downloader and corpus builder for WMT metrics from 2015 to the latest
python corpus_builder.py [arguments]--train_path --cache_path --downloaded_dir ...etc. check out the help of the arguments
The source, reference and system translations are only available from the google drive folder, WMT20_data. It's not downloadable from the script. You must prepare the folder in the --downloaded_dir beforehand.