环境变量

vim ~/.bashrc  # 输入i，变为insert插入模式
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda/lib64
export PATH=/usr/local/cuda-10.0/bin${PATH:+:${PATH}}
# 按esc，然后输入：wq！保存并退出
source ～/.bashrc #更新

之前报错的信息是：

2019-12-18 17:21:44.371903: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcufft.so.10.0'; dlerror: libcufft.so.10.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: :/usr/local/cuda/extras/CUPTI/lib64
2019-12-18 17:21:44.371936: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcurand.so.10.0'; dlerror: libcurand.so.10.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: :/usr/local/cuda/extras/CUPTI/lib64
2019-12-18 17:21:44.371981: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcusolver.so.10.0'; dlerror: libcusolver.so.10.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: :/usr/local/cuda/extras/CUPTI/lib64
2019-12-18 17:21:44.372015: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcusparse.so.10.0'; dlerror: libcusparse.so.10.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: :/usr/local/cuda/extras/CUPTI/lib64

问题根源首先是cuda10.2与tensorflow2.0不匹配所致。安装了cuda10.0,
然后手动找到以上缺失的包的位置，目前看或者在:/usr/local/cuda/extras/CUPTI/lib64,或者在:/usr/local/cuda/lib64,根据所在位置更新以上的环境变量后重新安装tensorflow-gpu

查看cuda与cudnn版本的命令

nvcc --version
nvidia-smi
cat /usr/local/cuda/version.txt
cat /usr/include/cudnn.h | grep CUDNN_MAJOR -A 2

python中检查tf2是否可用gpu：

python
import tensorflow as tf
physical_devices = tf.config.experimental.list_physical_devices('GPU')
print(physical_devices)

多版本cuda并存

sudo rm -rf cuda
sudo ln -s /usr/local/cuda-9.1 /usr/local/cuda

https://www.cnblogs.com/kevingrace/p/8072860.html

下载：
https://developer.nvidia.com/rdp/cudnn-download

vim命令行

先按ESC进入Command模式，然后输入“：wq”，回车就可以保存并退出了。
vi的其他保存/退出命令：
:w 保存；
:w filename 另存为filename；
:wq! 保存并强制退出；
:wq! filename 注：以filename为文件名保存后强制退出；
:q! 强制退出；
:x 保存并退出（仅当文件有变化时保存）

访问远程tensorboard

# 远程登陆服务器
ssh -L 16006:127.0.0.1:6006 account@server.address
tensorboard --logdir=./models
# 本地浏览器打开
http://127.0.0.1:16006/

vscode

autoDocstring
Better Comments
vscode-icons
kite