This data set contains 14k news articles which have been taken from here.

The data set has been cleaned and contains train and test set to benchmark your classification and summarization models.