A BERT Baseline for the Natural Questions

Benchmark Model Rank Results
question-answering-on-natural-questions-longBERTjoint#4F1: 64.7