CharacterBERT (base, medical)

Reported on 4 benchmarks across 4 tasks · 1 paper · 1 SOTA

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Natural Language Processing2 results

Natural Language InferenceonMedNLI
Accuracy· 2020-10-20
84.95
best: 86.59 (ClinicalMosaic)
SOTA
CharacterBERT: Reconciling ELMo and BERT for Word-Level Open-Vocabulary Representations From Characters arXiv:2010.10392
Relation ExtractiononChemProt
Micro F1· 2020-10-20
73.44
best: 79.98 (BioLinkBERT (large))
CharacterBERT: Reconciling ELMo and BERT for Word-Level Open-Vocabulary Representations From Characters arXiv:2010.10392

Medical2 results

Information ExtractiononDDI extraction 2013 corpus
Micro F1· 2020-10-20
80.38
best: 84.08 (DESC+MOL+SciBERT)
CharacterBERT: Reconciling ELMo and BERT for Word-Level Open-Vocabulary Representations From Characters arXiv:2010.10392
Clinical Concept Extractionon2010 i2b2/VA
Exact Span F1· 2020-10-20
89.24
best: 90.25 (BERTlarge (MIMIC))
CharacterBERT: Reconciling ELMo and BERT for Word-Level Open-Vocabulary Representations From Characters arXiv:2010.10392