README.md

metadata

language:
  - en
dataset_info:
  features:
    - name: id
      dtype: string
    - name: title
      dtype: string
    - name: context
      dtype: string
    - name: question
      dtype: string
    - name: answers
      sequence:
        - name: text
          dtype: string
        - name: answer_start
          dtype: int32
    - name: embedding
      sequence: float64
  splits:
    - name: train
      num_bytes: 6986291.79299258
      num_examples: 1759
    - name: validation
      num_bytes: 884677.1415817401
      num_examples: 218
  download_size: 1512557
  dataset_size: 7870968.93457432
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: validation
        path: data/validation-*

This dataset is a filtered version of the SQuAD v2 dataset filtered using embedding similarity with Space Engineering queries. https://huggingface.co/datasets/rajpurkar/squad_v2