m0javad's picture
Update README.md
4646a7d verified
---
task_categories:
- text-generation
- conversational
- text2text-generation
language:
- fa
size_categories:
- 100M<n<1B
---
# Degarbayan-SC: A Colloquial Paraphrase Farsi using pre-trained mT5
This is the Dataset of [Degarbayan-SC paper](https://ieeexplore.ieee.org/abstract/document/9959983).
You can Finetune with this dataset on the transformers models using [Github](https://https://github.com/m0javad/Degarbayan-SC).
```python
from datasets import load_dataset
dataset = load_dataset("m0javad/Degarbayan-SC-dataset")
```
### Statistic
![Lenght of sentences](https://i.ibb.co/C1RJhTZ/lenght.jpg")
our sentence length distribution is between 3 and 19 words and sentences are an average of 8 words. This makes sense because in the movie subtitles, sentences are shown in a range of time and we matched them with timespans. Humans can say a certain number of words in a certain period of time. Our collected sentences have 128,699 unique words.
as you see in the table above, our dataset contains a large number of paraphrasing sentences in various forms such syntactic, semantic and conceptual paraphrases.
### contact
contact me for contribution and future possible works at: mjaghajani.ai@gmail.com