---
license: mit
language:
- ne
- en
tags:
- nepali
- summarization
- textrank
- speech-summarization
pipeline_tag: summarization
---

# Nepali Summarization

This repository contains the local TextRank-based summarization implementation used in Sulav's Nepali Speech project.

It is not a trained Transformers checkpoint yet. The project backend uses these files directly for extractive summarization:

- `summarizer.py`
- `tokenizer.py`
- `ranker.py`
- `text_rank/`

To use this as a Transformers model later, upload a trained summarization checkpoint with files such as `config.json`, `model.safetensors` or `pytorch_model.bin`, tokenizer files, and `spiece.model`.