Today on the server fiddle, and certainly is a long way. Read the readme, First Install nvidia-docker, jump into it:
. The Make the Sure you have have Installed at The NVIDIA Driver and A Supported Version of Docker for your Distribution's (See the Prerequisites)
Be sure to install the NVIDIA driver, a supported version of Docker. See the Prerequisites:
the Prerequisites
at The List of the Prerequisites for running NVIDIA-Docker IS Described below 2.0.
The For Information ON How to install for your Linux Distribution's Docker, REFER to Please at The Docker Documentation.
The GNU / the x86_64 with the Linux Kernel Version> 3.10
Docker> = 1.12
the NVIDIA with the GPU Architecture> the Fermi (2.1)
the NVIDIA Drivers ~ = 361.93 (UNTESTED ON older versions)
Your Driver Version Might your limit the CUDA Capabilities (See the CUDA requirements)
. 1. Referring kernel version currently in use:
the uname -R & lt
4.15.0-47 Generic-
2. NVIDIA driver version:
view previously saved links: Driver Version: 418.56
means that the installation nvidia-docker must be installed Docker.
So the next installation Docker:
https://www.runoob.com/docker/ubuntu-docker-install.html
when you run the installation script Docker, error:
(1)
ZBP @ ZBP-Precision-Tower-7910: ~ $ wget - qO- https://get.docker.com/ | SH
#Executing Docker install Script, the commit: 2f4ae48
- -E-SH -c APT sudo GET Update -QQ> / dev / null
W: warehouse " http://extras.ubuntu.com/ubuntu xenial Release" Release file no.
E: Unable to download http://extras.ubuntu.com/ubuntu/dists/xenial/main/source/Sources 404 Not Found [IP: 2001: 67c: 1360: 8c01 23 :: 80]
E: section index files failed to download . If you ignore them, they will revert to the old index file.
E: Unable to parse package file /var/lib/apt/lists/extras.ubuntu.com_ubuntu_dists_xenial_main_binary-amd64_Packages (1)
E: Unable to parse package file /var/lib/apt/lists/extras.ubuntu.com_ubuntu_dists_xenial_main_binary-i386_Packages ( 1)
E: unable to parse package file /var/lib/apt/lists/extras.ubuntu.com_ubuntu_dists_xenial_main_binary-all_Packages (1)
E: unable to parse package file /var/lib/apt/lists/extras.ubuntu.com_ubuntu_dists_xenial_main_i18n_Translation- zh% 5fCN (1)
E: Unable to parse package file /var/lib/apt/lists/extras.ubuntu.com_ubuntu_dists_xenial_main_i18n_Translation-zh (1)
E: Unable to parse package file /var/lib/apt/lists/extras.ubuntu.com_ubuntu_dists_xenial_main_i18n_Translation-en% 5fUS (1)
E: unable to parse package file /var/lib/apt/lists/extras.ubuntu.com_ubuntu_dists_xenial_main_i18n_Translation-en (1)
W: You may want to run apt-get update to solve these problems
E: the package cache file damage to the
server (2) immediately turned teacher, run uname -r, run the same command as above, after successful. Docker display version enter docker -v.
https://blog.csdn.net/zbrwhut/article/details/80795073
I think docker installed, although docker run hello world, being given.
Installation began nvidia-docker. So then the following official link to start the whole:
start typing commands at The Package Penalty for Repositories directly from the Add:
https://github.com/NVIDIA/nvidia-docker
in the implementation of docker run, and just reported the same mistake, or permissions problems. We began this whole rights issue:
sudo groupadd docker # Add docker user group
sudo gpasswd -a zbp docker # docker will be added to the login user group
sudo systemctl restart docker docker restart the service
su root to root
su zbp then switch to the original application user above command will take effect
after the modification like this, it seems, anyway, before the report is wrong and not the same. Then again run the last sentence, to run successfully.
Afternoon come get another bar. . .
Nvidia-docker seems to be installed, and then run the following.
File Docker Build: https://blog.csdn.net/jiangyu1013/article/details/79967440
and links exactly the same.
Implementation of the docker build -t second:. V1.0 command is executed for a long time, there is a timeout interrupt, and once installed panda version did not succeed, because the wrong version numpy. After then.
Next: Launch the docker container
reference link https://www.cnblogs.com/linjiqin/p/8608975.html in a startup mode docker embodiment
performed after the make clean and make:
given 1: No opencv, enter the command sudo apt-get install libopencv-dev
being given 2: the nvcc fatal: Unsupported GPU Architecture 'compute_20'
https://blog.csdn.net/weixin_37983220/article/details/85341058
different rows according to their release notes in CUDA Makefile.config file.
After successfully make up.
The next execution TRAIN (gpu_machine):
error 1: syntax error. python3 in print must be enclosed in parentheses.
Given 2: ModuleNotFoundError: No module named ' osgeo'
It seems to be related to the installation GDAL. But when they installed the error. fatal error: cpl_vsi_error.h: No such file or directory
I a God, ah, finally solved the problem. Reference links: https://yq.aliyun.com/articles/592400
a prompt start no gdal-config, then use sudo apt-get install gdal-config, and later install GDAL various unsuccessful. Finally, still follow the above link to install the GDAL: pip install GDAL == 1.11.2
and later run python yolt2 ... finally just do not report the wrong.
Given 3: FileNotFoundError: [Errno 2] No such file or directory: '/ raid / local / src / yolt2 /'
the inner path and program issues.
Solution: https://www.fujieace.com/python/str-bytes.html
Solution:
Ha ha ha, finally run out. Shots are as follows:
the network structure are out, ha ha ha.
Online download a yolo.weights,
https://download.csdn.net/download/sunyuan0064/10286465
upload up, and error:
After the file name changed, error:
here, seems to be because there is no data set.
Anyway, you can always run up. After the data is set, take a look at what the code. This example must take a good clear.