Yifan Peng, Shankai Yan, Zhiyong Lu
| Task | Dataset | Metric | Value | Model |
|---|---|---|---|---|
| Relation Extraction | ChemProt | F1 | 74.4 | NCBI_BERT(large) (P) |
| Natural Language Inference | MedNLI | Accuracy | 84 | NCBI_BERT(base) (P+M) |
| Language Modelling | MedSTS | Pearson Correlation | 0.848 | NCBI_BERT(base) (P+M) |
| Language Modelling | BIOSSES | Pearson Correlation | 0.9159999999999999 | NCBI_BERT(base) (P+M) |
| Medical Relation Extraction | DDI extraction 2013 corpus | F1 | 79.9 | NCBI_BERT(large) (P) |
| Named Entity Recognition (NER) | BC5CDR-chemical | F1 | 93.5 | NCBI_BERT(base) (P) |
| Named Entity Recognition (NER) | BC5CDR-disease | F1 | 86.6 | NCBI_BERT(base) (P) |
| Named Entity Recognition (NER) | ShARe/CLEF eHealth corpus | F1 | 0.792 | NCBI_BERT(base) (P+M) |
| Text Classification | HOC | F1 | 87.3 | NCBI_BERT(large) (P) |
| Sentence Pair Modeling | MedSTS | Pearson Correlation | 0.848 | NCBI_BERT(base) (P+M) |
| Sentence Pair Modeling | BIOSSES | Pearson Correlation | 0.9159999999999999 | NCBI_BERT(base) (P+M) |
| Document Classification | HOC | F1 | 87.3 | NCBI_BERT(large) (P) |
| Classification | HOC | F1 | 87.3 | NCBI_BERT(large) (P) |
| Semantic Similarity | MedSTS | Pearson Correlation | 0.848 | NCBI_BERT(base) (P+M) |
| Semantic Similarity | BIOSSES | Pearson Correlation | 0.9159999999999999 | NCBI_BERT(base) (P+M) |