README.md

This is a copy of the Multi-News dataset, with its input source documents replaced by a sparse retriever. The retrieval pipeline used:

• query: The summary field of each example
• corpus: The union of all documents in the train, validation and test sets`
• retriever: BM25 via PyTerrier with default settings except for no stemming was applied.

Retrieval results on the test set:

ndcg	recall@100	recall@1000	Rprec
0.8222	0.8242	0.8492	0.7233