IR-Datasets Integration ======================= Datamaestro-text provides an interface to the `ir-datasets `_ library, giving access to hundreds of IR benchmarks through a unified API. Install ir-datasets: .. code-block:: bash pip install ir-datasets Usage: .. code-block:: python from datamaestro import prepare_dataset # Load any ir-datasets collection via the irds namespace dataset = prepare_dataset("irds.msmarco-passage") # Same API as native datasets for doc in dataset.documents.iter_documents(): print(doc) The list below is auto-generated and may not reflect the exact version of ir-datasets installed on your system. Data Types ---------- These wrapper types provide the datamaestro interface for ir-datasets data: .. autoxpmconfig:: datamaestro_text.datasets.irds.data.Topics .. autoxpmconfig:: datamaestro_text.datasets.irds.data.Documents .. autoxpmconfig:: datamaestro_text.datasets.irds.data.AdhocAssessments See also :class:`~datamaestro_text.datasets.irds.data.LZ4DocumentStore` in the :doc:`/api/ir` section. Available Datasets ------------------ .. dm:repository:: irds