Bert2bert Turkish Paraphrase Generation

🚀 Bert2Bert Turkish Paraphrase Generation

This project focuses on Turkish paraphrase generation, comparing different models and providing a practical solution for text paraphrasing.

🚀 Quick Start

To start using the Turkish paraphrase generation model, follow these steps:

from transformers import BertTokenizerFast,EncoderDecoderModel
tokenizer=BertTokenizerFast.from_pretrained("dbmdz/bert-base-turkish-cased")
model = EncoderDecoderModel.from_pretrained("ahmetbagci/bert2bert-turkish-paraphrase-generation")

text="son model arabalar çevreye daha mı az zarar veriyor?"
input_ids = tokenizer(text, return_tensors="pt").input_ids
output_ids = model.generate(input_ids)
print(tokenizer.decode(output_ids[0], skip_special_tokens=True))
#sample output
#son model arabalar çevre için daha az zararlı mı?

💻 Usage Examples

Basic Usage

from transformers import BertTokenizerFast,EncoderDecoderModel
tokenizer=BertTokenizerFast.from_pretrained("dbmdz/bert-base-turkish-cased")
model = EncoderDecoderModel.from_pretrained("ahmetbagci/bert2bert-turkish-paraphrase-generation")

text="son model arabalar çevreye daha mı az zarar veriyor?"
input_ids = tokenizer(text, return_tensors="pt").input_ids
output_ids = model.generate(input_ids)
print(tokenizer.decode(output_ids[0], skip_special_tokens=True))

📚 Documentation

Dataset

The dataset used in model training was created with the combination of the translation of the QQP dataset and manually generated dataset. Dataset Link

Cite

@INPROCEEDINGS{9548335,  
author={Bağcı, Ahmet and Amasyali, Mehmet Fatih},  
booktitle={2021 International Conference on INnovations in Intelligent SysTems and Applications (INISTA)},   
title={Comparison of Turkish Paraphrase Generation Models},   
year={2021},  
volume={},  
number={},  
pages={1-6},  
doi={10.1109/INISTA52262.2021.9548335}
}