#WEB SERVER FOR DEEP NEURAL NETWORKS

##Step 1 – Install Linux, Python, Pip, Git You can use any web server which includes a linux distribution.

Python version 2.7 comes preinstalled on many Linux OS distributions, if it is not installed, it can be downloaded from https://www.python.org/downloads/

Pip comes preinstalled on many Linux OS distributions, if it is not installed, it can be downloaded from https://pip.pypa.io/en/stable/installing/

GIT comes preinstalled on many Linux OS distributions, if it is not installed, it can be downloaded using apt-get (example for Ubuntu) using:

$ sudo apt-get install git-all 

##Step 2 – Clone GitHub repository Navigate to your working directory, and then run:

$ git clone https://github.com/yplatner/dnnserver

This will create a GIT repository with all the files needed to run a DNN server.

##Step 3 – Add requirements Inside your working folder, run:

$ pip install –r requirements.txt

This will install Django and Lasagne frameworks, and all pre-requirements.

##Step 4 – Run with integrated web server and database In order to run on localhost port 8080, run the following command –

$ python manage.py runserver 0.0.0.0:8080

##Step 5 (optional) – Install and run GUNUNICORN web server Gunicorn (‘Green Unicorn’) is a pure-Python WSGI server for UNIX. It has no dependencies and is easy to install and use. To install it:

$ pip install gununicorn

To run your server:

$ gununicorn dnnserver.wsgi

##Step 6 (optional) – Install and run POSTGRES database In order to install and configure postgres, run the following commands in your terminal:

$ sudo apt-get install libpq-dev postgresql postgresql-contrib
$ sudo su - postgres
$ psql
$ CREATE DATABASE myproject;
$ CREATE USER <user> WITH PASSWORD '<password>';
$ ALTER ROLE <user> SET client_encoding TO 'utf8';
$ ALTER ROLE <user> SET default_transaction_isolation TO 'read committed';
$ ALTER ROLE <user> SET timezone TO 'UTC';
$ GRANT ALL PRIVILEGES ON DATABASE dnnserver TO <user>;
$ \q
$ Exit

Now we need to edit the Django settings file in order to enable us to connect to the POSTGRES database. The relevant file is: dnnserver/settings.py

Replace the following lines :

$ DATABASES = {
$     'default': {
$         'ENGINE': 'django.db.backends.sqlite3',
$         'NAME': os.path.join(BASE_DIR, 'db.sqlite3'),
$     }
$ }

With :

$ DATABASES = {
$     'default': {
$         'ENGINE': 'django.db.backends.postgresql_psycopg2',
$         'NAME': 'dnnserver',
$         'USER': '<user>',
$         'PASSWORD': '<password>',
$         'HOST': 'localhost',
$         'PORT': '',
$     }
$ }

Now, run the following commands :

$ python manage.py makemigrations
$ python manage.py migrate
$ python manage.py createsuperuser
<create user + password>
$ python manage.py runserver 0.0.0.0:8000

##Step 7 (optional) – Enable GPU optimizations Install the latest CUDA Toolkit and possibly the corresponding driver available from NVIDIA: https://developer.nvidia.com/cuda-downloads

Closely follow the Getting Started Guide linked underneath the download table to be sure you don’t mess up your system by installing conflicting drivers.

After installation, make sure /usr/local/cuda/bin is in your PATH. Also make sure /usr/local/cuda/lib64 is in your LD_LIBRARY_PATH, so the toolkit libraries can be found.

NVIDIA also provides a library for common neural network operations that especially speeds up Convolutional Neural Networks (CNNs). It can be obtained from NVIDIA (after registering as a developer): https://developer.nvidia.com/cudnn

To install it, copy the .h files to /usr/local/cuda/include and the lib files to /usr/local/cuda/lib64.

To configure Theano to use the GPU by default, create a file .theanorc directly in your home directory, with the following contents:

$ [global]
$ floatX = float32
$ device = gpu

Optionally add “allow_gc = False” for some extra performance at the expense of (sometimes substantially) higher GPU memory usage.