README.md

metadata

language:
  - ja
license: cc-by-sa-4.0
dataset_info:
  features:
    - name: passage_id
      dtype: int64
    - name: anc
      dtype: string
    - name: pos_ids
      sequence: int64
    - name: neg_ids.me5-large
      sequence: int64
    - name: neg_ids.bm25
      sequence: int64
    - name: id
      dtype: int64
    - name: pos_rel
      sequence: float64
    - name: neg_rel.me5-large
      sequence: float64
    - name: neg_rel.bm25
      sequence: float64
  splits:
    - name: train
      num_bytes: 82174830
      num_examples: 20000
  download_size: 58343829
  dataset_size: 82174830
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*