This repository contains a Jupyter notebook for fine-tuning a large language model (LLM) to perform long-text summarization in Persian. The project uses the Unsloth library for efficient fine-tuning of the Qwen-1.5-4B-Chat model on a Persian summarization dataset.

Persian text summarization using a pretrained mT5-XLSum model with a clean inference pipeline.

Abstractive summarization of Persian texts using a fine-tuned mT5 model and a Streamlit web interface

Abstractive summarization of Persian texts using a fine-tuned mT5 model and a Streamlit web interface

📊 Access a structured Persian forum dataset to enhance NLP models for text classification, summarization, and semantic search in Farsi.

This notebook implements a pipeline for creating a Persian text summarization dataset, encompassing data acquisition, cleaning, clustering, summary generation using a Persian MT5 model, and evaluation with ROUGE metrics.

An AI-based Persian text summarization tool built using the mT5 model fine-tuned on the XL-Sum dataset. It automatically generates concise, coherent summaries of Persian articles, enabling faster information extraction for researchers, journalists, and developers.

YT DeepSeek Overview is a repo that contains files of a YT video tutorial and it contains PDF summarizer that extracts and translates text from PDFs into Persian. It offers both API-based (DeepSeek-R1-Distill-Qwen-32B) and local (DeepSeek-R1:8B) models, all with a user-friendly Streamlit interface for easy PDF uploads and summary viewing.