Learning to Ignore: Long Document Coreference with Bounded Memory Neural Networks

Open paper

Benchmark	Model	Rank	Results
coreference-resolution-on-conll-2012	U-MEM* + SpanBERT-large	#10	Avg F1: 79.6
coreference-resolution-on-ontonotes	U-MEM* + SpanBERT	#11	F1: 79.6