CodaLab Bundle Service and Comand-Line Interface

This repository provides the Python backend for operating on bundles (creating, running, uploading, etc.). The API exposed is used by both the website and the command-line utility (provided in this package).

Getting started

1. After cloning the repository, install the appropriate packages:

   virtualenv codalab_env --no-site-packages

   codalab_env/bin/pip-2.7 install sqlalchemy

2. Setup the interface to

   export PATH=$PWD/codalab/bin

   cl # This should print out a help message.

   cl worker # Start up a worker in another shell

3. Let's populate your CodaLab service actually with some initial content. Normally, you would have the content on your local machine and just upload it to CodaLab. Just to get some basic programs and datasets into your system, you will first run some scripts to download them to your local machine and then upload them into the CodaLab instance running on your local machine.

For machine learning:

(cd examples/weka && ./download.sh)
(cd examples/weka && ./upload.sh)
(cd examples/uci_arff && ./download.sh)
(cd examples/uci_arff && ./upload.sh)
(cd examples && ./basic_ml.sh)  # Run basic ML pipeline

For NLP:

(cd examples/stanford_corenlp && ./download.sh)
(cd examples/stanford_corenlp && ./upload.sh)
(cd examples/lewis_carroll_poems && ./upload.sh)
cl run stanford_corenlp lewis_carroll_poems 'program/run input output' --auto