Torch series: How to use Tensorboard in torch

Torch can also be used Tensorboard, through the installation of crayon can be

Below under the installation of Ubuntu for example to explain the installation process will still encounter a lot of twists and turns of the process

mainly for

① installation crayon will prompt for libssl.so files to be found, so you can add openssl_libdir=/usr/lib/x86_64-linux-gnu/to tell Luarocks,ssl the path to the library file.

②docker will use foreign addresses when downloading mirrors, so you need to replace the Docker mirror.
The installation process is given below: (1) Installing the Crayon Torch End
Luarocks Install Crayon openssl_libdir=/usr/lib/x86_64-linux-gnu/


(2) Installation Docker
sudo apt install Docker.io


(3) Set the mirror of Docker for the domestic mirror https://registry.docker-cn.com
sudo vim/etc/docker/daemon.json
Posted as follows:
{
"Registry-mirrors": ["https://registry.docker-cn.com"]
}

(4) Restart Docker service
sudo service docker restart

(5) Image of the service side of the download mirror crayon
Sudo-h Docker run-d-P 8888:8888-p 8889:8889--name Crayon Alband/crayon

(6) Add code in LUA to record loss or accuracy

Here's a sample: Https://github.com/torrvision/crayon/blob/master/client/lua/README.md#usage-example

Reference:

Https://github.com/torrvision/crayon