xsum

Simple text summarizer using T5-small model, trained on CNN/DailyMail and XSum datasets.

Offer straightforward guidance to reproduce the results of NLP papers.

fast and exact summation

This project implements a headline summarization pipeline using the T5 transformer model fine-tuned on the XSum dataset. It provides a user-friendly interface using Gradio to generate concise summaries for long headlines or paragraphs. The project showcases end-to-end machine learning capabilities, from preprocessing and fine-tuning to deployment

This is a Pytorch implementation of a summarization model that is fine-tuned on the top of Google-T5 pre-trained model.

4th-year HSE coursework on data-centric abstractive summarization: filtering XSum examples and fine-tuning BART.

Factuality-aware reranking for XSum summarization with reproducible offline fallback artifacts.

Transfer Learning for Text Summarization

Code for 'The Continued Value of Classic Summarization Models' (IEEE/ACM TASLP 2026). GXSum: LLM-generated references for XSum.

fast and exact summation