@article{tydiqa,
title = {TyDi QA: A Benchmark for Information-Seeking Question Answering in Typologically Diverse Languages},
author = {Jonathan H. Clark and Eunsol Choi and Michael Collins and Dan Garrette and Tom Kwiatkowski and Vitaly Nikolaev and Jennimaria Palomaki}
year = {2020},
journal = {Transactions of the Association for Computational Linguistics}
}
[null,null,["อัปเดตล่าสุด 2022-12-06 UTC"],[],[],null,["# tydi_qa\n\n\u003cbr /\u003e\n\n- **Description**:\n\nTyDi QA is a question answering dataset covering 11 typologically diverse\nlanguages with 204K question-answer pairs. The languages of TyDi QA are diverse\nwith regard to their typology -- the set of linguistic features that each\nlanguage expresses -- such that we expect models performing well on this set to\ngeneralize across a large number of the languages in the world. It contains\nlanguage phenomena that would not be found in English-only corpora. To provide a\nrealistic information-seeking task and avoid priming effects, questions are\nwritten by people who want to know the answer, but don't know the answer yet,\n(unlike SQuAD and its descendents) and the data is collected directly in each\nlanguage without the use of translation (unlike MLQA and XQuAD).\n| **Important:** Please choose your training split carefully.\n\nTraining splits:\n\n'train': This is the GoldP task from the original TyDi QA paper\n\\[\u003chttps://arxiv.org/abs/2003.05002\u003e\\] that has original-language labeled training\ndata.\n\n'translate-train-\\*': These splits are the automatic translations from English to\neach target language used in the translate-train baselines in the XTREME paper\n\\[\u003chttps://arxiv.org/abs/2003.11080\u003e\\]. This purposefully ignores the non-English\nTyDiQA-GoldP training data to simulate the transfer learning scenario where\noriginal-language data is not available and system builders must rely on labeled\nEnglish data plus existing machine translation systems.\n\nTypically, you should use EITHER the train or translate-train split, but not\nboth.\n\n- **Additional Documentation** :\n [Explore on Papers With Code\n north_east](https://paperswithcode.com/dataset/tydi-qa)\n\n- **Config description** : Gold passage (GoldP) task\n (\u003chttps://github.com/google-research-datasets/tydiqa/tree/master/gold_passage_baseline\u003e).\n\n- **Homepage** :\n \u003chttps://github.com/google-research-datasets/tydiqa\u003e\n\n- **Source code** :\n [`tfds.question_answering.TydiQA`](https://github.com/tensorflow/datasets/tree/master/tensorflow_datasets/question_answering/tydi_qa.py)\n\n- **Versions**:\n\n - **`3.0.0`** (default): Fixes issue with a number of examples where answer spans are misaligned due to context white-space removal. This change impacts roughly 25% of train and dev examples.\n- **Download size** : `121.30 MiB`\n\n- **Dataset size** : `98.35 MiB`\n\n- **Auto-cached**\n ([documentation](https://www.tensorflow.org/datasets/performances#auto-caching)):\n Yes\n\n- **Splits**:\n\n| Split | Examples |\n|------------------------|----------|\n| `'train'` | 49,881 |\n| `'translate-train-ar'` | 3,661 |\n| `'translate-train-bn'` | 3,585 |\n| `'translate-train-fi'` | 3,670 |\n| `'translate-train-id'` | 3,667 |\n| `'translate-train-ko'` | 3,607 |\n| `'translate-train-ru'` | 3,394 |\n| `'translate-train-sw'` | 3,622 |\n| `'translate-train-te'` | 3,658 |\n| `'validation'` | 5,077 |\n| `'validation-ar'` | 921 |\n| `'validation-bn'` | 113 |\n| `'validation-en'` | 440 |\n| `'validation-fi'` | 782 |\n| `'validation-id'` | 565 |\n| `'validation-ko'` | 276 |\n| `'validation-ru'` | 812 |\n| `'validation-sw'` | 499 |\n| `'validation-te'` | 669 |\n\n- **Feature structure**:\n\n FeaturesDict({\n 'answers': Sequence({\n 'answer_start': int32,\n 'text': Text(shape=(), dtype=string),\n }),\n 'context': Text(shape=(), dtype=string),\n 'id': string,\n 'question': Text(shape=(), dtype=string),\n 'title': Text(shape=(), dtype=string),\n })\n\n- **Feature documentation**:\n\n| Feature | Class | Shape | Dtype | Description |\n|----------------------|--------------|-------|--------|-------------|\n| | FeaturesDict | | | |\n| answers | Sequence | | | |\n| answers/answer_start | Tensor | | int32 | |\n| answers/text | Text | | string | |\n| context | Text | | string | |\n| id | Tensor | | string | |\n| question | Text | | string | |\n| title | Text | | string | |\n\n- **Supervised keys** (See\n [`as_supervised` doc](https://www.tensorflow.org/datasets/api_docs/python/tfds/load#args)):\n `None`\n\n- **Figure**\n ([tfds.show_examples](https://www.tensorflow.org/datasets/api_docs/python/tfds/visualization/show_examples)):\n Not supported.\n\n- **Examples**\n ([tfds.as_dataframe](https://www.tensorflow.org/datasets/api_docs/python/tfds/as_dataframe)):\n\nDisplay examples... \n\n- **Citation**:\n\n @article{tydiqa,\n title = {TyDi QA: A Benchmark for Information-Seeking Question Answering in Typologically Diverse Languages},\n author = {Jonathan H. Clark and Eunsol Choi and Michael Collins and Dan Garrette and Tom Kwiatkowski and Vitaly Nikolaev and Jennimaria Palomaki}\n year = {2020},\n journal = {Transactions of the Association for Computational Linguistics}\n }\n\ntydi_qa/goldp (default config)\n------------------------------"]]
