pn-summary-mt5-small Open-source Model - Optimized for Persian Text Summarization, Free Deployment, Highly Practical

Home

Pn Summary Mt5 Small

Developed by HooshvareLab

An optimized mT5-small model for Persian text summarization tasks, trained on pn_summary dataset

Text Generation

Transformers

Other#Persian summarization #Multilingual MT5 model #Low-resource optimization

Downloads 409

Release Time : 3/2/2022

Model Overview

This model is based on the mT5-small architecture, specifically designed for Persian text summarization tasks, capable of extracting key information from long texts to generate concise summaries

Model Features

Persian optimization

Specifically trained and optimized for Persian text characteristics

Multi-granularity ROUGE evaluation

Provides ROUGE evaluation results at low, medium, and high granularity levels to comprehensively reflect model performance

Lightweight architecture

Based on mT5-small architecture, reducing resource requirements while maintaining performance

Model Capabilities

Persian text comprehension

Key information extraction

Text summarization generation

Long text compression

Use Cases

News media

News article summarization

Automatically generate key summaries of news articles

ROUGE-1 F1 score approximately 43-44%

Academic research

Paper abstract generation

Generate concise abstracts for Persian academic papers

🚀 mT5-small Model for Article Summarization

This is an mT5-small model trained on the pnSummary dataset, designed to summarize articles effectively.

✨ Features

Utilizes the mT5-small architecture for summarization tasks.
Trained on the pnSummary dataset to provide accurate summaries.

📚 Documentation

Eval results

The following table summarizes the ROUGE scores obtained by the model for the validation set.

+-----------+------+-----------+--------+-----------+
|   Score   | Type | Precision | Recall | F-Measure |
+-----------+------+-----------+--------+-----------+
|   rouge1  | low  |   45.13   | 42.92  |   42.52   |
|   rouge1  | mid  |   45.67   | 43.47  |   43.04   |
|   rouge1  | high |   46.26   | 44.04  |   43.57   |
|           |      |           |        |           |
|    ***    | ***  |    ***    |  ***   |    ***    |
|           |      |           |        |           |
|   rouge2  | low  |   26.08   | 24.73  |   24.56   |
|   rouge2  | mid  |   26.65   | 25.29  |   25.10   |
|   rouge2  | high |   27.25   | 25.84  |   25.63   |
|           |      |           |        |           |
|    ***    | ***  |    ***    |  ***   |    ***    |
|           |      |           |        |           |
|   rougeL  | low  |   38.72   | 36.86  |   36.55   |
|   rougeL  | mid  |   39.28   | 37.38  |   37.05   |
|   rougeL  | high |   39.86   | 37.96  |   37.60   |
|           |      |           |        |           |
|    ***    | ***  |    ***    |  ***   |    ***    |
|           |      |           |        |           |
| rougeLsum | low  |   38.67   | 36.80  |   36.50   |
| rougeLsum | mid  |   39.25   | 37.36  |   37.02   |
| rougeLsum | high |   39.88   | 37.94  |   37.59   |
+-----------+------+-----------+--------+-----------+

Test results

The following table summarizes the ROUGE scores obtained by the model for the test set.

+-----------+------+-----------+--------+-----------+
|   Score   | Type | Precision | Recall | F-Measure |
+-----------+------+-----------+--------+-----------+
|   rouge1  | low  |   45.09   | 42.67  |   42.43   |
|   rouge1  | mid  |   45.63   | 43.19  |   42.87   |
|   rouge1  | high |   46.20   | 43.78  |   43.40   |
|           |      |           |        |           |
|    ***    | ***  |    ***    |  ***   |    ***    |
|           |      |           |        |           |
|   rouge2  | low  |   26.10   | 24.57  |   24.49   |
|   rouge2  | mid  |   26.68   | 25.13  |   25.04   |
|   rouge2  | high |   27.30   | 25.73  |   25.63   |
|           |      |           |        |           |
|    ***    | ***  |    ***    |  ***   |    ***    |
|           |      |           |        |           |
|   rougeL  | low  |   38.60   | 36.49  |   36.29   |
|   rougeL  | mid  |   39.15   | 37.06  |   36.82   |
|   rougeL  | high |   39.75   | 37.63  |   37.37   |
|           |      |           |        |           |
|    ***    | ***  |    ***    |  ***   |    ***    |
|           |      |           |        |           |
| rougeLsum | low  |   38.57   | 36.49  |   36.27   |
| rougeLsum | mid  |   39.15   | 37.07  |   36.82   |
| rougeLsum | high |   39.76   | 37.56  |   37.36   |
+-----------+------+-----------+--------+-----------+

Property	Details
Model Type	mT5-small
Training Data	pnSummary dataset

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