User Guide

• 1. Set up your own server to run RAMP
  • 1.1. Set up the RAMP database
  • 1.2. Set up the RAMP deployment for kits, data, and submissions
  • 1.3. Create an admin user
  • 1.4. Launching a test instance of the RAMP website
    • 1.4.1. Using Gunicorn
• 2. Personalize your RAMP instance
  • 2.1. Front page
  • 2.2. Privacy Policy page
  • 2.3. Sign up and login pages
• 3. Set up a new RAMP event
  • 3.1. Deploy a specific RAMP event
  • 3.2. Launch the dispatcher to train and evaluate submissions
  • 3.3. Launch several dispatchers at once
• 4. Create New Challenge
  • 4.1. requirements.txt file
  • 4.2. problem.py Python file
  • 4.3. submission directory
    • 4.3.1. example submission directory and its content
  • 4.4. data directory
  • 4.5. your_challenge_starting_kit.ipynb Jupiter notebook file
  • 4.6. examplar RAMP kits
• 5. Notes regarding the RAMP workers
  • 5.1. Running submissions in a Conda environment
    • 5.1.1. Event configuration
  • 5.2. Running submissions on Amazon Web Services (AWS)
    • 5.2.1. Prepare AWS image
      • 5.2.1.1. Launching AWS instance
      • 5.2.1.2. Prepare the instance
    • 5.2.2. Prepare AWS Pipeline
    • 5.2.3. Event configuration
  • 5.3. Running submissions with Dask
    • 5.3.1. DaskWorker setup of the main server
    • 5.3.2. DaskWorker setup with a local cluster
    • 5.3.3. DaskWorker setup with a remote cluster
  • 5.4. Create your own worker
• 6. Interactions with the RAMP website
  • 6.1. Interactions as an administrator
    • 6.1.1. Approve users’ access
    • 6.1.2. Event menu
  • 6.2. Interactions as a user
    • 6.2.1. Problems & events
    • 6.2.2. Challenge menu
• 7. Useful tips using RAMP
  • 7.1. How to restart a failed submission manually
  • 7.2. Running a standalone worker without connection to database
  • 7.3. Create automatically a conda environment linked to an event
  • 7.4. Update automatically a conda environment linked to an event
• 8. Server maintenance
  • 8.1. Database migrations