Context Quality Matters in Training Fusion-in-Decoder for Extractive Open-Domain Question Answering

Kosuke Akimoto, Kunihiro Takeoka, Masafumi Oyamada