深度学习本地环境配置步骤(pytorch、tensorflow框架)
安装步骤
-
Vscode (可在anaconda里面install或者launch,先vs再cuda)
-
Anaconda
-
Cuda(前提 删掉本机的显卡驱动)
-
Cudnn
-
Pytorch(tensorflow)
-
pycharm添加运行环境
3、4、5应考虑版本
在anaconda中安装pytorch的命令(pytorch与cuda的版本):https://pytorch.org/get-started/previous-versions/
其他整理可以借鉴的网站:
1.cuda与本机对应版本选择参考下文,本机选择cuda9.0
https://blog.csdn.net/qq_40992227/article/details/113705112
2.Windows安装tensorflow,配置vs2013, anaconda3.4, cudn9.0,cudnn7.0和pycharm
https://www.cnblogs.com/x739400043/p/9340919.html
3.tensorflow pytorch 与cuda cudnn各版本关系
https://blog.csdn.net/caiguanhong/article/details/112184290
4.Win10 Anaconda下TensorFlow-GPU环境搭建详细教程(包含CUDA+cuDNN安装过程)
https://blog.csdn.net/monk1992/article/details/89514645?utm_medium=distribute.wap_relevant.none-task-blog-baidujs_title-2
5.需要注意的是,安装CUDA和VS时,应该先安装VS,再安装CUDA。
https://blog.csdn.net/qq_45792899/article/details/108789898
6.Anaconda+pycharm+tensorflow安装和环境配置(win10)
https://blog.csdn.net/qq_43629743/article/details/105399552
7.python3.7+Anaconda3+CUDA10.0+cuDNN7.5.0 安装TensorFlow-GPU1.13.1详细步骤(卸载本机cuda)
https://blog.csdn.net/CH_monsy/article/details/109130026
8.检验是否能用gpu