The dataset used contains multilingual questions and respective con- texts (English) as well as answers if applicable. The dataset includes 2379 data points (training and validation) for Russian, 2654 for Finnish, and 2757 for Japanese.
Different NLP techniques were utlized such as LSTM, RoBERTa,and n-grams.