metadata

dataset_info:
  features:
    - name: texts
      list: string
    - name: embeds
      list:
        list: float32
    - name: avg_similarity
      dtype: float32
  splits:
    - name: train
      num_bytes: 6669544815
      num_examples: 19731
  download_size: 4030544023
  dataset_size: 6669544815
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
license: cc
pretty_name: Semantically Clustered PubMed Central (PMC) Open Access in Markdown

Semantically Clustered PubMed Central

This is a subset of casperhansen/pmc-oa-markdown-embeddings

Details:

3.82% of data was preserved from the original dataset.
Each sample in this dataset contains 5 texts and embeddings with average similarity > 0.75.