Relation Extraction on DocRED human-annotated (test) benchmark leaderboard · SOTA2 Research

Relation Extraction on DocRED human-annotated (test)

59.5Micro F1FineCL

Evaluation Results

Method	Links
FineCL2022.05		59.5	57.1
ERICA_RoBERTa2022.05		59.1	56.9
WCL_RoBERTa2022.05		58.5	56.2
RoBERTa2022.05		58.2	55.9
ERICA_BERT2022.05		57.4	55.2
SpanBERT*2022.05		57.3	55
CorefBERT*2022.05		57	54.5
MTB*2022.05		56.9	54.3
ERNIE*2022.05		56.6	54.2
BERT2022.05		56.6	54.4
HINBERT*2022.05		55.6	53.7
CP*2022.05		55.2	52.7
BiLSTM*2022.05		51.1	50.3
FineCL2022.05		50.3	48.3
ERICA_RoBERTa2022.05		50.1	48.1
WCL_RoBERTa2022.05		49.4	47.5
ERICA_BERT2022.05		48.5	46.4
RoBERTa2022.05		47.6	45.7
ERNIE*2022.05		46.7	44.2
SpanBERT*2022.05		46.4	44.5
MTB*2022.05		46.1	44.1
CorefBERT*2022.05		46	43.7
BERT2022.05		45.2	43.1
CP*2022.05		44.8	42.6
CNN*2022.05		42.3	40.3
FineCL2022.05		33.2	31.2
CorefBERT*2022.05		32.8	31.2
SpanBERT*2022.05		32.2	30.4
CP*2022.05		30.3	28.7
ERICA_RoBERTa2022.05		30	28.2
RoBERTa2022.05		29.6	27.9
MTB*2022.05		29	27.6
ERNIE*2022.05		26.7	25.5
ERICA_BERT2022.05		22.9	21.7
WCL_RoBERTa2022.05		22.3	20.8
BERT2022.05		19.9	18.8