I am trying to build a question-answering model that contains multiple words as answers. The details of the dataset are as below.
Vocabulary size = 198 traini