You could for example use the DailyMail/CNN dataset: cnn_dailymail · Datasets at Hugging Face
from datasets import load_dataset
dataset = load_dataset("cnn_dailymail")
Note that the columns in this dataset are called "article" and "highlights".