HyperAI초신경

Relation Extraction On Ade Corpus

평가 지표

NER Macro F1

RE+ Macro F1

평가 결과

이 벤치마크에서 각 모델의 성능 결과

			Paper Title
ITER	92.63 ± 0.89	85.6 ± 1.42	ITER: Iterative Transformer-based Entity Recognition and Relation Extraction
PFN (ALBERT XXL, average aggregation)	91.5	83.9	An Information Extraction Study: Take In Mind the Tokenization!
Deeper	89.48	83.74	Deeper Task-Specificity Improves Joint Entity and Relation Extraction
PFN (ALBERT XXL, no aggregation)	91.3	83.2	A Partition Filter Network for Joint Entity and Relation Extraction
SpERT.PL (without overlap and BioBERT)	91.14	82.39	Joint Entity and Relation Extraction from Scientific Documents: Role of Linguistic Information and Entity Types
REBEL (including overlapping entities)	-	82.2	REBEL: Relation Extraction By End-to-end Language generation
SpERT.PL (with overlap and BioBERT)	91.17	82.03	Joint Entity and Relation Extraction from Scientific Documents: Role of Linguistic Information and Entity Types
CMAN	89.40	81.14	Modeling Dense Cross-Modal Interactions for Joint Entity-Relation Extraction
Table-Sequence	89.7	80.1	Two are Better than One: Joint Entity and Relation Extraction with Table-Sequence Encoders
CLDR + CLNER	88.3	79.97	Imposing Relation Structure in Language-Model Embeddings Using Contrastive Learning
SpERT (without overlap)	89.25	79.24	Span-based Joint Entity and Relation Extraction with Transformer Pre-training
SpERT (with overlap)	89.28	78.84	Span-based Joint Entity and Relation Extraction with Transformer Pre-training
Relation-Metric	87.02	77.19	Neural Metric Learning for Fast End-to-End Relation Extraction
multi-head + AT	86.73	75.52	Adversarial training for multi-context joint entity and relation extraction
multi-head	86.40	74.58	Joint entity recognition and relation extraction as a multi-head selection problem

0 of 15 row(s) selected.