Using RAMP starting-kits¶
To get started working on an existing RAMP challenge:
Clone the starting kit for the challenge from GitHub:
$ git clone https://github.com/ramp-kits/<ramp_kit_name>.git $ cd <ramp_kit_name>
If the starting kit does not include the data (i.e. a
data/folder) you will need to download the data using the
$ python download_data.py
with conda to create a virtual environment with all the required dependencies:
# to create the environment $ conda env create -f environment.yml # to activate the environment $ source activate <ramp_kit_name>
$ pip install -r requirements.txt # install ramp-workflow $ pip install https://api.github.com/repos/paris-saclay-cds/ramp-workflow/zipball/master
Test that the starting kit works:
$ ramp_test_submission --quick-test
Alternatively you can test the kit from a Python shell environment using:
from rampwf.utils.testing import assert_submission assert_submission()
assert_submission()source code for more details.
Now you are ready to write your own solution for the prediction problem. It is
useful to read the starting kit notebook which will introduce the problem,
provide some data analysis and visualisation and walk you through a simple
example solution. There will also be at least one example submission in
submissions/, which show you which files are required for each submission,
how they need to be named and how each file should be structured.
Submiting to a RAMP event¶
To submit your solution to RAMP studio:
Log in at RAMP studio. If it is your first time, you will need to first register.
Find an open event for your RAMP challenge. Events for RAMP challenges can be found underneath the challenge title:
To sign up for an event, click on the open event (not on the challenge). This should take you to a page with the challenge and event as the title. Click on the yellow ‘Sign up’ button to sign up for this event:
Note that registering for RAMP studio and signing up for events are controlled by RAMP administrators, so there can be a delay between asking to sign up and being able to submit.
If you cannot find an open event for the challenge you are interested in, you can request a new event by clicking on ‘Click here if you would like to use this problem for teaching’, on the welcome page of the challenge. Note you need to click on the RAMP challenge and not an event of the challenge to find this button.
Once your sign up as been approved you will have access to a number of menu options on the right hand side:
leaderboard - table of all submissions from all participants of the event. The score metrics, contributivity, time taken to train and test the model, max RAM usage and time of submission will all be shown. By default it will show the most recent submissions first, but you can change this by clicking on the up and down arrows next to the column names.
competition leaderboard - table of the final submission of each participant. It will show only the ‘official’ score metric (see score types), time taken to train and test the model and submission time. This is ordered by the best score, by default.
sandbox - you can make your submissions here. See below for more details.
my submissions - shows all your previous submissions, with separate tables for submissions which are pending and submissions which have finished computing.
To make a submission go to your sandbox. Here, you can edit the required submission files (e.g.,
classifier.py), available in the middle of the page, then save the files. Alternatively, you can upload the required files from your computer, on the right hand side:
Submit by providing a name for your submission (it is a good idea to start the name with your username, so you can easily identify your submission on the leaderboard), then clicking ‘submit’ at the bottom of the page:
The submission is trained and tested on our backend in the same way as
ramp_test_submission does it locally. When your submission is waiting in
the queue and being trained, you can find it in the
‘New submissions (pending training)’ table in ‘my submissions’. Once it is
trained, you get will get an email, and your submission will show up on the
public leaderboard. If there is an error (note you should always test your
submission locally with
ramp_test_submission), it will show up in the
‘Failed submissions’ table in ‘my submissions’. You can click on the error to
see part of the trace. The data set we use at the backend is usually different
from what you find in the starting kit, so the score may be different.