Improving Pre-trained Language Model Sensitivity via Mask Specific Losses: Biomedical NER Case Study

Efficiently improving Language Model sensitivity by weighting domain-specific terms during fine-tuning.