README.md

metadata

dataset_info:
  features:
    - name: sentence
      dtype: string
    - name: entities
      list:
        - name: offsets
          sequence: int64
        - name: text
          dtype: string
        - name: type
          dtype: string
  splits:
    - name: train
      num_bytes: 115937
      num_examples: 632
    - name: validation
      num_bytes: 16540
      num_examples: 101
    - name: test
      num_bytes: 137794
      num_examples: 738
  download_size: 153216
  dataset_size: 270271
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: validation
        path: data/validation-*
      - split: test
        path: data/test-*

The files in this repo fall under two extensions:

• .tsv files, which contain the BIO files for clinical entities
• .txt files, which contain the PubTator version for relation extraction