Persian - t5 - paraphraser Open - source Model - Free Generation of Multiple Paraphrase Versions of Persian Text

Persian T5 Paraphraser

Developed by erfan226

This is a text paraphrasing model for Persian language, built upon a Persian monolingual T5 model, capable of generating multiple paraphrased versions of Persian texts.

Text Generation

Transformers

Other#Persian text paraphrasing #T5 architecture optimization #Multi-sentence generation

Downloads 290

Release Time : 3/2/2022

Model Overview

This model is primarily used for Persian text paraphrasing tasks, able to generate semantically similar but differently expressed sentence variants.

Model Features

Persian-specific

A paraphrase model specifically optimized for Persian texts

Multi-version generation

Capable of generating multiple different paraphrased versions for the same input

T5-based architecture

Utilizes the powerful T5 text-to-text transfer architecture for high-quality paraphrasing

Model Capabilities

Text paraphrasing

Text generation

Semantic-preserving transformation

Use Cases

Content creation

Wikipedia content expansion

Helps expand the expression methods of Wikipedia articles

Generates various expressions to enrich article content

Language learning

Persian learning assistance

Provides different expressions of the same meaning for Persian learners

Helps learners understand linguistic diversity

🚀 Persian-t5-paraphraser

This is a paraphrasing model for the Persian language, leveraging the power of the monolingual T5 model for Persian to rephrase text effectively.

🚀 Quick Start

This is a paraphrasing model for the Persian language. It is based on the monolingual T5 model for Persian.

💻 Usage Examples

Basic Usage

>>> pip install transformers
>>> from transformers import (T5ForConditionalGeneration, AutoTokenizer, pipeline)
>>> import torch

model_path = 'erfan226/persian-t5-paraphraser'
model = T5ForConditionalGeneration.from_pretrained(model_path)
tokenizer = AutoTokenizer.from_pretrained(model_path)
pipe = pipeline(task='text2text-generation', model=model, tokenizer=tokenizer)

def paraphrase(text):
  for j in range(5):
    out = pipe(text, encoder_no_repeat_ngram_size=5, do_sample=True, num_beams=5, max_length=128)[0]['generated_text']
    print("Paraphrase:", out)

text = "این یک مقالهٔ خرد آلمان است. می‌توانید با گسترش آن به ویکی‌پدیا کمک کنید."
print("Original:", text)
paraphrase(text)

# Original: این یک مقالهٔ خرد آلمان است. می‌توانید با گسترش آن به ویکی‌پدیا کمک کنید.
# Paraphrase: این یک مقالهٔ کوچک است.
# Paraphrase: این یک مقالهٔ کوچک است.
# Paraphrase: شما می توانید با گسترش این مقاله، به کسب و کار خود کمک کنید.
# Paraphrase: می توانید با گسترش این مقالهٔ خرد آلمان کمک کنید.
# Paraphrase: شما می توانید با گسترش این مقالهٔ خرد، به گسترش آن کمک کنید.

📚 Documentation

Training data

This model was trained on the Persian subset of the Tapaco dataset. It should be noted that this model was trained on a very small dataset and therefore the performance might not be as expected, for now.

Property	Details
Model Type	Paraphrasing model for Persian language
Training Data	Persian subset of the Tapaco dataset

⚠️ Important Note

This model was trained on a very small dataset and therefore the performance might not be as expected, for now.

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご