File size: 3,881 Bytes
9b215a0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9dd7479
 
 
 
 
 
 
9b215a0
9dd7479
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9b215a0
fec6f45
 
 
 
 
9e14b8f
 
 
 
 
 
 
 
 
 
 
 
fec6f45
 
71c6fde
 
 
 
 
 
 
 
 
 
9b215a0
71c6fde
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9e14b8f
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
---
language:
- en
- hi
- de
- ar
- bn
- fi
- ja
- zh
- id
- sw
- ta
- gr
- ru
- es
- th
- tr
- vi
- multilingual
datasets:
- squad_v2
- tydiqa
- mlqa
- xquad
- germanquad
widget:
- text: 'Hugging Face has seen rapid growth in its popularity since the get-go. It
    is definitely doing the right things to attract more and more people to its platform,
    some of which are on the following lines: Community driven approach through large
    open source repositories along with paid services. Helps to build a network of
    like-minded people passionate about open source. Attractive price point. The subscription-based
    features, e.g.: Inference based API, starts at a price of $9/month.'
  example_title: English
- text: 'A un a�o y tres d�as de que el bal�n ruede en el Al Bayt Stadium inaugurando
    el Mundial 2022, ya se han dibujado los primeros bocetos de la pr�xima Copa del
    Mundo.13 selecciones est�n colocadas en el mapa con la etiqueta de clasificadas
    y tienen asegurado pisar los verdes de Qatar en la primera fase final  oto�al.
    Serbia, Dinamarca, Espa�a, Pa�ses Bajos, Suiza, Croacia, Francia, Inglaterra,
    B�lgica, Alemania, Brasil, Argentina y Qatar, como anfitriona, entrar�n en   el
    sorteo del 1 de abril de 2022 en Doha en el que 32 pa�ses ser�n repartidos en
    sus respectivos grupos. '
  example_title: Spanish
---
# 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. Lower resource languages will likely have lower quality questions.


## Using the model

#### PyTorch version
```python
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
  
tokenizer = AutoTokenizer.from_pretrained("nbroad/mt5-base-qgen")
model = AutoModelForSeq2SeqLM.from_pretrained("nbroad/mt5-base-qgen")

text = "Hugging Face has seen rapid growth in its \
popularity since the get-go. It is definitely doing\
 the right things to attract more and more people to \
 its platform, some of which are on the following lines:\
Community driven approach through large open source repositories \
along with paid services. Helps to build a network of like-minded\
 people passionate about open source. \
Attractive price point. The subscription-based features, e.g.: \
Inference based API, starts at a price of $9/month.\
"

inputs = tokenizer(text, return_tensors="pt")
output = model.generate(**inputs, max_length=40)

tokenizer.decode(output[0], skip_special_tokens=True)
# What is Hugging Face's price point?
```

Model trained on Cloud TPUs from Google's TPU Research Cloud (TRC)