|
--- |
|
license: mit |
|
datasets: |
|
- ntphuc149/ViBidLQA |
|
language: |
|
- vi |
|
metrics: |
|
- bleu |
|
- rouge |
|
- meteor |
|
- bertscore |
|
base_model: |
|
- VietAI/vit5-base |
|
pipeline_tag: question-answering |
|
library_name: transformers |
|
tags: |
|
- legal |
|
- vietnamese |
|
- question-answering |
|
- text-generation |
|
--- |
|
|
|
# ViBidLAQA_base: A Vietnamese Bidding Legal Abstractive Question Answering Model |
|
|
|
## Overview |
|
ViBidLAQA_base is an abstractive question-answering (AQA) model specifically developed for the Vietnamese bidding law domain. Built upon the VietAI/vit5-base architecture and fine-tuned with a specialized bidding law dataset, this model demonstrates strong performance in generating natural and accurate responses to legal queries. |
|
|
|
## Model Description |
|
|
|
- **Downstream task**: Abstractive Question Answering |
|
- **Domain**: Vietnamese Bidding Law |
|
- **Base Model**: VietAI/vit5-base |
|
- **Approach**: Fine-tuning |
|
- **Language**: Vietnamese |
|
|
|
## Dataset |
|
|
|
The ViBidLQA dataset features: |
|
- **Training set**: 5,300 samples |
|
- **Test set**: 1,000 samples |
|
- **Data Creation Process**: |
|
- Training data was automatically generated by Claude 3.5 Sonnet and validated by two legal experts |
|
- Two Vietnamese legal experts manually created test set |
|
|
|
## Performance |
|
|
|
| Metric | Score | |
|
|--------|-------| |
|
| ROUGE-1 | 75.09 | |
|
| ROUGE-2 | 63.43 | |
|
| ROUGE-L | 65.72 | |
|
| ROUGE-L-SUM | 65.79 | |
|
| BLEU-1 | 53.61 | |
|
| BLEU-2 | 47.51 | |
|
| BLEU-3 | 43.40 | |
|
| BLEU-4 | 39.54 | |
|
| METEOR | 64.38 | |
|
| BERT-Score | 86.65 | |
|
|
|
## Usage |
|
|
|
```python |
|
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM |
|
|
|
# Load model and tokenizer |
|
tokenizer = AutoTokenizer.from_pretrained("ntphuc149/ViBidLAQA_base") |
|
model = AutoModelForSeq2SeqLM.from_pretrained("ntphuc149/ViBidLAQA_base") |
|
|
|
# Example usage |
|
question = "Thế nào là đấu thầu hạn chế?" |
|
context = "Đấu thầu hạn chế là phương thức lựa chọn nhà thầu trong đó chỉ một số nhà thầu đáp ứng yêu cầu về năng lực và kinh nghiệm được bên mời thầu mời tham gia." |
|
|
|
# Prepare input |
|
inputs = tokenizer(f"question: {question} context: {context}", return_tensors="pt", max_length=512, truncation=True) |
|
|
|
# Generate answer |
|
outputs = model.generate(inputs.input_ids, max_length=128, min_length=10, num_beams=4) |
|
answer = tokenizer.decode(outputs[0], skip_special_tokens=True) |
|
print(answer) |
|
``` |
|
|
|
## Applications |
|
|
|
This model is advantageous for: |
|
- Bidding law information retrieval systems |
|
- Legal advisory chatbots in the bidding domain |
|
- Automated question-answering systems for bidding law queries |
|
|
|
## Limitations |
|
|
|
- The model is specifically trained for the Vietnamese bidding law domain and may not perform well on other legal domains |
|
- Performance may vary depending on the complexity and specificity of the questions |
|
- The model should be used as a reference tool and not as a replacement for professional legal advice |
|
|
|
## Citation |
|
|
|
If you use this model in your research, please cite: |
|
``` |
|
comming soon... |
|
``` |
|
|
|
## Contact |
|
|
|
For questions, feedback, or collaborations: |
|
- Email: nguyentruongphuc[email protected] |
|
- GitHub Issues: [@ntphuc149](https://github.com/ntphuc149) |
|
- HuggingFace: [@ntphuc149](https://huggingface.co/ntphuc149) |
|
|
|
## License |
|
|
|
This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details. |
