README.md

metadata

dataset_info:
  features:
    - name: title
      dtype: string
    - name: context
      dtype: string
    - name: question
      dtype: string
    - name: answers
      struct:
        - name: answers_start
          sequence: int64
        - name: text
          sequence: string
    - name: is_impossible
      dtype: bool
  splits:
    - name: test
      num_bytes: 865505
      num_examples: 800
    - name: valid
      num_bytes: 217746
      num_examples: 200
    - name: test_hasAns
      num_bytes: 458114
      num_examples: 400
    - name: valid_hasAns
      num_bytes: 113725
      num_examples: 100
  download_size: 785547
  dataset_size: 1655090
license: apache-2.0
task_categories:
  - question-answering
  - feature-extraction
  - sentence-similarity
language:
  - fr
size_categories:
  - n<1K

Dataset Card for "Fquad2_test"

This dataset is released as part of FrenchBench, a benchmarking initiative for French Language Model evaluation.

It can be used for extractive QA, binary classifcation or infiormation retrieving evaluation !

Cite

@misc{faysse2024croissantllm,
      title={CroissantLLM: A Truly Bilingual French-English Language Model}, 
      author={Manuel Faysse and Patrick Fernandes and Nuno M. Guerreiro and António Loison and Duarte M. Alves and Caio Corro and Nicolas Boizard and João Alves and Ricardo Rei and Pedro H. Martins and Antoni Bigata Casademunt and François Yvon and André F. T. Martins and Gautier Viaud and Céline Hudelot and Pierre Colombo},
      year={2024},
      eprint={2402.00786},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}

@misc{heinrich2021fquad20,
      title={FQuAD2.0: French Question Answering and knowing that you know nothing}, 
      author={Quentin Heinrich and Gautier Viaud and Wacim Belblidia},
      year={2021},
      eprint={2109.13209},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}

@misc{dhoffschmidt2020fquad,
      title={FQuAD: French Question Answering Dataset}, 
      author={Martin d'Hoffschmidt and Wacim Belblidia and Tom Brendlé and Quentin Heinrich and Maxime Vidal},
      year={2020},
      eprint={2002.06071},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}