# nlpyang/BertSum

Code for paper Fine-tune BERT for Extractive Summarization

Repository: https://github.com/nlpyang/BertSum
Canonical: https://ross.abutalabs.com/products/bertsum
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2022-01-11T07:58:23+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 2719, "days_push": 1695, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1506, forks 410 (observed 2026-08-28T04:04:55.060033+00:00)

## What it is
BertSum is the official PyTorch implementation of the paper 'Fine-tune BERT for Extractive Summarization', providing preprocessing pipelines and training code for BERT-based extractive summarization models. It includes data preparation for CNN/DailyMail, multiple summarization head architectures (classifier, Transformer, LSTM), and pretrained model results.

## Use cases
- fine-tune BERT for extractive text summarization
- reproduce ROUGE benchmark results on CNN/DailyMail
- train a sentence-level summarization classifier
- preprocess news articles for summarization research
- compare summarization model architectures like Transformer and LSTM heads

## When to choose
- you need a research-grade extractive summarization baseline with published ROUGE scores
- you want to fine-tune BERT specifically for sentence selection in summarization
- you are working with the CNN/DailyMail dataset and need preprocessing tooling

## When to avoid
- you need abstractive summarization rather than extractive
- you want a production-ready, actively maintained library with modern dependency versions
- you need support for newer transformer models beyond BERT or easy installation

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, nlp, llm-training
- domain: deep-learning, machine-learning
- platform: python
- tags: extractive-summarization, bert, pytorch, research-code, text-summarization, natural-language-processing, linux

## Member repositories
- nlpyang/BertSum (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:55.060033+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:32:40.493328+00:00, confidence not recorded.
  - readme: https://github.com/nlpyang/BertSum (fetched 2026-08-28T04:04:55.060033+00:00, sha e5f362c9299b)
- Data as of 2026-08-30T08:39:29.467469+00:00.