jiacheng-xu/sum-interpret
Code for Dissecting Generation Modes for Abstractive Summarization Models via Ablation and Attribution (ACL2021)
This project helps natural language processing researchers and practitioners understand how abstractive summarization models work. By analyzing which parts of the input text contribute most to specific phrases in the summary, it reveals different 'generation modes' like extraction or paraphrasing. The input is a document and its abstractive summary, and the output is an attribution score showing the importance of each input token to each output token.
No commits in the last 6 months.
Use this if you need to debug, explain, or gain insights into the behavior of your abstractive summarization models.
Not ideal if you are looking for a tool to generate summaries or evaluate summary quality directly.
Stars
13
Forks
—
Language
Jupyter Notebook
License
—
Category
Last pushed
Jun 02, 2021
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/nlp/jiacheng-xu/sum-interpret"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
sildar/potara
Multi-document summarization tool relying on ILP and sentence fusion
hyunwoongko/summarizers
Package for controllable summarization
uoneway/Text-Summarization-Repo
텍스트 요약 분야의 주요 연구 주제, Must-read Papers, 이용 가능한 model 및 data 등을 추천 자료와 함께 정리한 저장소입니다.
yuhaozhang/summarize-radiology-findings
Code and pretrained model for paper "Learning to Summarize Radiology Findings"
Huffon/factsumm
FactSumm: Factual Consistency Scorer for Abstractive Summarization