--- license: apache-2.0 task_categories: - question-answering language: - en pretty_name: CopyColors MCQA size_categories: - n<1K configs: - config_name: 2_answer_choices data_files: - split: train path: 2_answer_choices/train.jsonl - split: validation path: 2_answer_choices/validation.jsonl - split: test path: 2_answer_choices/test.jsonl - config_name: 3_answer_choices data_files: - split: train path: 3_answer_choices/train.jsonl - split: validation path: 3_answer_choices/validation.jsonl - split: test path: 3_answer_choices/test.jsonl - config_name: 4_answer_choices data_files: - split: train path: 4_answer_choices/train.jsonl - split: validation path: 4_answer_choices/validation.jsonl - split: test path: 4_answer_choices/test.jsonl - config_name: 5_answer_choices data_files: - split: train path: 5_answer_choices/train.jsonl - split: validation path: 5_answer_choices/validation.jsonl - split: test path: 5_answer_choices/test.jsonl - config_name: 6_answer_choices data_files: - split: train path: 6_answer_choices/train.jsonl - split: validation path: 6_answer_choices/validation.jsonl - split: test path: 6_answer_choices/test.jsonl - config_name: 7_answer_choices data_files: - split: train path: 7_answer_choices/train.jsonl - split: validation path: 7_answer_choices/validation.jsonl - split: test path: 7_answer_choices/test.jsonl - config_name: 8_answer_choices data_files: - split: train path: 8_answer_choices/train.jsonl - split: validation path: 8_answer_choices/validation.jsonl - split: test path: 8_answer_choices/test.jsonl - config_name: 9_answer_choices data_files: - split: train path: 9_answer_choices/train.jsonl - split: validation path: 9_answer_choices/validation.jsonl - split: test path: 9_answer_choices/test.jsonl - config_name: 10_answer_choices data_files: - split: train path: 10_answer_choices/train.jsonl - split: validation path: 10_answer_choices/validation.jsonl - split: test path: 10_answer_choices/test.jsonl --- This dataset consists of formatted n-way multiple choice questions, where n is in [2,10]. The task itself is simply to copy the prototypical color from the context and produce the corresponding color's answer choice letter. The "prototypical colors" dataset instances themselves come from [Memory Colors](https://aclanthology.org/2021.blackboxnlp-1.10/) (Norland et al. 2021) and [corypaik/coda](https://huggingface.co/datasets/corypaik/coda) (instances whose object_group is 0, indicating participants agreed on a prototypical color of that object).