Overcoming Input Length Constraints of TransformersUsing extractive summarization to train Transformers on long documents efficiently.