1 paper · 1 filter
Christophe Ropers, David Dale, Prangthip Hansanti +9
Assessing performance in Natural Language Processing is becoming increasingly complex. One particular challenge is the potential for evaluation datasets to overlap with training da…