BeIR/msmarco

Dataset Card for BEIR Benchmark Dataset Summary BEIR is a heterogeneous benchmark built from 18 diverse datasets representing 9 information retrieval tasks. This msmarco subset is part of BEIR. Languages All tasks are in English (en). Dataset Structure This dataset uses the standard BEIR retrieval layout and includes: corpus: one row per document with _id, title, text queries: one row per query with _id, title, text Data… See the full description on the dataset page: https://huggingface.co/datasets/BeIR/msmarco.