sum-any-news-dataset

This dataset is a set for fine-tuning the summarization task on the “google/mt5-base” model and its derivatives. The set is based on the well-known dataset “csebuetnlp/xlsum”, but is a limited 20,000 examples of news articles from it. In addition, about 20 thousand news articles in Russian for the period from 2019-01-01 to 2024-08-01 have been added to the model (multiple sources).