--- dataset_info: features: - name: question dtype: string - name: types_of_harm dtype: string - name: risk_area dtype: string - name: specific_harms dtype: string - name: id dtype: string splits: - name: train num_bytes: 62711 num_examples: 150 download_size: 15849 dataset_size: 62711 configs: - config_name: default data_files: - split: train path: data/train-* license: cc-by-nc-sa-4.0 --- kunishowさんが整備してくださってるdo-not-answer-jaからむしろ創作能力という意味では回答してくれた方が有用なカテゴリ5つからおおむね30問ずつ、合わせて150問を抽出したものです。 do-not-answer-ja https://github.com/kunishou/do-not-answer-ja do-not-answer https://github.com/Libr-AI/do-not-answer