| # Multi-lingual Question Generating Model (mt5-base) | |
| Give the model a passage and it will generate a question about the passage. | |
| ## Trained on the following datasets: | |
| - [SQuAD (English)](https://rajpurkar.github.io/SQuAD-explorer/) | |
| - [TyDiQA-GoldP (Arabic, Bengali, Finnish, Japanese, Indonesian, Kiswahili, Korean, Russian, Telugu, Thai)](https://github.com/google-research-datasets/tydiqa) | |
| - [MLQA (Arabic, Chinese, English, German, Hindi, Spanish, Vietnames)](https://github.com/facebookresearch/MLQA) | |
| - [XQuAD (Arabic, Chinese, German, Greek, Hindi, Russian, Spanish, Thai, Turkish Vietnamese)](https://github.com/deepmind/xquad) | |
| - [GermanQuAD (German)](https://huggingface.co/datasets/deepset/germanquad) | |
| - [Persian QA (Persian)](https://www.kaggle.com/sajjadayobi360/persianqa) | |
| - [Bengali QA (Bengali)](https://www.kaggle.com/mayeesha/bengali-question-answering-dataset) | |
| - [chaii (Hindi, Tamil)](https://www.kaggle.com/c/chaii-hindi-and-tamil-question-answering/data) | |
| ## Training details | |
| I used [flax summarization script](https://github.com/huggingface/transformers/tree/master/examples/flax/summarization) and a TPU v3-8. Summarization expects a text column and a summary column. For question generation training, use the context column instead of text column and question instead of summary column. | |
| There is no guarantee that it will produce a question in the language of the passage, but it usually does. | |
| Model trained on Cloud TPUs from Google's TPU Research Cloud (TRC) |