What are all the files?

Implementation

  • src/: directory for storing code your algorithm needs.
  • src/submission.py or src/evaluation.py: Python file for developing the submission or evaluation code. Can be run locally on a copy of the benchmark data.
  • src/run_submission.py or src/run_evaluation.py: Python file containing the code for running a submission or evaluation inside the Docker container.
  • requirements.txt: file for listing the Python packages your algorithm depends on.

Local data handling

  • data/input/: local directory used to read the input data from.
  • data/input/test_data, data/input/ground_truth, and/or data/input/implementation_output: example data for the boilerplate submission or evaluation. These files need to be replaced with data from the benchmark.
  • data/: local directory for storing the output of the algorithm.
  • data/output_data_appears_here.txt: example file to show where the local output is stored (feel free to delete this file).
  • data/.gitignore and data/input/.gitignore: configuration files to make sure the data directories are created on your computer. If you are using git for version control, these files also ensure that data files are not stored in your repository.

Scripts

  • build.sh: script for building the container (without running the code).
  • test.sh: script for building the container and running the code.
  • push.sh: script for submitting the algorithm to the EYRA Benchmark Platform.
  • export.sh: script for exporting the image of your container to tar file.

Note

export.sh is currently not used. Alternative ways of submitting Docker containers will be added to the EYRA Benchmark Platform later.

Miscellaneous

  • Dockerfile: specifies the Docker container. If your algorithm only uses Python packages, there is no need to change this file.