Frustratingly Easy Natural Question Answering

Benchmark Model Rank Results
question-answering-on-natural-questions-longBERTwwm + SQuAD 2–F1: 68.2