DLMonitor: Monitoring all things happening in deep learning

Purpose

This project aims to save time and energy for deep learning folks. It monitors new things on multiple sources and find out those important to you. Currently, the data sources include:

• Arxiv papers
• Tweets
• Reddit posts

Take a look at the public server: https://deeplearn.org

Install

1. Install postgres server
2. pip install -r requirements.txt
3. sudo apt-get install poppler-utils

Setup database

1. Create a .env file in the project root.

DATABASE_USER=dlmonitor
DATABASE_PASSWD=something

TWITTER_CONSUMER_KEY=something
TWITTER_CONSUMER_SECRET=something
TWITTER_ACCESS_TOKEN=something
TWITTER_ACCESS_SECRET=something

SUPERVISORD_PASSWD=something

2. Create database

Run bash bin/create_db.sh

Install Quick Read dependencies

1. install cpan
2. install text::Unidecode in cpan
3. git clone https://github.com/brucemiller/LaTeXML
4. perl Makefile.PL; make; make install

Fetch resources

Fetch Arxiv papers and tweets.

python bin/fetch_new_sources.py all

Run test server

PYTHONPATH="." python dlmonitor/webapp/app.py

Setup production server

1. Install nginx

2. Copy configuration files for supervisord and nignx

bash bin/config_server.sh

3. Start Gunicorn processes through supervisord

bash bin/start_supervisord.sh

4. Start arxiv source loading worker

PYTHONPATH="." python bin/auto_load_arxiv.py --forever