Keep Looking, Don't Settle

install docker, dl package, ssh access for ubuntu

1. install vim

sudo apt remove vim-tiny
sudo apt update
sudo apt install vim

2. change resolution

it is like 1604_2 has fixed this. it can be full screen now. skip this. vi

xrandr --newmode "1920x1080"  173.00  1920 2048 2248 2576  1080 1083 1088 1120 -hsync +vsync
xrandr --addmode Virtual1 1920x1080
xrandr --output Virtual1 --mode 1920x1080

3. install docker

4. install the dl-docker image

4.1. install based on the docker file for particular version 
` docker build -t shmhub/dl-class:cpu -f Dockerfile.cpu.dl_class . `

4.2. change keras to use theano by default
`vi .keras/keras.json`
    "image_dim_ordering": "tf",
    "epsilon": 1e-07,
    "floatx": "float32",
    "backend": "tensorflow"

4.3. test and commit 
`docker run -it -p 8888:8888 -p 6006:6006 -v /$(pwd):/root/sharedfolder shmhub/dl-class:cpu bash`
to commit, please refer to docker document.
`docker commit -m "change keras to use theano by default" -a "author: shm" 62fa48f18e4f shmhub/dl-class:cpu`

5. set up remote access

5.1. install openssh    `sudo apt-get install openssh-server`
5.2. set up static ip for ubuntu 
5.3. port forwarding and dns

6. install the others, like conda on ubuntu

7. install/setup git on ubuntu

sudo apt-get update
sudo apt-get install git

7.1. set up git

7.2. Generating a new SSH key

For all the steps on video, please click:


8. remote access to jupyter notebook

Scenario: On your local computer, you want to open and manipulate an IPython notebook running on a remote computer. We will do this by opening an SSH tunnel. This tunnel will forward the port used by the remotely running IPython instance to a port on the local machine, where it can be accessed in a browser just like a locally running IPython instance.

On ubuntu or any other linux:

First, make sure you install Jupyter notebook in both remote (working station in your offcie) and local (your home computer)

In remote host, open the terminal, change directory to where you have your notebooks and type:

jupyter notebook --no-browser --port=8888

In your local computer, open MS-DOS cmd (if using Windows) or Unix terminal, then type:

ssh -N -f -L localhost:8888:localhost:8888 remote_user@remote_host

If you use Windows, there is no built in ssh terminal, you can use tools such as Putty to build a ssh tunnel follow the following steps:

open putty client: Session -> put your hostname and port = 22, just as what you did normally to connect to you remote server.

Connection -> SSH -> Tunnels

In the destination port box, enter:, And select Local and Auto options.

In the Source port, enter: 8888, Then click the Add button, you will see L8888 in the box under the Remove button.

Now you can open the session, and on your local machine, type

You will be able to get access to the Ipython notebook server running on the remot

9. turn off jupyter notebook token and password

In the ~/.jupyter/ or ~/.jupyter/jupyter_notebook_config.json, add

c.NotebookApp.token = ''
c.NotebookApp.password = ''