any guidance on ontonotes prep?

[mprorock]

Since ontonotes requires direct licensing from source are there any pointers or scripts to prep for how to convert the corpus format over to the expected train / dev / test splits so that ud_ontonotes.tar.gz can be properly replicated locally?

[dakinggg]

I unfortunately don't have the exact details to provide you, @DeNeutoy might remember some more details, but I believe it was the same splits/processing spacy uses which appear to be referenced in a couple places (explosion/spaCy#5276, explosion/spaCy#3587 (comment)).