File size: 1,798 Bytes
2b6bc4f
 
 
 
 
 
 
 
 
94334eb
 
2b6bc4f
94334eb
 
 
a830ac0
94334eb
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2b6bc4f
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
---
datasets:
- boun-tabi/squad_tr
language:
- tr
metrics:
- exact_match
- f1
library_name: transformers
base_model:
- dbmdz/convbert-base-turkish-cased
pipeline_tag: question-answering
tags:
- Turkish Question-Answering
---
# 🇹🇷 ConvBERTurkQA for Turkish Question-Answering
This model is a fine-tuned version of [ConvBERTurk Base](https://huggingface.co/dbmdz/convbert-base-turkish-cased) on the [SQuAD-TR](https://huggingface.co/datasets/boun-tabi/squad_tr), a machine‑translated Turkish version of the original [SQuAD 2.0](https://huggingface.co/datasets/rajpurkar/squad_v2). For more details about the dataset, methodology, and experiments, you can refer to the corresponding [research paper](https://dergipark.org.tr/en/pub/bsengineering/issue/88008/1596832).

---
## Citation
If you use this model in your research or application, please cite the following paper:

```
@article{incidelen8performance,
  title={Performance Evaluation of Transformer-Based Pre-Trained Language Models for Turkish Question-Answering},
  author={{\.I}ncidelen, Mert and Aydo{\u{g}}an, Murat},
  journal={Black Sea Journal of Engineering and Science},
  volume={8},
  number={2},
  pages={15--16},
  publisher={U{\u{g}}ur {\c{S}}EN}
}
```
---
## How to Use

You can use the model directly with 🤗 Transformers:

```python
from transformers import pipeline

qa = pipeline(
    "question-answering",
    model="incidelen/convbert-base-turkish-cased-qa"
)

result = qa(
    question="...",
    context="..."
)

print(result)
```
## Evaluation Results
| Exact Match (%)    | F1 Score (%)      |
|--------------------|-------------------|
| 57.82              | 71.59             |

---
## Acknowledgments
Special thanks to [maydogan](https://huggingface.co/maydogan) for their contributions and support.

---