sumbert 1.0.0

sumbert: A BERT-based summarizer with a frindly API

Abstractive BERT-based Text Summarizer

A BERT-based summarizer, centered towards a simple API: some text in, a summary out, both strings. Also at pypi.org as package sumbert.

Installation:

pip3 install sumbert

or if you want ot play with the code, after cloning this, and going into the resulting folder, install as an editable package with:

pip3 install -e .

Usage:

$ python3 -i
>>> from sumbert import summarize
>>> 
>>> text= """... your favorite story ..."""
>>> print(summarize(text))

This package is designed as a simple API for an otherwise fairy intricate set of components.

Credits:

The code is derived from the bertabs submodule in the HuggingFace transformers package at

https://github.com/huggingface/transformers

which credits the abstractive summarization described in the article Text Summarization with Pretrained Encoders by Yang Liu and Mirella Lapata.

The original code described in the article can be found on the Yang Liu's github repository.