BioBERT: a pre-trained biomedical language representation model for biomedical text mining

Benchmark Model Rank Results
drug-drug-interaction-extraction-on-ddiBioBERT#4F1: 0.8088Micro F1: 80.88
few-shot-learning-on-medconceptsqadmis-lab/biobert-v1.1#6Accuracy: 25.458
named-entity-recognition-ner-on-jnlpbaBioBERT#12F1: 77.59
named-entity-recognition-ner-on-ncbi-diseaseBioBERT#1F1: 89.71
named-entity-recognition-on-species-800BioBERT#1F1: 75.31
question-answering-on-medqa-usmleBioBERT (large)#18Accuracy: 36.7
question-answering-on-medqa-usmleBioBERT (base)#19Accuracy: 34.1
relation-extraction-on-chemprotBioBERT#8F1: 76.46
representation-learning-on-scidocsBioBERT#6Avg.: 58.8
zero-shot-learning-on-medconceptsqadmis-lab/biobert-v1.1#3Accuracy: 26.151