This is a continued pre-training PubMedBERT model (MLM) on BioRel unlabeled text data (train set).