1 - Run the setup file: ./setup.sh.
1 - Run ./update.sh to ensure you are running the latest version in master for each framework.
3 - Make sure the folder structure of the example you want to run is in place. You can find the folder name in the wrapper function run_experiment of your example. If it is not in place (it must have a folder in the results folder with a folder for all three representations, and a main.csv in each of these subfolders with the relevant column names), you can run the initialize.py programme: python initialize.py <folder_name>.
4 - Run the programme once: python run_example.py -e <example_name> -s <seed (int, optional)> -r <representation (int, optional)>, or run it for each representation with 30 seeds: parallel_comparison.sh -e <example_name>. If you want to run the cnn example, run the bash script with more memory and time: parallel_comparison_cnn.sh
Download the lastest run from: https://cirrus.ciencias.ulisboa.pt/owncloud/s/MpBKaPtJef2eTE5
Place those 2 parquet files in the folder results