The dataset was compiled using Wiktionary inflection tables. The resulting data is 2,686 lemmas and 241,323 word forms. It consists of mostly predicates, so the resulting lemmas are mainly verbs and a smaller number of adjectives. The scraped annotated paradigms turned out to be quite similar (mainly because the adjective paradigm is a reduced verb paradigm) and do not represent all forms of verbs and adjectives. It is important to note that different types of converbs were tagged consistently.
Maria Nepomniashchaya, Daria Rodionova, Anastasia Yemelina