inscite / saturn

Saturn: Web IDE for Machine Learning

Saturn

Saturn is opensource Machine Learning Framework.

Features

• Support multi-language: python, node, markdown
• Distribution your work using remote
• Unit test available, even though asynchronous loop
• Mailing on event (error, finish)

Quick Reference

• common usage
• chart demo
• crawler demo
• machine learning
• web server

Related Projects

• saturn
  • Web based GUI framework
• saturn-core
  • Core for running and compile
  • will be update language module for supporting more languages
• saturn-docker
  • Docker for easy install

---

Installation

Using Docker

• Installation & Run

  docker pull proin/saturn
  docker run -it --name saturn -v ~/workspace/saturn:/workspace -p 3000:3000 proin/saturn

• Run as Daemon

  docker run -d --name saturn -v ~/workspace/saturn:/workspace -p 3000:3000 proin/saturn
  docker kill saturn #if you want to kill process

• Restart

  docker restart saturn

• Attach Running Container

  docker exec -i -t saturn /bin/bash

Build from Source

• Clone & Build

  npm install -g lwot
  git clone https://github.com/proin/saturn 
  cd saturn
  bower install
  cp ./controller/express/config-sample.json ./controller/express/config.json
  lwot install
  lwot build
  lwot express run
  # then started at your browser, localhost:3000

• Run as Daemon

  lwot express forever start # after launched, press Ctrl+C
  lwot express forever stop # stop daemon
  lwot express forever log # show logs

• Update

  • you can easily update saturn in web. just click update.
  • or run sh update.sh in Unix systems

Configuration

• you can change configuration editing by ./controller/express/config.json file

  {
      "port": 3000, // server port to bind
      "home": "./saturn-project", // workspace home, start from USER HOME
      "dev": true, // display express log
      "log": true, // display thread log
      "hostname": "http://localhost", // direct url
      "smtp": { // emailjs configuration object
          "host": "smtp.your-site.net",
          "user": "[email protected]",
          "password": "password",
          "ssl": true
      },
      "mailingList": "[email protected], [email protected]", // who are received? 
      "mailingOn": {
          "error": true, // mail when error occured 
          "finish": true // mail when work finished
      },
      "readonly": true, // readable project for whom not signin users
      "user": "admin", // admin user
      "password": "admin123", // admin password
      "session": {
          "secret": "sAtuRN-seSsIon-sEcreT",
          "resave": false,
          "saveUninitialized": false
      },
      "MAX_LOG": 500, // display log size
      "encrypt": "sAtuRn"
  }

---

Overview

Machine Learning

• you can use machine learning libraries in saturn
• now saturn supports python (including tensorflow), R, torch
• in saturn-docker installation, all frameworks are embedded in there.

Code with Markdown

• saturn support markdown editor in project
• you can write down code more effectively

Mailing on Event

• you can receive email when the work is finished or occured warning or error
• you no longer have to wait for the job to finish

Graph

• you can use graph by making structure & calling function.
• for more usage, refer chart demo

Remote Server

• Distribute some work using remote features
• you can easily distribute in [project > remote]

System Terminal

• You can use system terminal where saturn is installed
• This allow to only GRANTALL user

---

Future Works

• Support Machine Learning Frameworks if it is possible
  • Torch
  • Caffe
  • R