Sparsity and Sentence Structure in Encoder-Decoder Attention of Summarization Systems

30908

Knowledge Triples

Source Entity	Relationship	Target Entity
Modified Encoder Architectures Method	includes	LED Method
Modified Encoder Architectures Method	includes	LoBART Method
Self Attention Method	comparesWith	Encoder Decoder Attention Method
Transformer Models Method	achieves	Summarization Task
Transformer Models Method	uses	Self Attention Method