaiyun.com/qq_39938666/article/details/89601658
4.https://blog.youkuaiyun.com/qq_42165939/article/details/86181585
5.https://blog.youkuaiyun.com/zzq060143/article/details/88042075
6.https://blog.youkuaiyun.com/qq_38671976/article/details/84332140
7.https://blog.youkuaiyun.com/WannaSeaU/article/details/88427010
8.https://blog.youkuaiyun.com/yucicheung/article/details/79095742

2.遇到的坑

1.Pytorch官网上没有CUDA8.0了，采用下面的方式安装

pip install https://download.pytorch.org/whl/cu80/torch-1.0.1.post2-cp36-cp36m-linux_x86_64.whl
pip install torchvision

2.直接安装巨慢无比，添加清华源

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 --set show_channel_urls yes

添加的时候出现有关于yaml的错误，解决方法：使用pip安装升级

3.最终加载模型的时候

import torchvision

1.出现：ModuleNotFoundError: No module named 'Image'错误
2.出现：ModuleNotFoundError: No module named 'six'错误

解决方法：1.pip install -U pillow将pillow包给更新了
2.pip install -U six 全部更新