This dataset contains the parsed data from the MedMentions dataset (st21pv version).
Dataset Description
The dataset provides directly the train / validation / test splits of the MedMentions dataset, merging together the titles and abstracts to a single text string.
The annotations only consist of the start and end token indices, marking the beginning and end of each entity. Entity types are not distinguished.
License: CC0
Dataset Sources… See the full description on the dataset page: https://huggingface.co/datasets/zameji/medmentions-st21pv.
