no code implementations • COLING (WANLP) 2020 • Ori Terner, Kfir Bar, Nachum Dershowitz
To measure the contribution of context to learning, we also tested word-shuffled data, for which the error rises to 2. 5%.
Transliteration