File size: 1,491 Bytes
fec6f45 9e14b8f fec6f45 9e14b8f |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 |
# 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) |