README.md

---
language:
- ar
library_name: transformers
pipeline_tag: text-generation
datasets:
- IBB-University/DATA_FATAWA
metrics:
- accuracy
- bleu
- bertscore
widget:
- text: ' اركان الاسلام'
- text: "ما حكم الاحتفال بالمولد النبوي"
- text: "ما هي الكتب السماوية"
---

# Answering Arabic Questions Using The Aragpt2 Transformer

  -Question-answering (QA) models are natural language processing (NLP) models that can answer questions in natural language. They are trained on a large corpus of text and use this data to understand the meaning of questions and find the best answer. QA systems are widely used in applications like customer service, knowledge management, and education, providing quick and accurate answers. Techniques like natural language processing, information retrieval, and machine learning aid the process. Transformers are widely used for QA due to their contextual embeddings, pre-training, attention mechanisms, transfer learning, and performance. Transformers have demonstrated state-of-the-art performance on a wide range of QA tasks, outperforming traditional methods in terms of accuracy and speed.

Generative QA models, which generate text directly to answer questions, are a powerful approach to QA, as they can generate answers to questions not explicitly mentioned in the context, making them well-suited for open-domain tasks. Transformers have demonstrated state-of-the-art performance on a wide range of QA tasks, making them a popular choice in many applications.

![image/png](https://cdn-uploads.huggingface.co/production/uploads/6484c52c906e0dce8c904a4a/33Ulbsf2DkzlmJV9lwx-L.png)
# AraGPT2
@inproceedings{antoun-etal-2021-aragpt2,
    title = "{A}ra{GPT}2: Pre-Trained Transformer for {A}rabic Language Generation",
    author = "Antoun, Wissam  and
      Baly, Fady  and
      Hajj, Hazem",
    booktitle = "Proceedings of the Sixth Arabic Natural Language Processing Workshop",
    month = apr,
    year = "2021",
    address = "Kyiv, Ukraine (Virtual)",
    publisher = "Association for Computational Linguistics",
    url = "https://www.aclweb.org/anthology/2021.wanlp-1.21",
    pages = "196--207",
}
## Testing the model using `transformers`:

```python
from transformers import GPT2TokenizerFast, pipeline
#for base and medium
from transformers import GPT2LMHeadModel
#for large and mega
# pip install arabert
from arabert.aragpt2.grover.modeling_gpt2 import GPT2LMHeadModel

from arabert.preprocess import ArabertPreprocessor

MODEL_NAME='IBB-University/Question_answer_model'
arabert_prep = ArabertPreprocessor(model_name=MODEL_NAME)

text=""
text_clean = arabert_prep.preprocess(text)

model = GPT2LMHeadModel.from_pretrained(MODEL_NAME)
tokenizer = GPT2TokenizerFast.from_pretrained(MODEL_NAME)
generation_pipeline = pipeline("text-generation",model=model,tokenizer=tokenizer)

#feel free to try different decoding settings
generation_pipeline(text,
    pad_token_id=tokenizer.eos_token_id,
     max_length=512,
    penalty_alpha=0.6,
    top_k=4 )[0]['generated_text']
```
## evaluate the quality of the generated answer`:

![image/png](https://cdn-uploads.huggingface.co/production/uploads/6484c52c906e0dce8c904a4a/aiQBC6lBlrfoEx_yrmLOf.png)


![image/png](https://cdn-uploads.huggingface.co/production/uploads/6484c52c906e0dce8c904a4a/UcwxD5AlnWBqRpbSp9a9v.png)

Training_loss=0.5 , rate = 5e-5,epoch=1
Metrics	value
BERTScore	0.736
Levenshtein distance	0.226
BLEU	0.0099